Materials

All DNA oligonucleotides were synthesized by Integrated DNA Technologies (Coralville, IA, USA) with the exception of those containing a fluorescein attached via a phosphorothioate linkage, which were synthesized by Sigma-Genosys (St Louis, MO, USA). T4 polynucleotide 3?-phosphatase (PNK) and the 3?-phosphatase minus mutant T4 PNK [PNK(?)] were purchased from New England BioLabs (Cambridge, MA, USA). Terminal deoxynucleotidyl transferase (TdT) was purchased from Calbiochem (La Jolla, CA, USA).

Expression and purification of Tdp1

Human Tdp1 was expressed in Escherichia coli BL21 (DE3) cells and purified as described earlier (¹⁸).

Preparation of radiolabeled oligonucleotides

DNA oligonucleotides were either 5?-end labeled using PNK and [?-³²P] ATP or 3?-end labeled using TdT and [?-³²P] cordycepin-5?-triphosphate, respectively. Unincorporated radioactive nucleotides were removed using a mini Quick Spin Oligo column (Roche Diagnostics, Indianapolis, IN, USA).

Tdp1 and PNK enzymatic assays

One nanomolar of ³²P-labeled oligonucleotide substrate in a 10 ?l reaction volume was incubated with the indicated concentrations of recombinant Tdp1 for 30 min at 25?C in a buffer containing 50 mM Tris?HCl (pH 7.5), 25 mM KCl and 2 mM EDTA. In reactions containing PNK or PNK(?), the reaction buffer was supplemented with 10 mM MgCl₂. Reactions were terminated by the addition of two volumes of gel loading buffer [96% (v/v) formamide, 10 mM EDTA, 1% (w/v) xylene cyanol and 1% (w/v) bromphenol blue]. The samples were subsequently heated to 95?C for 5 min and subjected to 20% sequencing gel electrophoresis. Imaging was performed using the Typhoon 8600.

Fluorescence anisotropy

Equilibrium binding of Tdp1 to fluorescein (6-FAM)-labeled oligonucleotides was monitored by fluorescence anisotropy using previously developed methods (²⁴). Briefly, serial dilutions of recombinant Tdp1 were mixed with 10 nM of fluorescein-labeled oligonucleotides (Table 1) in a binding buffer containing 50 mM Tris?HCl (pH 7.5), 25 mM KCl and 2 mM EDTA in a final volume of 50 ?l. Aliquots (40 ?l) were transferred into 384-well Costar polypropylene plates (Corning, NY, USA) and read using a Tecan Ultra plate reader (Durham, NC, USA) with excitation and emission wavelengths of 485 and 535 nm, respectively.

Fluorescence anisotropy competition

A constant concentration of F14p (10 nM) was mixed with 2-fold serial dilutions of unlabeled competitor oligonucleotide in 50 mM Tris?HCl (pH 7.5), 25 mM KCl, 2 mM EDTA and placed in a 96-well Costar polypropylene plates (Corning). A constant concentration of Tdp1 (100 nM) was then added to a final volume of 50 ?l and incubated at room temperature for 20 min. Aliquots (40 ?l) were transferred into 384-well Costar polypropylene plates (Corning) and read using a Tecan Ultra plate reader (Durham) with excitation and emission wavelengths of 485 and 535 nm, respectively.

Fluorescence anisotropy data analysis

Equilibrium dissociation constants (K_D) were estimated from the anisotropy in three steps. First, the total intensity of the fluorophore was examined and a correction parameter, R, was estimated (²⁵). Next, an equation relating fractional binding (f_B) and anisotropy was selected [Equation (¹)],

where f_B is fractional binding; r_B is anisotropy signal of saturated binding; r_F is anisotropy signal of no binding and R is ratio of maximum and minimum total fluorescence intensity.

Finally, the fractional binding model is fit to the anisotropy signal using a nonlinear least squares algorithm. For this, the simple Langmuir model was selected [Equation (²)], where K_D refers to the dissociation constant.

Equations (¹) and (²) are combined and solved numerically. The concentration of DNA_free and Tdp1_free were solved numerically using a conservation law. The system is then fit to experimental anisotropy data using a non-linear regression operator based on a Marquardt compromise. When the K_D value is small relative to the concentration titration range, the binding may be assumed to be stoichiometric. In this case, a plot of the molar ratio of the Tdp1 and DNA versus fractional binding will achieve half saturation binding at a ratio of unity. This is checked when the DNA binds tight enough for this test to be completed.

Time resolved fluorescence

Fluorescence lifetimes were measured time-correlated single-photon counting using a PicoQuant FluoTime200 instrument. The excitation wavelength was 470 nm with a repetition rate of 10 MHz. The emission was collected at a wavelength of 535 nm through a 0.5 nm band pass under magic angle conditions. The instrument response signal was measured with a scattering solution of Ludox. All measurements were performed at ambient temperature. Fluorescein-labeled oligonucleotides were prepared in 50 mM Tris?HCl (pH 7.5), 25 mM KCl and 2 mM EDTA at a concentration of 150 nM. Recombinant Tdp1 was added (37.5?450 nM) to each oligonucleotide and the intensity decay curve was recorded. For each titration the decay curves were fit globally using the FluoFit software (PicoQuant) using a bi-exponential function. The quality of the fits were evaluated by the reduced chi-squared and randomness of the residuals.

Surface plasmon resonance binding assays

Surface plasmon resonance (SPR) binding experiments were performed on a Biacore 2000 instrument (GE, Piscatawy, NJ, USA). Neutravidin was coupled to the carboxymethylated dextran surface of a CM5 sensor chip (GE, Piscatawy, NJ, USA) using standard amine coupling chemistry. In brief, the surface was activated with 0.1 M N-hydroxysuccinimide and 0.4 M N-ethyl-N?-(3-dimethylaminopropyl) carbodiimide at a flow rate of 20 ?l/min. Neutravidin was diluted to 200 ?g/ml in 10 mM sodium acetate (pH 4.5) and injected until a density of ?1800 RU was attached. Activated amine groups were quenched with an injection of 1 M ethanolamine (pH 8.0). 5?- and 3?-biotinylated DNA oligonucleotides (see B14p and p14B in Table 1) were immobilized to the neutravidin-coated sensor chips at 37 and 28 RU, respectively. Recombinant Tdp1 was diluted in running buffer [20 mM HEPES (pH 7.5), 150 mM NaCl, 2 mM EDTA, 0.05% (w/v) Tween-20, 5 mg/ml carboxymethyl dextran, 0.05% (w/v) polyethylene glycol 20 000] in nine successive 2-fold dilutions from 500 nM and injected over the immobilized oligonucleotides for 1 min at 20 ml/min at 25?C. At the end of the injection, Tdp1-oligonucleotide complexes were allowed to dissociate for 3 min. A 30 s pulse of 1 M NaCl removed any material that remained bound to the surface. Each cycle of Tdp1 injection was followed by a running buffer cycle for referencing purposes. To ensure the integrity of the oligonucleotide surface was not compromised by Tdp1 binding, five injections of 50 nM HIV-1 nucleocapsid protein were made through the course of the experiment. The binding kinetics of Tdp1 could not be fit with a simple Langmuir binding model. However, the equilibrium binding response was fit with a Langmuir binding model using BiaEvaluation software.

Human Tdp1 can remove fluorescein from the 3?-end, but not the 5?-end of DNA

Based on the structural diversity of substrates that are cleaved by Tdp1 (¹,⁵,⁶), we reasoned that a fluorescein molecule could also serve as a convenient and quantitative probe to investigate the DNA binding and processing of human Tdp1. To address the end-specific cleavage (3? versus 5?) of human Tdp1, 14-mer oligonucleotide conjugates were synthesized containing the fluorescein derivative, 6-FAM (chemical structure shown in Figure 1A), attached via the terminal phosphodiester at the 3?- or 5?-end of the oligonucleotide (14F and F14 in Table 1, respectively). As expected, the 3?-fluorescein-phosphodiester bond of the 14F oligonucleotide was hydrolyzed by human Tdp1 to generate the 3?-phosphate-terminated 14-mer product (Figure 1A). However, this reaction was less efficient than the 3?-phosphotyrosine substrate (14Y in Table 1) (Figure 1A and D).

Having established that 3?-fluorescein is a substrate for Tdp1, we tested whether Tdp1 could also process 5?-fluorescein-DNA (F14). Figure 1B shows that a single cleavage product was observed with the F14 oligonucleotide. However, this product exhibited a reduced migration relative to the 5?-phosphate-terminated marker (Figure 1B), suggesting that Tdp1 is unable to simply cleave the fluorescein adduct linked to the 5?-end of DNA. Given that the TdT-mediated 3?-end ³²P-radiolabeling of the F14 oligonucleotide leads to the addition of a 3?-terminal 3?-deoxyadenosine (cordycepin), we considered the possibility that the detected cleavage product may be a consequence of the 3?-nucleosidase activity of Tdp1. To establish the occurrence of this 3?-nucleolytic product, we utilized the 3?-phosphatase activity of polynucleotide kinase from bacteriophage T4 (PNK) to show that the product of the Tdp1 reaction contained a 3?-terminal ³²P-radiolabel. As shown in Figure 1C, the combination of Tdp1 and PNK resulted in the disappearance of the product (compare lanes 3 and 5), which indicates the removal of the 3?-deoxynucleoside by Tdp1 followed by removal of the resulting 3?-terminal ³²P-radiolabel by PNK (Figure 1E). This result was further corroborated in control reactions, which employed a 3?-phosphatase-minus T4 PNK mutant [PNK^?] (see lane 7). Similar results were obtained using a 5?-biotin-DNA substrate (data not shown). Taken together, these results demonstrate that human Tdp1 can hydrolyze a 3?-terminal phosphodiester-linked fluorescein or nucleoside leading to a 3?-phosphate DNA end under conditions where it has no detectable activity towards a 5?-terminal-phosphodiester-linked fluorescein.

Fluorescence anisotropy analysis of binding of human Tdp1 to fluorescein-labeled oligonucleotides

To characterize the DNA binding properties of human Tdp1 depending on the presence of fluorescein at the 5?- or 3?-end of the substrate, we measured changes in the anisotropy of fluorescein-labeled oligonucleotides upon titration with Tdp1. Fluorescence anisotropy is a spectroscopic technique that monitors macromolecular associations predominantly through changes in the rotational motion of fluorescent molecules (²⁵). As the molecular weight of the fluorophore increases, due to Tdp1?DNA binding, the rotational freedom of the DNA is reduced, leading to an increase in anisotropy (i.e. retention of polarization). This is illustrated schematically in Figure 2A.

For these experiments, we employed the two fluorescein-labeled oligonucleotides previously employed in the enzymatic activity assays (14F and F14) as well as an additional 5?-fluorescein-labeled oligonucleotide containing a 3?-phosphate (Table 1, F14p), known not to be hydrolyzed by Tdp1 (⁶). Figure 2B depicts the fluorescent anisotropy with increasing concentration of Tdp1. No increase in anisotropy was observed with the 3?-labeled oligonucleotide (14F), which was anticipated, given that Tdp1 hydrolyzes the 3?-fluorescein (Figure 1A). More interestingly, we detected a concentration-dependent enhancement in anisotropy for both 5?-labeled oligonucleotides (F14 and F14p). Based on calculated K_D values (Table 1), Tdp1 bound with an ?3-fold higher affinity to F14p than it did to F14. There are two possible explanations for this difference. First, Tdp1 may bind better to a 3?-phosphate as compared to a 3?-hydroxyl; or second, Tdp1 may have a higher binding affinity for longer DNA sequences in vitro. The latter is based on the fact that Tdp1 can hydrolyze 3?-hydroxyl ends (Figure 1B), which would result in a one base shorter F13p oligonucleotide product. Consistent with that explanation, previous studies have shown that Tdp1 is more effective at processing DNA-peptide substrates with a longer DNA component (⁴). Further support for the DNA length dependent binding of human Tdp1 is shown in Figure 2C, wherein Tdp1 shows a roughly 10-fold higher affinity for a 50-mer 3?-phosphate containing oligonucleotide as compared to a 14p oligonucleotide using a fluorescence anisotropy competition assay. These results also suggest that Tdp1 does not have to initially bind to the 3?-end of the DNA. If Tdp1 was directly binding the 3?-end of the DNA, then one would expect the binding curves to be indistinguishable given that both oligonucleotides have a single 3?-end available for binding. Thus, these results imply that Tdp1 may initially interact electrostatically with the DNA prior to reaching its site of action (i.e. the 3?-end).

3?-end specific binding of human Tdp1 to a non-hydrolyzable, phosphorothioate-linked 3?-fluorescein-containing oligonucleotide

Previous studies have demonstrated that guanosine residues have the propensity to quench the fluorescence intensity of fluorescein-oligonucleotide conjugates (25?7). More specifically, it has been shown that the time-resolved fluorescence decay of a fluorescein located proximal to a guanosine residue in a single-stranded oligonucleotide is fitted to a bi-exponential function with two apparent decay lifetimes (²⁷). This indicates that the fluorescein is found in two distinct conformations. The longer lifetime corresponds to an unquenched state or direct excitation of the fluorescein, whereas the shorter lifetime refers to a quenched state or rapid charge transfer reaction between the fluorescein and the guanosine residue (Figure 3A). Furthermore, the bi-exponential fit allows for determination of the relative abundance or amplitude of the two decay lifetimes.

To assess the end-specific binding of human Tdp1 using the aforementioned guanosine quenching method, we designed two oligonucleotides wherein a guanosine base was placed adjacent to the 5?- or 3?-terminal fluorescein. We envisioned that if human Tdp1 possesses a 3?-end specific binding activity, then we would observe a decrease in quenching, as Tdp1 would occlude the 3?-fluorescein from interacting with the adjacent guanosine residue (Figure 3A).However, prior to testing our hypothesis a non-hydrolyzable Tdp1 substrate had to be developed given that Tdp1 readily removes a phosphodiester-linked 3?-fluorescein (Figure 1A). Phosphorothioate linkages are commonly used to reduce oligonucleotide sensitivity toward nucleases. In addition, phosphorothioate modifications have been used to study the action of other phospholipase D enzymes (²⁸,²⁹). Therefore, we reasoned that attaching the 3?-fluorescein via a phosphorothioate linkage might result in a non-hydrolyzable Tdp1 substrate, which subsequently could be exploited for end-specific DNA binding studies.

As shown in Figure 3B, the phosphorothioate-linked 3?-fluorescein [15(pS)F] is resistant to Tdp1 hydrolysis under conditions where Tdp1 is able to remove phosphodiester-linked 3?-fluorescein (15F). These results enabled us to use the phosphorothioate-linked 3?- and 5?-fluorescein-labeled oligonucleotides [15(pS)F and F(pS)15 in Table 1, respectively] as probes for the time-resolved fluorescence quenching studies. We observed a concentration-dependent increase in the unquenched state of the 3?-fluorescein upon titration with recombinant human Tdp1. By contrast only minimally changes were discerned for the 5?-fluorescein (Figure 3C, lower curve). In addition, we found that the binding of Tdp1 to the non-hydrolyzable, 3?-fluorescein-containing oligonucleotide reached a plateau at a 1: 1 molar ratio. These results provide additional data on the distribution of different species in solution and are consistent with the 3?-end DNA binding of human Tdp1.

Fluorescence anisotropy was then used to quantify the binding of human Tdp1 to the fluorescent probes. The attachment of 5?-fluorescein via a phosphodiester or phosphorothioate linkage produced similar Tdp1 binding [compare F15 and F(pS)15 in Figure 3D), which was expected given that both oligonucleotides have identical 3?-termini (i.e. 3?-hydroxyl) (Table 1). Conversely, binding of Tdp1 to the non-hydrolyzable 15(pS)F oligonuceotide could now be detected by fluorescence anisotropy and was at least 10-fold more efficient than the 5?-fluorescein-linked oligonucleotides (Figure 3D and K_D values in Table 1). Taken together, these results suggest that unnatural 3?-adducts may create a high affinity binding site for human Tdp1.

Preferential binding of human Tdp1 to the 3?-end of DNA using SPR

To gain further evidence that human Tdp1 binds to the 3?-end of DNA, we utilized SPR wherein oligonucleotides were biotinylated at either the 5?- or 3?-end for end-specific immobilization to a neutravidin-coated sensor chip (B14p and p14B in Table 1). As shown in Figure 4A by the increase in response units from the initial baseline, Tdp1 binds with high affinity to the 5?-biotin immobilized, 3?-phosphate containing oligonucleotide (B14p) in a concentration-dependent manner. The Tdp1 binding kinetics to the B14p oligonucleotide could not be fit assuming a single association and dissociation rate constant. The complex kinetics may be due to non-specific binding of Tdp1 to the surface, kinetic heterogeneity in the Tdp1 protein sample, or transport effects. However, Tdp1 does form a stable complex with the B14p oligonucleotide at each Tdp1 concentration (Figure 4A). This equilibrium binding response can be fit to a Langmuir binding model assuming a 1: 1 binding interaction (inset in Figure 4A) with a K_D value of 57 nM (Table 1). In contrast, SPR analysis of the oligonucleotide immobilized in the opposite orientation (p14B) results in a complete loss of Tdp1 DNA binding activity (Figure 4B), indicating that the interaction between the DNA and Tdp1 requires an exposed 3?-end. Given the previously described ability of Tdp1 to hydrolyze 3?-linked biotin adducts (⁶), it is worth noting that the maintenance of the immobilized oligonucleotides subsequent to assaying the DNA binding activity of Tdp1 was ensured based on reproducible binding of HIV nucleocapsid protein (data not shown) (²⁴).