DNA Binding and Cleavage by the Human Parvovirus B19 NS1 Nuclease Domain

Abstract

Infection with human parvovirus B19 (B19V) has been associated with a myriad of illnesses, including erythema infectiosum (Fifth disease), hydrops fetalis, arthropathy, hepatitis, and cardiomyopathy, and also possibly the triggering of any number of different autoimmune diseases. B19V NS1 is a multidomain protein that plays a critical role in viral replication, with predicted nuclease, helicase, and gene transactivation activities. Herein, we investigate the biochemical activities of the nuclease domain (residues 2–176) of B19V NS1 (NS1-nuc) in sequence-specific DNA binding of the viral origin of replication sequences, as well as those of promoter sequences, including the viral p6 and the human p21, TNFα, and IL-6 promoters previously identified in NS1-dependent transcriptional transactivation. NS1-nuc was found to bind with high cooperativity and with multiple (five to seven) copies to the NS1 binding elements (NSBE) found in the viral origin of replication and the overlapping viral p6 promoter DNA sequence. NS1-nuc was also found to bind cooperatively with at least three copies to the GC-rich Sp1 binding sites of the human p21 gene promoter. Only weak or nonspecific binding of NS1-nuc to the segments of the TNFα and IL-6 promoters was found. Cleavage of DNA by NS1-nuc occurred at the expected viral sequence (the terminal resolution site), but only in single-stranded DNA, and NS1-nuc was found to covalently attach to the 5′ end of the DNA at the cleavage site. Off-target cleavage by NS1-nuc was also identified.

Graphical abstract

Human parvovirus B19 (B19V) is a ubiquitous virus infecting the majority of the human population.^1,2 B19V, a Parvoviridae family member of the genus Erythrovirus, so named for its tropism for erythroid precursor cells,³ has been associated with a myriad of different illnesses. It was first discovered as the cause of aplastic crisis in patients with chronic hemolytic anemia⁴ and then identified in 1983 as the causative agent of erythema infectiosum (Fifth disease),⁵ which results in mild fever and a distinctive rash in children and fever often with hepatitis and arthralgia in adults. B19V infection is also associated with pure red cell aplasia from persistent infection in immune-compromised patients and hydrops fetalis in pregnant women.¹ In addition, B19V infection has also been associated with several other serious conditions such as inflammatory cardiomyopathy and the induction of autoimmune or autoimmune-like disease (short or long term).^6–9

B19V is a single-stranded nonenveloped DNA virus of 5594 nucleotides, with an internal coding region flanked by palindromic sequences capable of forming terminal hairpin structures (NCBI entry NC_000883.1). The viral genome encodes five known, and a possible sixth, protein products: VP1 and VP2 that compose the viral capsid, NS1, the main replicative protein, and two or three smaller proteins of unclear function (7.5 kDa, 11 kDa, and an open reading frame for the putative X protein).^10–12 In addition to erythroid precursor cells, which are permissive for viral replication, B19V can infect a number of different cell types, utilizing the globoside or blood group P-antigen as a receptor¹³ along with α5β1 integrin.¹⁴ After entry, the viral DNA is shuttled to the nucleus with the host-encoded DNA repair protein Ku80,¹⁵ where replication and transcription occur concurrently following initial conversion of the viral DNA to double-stranded DNA.^16,17 Expression is thought to occur utilizing a single viral promoter, with mRNA for the different gene products produced from different splice variants.^11,18,19 Binding sites for several cellular transcription factors are found in the viral promoter,^20,21 which can be transactivated by host factors as well as by the viral NS1 protein.^20–22

Viral replication utilizes cellular proteins, proposed to be coordinated by the viral NS1 protein,^23–25 and is thought to make use of the terminal hairpins^23,25 (Figure 1). Extension of the 3′ end allows for replication of the majority of the viral genome, and NS1 is thought to cleave the opposite strand [at the trs (Figure 1)], producing a new 3′ end that can be used to prime synthesis of the remaining viral DNA.²⁴ B19V NS1 is predicted to contain both nuclease and helicase activities, likely required in B19V replication, and also contains a C-terminal domain that may be involved in the protein’s gene transcriptional transactivation activity. In addition to the transactivation of its own viral promoter, p6, NS1 has also been implicated in the transactivation of several host promoters,^26–30 some of which induce the S phase of the cell cycle (in permissive cells) for the production of host replicative proteins, followed by G2/M stalling, and finally apoptosis for viral release.^16,29 In addition, the genes for the pro-inflammatory cytokines TNFα and IL-6 have also been shown to be directly transactivated by B19V NS1 and have been related to models of autoimmune disease induction.^27,30

Figure 1.

Hairpin primed replication mechanism of B19V. Dotted lines represent the newly synthesized DNA, and trs denotes the terminal resolution site (predicted NS1 cleavage site).

Despite the prevalence of B19V and its multiple associated illnesses, relatively few biochemical and no structural studies of B19V NS1 have been reported. Herein, we investigate the DNA binding and cleavage activity of the nuclease domain of human parvovirus B19 NS1 (NS1-nuc). We show that NS1-nuc binds specifically and cooperatively to GC-rich sequences in the human p21 promoter, as well as to those known as NSBE (for NS1 binding elements) located in the viral origin of replication (Ori), which also overlap with the viral p6 promoter. Stoichiometric and sedimentation measurements indicate that three copies of NS1-nuc bind to the GC-rich regions of the p21 promoter and five to seven to the Ori region containing all four NSBE sequences. Only weak or nonspecific binding of NS1-nuc to the sequences derived from the human TNFα and IL-6 promoters was found. We also show specific cleavage at the terminal resolution site (trs) within the Ori (Figure 1), resulting in covalent attachment of NS1-nuc to the 5′ end of the DNA. We find that NS1-nuc cleaves the trs only when single-stranded, and with limited sequence specificity. The limited specificity results in off-target DNA cleavage, as well, which may pertain to one proposed mechanism of autoimmune disease induction by B19V involving epitope spreading.³¹

MATERIALS AND METHODS

Protein Preparation

A synthetic codon-optimized gene for NS1 (coding sequence based on GenBank entry ABN45789.1) was purchased (Epoch, Inc.) and used to amplify NS1 residues 2–176 (hereafter NS1-nuc), followed by ligation into the pMAL-c4E vector (New England Biolabs, Inc.) using the SacI and EcoRI restriction sites to create the expression vector for MBP-NS1-nuc. This expression construct was further modified to include (1) a cleavage site for the protease from tobacco etch virus (TEV), inserted between the enterokinase cleavage site and the N-terminus of NS1-nuc, (2) the addition of six histidine residues inserted just after the first residue of the fusion protein, and (3) three glycine codons inserted between the TEV protease cut site (TEVcs) and the N-terminus of NS1-nuc to create the expression vector for hMBP-TEVcs-GGG-NS1-nuc. The glycine residues were found to be necessary to obtain full protease cleavage of the chimeric protein but result in four additional amino acid residues (SGGG) on the N-terminus of NS1-nuc following cleavage by TEV protease. TEV protease was prepared as described previously.³²

Purification of NS1-nuc free of the purification tag was prepared using the hMBP-TEVcs-GGG-NS1-nuc construct and Talon resin (Clonetech, Inc.) chromatography, followed by overnight cleavage (at 4 °C) with TEV protease, and finally DEAE and Heparin FPLC (GE, Inc.) to separate NS1-nuc from TEV protease and hMBP. Figure S1 shows the purity of the protein preparation using Coomassie-stained SDS–PAGE.

DNA Preparation

The oligonucleotides were prepared synthetically by a commercial source and purified using C18 reverse phase high-performance liquid chromatography or PAGE and extraction. The concentration was measured spectrophotometrically, with an extinction coefficient calculated from standard values for the nucleotides,³³ and where appropriate including that for fluorescein. Equimolar quantities of complementary DNA were annealed by heating to 90 °C for 10 min at a concentration of 1 mM, followed by slow cooling to room temperature. The different DNA substrates used in binding, sedimentation, and cleavage assays are shown in Figure 2 and Figure S2.

Figure 2.

DNA sequences used in binding and cleavage assays. Boxed regions (NSBE1–4) are implicated in NS1 or transcription factor binding. ITR denotes an inverted terminal repeat or sequences that form the fold-over hairpin structures at either end of the B19V genome.

DNA Binding Affinity and Cooperativity Measurements

The gel shift assay³⁴ was used to measure the affinity and cooperativity of binding of NS1-nuc to DNA. The DNA was 5′ end-labeled with ³²P using T4 polynucleotide kinase (Thermo Fisher Scientific Inc.) and [γ-³²P]ATP (PerkinElmer, Inc.) and then purified using gel filtration (Micro Bio-30 Spin columns, Bio-Rad Laboratories, Inc.). Assays used a constant concentration of 1 nM ³²P-labeled DNA in 20 µL of binding buffer [100 mM Tris (pH 8), 150 mM NaCl, 1 mM EDTA, 1 mM 2-mercaptoethanol, and 10% glycerol], and independent incubations were performed with varied concentrations of NS1-nuc (0.1 nM to 15 µM). These concentrations were carefully chosen to give a well-defined binding curve. Native PAGE [10% 29:1 acrylamide:bis(acrylamide), 89 mM Tris base, 89 mM boric acid, and 2 mM EDTA] was used to separate the bound and unbound DNA. Care was taken to prevent heating of the gel, by electrophoresing at 4 °C and at low voltage (190 V). Gels were loaded while undergoing electrophoresis at 300 V, and the voltage was returned to 190 V 5 min after the loading of the last sample. Gels were then electrophoresed for an additional 2–3 h at 4 °C. Autoradiography of gels was performed without drying with a phosphor image plate exposed at 4 °C for 12–17 h. Densitometry of phosphor image plates was performed with a PharosXF imager (Bio-Rad, Inc.), and integration using Image Lab software (Bio-Rad, Inc.). The data were fit using the Hill equation³⁵ and the software Kaleidagraph (Synergy Software):

$$A=A_{\text{min}}+(A_{\text{max}}-A_{\text{min}})[P^n/({K_{1/2}}^n+P^n)]$$

where A is the fraction of DNA that is shifted (calculated from the integrated densities of shifted and unshifted bands) at a given NS1-nuc monomer concentration, A_max is the fitted fraction of shifted DNA upon saturation of all binding sites (at very high NS1-nuc concentrations), A_min is the fitted fraction of shifted DNA in incubations without added NS1-nuc, P is the concentration of free NS1-nuc (estimated here using the total concentration of NS1-nuc), K_1/2 is related to the apparent equilibrium dissociation constant K_d (K_d = K_1/2ⁿ, and is the concentration of NS1-nuc giving half-maximal binding), and n is the Hill coefficient and a measure of cooperativity in the binding of NS1-nuc to DNA. All measurements were performed in triplicate and reported as the average ± the standard deviation. Examples of gel images and data fitting are given in Figures S3 and S4.

In some cases, as indicated, the equilibrium dissociation constant for NS1-nuc and DNA was also measured using fluorescence polarization.³⁶ Fluorescein 5′ end-labeled DNA [labeled on the top strand only (Figure 2)] at 1 nM in 2 mL of binding buffer [100 mM Tris-HCl (pH 8.0 at room temperature), 50–150 (as noted) mM NaCl, 1 mM EDTA, and 1 mM 2-mercaptoethanol, with 10% glycerol] at 4 °C was titrated with increasing amounts of NS1-nuc (from 1 nM to 2 µM), and the anisotropy of the emitted fluorescence was monitored. Excitation occurred at 494 nm in a PC1 (ISS) fluorimeter with temperature control. The emitted intensities were measured using a 50.8 mm diameter 570 nm cutoff filter with a 580–2750 nm transmittance range (ThermoOriel Inc., catalog no. 59510) and 1 mm slit widths. The anisotropy of the emitted light as a function of added NS1-nuc was fit to the Hill equation as described above for the gel shift assay, with the exception that A is the anisotropy at a given protein concentration, A_max is the predicted anisotropy of fully bound DNA, and A_min is the anisotropy with no NS1-nuc binding.

Stoichiometric Measurements of DNA Binding Using the Gel Shift Method

Measurement of the stoichiometry of binding of NS1-nuc to various DNAs was performed using the gel shift assay as described above but with 10 µM ³²P-labeled DNA, and NS1-nuc concentrations of ≤240 µM. Gels were imaged as described above, and the stoichiometry was determined from the break or end point in NS1-nuc concentration after which no further binding occurred: the percent bound versus NS1-nuc concentration was fit to two straight lines, one before and one after saturation of binding of the DNA, and their intersection used to determine the concentration of NS1-nuc that fully saturated the 10 µM DNA in the assay. Examples of gel images and data fitting are given in Figures S5 and S6.

DNA Cleavage Assays

Single-turnover kinetic measurements of DNA cleavage were performed using 5′ end ³²P-labeled oligonucleotide substrates (1 nM, unless otherwise noted), under conditions of excess NS1-nuc (1 µM, unless otherwise noted). All reactions were performed at 37 °C in 50 mM HEPES-NaOH (pH 7.0 at room temperature), 150 mM NaCl, and 10 mM CoCl₂ (unless otherwise noted). Aliquots of 5 µL were withdrawn at specific time intervals after mixing NS1-nuc and ³²P-labeled DNA (100 µL total reaction volume), quenched by the addition of 5 µL of quench (80% formamide, 50 mM EDTA, 1 mg/mL XCFF dye, and 1 mg/mL BPB dye), and electrophoresed on denaturing polyacrylamide [20% acrylamide:bis(acrylamide) (19:1 ratio), 4 M urea, 89 mM Tris base, 89 mM boric acid, and 2 mM EDTA] gels. Autoradiography of gels was performed without drying using a phosphor image plate exposed at 4 °C for 12–17 h. Densitometry of phosphor image plates was performed with a PharosXF imager (Bio-Rad, Inc.), and integration using Image Lab software (Bio-Rad, Inc.). The percent of cleaved DNA product formed as a function of time was determined by integrating the density of both cleaved and uncleaved DNA bands and normalizing to the total amount cleaved. The fraction of cleaved DNA was then fit to a single-exponential function to determine the single-turnover rate constant of DNA cleavage using Kaleidagraph (Synergy Software):

$$\text{percentage of product}=C_1+C_2(1-{\mathrm{e}}^{-\mathit{\text{kt}}})$$

where C₁ is a constant fitting the baseline, C₂ is the total percent of DNA predicted to be cleaved, k is the observed rate constant (k_obs), and t is the length of incubation in minutes. Measurements were performed at least three independent times and are presented as the average ± standard deviation.

5′ Attachment of NS1-nuc to DNA

A 3′-fluorescein-labeled version of Ori1–67-top (Figure 2B, top strand) was purchased (Sigma-Genosys, Inc.) with PAGE purification and used in DNA cleavage assays to determine if NS1-nuc covalently attaches to the 5′ end upon DNA cleavage. A reaction mixture consisting of 2.6 µM Ori1–67-top-3′-fluorescein DNA, 27 µM NS1-nuc, 10 mM bis-tris propane (pH 9.5), 150 mM NaCl, and 8.5 mM MnCl₂ in a 414 µL total volume was incubated overnight at 37 °C. The reaction was quenched with 100 µL of 5× SDS–PAGE loading buffer [0.4 M Tris-HCl (pH 6.8), 0.5 M DTT, 10% SDS, and 50% glycerol] followed by heating to 95 °C for 5 min, cooling on ice, and then electrophoresis on a 12% acrylamide SDS–PAGE gel. Visualization of the fluorescein utilized a Typhoon scanner (GE, Inc.) with the Y520 filter setting (band-pass filter, 500–540 nm, centered on 520 nm) and the blue (488 nm) excitation laser. SDS–PAGE gels were also stained with silver stain (staining protein and nucleic acids) and Coomassie stain (staining protein only).

Analytical Ultracentrifugation

Sedimentation velocity experiments were performed using a Beckman Coulter XL-I instrument with monochromator and interference scanning optics, automated scanning capability, and a Ti-50 rotor. For NS1-nuc without DNA, approximately 400 µL of sample containing 29 µM NS1-nuc was loaded in one of the sectors of the two-sector sedimentation velocity cells. The other sector was loaded with 425 µL of buffer [10 mM Tris-HCl (pH 8.0 at room temperature), 150 mM NaCl, 1 mM EDTA, and 1 mM DTT]. The sample was allowed to equilibrate at 4 °C for at least 1 h in the mounted rotor. The sample was centrifuged at 40000 rpm (115000g), and absorbance scans were taken continuously at 280 nm for 10 h. Sedimentation velocity measurements performed with Flo-NSBE_DNA alone, Flo-p21 dsDNA alone, and mixtures of NS1-nuc with these DNAs were performed similarly but using concentrations of 17 µM (Flo-NSBE_DNA alone), 14 µM (Flo-p21 dsDNA alone), 107 and 15 µM (NS1-nuc and Flo-NSBE_DNA, respectively), and 130 and 16 µM (NS1-nuc and p21 dsDNA, respectively). Detection utilized the absorption of fluorescein at 495 nm. Data obtained from the scans were fit to a sedimentation coefficient distribution, c(s), using SEDFIT.^37,38 The viscosity and density of sample buffers were estimated using SEDNTERP.³⁹ Fitted values of f/f₀ were converted to diffusion coefficients and used in the Svedberg equation, along with estimated values from SEDNTERP for partial specific volume (v-bar) to compute a molecular mass (Mw). Calculated partial specific volumes of 0.72 and 0.55 mL/g were used to calculate Mw from data of NS1-nuc alone and DNA alone, respectively. However, in the case of protein/DNA mixtures, and because only a single value for v-bar can be used with the c(s) model in SEDFIT, a v-bar of 0.72 mL/g was used and the stoichiometry determined using the fitted Mws for the complexes, NS1-nuc alone, and DNA alone determined via sedimentation using the same value for v-bar (i.e., 0.72 mL/g). An additional analysis was used for the NS1-nuc/DNA data, where the fitting of different frictional ratios (f/f₀) was allowed for different s values [the bimodal fits (Table 2)]. Table S1 shows the final fitted parameters from SEDFIT. To determine the error limits on the fitted Mw of the NS1-nuc/DNA complexes, the following analysis was performed. First, the sedimentation data were fit using SEDFIT to give the lowest root-mean-square deviation (RMSD). Then, the critical RMSD was determined (at the 1σ level) using the statistical analysis component of SEDFIT, which is the RMSD below which different fits to the data can be considered equivalent, within 1σ or standard deviation (i.e., at a 68% confidence level). Then different values of the frictional ratio, f/f₀, were systematically altered and held fixed during SEDFIT fitting of the data. The resulting calculated apparent Mw and RMSD for only those fits (and corresponding f/f₀) below the critical RMSD are listed in Tables S2 and S3. Because one peak bears an apparent Mw equivalent to that of dsDNA, this value was used in the estimate of NS1-nuc/DNA stoichiometry, along with the apparent Mw of NS1-nuc also determined using AUC sedimentation velocity.

Table 2.

AUC Sedimentation Velocity Results with NS1-nuc and Fluorescein-Labeled DNA

sample	v-bar (partial specific volume, mL/g)	apparent Mw (kDa) (peak size)	interpretation
NS1-nuc	0.72	22.5 (95%)	monomeric (20.1 kDa, theoretical)
Flo-NSBE_DNA	0.55	29 (89%)	dsDNA (27 kDa, theoretical)
		13 (10%)	ssDNA (14 kDa, theoretical)
	0.72	48 (90.5%)	N/Aa
		21 (8.8%)
NS1-nuc/Flo-NSBE_DNA	0.72	166 (65%)	5.4:1 NS1-nuc:dsDNA
		44 (23%)	dsDNA
NS1-nuc/Flo-NSBE_DNA, bimodal fit	0.72	159 (67%)	5.3:1 NS1-nuc:dsDNA
		92 (7%)	2.4:1 NS1-nuc:dsDNA
		38 (23%)	dsDNA
Flo-p21 dsDNA	0.55	34 (82%)	dsDNA (33 kDa, theoretical)
		14 (6%)	ssDNA (16 kDa, theoretical)
	0.72	57 (81%)	N/Aa
		23 (6%)
NS1-nuc/Flo-p21 dsDNA	0.72	115 (83%)	2.9:1 NS1-nuc:dsDNA
		50 (13%)	dsDNA
		266 (2%)	9.6:1 NS1-nuc:dsDNA
NS1-nuc/Flo-p21 dsDNA, bimodal fit	0.72	110 (76%)	2.6:1 NS1-nuc:dsDNA
		51 (12%)	dsDNA
		266 (2%)	9.6:1 NS1-nuc:dsDNA

^aNot applicable.

RESULTS

Purification of Recombinant B19V NS1-nuc

The nuclease domain (residues 2–176) of B19V NS1 (NS1-nuc) was successfully purified using a synthetic codon-optimized (for Escherichia coli) gene in E. coli and an E. coli maltose binding protein (MBP) fusion expression system. Following initial capture of the MBP fusion protein, the NS1 nuclease domain was cleaved free from the MBP domain using recombinant TEV protease and the cleavage site engineered between the MBP and NS1 nuclease domains. In earlier versions of the fusion protein, cleavage was successful in only 50% of the protein preparation; hence, several (three) glycine residues were inserted between the cleavage site and the amino terminus of the nuclease sequence. This modification resulted in complete cleavage of the fusion protein (overnight incubation at 4 °C), and the MBP was subsequently purified from NS1-nuc using ion exchange chromatography. An overloaded lane in the SDS–PAGE gel shown in Figure S1 demonstrates the purity of NS1-nuc using Coomassie stain. Purified NS1-nuc was then dialyzed, concentrated, and stored with 50% glycerol in small aliquots at −80 °C until it was used in assays. Use of buffer at high pH (9.5) greatly reduced the number of solubility problems encountered with the protein, allowing relatively high concentrations (e.g., 10 mg/mL) to be achieved.

NS1-nuc Is Monomeric in the Absence of DNA

To determine the oligomeric state of NS1-nuc, analytical ultracentrifugation sedimentation velocity was used. Figure 3A shows the resulting c(s) distribution and calculated apparent Mw (22.5 kDa), which is very close to the Mw of 20.1 kDa calculated from the protein sequence (Table 2 and Figure S8; see also Table S1 for fixed and fitted parameters). Hence, NS1-nuc is monomeric in the absence of DNA.

Figure 3.

Sedimentation velocity analytical ultracentrifugation of NS1-nuc. (A) c(s) distribution calculated for the NS1-nuc sedimentation velocity data using SEDFIT, with the apparent Mw determined from the fit given above the c(s) peak. The theoretical Mw determined from the protein sequence is 20.1 kDa. (B) As in panel A, but with NS1-nuc/NSBE_DNA data. (C) As in panel A, but with NS1-nuc/p21 dsDNA data.

Binding of NS1-nuc to Viral Origin of Replication DNA

The affinity of NS1-nuc for DNA containing sequences from the B19V origin of replication (Ori) was measured using the gel shift and, in some cases, the fluorescence polarization assay (Materials and Methods). The resulting binding isotherms show sigmoidicity characteristic of cooperative binding (Figure 4), and therefore, the Hill equation was used to fit the data and determine K_1/2 and Hill coefficient n (Materials and Methods). K_1/2 is a measure of affinity, while Hill coefficient n is a measure of cooperativity. Table 1 gives the results of triplicate measurements for binding of NS1-nuc to the synthetic Ori containing DNA constructs Ori2–75 and NSBE_DNA (and its “knockouts” or KO) (see Figure 2 and Figure S2 for sequences). Ori2–75 is a 75 bp double-stranded DNA that contains all four of the NSBE (for NS1 binding elements) sequences as well as the trs (for terminal resolution site) where NS1 is implicated in binding and cleavage, respectively (Figures 1 and 2). NSBE_DNA is a shortened form of Ori2–75 containing only the NSBE sequences (Figure 2). In the case of NSBE_DNA and the KO versions [where each NSBE is replaced with a random sequence (Figure S2)], the stoichiometry of binding was also determined, which gives the number of NS1-nuc bound per DNA. From the data listed in Table 1, it is apparent that NS1-nuc binds relatively weakly (K_1/2 ~ 1 µM) but highly cooperatively (Hill coefficient n ~ 2–4) to the Ori DNA. In addition, little affinity is lost with the decrease in the size of the DNA sequence from Ori2–75 to NSBE_DNA, which eliminates the trs and surrounding sequences but retains four of the NSBE sequences (Figure 2B,C). Binding to the NSBE_DNA and KO DNA constructs was measured with both the gel shift and the fluorescence assay [GS and FPA, respectively (Table 1)], and the results show very good agreement, with the exception of a lower K_1/2 for the NSBE_DNA using FPA (Table 1). In general, the K_1/2 values for the four KO versions of NSBE_DNA tend to be higher (weaker affinity) but retain high cooperativity (Hill coefficient n > 1) (Table 1). The stoichiometry of binding of NS1-nuc to NSBE_DNA and its KO sequences was also determined using the gel shift assay and found to be seven copies of NS1-nuc bound in the case of NSBE_DNA, reduced by one or two copies in the case of the different KO sequences (Table 1). This suggests two copies of NS1-nuc bind per NSBE, with the exception of NSBE4, which binds only one.

Figure 4.

DNA binding by NS1-nuc. (A) Titration of 1 nM ³²P-labeled NSBE_DNA dsDNA and varied NS1-nuc protein (concentration increasing from left to right) using the gel shift assay [100 mM Tris-HCl (pH 8), 150 mM NaCl, 1 mM EDTA, 1 mM 2-mercaptoethanol, and 10% glycerol at 4 °C]. (B) Fit (line) to integrated densities (●) from panel A and the Hill equation giving a K_1/2 of 930 nM and an n of 2.8. (C) Binding data (●) and fit using the Hill equation (—) for fluorescein-labeled NSBE_DNA [100 mM Tris-HCl (pH 8), 150 mM NaCl, 1 mM EDTA, 1 mM 2-mercaptoethanol, and 10% glycerol at 4 °C] using fluorescence polarization anisotropy. The fit gives a K_1/2 of 550 nM and an n of 3.6.

Table 1.

Fitted Binding Constants and Stoichiometries for NS1-nuc with DNA Sequences Derived from the B19V Origin of Replication/p6 Promoter

DNA	K1/2 (µM)	Hill coefficient n	stoichiometry	methoda
Ori2–75 dsDNA	0.937 ± 0.015	4.4 ± 0.8	NDb	GS
NSBE_DNA dsDNA	0.930 ± 0.05	2.60 ± 0.13	6.94 ± 0.08	GS
Flo-NSBE_DNA dsDNA	0.61 ± 0.14	3.1 ± 0.9	NDb	FPA
NSBE KO1 dsDNA	1.09 ± 0.14	3.4 ± 0.5	5.27 ± 0.08	GS
NSBE KO2 dsDNA	1.58 ± 0.18	2.7 ± 0.3	5.6 ± 0.4	GS
NSBE KO3 dsDNA	1.39 ± 0.02	3.7 ± 0.6	5.3 ± 0.1	GS
NSBE KO4 dsDNA	1.27 ± 0.12	3.0 ± 0.19	6.0 ± 0.4	GS

^aMethod used to measure binding: GS, gel shift assay; FPA, fluorescence polarization assay.

^bNot determined.

Size of the NS1-nuc/Viral Origin of Replication DNA Complexes

Analytical ultracentrifugation (AUC) sedimentation velocity was used to investigate the Mw of the complex formed by NS1-nuc and DNA containing the viral origin of replication sequence. NSBE_DNA dsDNA was purchased with 5′ fluorescein [top strand labeled only (Figure 2A, Flo-NSBE_DNA)] to allow detection during ultracentrifugation of only species containing the fluorescein-labeled DNA. Figures S9–S12 show the results with Flo-NSBE_DNA dsDNA alone and NS1-nuc mixed with this DNA. In the case of the Flo-NSBE_DNA dsDNA alone, species with apparent Mw values of 29 and 13 kDa were identified when a value of 0.55 mL/g is used for the partial specific volume (v-bar) predicted for this DNA in the data analysis (Table 2, Table S1, and Figure S9). The predicted Mw of the duplex form is 27 kDa, which agrees well with the major peak, and the peak with half the size is likely the single-stranded DNA (the fluorescein-labeled strand). When the sample was mixed with NS1-nuc, two peaks were found in the calculated c(s) distribution: 166 and 44 kDa (in order of peak size). When the data were fit using a bimodal model, allowing for different frictional ratios to be fitted for species with different s values (to accommodate the possibility of different shapes of the DNA and the NS1-nuc/DNA complexes), apparent Mw values of 159, 92, and 38 kDa were found (Figure S12, Table 2, and Table S1).

To determine the stoichiometry of NS1-nuc and DNA in the complexes observed in the AUC data, the following procedure was used. First, because analysis of the data with SEDFIT required the use of only a single value for the partial specific volume of the macromolecular species (i.e., v-bar), and protein and DNA have very different values of v-bar, a v-bar corresponding to that of NS1-nuc (0.72 mL/g) was used. This required the estimation of the apparent Mw of the DNA when this v-bar is assumed, resulting in a value of 48 kDa (Table 2 and Figure S10). It was also noticed that the apparent Mw of one of the peaks of the c(s) distribution of the NS1-nuc/Flo-NSBE_DNA mixtures was very similar to this value [38–44 kDa (Table 2); also 37–57 kDa in the analysis depicted in Table S2]. Hence, stoichiometries were calculated using the apparent Mw of the larger species in the NS1-nuc/DNA mixtures less the apparent Mw of the DNA, and divided by the apparent Mw of NS1-nuc. Using the apparent Mw of 166 kDa for the Mw of the NS1-nuc/DNA complex (Table 2) and a value of 48 kDa for the DNA (Table 2), a 5.2:1 NS1-nuc:Flo-NSBE_DNA stoichiometry is found. If the apparent Mw of 44 kDa is used for the DNA instead [as found in the NS1-nuc/Flo-NSBE_DNA c(s) distribution (Table 2)], a ratio of 5.4:1 is found. Data from the bimodal fit result in stoichiometric ratios of 5.0:1 and 5.4:1 (Table 2).

To estimate the error limits of the stoichiometric ratios so derived, an analysis investigating the dependence of the apparent Mw (determined by SEDFIT³⁷) on different values of the frictional ratio, f/f₀, was performed (Table S2). The frictional ratio, f/f₀, is a measure of the effect of molecular size and shape on sedimentation, and values significantly greater than 1 indicate deviations from perfect spherical shape and/or from hydration. Sedimentation velocity experiments measure the s value of sedimenting species, and to estimate a Mw from that s value, an estimate of f/f₀ must first be determined. SEDFIT estimates f/f₀ from diffusion-dependent processes found in the experimental data,³⁷ thus allowing the estimation of an apparent Mw. SEDFIT also calculates the RMSD for each fitting analysis, as well as a critical RMSD below which any fit can be considered equivalent, within a set error limit. The analyses of Table S2 show results from different fits of the experimental data when f/f₀ was held constant at different values, and which showed RMSD below the critical RMSD (at 1σ). The calculated apparent Mw of the largest component of the c(s) distribution ranges from 146 to 209 kDa (Table S2). The peak identified as that of the free dsDNA [Mw3 (Table S2)] also varies, from 37 to 57 kDa, in its apparent Mw. An intermediate-sized complex is identified in some of the fits and ranges from 88 to 93 kDa in size. The stoichiometric ratios listed in Table S2 were calculated using the apparent Mw of the free dsDNA (from the same fit, i.e., Mw3), and a value of 22.5 kDa for the apparent Mw of NS1-nuc [determined via AUC (Table 2)]. These ratios range from 4.8 to 6.7 copies of NS1-nuc per single copy of Flo-NSBE_DNA. Hence, the stoichiometry determined by AUC sedimentation velocity is 5–7 copies of NS1-nuc per Flo-NSBE_DNA for the major NS1-nuc/DNA complex. It should be noted that this estimate assumes only a single copy of the Flo-NSBE_DNA per complex. If two copies of the DNA were present in the complex, three copies of NS1-nuc would be predicted to occur in the complex.

Binding of NS1-nuc to p21 Promoter DNA

The binding of NS1-nuc to a sequence derived from the human p21 promoter was also investigated for affinity and cooperativity (Table 3). This 53 bp segment of the p21 promoter that was the minimum sequence implicated in gene transactivation by NS1²⁶ (Figure 2D), and also contains several Sp1 binding sites, was found to bind to NS1-nuc with a K_1/2 of 5.7 ± 0.8 µM and with high cooperativity [Hill coefficient of 2.3 ± 0.3 (Table 3)].

Table 3.

Fitted Binding Constants for Binding of NS1-nuc to DNA Sequences Derived from the p21, TNFα, and IL-6 Promoters, as Well as Random Sequence DNA, Using the Gel Shift Method

DNA	K1/2 (µM)	Hill coefficient n
p21 dsDNA	5.7 ± 0.8	2.3 ± 0.2
TNFα AP-1 43 bp dsDNA	4 ± 3	2 ± 1
IL-6 NFkB 43 bp dsDNA	28 ± 18	1.32 ± 0.14
random 62 bp with 72.6% GC content	30 ± 4	6 ± 2
random 62 bp with 50% GC content	33 ± 6	1.6 ± 0.4

Size of the Complex Formed from NS1-nuc Binding to p21 Promoter DNA

In the case of the fluorescein-labeled p21 dsDNA (Flo-p21 dsDNA) alone, the AUC sedimentation velocity data analysis (using the calculated v-bar of 0.55 mL/g) indicates species with apparent Mw values of 34 and 14 kDa (Figure S13, Table 2, and Table S1). The predicted Mw is 32.6 kDa, in good agreement with the estimated Mw of the major peak in the c(s) distribution (Table 2). Analysis of the sedimentation velocity data of a mixture of NS1-nuc and Flop21 dsDNA resulted in the determination of species with apparent Mw values of 115, 50, and 266 kDa [in order of peak size (Table 2, Figure 3C, and Figure S15)]. Using the apparent Mw for the DNA when a v-bar of 0.72 mL/g is used [57 kDa (Table 2)] and 22.5 kDa for NS1-nuc (Table 2), NS1-nuc:Flo-p21 dsDNA stoichiometries of 2.6:1 and 9.3:1 are found for the two largest species. Using the Mw computed for the peak closest to 57 kDa in the NS1-nuc/Flo-p21 dsDNA analysis [50 kDa (Table 2)] as that of the free DNA gives stoichiometries of 2.9:1 and 9.6:1 for the two largest species. An analysis similar to that described above for the NS1-nuc and Flo-NSBE_DNA AUC data was performed to determine the 1σ error boundary limits of the stoichiometric ratios (Table S3), resulting in the ranges of 2.8–3.0:1 and 9.6–9.7:1. We interpret these values to indicate NS1-nuc:Flo-p21 dsDNA stoichiometries of 3:1 and 10:1. The 3:1 complex is by far the predominant complex, with the 10:1 complex being present at only an estimated 2% of the total DNA-bearing species. Alternative stoichiometries of the major complex (the 3:1) are not possible, because the major complex is too small to harbor more than one copy of the dsDNA with any copies of NS1-nuc. In addition, binding to single-stranded DNA is not anticipated because such binding was not seen in the gel shift data. [The K_1/2 for binding of NS1-nuc to either single strand in p21 dsDNA is estimated to be much greater than 50 µM from gel shift data using single-stranded p21 DNA (data not shown), which is 25-fold weaker than with double-stranded p21 DNA. Hence, binding would favor the double-stranded form, and also very little single-stranded DNA was observed in the sedimentation velocity experiment with Flop21 dsDNA alone (Figure S13).]

Binding of NS1-nuc to TNFα Promoter, IL-6 Promoter, and Random DNA Sequences

Finally, binding of NS1-nuc to sequences derived from the promoters of the human TNFα and IL-6 genes was investigated (Table 3). Unlike the B19V Ori/p6 and p21 sequences described above, binding of NS1-nuc to the TNFα and IL-6 sequences was much weaker [K_1/2 values of 4 ± 3 and 28 ± 18 µM, respectively (Table 3)] and with lower cooperativity [Hill coefficients n between 1 and 2 (Table 3)]. To determine if this binding behavior is non-sequence-specific, binding to a 62 bp duplex DNA containing random sequences of variable %GC [50 and 72.6% (Figure S2)] was also measured. In both cases, the K_1/2 was found to be ~30 µM, similar to that found for the IL-6 promoter element sequence (Table 3). However, this is 10-fold weaker than the binding affinity observed for the TNFα sequence. The cooperativity of binding to the random sequences differed, being low for the 50% GC DNA [Hill coefficient of 1.6 ± 0.4 (Table 3)] but higher for the 72.6% GC DNA [Hill coefficient of 6 ± 2 (Table 3)].

DNA Cleavage

NS1-nuc was used in assays to test for DNA cleavage (Figures 5 and 6). The assays made use of 5′ ³²P-labeled DNA containing segments of the viral origin of replication sequence [Ori1–67-top (Figure 2B)], where NS1 is presumed to cleave during viral replication (Figure 1). This 67 bp portion of the B19V genome has been found to be sufficient for NS1-mediated DNA replication.⁴⁰ The trs, terminal resolution site, is located 5′ to three or four copies of a repeat sequence designated as the NS1 binding element (NSBE)²²-(Figure 2A,B). The first two NSBEs contain exact copies of the sequence GCCGCCGG [two left-most boxes, NSBE1 and NSBE2 (Figure 2A–C)], while the sequence that follows contains GC-rich sequences that differ considerably but contain smaller (i.e., 4 bp) portions of the repeat sequence [shown as two additional boxed regions, NSBE3 and NSBE4 (Figure 2A,B)]. To test if NS1-nuc is capable of site-specific cleavage at the trs sequence, a synthetic oligonucleotide [Ori1–67-top (Figure 2B)] was 5′ end ³²P-labeled and incubated at a concentration of 1 nM with a high concentration of NS1-nuc (1 µM) under varied buffer conditions. Analysis of overnight incubations at 37 °C in buffer containing 50 mM HEPES-NaOH (pH 7.5), 150 mM NaCl, and 10 mM divalent cation (Ca²⁺, Mg²⁺, Zn²⁺, Co²⁺, Ni²⁺, or Mn²⁺) shows the presence of cleavage products with Mg²⁺ (red box, Figure 5), Co²⁺, Ni²⁺, and Mn²⁺ (Figure 5). The extent of cleavage with Mn²⁺, Co²⁺, and Ni²⁺ was greater than that with Mg²⁺, but the reaction also produced more than one cleavage product (Figure 5). Buffer conditions were varied and tested in NS1-nuc cleavage assays, with the optimal cleavage found to occur at pH 7.0–7.5 with 10 mM Mn²⁺ or Co²⁺ and found to be insensitive to NaCl concentrations between 0 and 200 mM (Figure S7). The cleavage sites with Mn²⁺ as the divalent cation were mapped using size markers containing the Ori sequence (Figure 5); the main cleavage site was determined to be 18 nucleotides (nt) from the 5′ end, corresponding to the trs (Figure 2B), and the minor cleavage site was found to be 30 nt from the 5′ end.

Figure 5.

Cleavage of viral origin of replication DNA. Cleavage reactions of NS1-nuc and ³²P-labeled Ori1–67-top, with 10 mM metal salts as indicated and size markers from 17–19 to 28–30 nt. The size in nucleotides of the main cleavage products is given in red adjacent to dotted lines indicating the migration pattern of the relevant size markers. The red box indicates a band corresponding to the 18 nt cleavage product expected from cleavage at the trs with MgCl₂. All reactions performed in buffer [50 mM HEPES-NaOH (pH 7.5) and 150 mM NaCl] at 37 °C for 18 h.

Figure 6.

DNA cleavage and binding assays with isolated NS1-nuc. (A–C) Constructs based on the viral origin of replication [Ori1-top-67 (Figure 2B)] with different degrees of truncation. trs denotes the terminal cleavage site. Gray boxes indicate putative NS1 binding sites (NSBE1–4); the arrow indicates cut site (trs), and numbers indicate nucleotides between the cut site and ends. (D) Denaturing PAGE of 5′ end-labeled DNA substrates from panels A–C before and after reaction with NS1-nuc [overnight at 37 °C, in 60 mM HEPES-NaOH (pH 7.5), 10 mM CoCl₂, 1 nM DNA, and 1 µM NS1-nuc]. Numbers indicate migration positions of different lengths of DNA. The asterisk indicates off-target (i.e., not at trs) cleavage.

To determine the minimal recognition sequence for trs cleavage by NS1-nuc, DNA oligonucleotides truncated at the 5′ or 3′ end of Ori1–67-top were tested in DNA cleavage assays (Figure 6). NS1-nuc was found to be capable of cleaving the trs sequence without the NSBE repeats, even when only 6 or 9 nt remain 3′ or 5′ of the trs, respectively (Figure 6B–D). Shorter oligonucleotides were not tested; instead, substitutions around the trs were investigated to map the importance of the nucleotides that surround the trs in recognition and cleavage by NS1-nuc (see below).

Timed measurements were also performed using conditions with the NS1-nuc concentration in excess of the DNA concentration (1 µM NS1-nuc and 1 nM ³²P-labeled DNA), and the production of cleaved DNA (Figure 7A) fit to a first-order rate constant, k_obs (Figure 7). Measurements were performed in triplicate (Table 4) and reveal a relatively slow rate constant compared to the rate constants for DNA cleavage by other nucleases (0.006 min⁻¹; for comparison, the type II restriction endonuclease EcoRV cleaves DNA with a single-turnover DNA cleavage rate constant⁴¹ of 70 min⁻¹), which was found to be slowed with the shorter DNAs Ori-20-top and NUC1 (Table 4; NUC1 contains the Ori1–67-top sequence but only 9 nt 5′ and 6 nt 3′ of the trs). Cleavage of NUC1 is approximately 10-fold slower than that of the others tested (Table 4). Interestingly, the cleavage rate constant of the off-target cleavage [30 nt from the 5′ end of Ori1–67-top (Figure 2B)] is very similar to that at the trs (Table 4).

Figure 7.

Timed cleavage assay using Ori1-top-67 DNA. (A) Image of a denaturing gel showing the increase in the level of cleaved DNA with time. (B) Integrated band intensities (points) and fits (lines) to data derived from panel A with cleavage at the trs (blue and blue filled squares) and off-target cleavage (red and red filled circles). Fits utilized the single-exponential function described in Materials and Methods.

Table 4.

Rate Constants for DNA Cleavage by NS1-nuc

DNAa	kobs (min−1)b	divalent cationc	cleavage site
Ori1–67-top	(2.3 ± 0.4) × 10−3	Co2+	trs
Ori1–67-top	(5.5 ± 1.2) × 10−3	Mn2+	trs
Ori1–67-top	(5.9 ± 0.8) × 10−3	Mn2+	off-target site
Ori1–67-top	(1.48 ± 0.17) × 10−3	Ni2+	trs
Ori-24-top	(3.2 ± 1.5) × 10−3	Co2+	trs
Ori-24-top	(1.32 ± 0.17) × 10−3	Ni2+	trs
Ori-20-top	(1.23 ± 0.4) × 10−3	Co2+	trs
Ori-20-top	(4.1 ± 0.2) × 10−4	Mn2+	trs
NUC1d	(5.4 ± 2.7) × 10−4	Mn2+	trs
NUC1d	(2.26 ± 0.5) × 10−5	Ni2+	trs

^aSee Figure 2, Figure 6, and Figure S2 for DNA sequences.

^bk_obs given as the average of three independent trials ± the standard deviation.

^cReaction conditions: 1 nM DNA, 1 µM NS1-nuc (unless otherwise listed), 50 mM HEPES-NaOH [pH 7 (Co²⁺) or pH 7.5 (Mn²⁺ or Ni²⁺)], 150 mM NaCl, 10 mM divalent cation (as listed), 37 °C, overnight.

^dNUC1 contains the Ori sequence [see Ori1–67-top (Figure 2B)] with only 9 nt 5′ and 6 nt 3′ of the trs.

To test the importance of each nucleotide around the trs in cleavage by NS1-nuc, a series of 29 nt oligonucleotides containing the Ori-top sequence (Figure 2A,B) 18 nt 5′ and 11 nt 3′ of the trs were tested in cleavage assays. Each contained a single-nucleotide substitution in one of the 6 nt on either side of the trs (substituted base colored red in Figure 8, Ori sequence colored black), and its cleavage by NS1-nuc was compared to that of the unsubstituted Ori sequence. Nucleotides −7 to −9 and +7 to +11 were also substituted collectively and tested for cleavage by NS1-nuc. Assays were performed with 1 µM NS1-nuc and 1 nM ³²P-labeled DNA in 50 mM HEPES-NaOH (pH 7.5), 150 mM NaCl, and 10 mM MnCl₂ for 29 h at 37 °C. Analysis of the cleavage rate (when cleavage occurred) gave no systematic pattern with the substitutions (data not shown), but the fraction of DNA cleaved by NS1-nuc was greatly affected by some substitutions. Therefore, the fraction of DNA cleaved, relative to that for the unsubstituted DNA, after a 29 h incubation was calculated, normalized to that cleaved in the case of the unsubstituted 29mer, and plotted in Figure 8. Substitutions just adjacent to the trs (1 nt 5′ and 2 nt 3′, i.e., −1, 1, and 2 in Figure 8) resulted in the most drastic reductions in the level of cleavage. In addition, substitutions in the 5 nt at the 3′ end of the 29mer also affected DNA cleavage by NS1-nuc (7 in Figure 8). Therefore, specificity directly at the trs appears to be limited to the A|CC sequence, with the vertical bar indicating the cleavage site. The other apparent specificity determinant occurred ≥7 nt from the trs, and just adjacent to the NSBE sequences (not present in the 29 nt DNA substrate used in the assays). In fact, inclusion of the NSBE sequences (i.e., cleavage of Ori1–67-top), while clearly not required for cleavage by NS1-nuc, did increase the total amount of DNA cleavage to a value of approximately 7 using the same scale as that in Figure 8 (data not shown).

Figure 8.

Plot of percent cleavage of DNA sequences containing single-site substitutions (red) around the trs within the Ori-29-top ssDNA. Cleavage occurs between −1 and 1. The wild-type sequence is colored black. Data are normalized to cleavage of unsubsituted Ori-29-top [a 29 nt single-stranded oligonucleotide with 18 nt 5′ and 11 nt 3′ of the trs (see Figure 2A for the sequence around the trs)]. Reactions performed with 1 µM NS1-nuc, 1 nM DNA, 50 mM HEPES-NaOH (pH 7.5), 150 mM NaCl, and 10 mM MnCl₂ at 37 °C for 29 h.

Figure 9 demonstrates the covalent attachment of NS1-nuc to the 5′ end of DNA after cleavage of the Ori sequence. Approximately 27 µM NS1-nuc was incubated with 2.6 µM Ori1–67-top DNA containing fluorescein at the 3′ end overnight at 37 °C in buffer containing 8.5 mM Mn²⁺. The reaction was quenched with SDS–PAGE loading buffer, and samples were electrophoresed using SDS–PAGE. Figure 9C shows the result of scanning the gel for fluorescein fluorescence. The fluorescein-labeled DNA is visible in the two right-most lanes that correspond to reactions with and without NS1-nuc, respectively. Only the lane with NS1-nuc contains a slower-migrating band that we identify as NS1-nuc covalently attached to the DNA (boxed region in Figure 9C). Panels A and B of Figure 9 correspond to Coomassie and silver staining, respectively, of similarly electrophoresed gels. Coomassie stains protein only, while silver stain stains both protein and nucleic acids. In Figure 9A, the free NS1-nuc is clearly visible only in lanes that contain NS1-nuc, and the lane with the sample incubated with the fluorescein-labeled DNA shows the additional band (boxed region in Figure 9A) running at the same position as that seen in the fluorescein scan (boxed region in Figure 9C). This region can also be identified in the silver-stained gel (boxed region in Figure 9B).

Figure 9.

Covalent attachment of NS1-nuc to 3′ fluorescein-labeled Ori1-top-67 following cleavage analyzed via 12% SDS–PAGE: (A) Coomassie stain, (B) silver stain, and (C) fluorescein scan.

DISCUSSION

Cooperative Binding of NS1-nuc at the Viral Origin of Replication/Viral p6 Promoter DNA Sequences

There are several reports in the literature regarding binding of B19V NS1 to DNA, either using direct binding analysis or via detection of NS1-mediated promoter transactivation. Among the purported DNA binding sites of B19V NS1 is the sequence found in the viral origin of replication (Ori), which also overlaps with the single B19V promoter (the p6 promoter).^20,22 NS1 has been shown to interact specifically with a series of 8 nt GC-rich regions within this region termed NSBE for NS1 binding elements (Figure 2A,B).^22,42 Because the Ori sequences overlap with the p6 sequences, it is possible that these same NSBE are bound by NS1 for both viral replication functions and p6 promoter transactivational functions. By analogy with the parvoviruses AAV and MVM, and their Rep/NS1 proteins that share homology with B19V NS1, binding of NS1 at the NSBE sequences is predicted to function in replication by producing DNA cleavage at the trs just adjacent to the NSBE (Figure 6A).^43–50 In p6 promoter transactivation, NS1 may bind to the NSBE boxes and transactivate directly or, alternatively, indirectly through cooperation with other transcription factors,^20,22 as has also been suggested for homologues of B19V NS1.⁵¹ Switching between the two functions, viral replication and promoter transactivation, may involve post-translational modifications of parvoviral NS1 proteins.^52–62 We found that the isolated B19V NS1-nuc, prepared recombinantly from E. coli and therefore without any post-translational modifications, was capable of binding to double-stranded DNA containing the NSBE sequences, in the absence of other proteins (Figure 4A–C and Table 1). NS1-nuc binds relatively weakly [K_1/2 ~ 1 µM (Table 1)] but with high cooperativity [Hill coefficient of ~2–4 (Table 1)] to this DNA. [We use the notation K_1/2, rather than K_D, to denote the midpoint of the transition of the binding isotherm (Figure 4B,C) because binding is cooperative with a stoichiometry greater than 1, as described below.] Weak, micromolar binding of an isolated parvoviral NS1 nuclease domain to DNA sequences derived from the viral origin of replication was also observed for MVM NS1.⁴⁸ It should be noted that although binding to the NSBE sequences is relatively weak, it is still 30-fold tighter than binding of NS1-nuc to nonspecific DNA (Table 3; binding to dsDNA of random sequence has a K_1/2 of ~30 µM). Also, the K_1/2 for binding of NS1-nuc to the NSBE containing DNA determined using the fluorescence anisotropy method is lower, indicating tighter binding, than that measured using the gel shift assay (Table 1). This difference is likely due to the differences in two techniques;⁶³ the gel shift assay involves separation of the bound DNA from unbound in measuring their relative concentrations, while the fluorescence anisotropy assay does not. Separation of the bound and unbound DNA disrupts their equilibrium, allowing further dissociation of the DNA-bound forms, hence potentially leading to an estimate for affinity that is shifted toward a higher K_1/2 (underestimating the strength of binding). This effect is expected to be greater when binding is weaker, as in the current case. In addition, the presence of the fluorophore itself may influence the binding of a protein to the labeled nucleic acid, typically strengthening it through nonspecific binding of the fluorophore to the protein.⁶³ Finally, the high cooperativity of DNA binding by NS1-nuc suggests protein–protein interactions on the DNA and/or alterations in the DNA structure that facilitate further binding of NS1-nuc. The K_1/2 values indicate weak affinity of individual monomeric NS1-nuc for the DNA; however, in the context of full length NS1, which is likely oligomeric like its parvoviral homologues,^64–70 binding should be much stronger because of cooperative binding from the multiple NS1-nuc domains found in an oligomeric NS1.

Five to Seven Copies of NS1-nuc Bind to Four NSBE Sequences in the Viral Origin of Replication/p6 Promoter

Because the Hill coefficient is greater than 1 (Table 1), cooperative binding of multiple copies of NS1-nuc to the DNA is implicated. Therefore, we also measured the stoichiometry of binding of NS1-nuc to the NSBE sequences. Using a gel shift method, a total of seven copies of NS1-nuc were found to bind per DNA containing all four NSBE sequences [NSBE_DNA (Table 1)]. In addition, AUC sedimentation velocity was used to measure the sizes of complexes formed with NS1-nuc and fluorescein-labeled NSBE_DNA (Table 2, Figure 3B, Figures S11 and S12, and Tables S1 and S2). Several peaks were identified in the c(s) distribution (Figure 3B), with the major peak being consistent with approximately five to seven copies of NS1-nuc bound per NSBE_DNA, in agreement with the stoichiometry determined by gel shift, which indicated seven. A second species was also identified, which is consistent with the binding of two copies of NS1-nuc bound to this DNA (Table 2).

Therefore, the cooperativity observed in binding (i.e., Hill coefficient n greater than 1), stoichiometry, and AUC data all indicate multiple copies of NS1-nuc binding per NSBE containing DNA. Because NS1-nuc is monomeric in the absence of DNA (Figure 3), a Hill coefficient above 2 suggests that at least two copies of NS1-nuc bind cooperatively. In some cases, the Hill coefficient is as high as 4, indicating cooperative binding of at least four copies of NS1-nuc (Table 1). The finding from the stoichiometry and AUC measurements that five to seven copies of NS1-nuc bind this DNA indicates that the cooperativity while high is not complete. Perfect or complete cooperativity would mean all copies of NS1-nuc bind simultaneously, and that no intermediate binding states exist with fewer. Instead, the incomplete cooperativity with a Hill coefficient of less than five to seven suggests that although binding of NS1-nuc to the DNA favors the binding of additional NS1-nuc, complexes with fewer than five to seven copies bound to the DNA are formed at subsaturating concentrations of NS1-nuc.

To investigate binding of NS1-nuc to the individual NSBE sequences, four “knockout” (KO) versions of the DNA (Figure S2B–E), in which each NSBE was replaced with random AT sequences, were tested for NS1-nuc binding. These showed somewhat decreased affinity [K_1/2 (Table 1)] for NS1-nuc, with variable effects on binding cooperativity [Hill coefficients (Table 1)]. The largest effect occurred with the KO2 substitution, where the sequences of the second NSBE were substituted (Figure S2C), and that resulted in a 65% increase in K_1/2, indicating weakened binding affinity (Table 1). However, overall, the effects of the KO substitutions on binding affinity and cooperativity are difficult to interpret due to the multiple, cooperative nature of binding of NS1-nuc to the DNA. In contrast, the effects of the KO substitutions on the total number of NS1-nuc copies bound to the DNA are striking, with reductions in one to two copies with each NSBE box substitution (Table 1).

Model of NSBE Binding by NS1-nuc

Combining the binding affinity, cooperativity, and stoichiometry data, along with the crystal structure of the AAV Rep nuclease domain bound to Ori DNA,⁷¹ results in the model for binding of NS1-nuc to NSBE shown in Figure 10. In the AAV Rep nuclease domain/DNA X-ray crystal structure, five nuclease domains are bound to the AAV Ori DNA containing the five direct RBE repeats.⁷¹ Because each RBE repeat contains 4 bp, yet each NSBE repeat contains 8 bp, it is possible that each NSBE half-site is analogous to each RBE repeat. This would predict the binding of two NS1-nuc domains per NSBE. Indeed our NSBE “knockout” data [NSBE KO1–KO4 dsDNA (Table 1)] do in fact indicate that two NS1-nuc molecules bind per NSBE, with the exception of the fourth NSBE, which appears to bind only one. The fourth NSBE sequence contains one GC-rich half-site and one AT-rich half-site (Figure 10). Each of the other NSBE half-sites is GC-rich; therefore, the model of Figure 10 shows the singular NS1-nuc bound to NSBE4 at the GC-rich half-site. This positioning also places it closer to the other DNA-bound NS1-nuc, such that favorable protein–protein interactions could potentially occur. In the AAV Rep nuclease domain/DNA structure, the nuclease domains interact sequence specifically with the RBE repeats, mostly with one of the two DNA strands, and therefore spiral around the DNA.⁷¹ The NS1-nuc forms shown schematically in Figure 10 are drawn to emphasize the relative spacing seen in the AAV Rep nuclease domain/DNA structure, where the closest approaches in three dimensions between proteins are not to those bound at adjacent DNA sequences, but to those two binding sites away (blue to blue and red to red, Figure 10). In addition, because every other NS1-nuc binding site is 10 bp apart, every other NS1-nuc would be one turn apart and on the same side of the DNA (if the DNA maintains the B-form conformation). Although the AAV Rep nuclease domains are not in close contact with each other in the crystal structure, the amino acid sequence of B19V NS1-nuc contains several inserts relative to that of AAV Rep, and in locations of the structure that could potentially create protein–protein contacts. Favorable protein–protein interactions between NS1-nuc molecules bound to the NSBE sequences would explain the observed cooperativity in DNA binding [Hill coefficient of >1 (Table 1)]. Finally, rules for sequence specificity at the NSBE half-sites are not obvious but, if similar to those for the AAV Rep nuclease domain, could involve a combination of NS1-nuc/DNA contacts, not all of which need to be satisfied by any one copy of NS1-nuc, and could derive from more than one copy of the DNA sequence repeat to any one NS1-nuc.⁷¹

Figure 10.

Model for binding of NS1-nuc to the NSBE sequences in the B19V Ori DNA.

Recently, another study that investigated the binding of B19V NS1-nuc to sequences at the viral origin of replication was reported.⁴² As described in the study presented here, NS1-nuc (residues 4–180) was expressed and purified from a recombinant system in E. coli. As in our study, it was found to bind to the region containing the NSBE sequences; however, a major conclusion from that study was that NS1-nuc bound to NSBE1 and NSBE2, but not to NSBE3 or NSBE4. The authors suggest that host factors may bind to NSBE3 and -4, because NSBE1–3 have been found to be critical for viral replication, and NSBE4 was required only for maximal viral replication.⁴⁰ The study also found that the base pairs at positions 1, 4, and 6–8 of NSBE1 or NSBE2 were the most important for sequence-specific binding by NS1-nuc, and the authors suggest a model in which NSBE1 and NSBE2 each bind one copy of NS1. We also found that NSBE1 and -2 were important for NS1-nuc binding but propose, on the basis of our stoichiometry data, that each NSBE in the first three boxes (i.e., NSBE1–3) binds to two copies of NS1-nuc, and NSBE4 binds a seventh. Also, unlike this previous study, we report the K_1/2, a measure of affinity, and the Hill coefficient cooperativity factor n for binding NSBE1–4 and the various NSBE KO by NS1-nuc.

AAV Rep68 (i.e., full length containing the helicase/ATPase domain) has been shown to bind to its Ori with five total copies,⁶⁷ but also to form rings containing six, seven, or eight copies.^64,66,70,72 These studies, along with structural studies, have inspired a model for Ori recognition, binding, and manipulation by parvoviral replication proteins wherein binding to the nearby repeats (i.e., NSBE in the case of B19V Ori, RBE in the case of AAV Rep) induces ring formation of the Rep protein around the double-stranded DNA. The ring structure then uses its helicase/ATPase activity to cause strand separation at the nearby trs sequence, thereby making the trs competent for cleavage by a nuclease domain of the oligomeric Rep⁷¹ (because the AAV Rep nuclease domain requires the trs to be single-stranded to be cleaved, as we have also observed here for B19V NS1-nuc). Our observation of five to seven copies of NS1-nuc binding to the NSBE sequences is consistent with this model. Alternatively, a model in which binding to DNA in a spiral fashion induces melting at the trs via supercoiling tension and/or DNA stretching, as has been proposed for the E. coli origin of replication by DnaA,⁷³ is an attractive alternative and is also consistent with the current data.

DNA Cleavage by NS1-nuc

B19V NS1-nuc belongs to the HUH (for histidine–hydrophobic–histidine) superfamily of single-stranded DNA nucleases, which are divalent cation-dependent and utilize two histidine residues in the active site, as well as a tyrosine residue that supplies the nucleophile in the phosphodiesterase reaction, resulting in a 5′-phosphotyrosine complex and a free 3′-OH.⁷⁴ B19V NS1 is proposed to function in cleaving replicating viral DNA (Figure 1), necessary to complete viral genome replication. The DNA cleavage activity of B19V NS1-nuc was tested using DNA containing the expected cleavage site, namely the B19V terminal resolution site or trs (Figures 1, 2A,B, and 6), and DNA length and cleavage buffer composition were varied and tested for their effects on DNA cleavage activity of B19V NS1-nuc (Figures 5 and 6 and Figure S7). The likely biologically relevant cofactor for the nuclease activity is Mg²⁺; however, Mg²⁺-dependent nucleases can often utilize other metal ions, and therefore, several were tested (Figure 5). Weak DNA cleavage activity was found with Mg²⁺ (red box in Figure 5), but more robust activity was found with Mn²⁺, Co²⁺, and Ni²⁺ (Figure 5). Substitution of Mn²⁺ in place of Mg²⁺ is known to “rescue” the Mg²⁺-dependent nuclease activity of mutant enzymes, doing so by increasing both the cleavage rate and the affinity of the enzyme for the DNA substrate.^75,76 In the case of NS1-nuc, truncation of the nuclease domain from the remainder of the full length protein may compromise its binding affinity for the DNA substrate, which could be partially compensated by using ions such as Mn²⁺, Co²⁺, and Ni²⁺. The metal ion dependence reported for other HUH superfamily endonucleases⁷⁴ shows DNA cleavage with Mg²⁺, Mn²⁺, and Ni²⁺, but also Ca²⁺, Zn²⁺, and Cu²⁺. However, work with other parvoviral NS1 enzymes^45,77 has shown that while the full length protein can utilize Mg²⁺, the isolated AAV Rep nuclease domain⁷⁷ strongly preferred Mn²⁺, and little activity⁷⁸ is found with Zn²⁺ or Ca²⁺, as seen in our assays.

Cleavage assays performed at different pHs and salt concentrations indicate that the nuclease activity of NS1-nuc is optimal at pH 7.0–7.5 with 10 mM Co²⁺ or Mn²⁺, and that a NaCl concentration between 0 and 200 mM had little effect on activity (Figure S7). Cleavage was also tested with double-stranded (data not shown) and single-stranded versions of the viral Ori sequences (Figure 2A,B). NS1-nuc cleaved the DNA only when it was single-stranded, consistent with earlier reports about related nucleases of increased nuclease activity on a nonduplexed substrate.^71,74,79,80 The precise cleavage site was mapped to the expected location [i.e., the trs (Figure 2A,B)] by comparison of the electrophoretic mobility of the product of cleavage with size markers containing the same sequence of the expected product DNA if cleaved at the trs [18 nucleotides from the 5′ end of Ori1–67-top (Figure 2B)] or one nucleotide shorter or longer [17 or 19 nucleotides, respectively (Figure 5)]. The main cleavage product runs at the same position as the 18 nt size marker (Figure 5), showing clearly that NS1-nuc targets specifically the trs in the DNA; however, at least one minor cleavage product occurs farther from the 5′ end, discussed below.

To identify the minimal DNA sequences required for recognition and cleavage by NS1-nuc, truncations of the trs containing DNA were tested in cleavage assays (Figure 6). First, Figure 6A shows the 67 nt single-stranded DNA, Ori1–67-top (sequence shown in Figure 2B), and Figure 6D shows that this DNA is cleaved specifically by NS1-nuc (second lane from the left, Figure 6D). Next, a DNA construct lacking the NSBE boxes and containing only 6 nt 3′ of the trs (Figure 6B) was found to be cleaved by NS1-nuc (fourth lane from left, Figure 6D). This indicates that the NSBE sequences are not necessary for recognition and cleavage of the trs by NS1-nuc in single-stranded DNA. A construct with the DNA 5′ of the cleavage site reduced to 9 nt was also cleaved by NS1-nuc (Figure 6C, and right-most lane, Figure 6D). Together, these data indicate that NS1-nuc can recognize the trs using all or a subset of the 17 nt (9 nt 5′ of the trs and 6 nt 3′) around the trs. Further truncations of the DNA were not tested; instead, to investigate which nucleotides were most important for recognition by NS1-nuc, substitutions in a 29 nt DNA containing 18 nt 5′ and 11 nt 3′ of the trs were next tested for cleavage by NS1-nuc (see below).

Rate of DNA Cleavage by NS1-nuc

The rate of DNA cleavage by NS1-nuc was also measured using timed cleavage assays (Figure 7). Figure 7A shows an autoradiogram of cleavage reaction products formed during a timed reaction, and Figure 7B shows the data plotted as a function of time. The data fit well to a single-exponential function giving an observed rate constant of ~0.006 min⁻¹ (Table 4). As can be seen in Table 4, similar rate constants were found for the different trs containing DNA substrates that differ in length, with the exception of the shortest construct, NUC1 DNA (Table 4). In addition, cleavage of an off-target (i.e., not the trs) site in the longer construct, Ori1–67-top, occurred with similar kinetics (Table 4). Although the data fit well to a single-exponential function, we hesitate to ascribe this rate constant to the rate of the “chemical step” of DNA cleavage by NS1-nuc, because the binding affinity of the nuclease domain for these DNA constructs is very weak [K_1/2 > 6 µM measured without a divalent cation (data not shown), which is similar to the NS1 concentration of 10 µM in the assays], and saturation of binding to the DNA in the assay may be incomplete. It is also interesting that the trs does not appear to be cleaved in the “off-target” cleaved DNA; this off-target cleavage product reaches a maximal value but then does not then decrease to produce the product cleaved at the trs (red, Figure 7B). This may be due either to loss of affinity (sequence-specific or non-sequence-specific) of NS1-nuc for the DNA missing the nucleotides 3′ of the off-target cleavage site (and shortening the number of nucleotides 3′ of the trs from 49 to 12) or alternatively to loss of enzyme activity over time, or perhaps both.

Sequence Specificity of DNA Cleavage by NS1-nuc

Because NS1-nuc cleaved relatively specifically at the trs sequence, substitutions around the cleavage site were tested for their effects on this cleavage activity (Figure 8). The substitutions were made in the context of a 29 nt oligonucleotide containing 18 nt 5′ of the trs and 11 nt 3′ (see Figure 2A,B for the DNA sequence around the trs). The reaction conditions included 50 mM HEPES-NaOH (pH 7.5), 150 mM NaCl, 10 mM MnCl₂, 1 nM ³²P-labeled DNA, and 1 µM NS1-nuc. Mn²⁺ was used in the assays because of the relatively robust cleavage activity it imparts to NS1-nuc (Figure 5); however, it is important to remember than the substitution of Mn²⁺ for Mg²⁺ may also lead to loss of some DNA sequence specificity.⁸¹ Substitutions were made either individually (the 6 nt either side of the trs) or in groups [nucleotides −7, −8, and −9 together and nucleotides +7 to +11 together (Figure 8)] and were made to alter the base identity to its base pairing complement; hence, G was changed to C, A to T, etc. The cleavage reaction was performed for 29 h at 37 °C, and the fraction of total DNA cleaved was determined for each substituted DNA and normalized to that for the unsubstituted DNA (dotted line, Figure 8). Most of the substituted oligonucleotides were cleaved to an extent similar to or greater than that of the unsubstituted DNA, with the exception of those substituted directly around the cleavage site, as well as those at the 3′ end (Figure 8). When either the C just 3′ of the trs (black sequence, position 1, Figure 8) or the next nucleotide, also a C (position 2, Figure 8), was changed to G, cleavage by NS1-nuc fell to nearly undetectable levels. Substitution of the base just 5′ of the cleavage site also resulted in a clear reduction in the level of cleavage by NS1-nuc.

The data show that the DNA cleavage sequence specificity of NS1-nuc for the trs is limited to very few nucleotides around the cleavage site (Figure 8), at least under the buffer conditions tested, which included Mn²⁺. Substitution of the 5 nt at the 3′ end of the 29mer, which are just before the NSBE sequences (absent in the 29mer), also greatly diminishes the level of cleavage by NS1-nuc. This cannot merely be due to changes in non-sequence-specific affinity, because the length of the oligonucleotide does not change in the substituted DNA, merely its sequence. Therefore, although not necessary for cleavage, these sequences have a positive effect on DNA cleavage by NS1-nuc. Similarly, the inclusion of the NSBE sequences, though clearly not necessary for trs recognition and cleavage, greatly increases the percentage of DNA cleavage using this assay [to a value of roughly 7 on the scale of Figure 8 showing % cleavage/% cleavage of unsubstituted Ori-29-top (data not shown)]. These effects may be a result of increased affinity of NS1-nuc for the DNA through sequence-specific (nucleotides +7 to +11 and/or NSBE) or non-sequence-specific interactions (the DNA with the NSBE is longer).

These results differ from those reported for AAV Rep,⁸² where effects on cleavage by substitutions of seven nucleotides around the trs (5 nt 5′ and 2 nt 3′ of the trs) were found. However, the studies with Rep were performed with the full length enzyme, rather than the isolated nuclease domain, and also used a double-stranded Ori construct containing the full RBE sequences (analogous to the NSBE sequences of B19V). Additionally, the studies with AAV Rep were conducted with Mg²⁺ as the divalent cation cofactor, rather than Mn²⁺ as in the current B19V NS1-nuc studies, and as previously mentioned, Mn²⁺ has been reported to affect sequence specificity in some Mg²⁺-dependent endonucleases.⁸¹

Covalent Attachment of NS1-nuc to DNA

Previous reports using nuclease enzymes homologous to NS1 show covalent attachment to the 5′ end of the DNA at the site of cleavage via an active site tyrosine residue.^{43,45,56,83–85} To test for this activity in recombinant B19V NS1-nuc, products of the reaction with 3′-fluorophore-labeled Ori1–67-top DNA (Figure 2B) were analyzed using SDS–PAGE with staining for protein (Coomassie stain, Figure 9A), protein and DNA (silver stain, Figure 9B), or DNA only (fluorescence, Figure 9C). In Figure 9, the boxed region identifies the inferred position of the NS1-nuc–DNA covalent adduct in these gels. Because the sample was treated with SDS and a reducing agent (DTT) and was boiled prior to electrophoresis, no noncovalent complexes should be present. The boxed species migrates more slowly than the protein alone (Figure 9A,B) as well as the DNA only (Figure 9C) and contains protein (Figure 9A) and DNA containing a 3′ fluorophore (Figure 9C), as expected for a covalent complex between NS1-nuc and DNA.

Off-Target DNA Cleavage by NS1-nuc

Cleavage of sequences other than the viral origin of replication is not implicated in any known biological function of NS1; however, cleavage at non-trs’s in the host genome has been implicated in pathogenic activities of B19V and NS1.^31,86 Such cleavages are damaging to the host genomic DNA, in terms of both the single-stranded break and the resulting covalent attachment of NS1 to the DNA. This damage can result in apoptosis and the production of apoptotic bodies that could stimulate the immune system to become autoreactive via the production of anti-DNA antibodies.^31,86 Such antibodies may also be a gateway to immune reaction via epitope spreading against additional nuclear targets, leading to further loss of self-tolerance.⁸⁷ This is one of several mechanisms implicated for the induction of autoimmune disease by B19V.³¹ In our DNA substrate, the off-target cleavage was mapped to 30 nucleotides from the 5′ end of the Ori1–67-top substrate (Figure 5), which also corresponds to cleavage just after the first base pair of the first NSBE sequence (Figure 2B). The cleavage sequence at this position is G|CC (the vertical bar indicating the cleavage site), while the sequence at the trs is A|CC. Cleavage at the G|CC sequence is consistent with the results of cleavage of DNA substituted at nucleotides around the trs (Figure 8), where the two C nucleotides just 3′ of the trs (i.e., the |CC nucleotides) were most important to recognition, followed by the position just 5′ of the cleavage site (A| in the trs sequence). While the identity of that base in the off-target cleavage is a G, rather than an A, its identity as a purine may confer some recognition by NS1-nuc (the substitution tested in Figure 8 at this position, −1, was a T, rather than a G). This off-target site would presumably be masked by NS1 binding at the first NSBE in double-stranded DNA (see above), and if the NS1 nuclease is similar to the homologous AAV Rep nuclease domain, the cleavage active site would be located on a face of the enzyme distinct from the DNA sequence-specific binding face.⁷¹ Interestingly, three other GCC sequences (and one TCC sequence) also occur in Ori1–67-top but are not cleaved by NS1-nuc in our assays. This suggests other recognition elements exist around the trs and off-target cleavage site or that these noncleaved sites are somehow masked, perhaps also by alternative modes of NS1-nuc binding.

Promoter DNA Binding by B19V NS1-nuc

B19V NS1 has been implicated in promoter transactivation of the B19V p6 promoter,²⁰ as well as those of several host genes (p21,²⁶ TNFα,³⁰ and IL-6²⁷). The transactivation of TNFα and IL-6 genes by NS1 is another possible mechanism by which B19V could modulate the host immune system and contribute to autoimmune disease,⁸⁸ while production of p21 cyclin-dependent kinase through promoter transactivation results in cell cycle arrest at G1.²⁶ In each case, cooperation of NS1 with a host transcription factor has been implicated: Sp1 and possibly others for the p6 promoter,^20,22 Sp1 for the host p21 promoter,²⁶ AP-1 for the TNFα promoter,³⁰ and NF-kB for the IL-6 promoter.²⁷ Direct interactions with NS1 have been shown only in the case of Sp1^22,26 and possibly AP-1.³⁰ In addition, in the cases of the host promoters, the exact DNA sequences necessary and sufficient for transactivation by NS1 have been mapped to those same transcription factor binding sites.^26,27,30 In the p6 promoter, the critical sequences were identified as those containing the NSBE, which also coincide with Sp1 binding sites.²⁰ Therefore, the model for NS1-mediated gene transactivation involves direct and/or indirect interaction of NS1 with the transcription factor binding sites of these promoters, as well as the possible interaction of NS1 with the host transcription factors Sp1, AP-1, and NF-κB. To investigate direct interactions of NS1 with the promoter sequences, the implicated DNA sequences were used in binding assays with NS1-nuc. Specific binding (i.e., stronger than that to random DNA sequences) was found between NS1-nuc and DNA sequences of the p6, p21, and TNFα promoters (see Figure 2 for sequences, Figure 4 for binding isotherms, and Tables 1 and 3 for binding constants). Because the implicated NS1 binding sites of the p6 promoter overlap with those of the viral origin of replication,^20,22 the binding study described above with DNA shown in Figure 2A–C relates also to binding at the p6 promoter. As discussed above, binding was found to be relatively weak [K_1/2 ~ 1 µM (Table 1)] but highly cooperative [Hill coefficients of 2–4 (Table 1)], with five to seven copies of NS1-nuc bound to the p6 promoter-derived DNA.

The binding affinity of NS1-nuc for the DNA containing p21 promoter-derived sequences was found to have a K_1/2 in the micromolar range [5.7 ± 0.8 µM (Table 3)] and also to be cooperative [Hill coefficients of 2.3 ± 0.2 (Table 3)]. Similarly, the K_1/2 and Hill coefficient for binding of NS1-nuc to the AP-1 site in the TNFα promoter were found to be 4 ± 3 µM and 2 ± 1, respectively (Table 3). In contrast, binding of NS1-nuc to the NF-κB site of the IL-6 promoter DNA was found to be noncooperative and much weaker, and in fact equivalent to that with a random DNA sequence [K_1/2 of 28 ± 18 µM (Table 3)]. The weak binding affinity for these three promoter element DNAs precluded stoichiometry measurements using the gel shift method. The observed weak binding suggests that interaction of NS1 with these DNA sequences may require the assistance of other proteins (host or viral), other parts of the NS1 protein (including those implicated in NS1 oligomerization), post-translational modifications,^52–58 modification of the DNA,⁸⁹ or other segments of the promoter sequences. Alternatively, NS1 may induce expression of these genes through some pathway that does not involve NS1 binding at their promoters.

Further investigation of the stoichiometry of binding of NS1-nuc to the p21 promoter element DNA was performed using AUC sedimentation velocity and resulted in the discovery of three copies of NS1-nuc per fluorescein-labeled p21 dsDNA (Figure 3C, Table 2, Tables S1 and S3, and Figures S15 and S16). An additional complex, although at a much lower concentration, contained approximately 10 copies of NS1-nuc (Table 2 and Tables S1 and S3). The observation of multiple copies of NS1-nuc binding to the p21 promoter element DNA is also consistent with the observed cooperativity in the DNA binding isotherms [Hill coefficient of >1 (Table 3)].

The finding that NS1-nuc protein is capable of direct DNA binding to sequences derived from the p6, and to a lesser extent p21 and TNFα promoters, indicates that NS1 does not necessarily require host transcription factors for promoter binding. It is even conceivable that NS1 could transactivate these promoters itself, as the C-terminal domain is implicated in transactivational activity in homologous proteins (athough the C-terminal domain sequences of these homologous NS1 proteins do not share significant sequence identity).^90,91 The current data do not, however, rule out the possibility that NS1 cooperates with transcription factors at these promoters at some level, either in recruitment to the promoter or in promoter transactivation. Further studies will be needed to investigate these potential interactions and activities.

The role of cooperative binding between NS1-nuc at the p6 and p21 promoter sequences is also interesting. This cooperativity could serve to enhance binding affinity, bring NS1 molecules together on the DNA, and/or have some role in altering the architecture of the DNA that could influence the binding or structure of other factors at the promoter and thereby influence replication or transcriptional events. It is also noteworthy that studies using an NS1 protein with a mutation that results in loss of its ATPase activity show that this activity is necessary for the gene transactivation activity of NS1 in some cases,²⁶ though not all.⁹² Similar results have been found with NS1 homologues.^49,91,93 Our studies utilize only the nuclease domain of NS1, which is missing the portion of the protein that contains the ATPase activity; however, the involvement of the ATPase activity of NS1 in transactivation does suggest some role in DNA and/or chromatin structural manipulation (or alternatively in a conformational change that may unmask a transactivational region of NS1). Such mechanisms have been suggested for related parvoviral replication proteins.^49,91,93 Future studies will be necessary to fully understand the mechanism of gene transactivation by B19V NS1.

ABBREVIATIONS

AAV

adeno-associated virus

AUC

analytical ultracentrifugation

BPB

bromophenol blue

DTT

dithiothreitol

EDTA

ethylenediaminetetraacetic acid

HEPES

4-(2-hydroxyethyl)-piperazine-1-ethanesulfonic acid

HEPES-NaOH

HEPES titrated to a desired pH with NaOH

KO

“knockout”, where a subset of the DNA sequence is substituted

MVM

minute virus of mice

Mw

molecular weight or molecular mass in daltons

NS1-nuc

human parvovirus B19 NS1 nuclease domain (residues 2–176)

NSBE

NS1 DNA binding element

nt

nucleotide(s)

OAc

acetate

PAGE

polyacrylamide gel electrophoresis

Rep

NS1 homologue from AAV, specifically Rep68 or Rep78

SDS

sodium dodecyl sulfate

Tris

tris-(hydroxymethyl)aminomethane

Tris-HCl

Tris titrated to a desired pH with HCl

TSS

transcription start site

v-bar

partial specific volume in milliliters per gram

XCFF

xylene cyanol FF