Crystal structure of the modification-dependent SRA-HNH endonuclease TagI

Abstract

TagI belongs to the recently characterized SRA-HNH family of modification-dependent restriction endonucleases (REases) that also includes ScoA3IV (Sco5333) and TbiR51I (Tbis1). Here, we present a crystal structure of dimeric TagI, which exhibits a DNA binding site formed jointly by the nuclease domains, and separate binding sites for modified DNA bases in the two protomers. The nuclease domains have characteristic features of HNH/ββα-Me REases, and catalyze nicks or double strand breaks, with preference for /RY and RYN/RY sites, respectively. The SRA domains have the canonical fold. Their pockets for the flipped bases are spacious enough to accommodate 5-methylcytosine (5mC) or 5-hydroxymethylcytosine (5hmC), but not glucosyl-5-hydroxymethylcytosine (g5hmC). Such preference is in agreement with the biochemical determination of the TagI modification dependence and the results of phage restriction assays. The ability of TagI to digest plasmids methylated by Dcm (C5mCWGG), M.Fnu4HI (G5mCNGC) or M.HpyCH4IV (A5mCGT) suggests that the SRA domains of the enzyme are tolerant to different sequence contexts of the modified base.

Figure 1.

TagI activity assays on (A) ^5hmC and ^5mC containing PCR products and (B,C) modified phage DNA in vitro and in vivo. (A) One μg of mixed PCR DNA (∼12 nM) made from modified dNTP mixtures was digested at 37°C for 1 h with 1 μg TagI (∼0.3 μM) in 2-fold serial dilutions in NEB buffer 2.1. The substrates for TagI digestion in 10 mM Mg²⁺ contained C (3 kb) and ^5hmC or ^5mC (2.2 kb). The assay in 1 mM Mn²⁺ was performed on C (3 kb), ^5hmC (2.2 kb) and ^5mC (1.2 kb) containing PCR DNA. The amount of TagI (μg) shown on top of each lane corresponds to 295, 147, 74, 37, and 18 nM of protein dimer, respectively. (B) Modified DNA from phage T4gt (^5hmC, ∼0.2 nM), T4 (^g5hmC, ∼0.2 nM) or XP12 (^5mC, 0.5 nM) was digested by TagI (∼0.3 μM) and control enzymes: tolerant to the presence of modified cytosines (MluCI (/AATT), 10 U), inhibited by cytosine modifications (HpaII (C/CGG), 10 U) and affected only by the presence of ^g5hmC (MspJI, 5U). (C) Late-log phase host cells were plated on soft agar to form a cell lawn, and diluted phages (Lambda, T4gt or T4) were spotted onto the cell lawns. Cell lysis and plaque formation indicated susceptibility to phage infection. No plaque formation indicated the restriction of T4gt phage by TagI expressing cells.

Figure 2.

Sequence logo for TagI ds cleavage (left) or nicking (right) activity. 16 double strand cleavage sites (A) and 31 nicking sites (B) were combined from pBR322, pBRFM⁺, and ^5hmC PCR DNA cleavage performed at 37°C. The arrow denotes the site of DNA cleavage or nicking. Note a systematic bias in the determination of the sequence logos. An A base immediately downstream of the cleavage site is detected with lower efficiency (because the polymerase incorporates the correct base, albeit in a template-independent manner).

Figure 3.

Experimentally determined structure of TagI in the absence of DNA (A) and a model of TagI with separate DNA fragments bound to the SRA and HNH domains (B). The TagI protomer in the asymmetric unit is shown in yellow (SRA domain) and green (HNH domain), the crystallographic symmetry mate that completes the TagI dimer is shown in light gray. The fragment of the structure that is disordered in the crystal is indicated by a dashed line. The unit cell and the directions of the crystallographic axes are shown in gray. The DNA molecules that have been modelled in complex with the SRA domains of the dimer are shown in gold and light gold color, and the single modelled DNA molecule that is bound to the HNH dimer is shown in dark grey color.

Figure 4.

Sequence alignment (A) and structure of TagI (B), UHRF1 (40) (C) and MspJI (31) (D) SRA domains. The models are in ribbon representation. Key loops and selected functionally important residues are highlighted. The DNA in panel B is not present in the crystals and was modeled based on the UHRF1-DNA complex shown in panel C.

Figure 5.

Specificity of the flipped base binding pockets in the SRA domains of TagI (A), UHRF1 (40) (B), UHRF2 (64) (C) and PvuRts1I (29) (D). Upper panels show the key residues forming the flipped base binding pockets. Lower panels depict the surface of the pockets. The TagI SRA domain surface was colored according to the sequence conservation calculated by the ConSurf server (57). The position of the flipped nucleotide in the TagI and PvuRts1I pockets is not based on experimental structure, but inferred from the binding mode of ^5hmC to human UHRF2. The ^g5hmC residue was modelled based on the NMR structure of β-d-glucosylated DNA (65). The binding modes of the ^5mC in UHRF1 and ^5hmC in UHRF2 are based on crystal structures.

Figure 6.

HNH domain of TagI (A) and of the Hpy99I REase with bound DNA (52) (B), and sequence alignment of selected HNH nucleases (C). The ββα-Me motif that is characteristic for the HNH nucleases is highlighted by more intense coloring. Selected active site residues and the ligands of the structural Zn²⁺ ion are indicated. In panel (B), only the dinucleotide around the scissile phosphodiester bond is shown.

Figure 7.

Efficacy of DNA TagI cleavage of substrates containing none to four ^5mC bases. 20 ng (∼32 nM) of annealed 48-mer oligonucleotides were digested by TagI (0.125 μg, ∼92 nM of TagI dimer) in NEB buffer 2.1 for 30 min. Cleavage products were resolved in 15% urea-PAGE, stained by SYBR Gold and imaged on Typhoon imager. The predicted TagI cleavage site GCS/GC at the center of the duplexes conforms to the GC/NGC consensus for cleavage by Fnu4HI (20 U), which was used as a positive control. ^5mCs in oligoduplexes are represented by small black dots in the diagrams above each lane.