Analysis of the costructure of the simian virus 40 T-antigen origin binding domain with site I reveals a correlation between GAGGC spacing and spiral assembly

Abstract

Polyomavirus origins of replication contain multiple occurrences of G(A/G)GGC, the high-affinity binding element for the viral initiator T-antigen (T-ag). The site I regulatory region of simian virus 40, involved in the repression of transcription and the enhancement of DNA replication initiation, contains two GAGGC sequences arranged head to tail and separated by a 7-bp AT-rich sequence. We have solved a 3.2-Å costructure of the SV40 origin-binding domain (OBD) bound to site I. We have also established that T-ag assembly on site I is limited to the formation of a single hexamer. These observations have enabled an analysis of the role(s) of the OBDs bound to the site I pentanucleotides in hexamer formation. Of interest, they reveal a correlation between the OBDs bound to site I and a pair of OBD subunits in the previously described hexameric spiral structure. Based on these findings, we propose that spiral assembly is promoted by pentanucleotide pairs arranged in a head-to-tail manner. Finally, the possibility that spiral assembly by OBD subunits accounts for the heterogeneous distribution of pentanucleotides found in the origins of replication of polyomaviruses is discussed.

Fig 1

(A) DNA sequence of the core origin of replication and the site I regulatory region. Site II is indicated by a cyan box, and site I is indicated by a yellow box. Black arrows indicate both the positions and the orientations of the GAGGC pentameric sequences. The locations of the AT-rich and EP regions are also indicated. (B) Site I-containing oligonucleotide used to determine the OBD-site I costructure. Note that the oligonucleotide has a 1-bp overhang; this is commonly used to promote formation of a pseudocontinuous helix within the crystal. (C) Twenty-eight-base-pair site I-containing oligonucleotide used in the ITC studies. DNA substrates used in the EMSAs included the 59-bp site I-containing oligonucleotide (D) and the 64-bp core origin oligonucleotide (E).

Fig 2

Costructure of the T-ag OBD bound to site I. (A, top) Ribbon diagrams of the two T-ag OBDs (yellow and cyan) bound to the site I oligonucleotide (shown as a surface representation). Pink arrows indicate the orientations of the GAGGC sequences; P5 and P6 are labeled. The GAGGC sequences are pink, and their complements are light green. The DNA-binding A1 (aa 147 to 159) and B2 (aa 203 to 207) motifs are shown in red. The N and C termini of the OBDs are labeled when visible. The B3 residues (aa 213 to 220) of the OBD bound to P5 are orange and shown as sticks. The C-terminal residues of the OBD bound to P6 are light blue and shown as sticks. (Bottom) Same as in panel A, but the view is rotated 90 degrees. Mol A and Mol B, molecules A and B, respectively. (B) Closeup of the protein-DNA interactions in which residues in the A1 and B2 motifs are selectively depicted. Residues in A1 making site-specific contacts with the GAGGCs include N153, R154, and T155. Residues in B2 making site-specific contacts include H203 and R204.

Fig 3

Set of interactions between the T-ag OBD and site I and related biophysical data. (A) Schematic of the protein-DNA interactions. Dashed red lines indicate interactions with the phosphate backbone; solid blue lines indicate base-specific interactions. Residues from the A1 motif extend between residues 147 and 159, while those from the B2 motif extend between residues 203 and 207. (B) Analysis of DNA structure within the complex. The minor groove width is shown on the y axis; the DNA sequence is presented on the x axis. (C) ITC-based measurement of the thermodynamic parameters for the interaction of the T-ag OBD with site I. The 28-bp duplex DNA used in these experiments, used at a concentration of ∼1.5 μM, is presented in Fig. 1C. Titration of the T-ag OBD into the site I-containing oligonucleotide took place at 25°C in reaction buffer (Materials and Methods). The protein concentration in the syringe was ∼37.5 μM. The actual calorimetric trace is shown in the top panel. The y axis of the isotherm is power in μcal/s; the x axis is time in minutes. The stoichiometry and association constants were determined from curve fitting the integrated calorimetric trace presented in the bottom panel.

Fig 4

Full-length T-ag forms single hexamers on site I-containing DNA. (A) EMSA of full-length SV40 T-ag bound to a 59-bp oligonucleotide containing site I (lanes 2 and 3) or the 64-bp core origin oligonucleotide (lanes 5 and 6). The presence (+) or absence (−) of T-ag is indicated at the bottom. Lanes 2 and 5, 1.5 pmol T-ag; lanes 3 and 6, 3 pmol T-ag. The reaction products include single hexamers (SH) and double hexamers (DH). The DNA that did not enter the gel is labeled “well,” and the unbound DNA substrates are labeled “input.” (B) In situ footprinting of full-length T-ag and the OBD, when complexed to site I. The footprints were obtained using the gel retardation 1,10-phenanthroline–copper ion footprinting technique (57). The initial EMSAs were conducted with the previously described 47-bp site I + wt 30 oligonucleotide (50). Free DNA (i.e., DNA obtained from reactions conducted in the absence of protein and used as a control) is presented in lane 1. The locations of sequence features, including P5 and P6, are indicated. (C) Structure-based modeling of hexamers of the T-ag helicase and OBD domains on site I. The OBDs initially bound to P5 and P6 are shown as yellow and cyan, respectively. The OBDs are represented by single spheres, which are centered at the geometric center of mass, and the radius is approximately the radius of gyration of the domain (i.e., ∼17.5 Å). The helicase domains are represented by 2 spheres; those helicase domains connected to the initially bound OBDs are also shown as yellow and cyan. Dotted black lines represent the flexible linkers connecting the N terminus of the helicase domains to the C terminus of the OBDs. Finally, an idealized three-dimensional model of site I DNA, positioned along the 6-fold screw axis of the OBD spiral, is shown as a ribbon representation. (D) Molecular modeling studies indicating that two independent T-ag hexamers cannot form on P5 and P6. Two models of T-ag hexamers were constructed; one nucleated at P5 and one at P6. In the resulting model, significant collisions occur between the helicase domains (the collisions are shown in green).

Fig 5

Spatial correlations between the T-ag OBDs site-specifically bound to pairs of GAGGC pentanucleotides, with OBD pairs within the left-handed spiral. (Top) Two views (rotated by ∼90°) of the OBDs bound to site I are presented as translucent surface representations; the DNA is shown as a ribbon diagram. Values describing the relationships between the OBDs (i.e., the translation and angular rotation) are indicated. (Bottom) Two views of the left-handed spiral structure of the T-ag OBD rotated by ∼90° (the subunits are labeled a to f). Spiral subunit a is yellow, while subunit f is cyan; they are offset by the same angle that separates the OBDs bound to P5 and P6 (i.e., spiral subunits a and f are also 60 degrees apart).

Fig 6

Simplified renderings of the OBD interactions with pairs of pentanucleotides. (A, top) Two views of the OBD interactions with site I. The OBDs are represented by spheres centered at the geometric center of mass. The T-ag OBDs are yellow on P5 and cyan on P6. The smaller red spheres represent the DNA-binding A1 and B2 motifs. (Bottom) Rendering of the left-handed spiral in which the OBDs proximal to the gap are represented by spheres colored as described above. This depiction serves to further illustrate that spiral subunits a and f have the same spatial relationship as the OBDs bound to P5 and P6. As in the top images, the red spheres indicate the relative locations of the A1 and B2 motifs. (B) Modeling studies indicate that the OBDs initially bound to P5 (yellow) and P6 (cyan) undergo a significant transition during spiral formation. This involves rearrangement of the A1 loop from the bound to the free form, subsequent rotation and translation away from the major groove of the dsDNA, and interaction(s) with other OBDs (reviewed in reference 13). (C, top) Two views of the OBD interactions with site II, rotated by ∼90°, used to illustrate the generality of the relationship (24, 26) (RCSB PDB ID 2ITL). P1 and P2 are separated by a 1-bp spacer; therefore, the OBDs bound to P1 (yellow) and P2 (cyan) are ∼156 degrees apart and separated by a translation of ∼19 Å. (As in previous examples, the DNA-binding A1 and B2 motifs are red and magenta, respectively). (Bottom left) Side view of the left-handed OBD spiral hexamer with DNA modeled within the central channel (the 6 subunits are labeled a to f). It is apparent that in this instance, spiral subunits a and d (yellow and cyan, respectively) have a spatial relationship that is analogous to that of the OBD subunits bound to P1 and P2. (Bottom right) Helicase-proximal view of the left-handed spiral structure with duplex DNA modeled in the central channel.

Fig 7

Model depicting hexamer-dependent OBD spiral formation on site I. (A) Initially, a monomer of T-ag, shown in yellow, binds to site I (the J-domain is not shown). The flexible linker that connects the two domains is shown as a dashed line. (B) Intermediates in hexamer formation. (Left) Following the binding of the initial monomer, additional T-ag molecules are recruited to site I owing to the high affinity of the helicase domains for each other. Upon recruitment, the helicase domains assemble into a hexamer (18). The five newly recruited monomers are shown in gray; the OBDs are symbolized by the small gray spheres. The black line represents one strand of DNA going over the outer surface of the helicase domain; the second strand is routed through the central channel. (Right) An alternative possibility is that a second monomer of T-ag binds to the second pentanucleotide in site I. (C) Formation of an immature hexamer. In this intermediate, the helicase domains have formed a hexamer over the flanking region, and P5 and P6 are bound by two of the OBDs. The four unbound OBDs are shown as gray spheres. The routing of the DNA is as described above. (D) Maturation of the hexamer to form the spiral structure. It is proposed that the four free OBDs shown in panel C interact with the OBDs bound to P5 and P6. As a result of this interaction, the OBDs adopt the left-handed spiral structure. (A left-handed helix is defined as turning in a counterclockwise fashion as it progresses forward. The N termini of the OBDs in the spiral are indicated.) Once formed, the spiral has a gap though which ssDNA may pass. Finally, it is noted that the isolated T-ag OBDs are monomeric in solution even at high concentrations (74). This is one indication that the protein-protein interactions that occur in the OBD spiral are not strong; therefore, the spiral structure is presumably dynamic in nature.

Fig 8

OBD oligomerization in the context of T-ag double hexamers formed on the core origin. (A) The PDB coordinates for the helicase domain (1SVM) and the OBD (2FUF) were fit into the EM density map (EMD-1681 [36]). The figure was generated using the program CHIMERA (103); the helicase domains were fit using the CHIMERA fit-in-map option. A double-hexamer spiral of the OBDs, shown as a ribbon representation, was generated and manually positioned into the central density. (B) Model of a T-ag OBD double spiral. The individual spirals are shown in yellow and cyan. The smaller arrow illustrates that its diameter (∼66 Å) is similar to the area protected in previous footprinting studies (see, e.g., reference 38).

Fig 9

T-ag hexamers cannot exist simultaneously on site I and the proximal side of site II. As reported herein, a T-ag hexamer nucleated through the OBDs bound at P5 and P6 positions the helicase domains over the EP region. Furthermore, when T-ag assembles into a double hexamer on site II, the helicase domains span both the AT-rich and EP regions (19). Therefore, if both site I and the EP-proximal side of site II were occupied by T-ag hexamers, the helicase domains would completely overlap and suffer steric clashes.