Structural Basis for the Propagation of Homing Endonuclease-Associated Inteins

Abstract

Inteins catalyze their removal from a host protein through protein splicing. Inteins that contain an additional site-specific endonuclease domain display genetic mobility via a process termed "homing" and thereby act as selfish DNA elements. We elucidated the crystal structures of two archaeal inteins associated with an active or inactive homing endonuclease domain. This analysis illustrated structural diversity in the accessory domains (ACDs) associated with the homing endonuclease domain. To augment homing endonucleases with highly specific DNA cleaving activity using the intein scaffold, we engineered the ACDs and characterized their homing site recognition. Protein engineering of the ACDs in the inteins illuminated a possible strategy for how inteins could avoid their extinction but spread via the acquisition of a diverse accessory domain.

Keywords: DNA recognition; homing endonuclease; horizontal gene transfer; intein; intein structures; meganuclease; protein splicing.

FIGURE 1

Schematic mechanisms of homing endonuclease (HEN) and Hedgehog/INTein (HINT) domains in inteins. (A) The HINT domain catalyzes self-excision of the intein (here consisting of HINT, HEN, and an accessory domain (ACD)) while covalently ligating the flanking extein sequences of the host protein during the protein cis-splicing reaction. (B) The nested HEN domain of the intein promotes gene conversion by introducing DNA double-strand breaks at the homing site into a vacant host allele followed by invasion via horizontal gene transfer (HGT) and fixation into the organism or population. Saturation of occupied alleles may cause HEN degeneration and loss.

FIGURE 2

Structures of degenerated and active VMA inteins. (A) General domain organization and conservation in inteins. The HEN domain resides within the intein while the intein resides within a host protein (N- and C-exteins). Conserved sequence Blocks A–H are indicated. Host protein, black; intein, gray; HEN, yellow. (B) Sequence comparison around Blocks C and E corresponding to the active site-carrying LAGLIDADG helices of the HEN domains. Comparison of the intein orthologs of Pyrococcus horikoshii (PhoVMA), Thermococcus litoralis (TliVMA), Pyrococcus furiosus (PfuVMA), and Pyrococcus abyssi (PabVMA). The position of the catalytic aspartates in Blocks C and E are highlighted in red. (C) Crystal structure of PhoVMA intein. (D) Crystal structure of TliVMA intein. For (C,D), the locations of the active sites are highlighted by red circles. The close-ups of the active sites are shown to the left with electron density maps at 1.0 σ contour level. (E) The previously reported three crystal structures of PI-PfuI (PDB: 1dq3) (Ichiyanagi et al., 2000)(Ichiyanagi et al., 2000), PI-TkoII (PDB: 2cw7) (Matsumura et al., 2006), and PI-SceI (PDB: 1lws) (Moure et al., 2002). In (C–E), HINT, HEN, and ACD domains are colored in gray, yellow, and blue, respectively. PI-TkoII contains an additional domain IV indicated in orange. Int_N and Int_C indicate the N- and C-terminal parts of the HINT domain, which are separated by the HEN. The domain arrangement is schematically illustrated below each structure. N and C denote the termini.

FIGURE 3

DNA-binding and cleavages of the theoretical homing sites by TliVMA and PhoVMA intein variants. (A) PCR-based construction of linear DNA substrates from Tli and Pho genomic DNA and expected cleavage pattern. The homing site was reconstituted by deleting the intein coding sequence from the vma gene while adjusting the size to 750 bp asymmetrically harboring the homing sites to generate the expected 250- and 500-bp cleavage products. (B) DNA-binding and cleavage of the 750-bp Tli DNA substrate by incubation with increasing concentrations of TliVMA intein at 80°C for 2 h. The electrostatic surface potential with an isoelectric point of 10.23 is shown below with a view of the DNA-binding interface. Positive, blue; neutral, white; negative, red. (C) Experiment as in (B) but using the 750-bp Pho DNA substrate and PhoVMA intein. The isoelectric point of the electrostatic surface potential is 5.45. The electrostatic surface potential model was generated using an alternative coordinate file without gaps in the HEN domain. (D) Activity test of the reactivated PhoVMA_Act intein. Experiment as in (C) but using PhoVMA_Act intein. The reconstitution of active site regions in the PhoVMA intein by grafting the sequence from Block C and E regions of the TliVMA intein is illustrated below. In panels (B–D), S, substrate; P1, 500-bp product, P2, 250-bp product, P2′, ∼200-bp product, P3, ∼50-bp product. M stands for the DNA ladder, “stop” indicates the addition of SDS-containing stop solution after incubation. (E) Distance of homing site (HS) and alternative site (AS) in the T. litoralis DNA substrate. Arrowheads indicate the positions where strand cleavage occurs. (F) The alignment of TliVMA intein homing (HS) and alternative (AS, reversed) sites. A region of 27 bp with high sequence similarity between the HS and the reversed AS is indicated by a box. The central four base pairs where cleavage occurs are highlighted in red (HS) and blue (AS).

FIGURE 4

Deletion and grafting of the ACD in the TliVMA intein. (A) The crystal structures of TliVMA intein (left) and TliVMA^ΔACD intein (right) without ACD, connecting the HEN domain and the C-terminal part of the HINT domain. HEN, ACD, and HINT are colored in yellow, blue, and gray, respectively. (B) DNA-binding and cleavages of the reconstituted 750-bp DNA fragment with the homing site from Thermococcus litoralis genome (Tli) by TliVMA^ΔACD intein. The cleavages were analyzed on agarose gels after the incubation with increasing concentrations at 80°C for 2 h. (C) Comparison of DNA cleavages of the reconstituted 750-bp Tli DNA fragment by TliVMA, TliVMA^ΔACD, and PhoVMA_Act inteins. (D) Comparison of the DNA cleavages of the 2037 bp DNA fragment, including the TliVMA intein coding sequence at the homing site by the three inteins as in C. This fragment lacked the reconstituted homing site (HS) due to the TliVMA intein coding sequence, hence only possessed the alternative site (AS). (E) DNA cleavages of the 750-bp Pho DNA fragment by the reactivated PhoVMA_Act intein with the ACD from TliVMA intein (PhoVMA_Act-ACD(Tli)) and the comparison with TliVMA and PhoVMA_Act inteins. In (B–E), M stands for the DNA ladder, “stop” indicates the addition of SDS-containing stop solution after incubation. The migration height for the 750-bp substrate (S) and the 500 bp (P1) and 250 bp (P2) products are indicated in (B), P2’ (200 bp), and P3 (50 bp) are shown in (C,D). (F) The Sanger sequencing chromatogram of the 250-bp cleavage product generated by the TliVMA^ΔACD intein lacking the ACD.

FIGURE 5

Effect of swapping ACDs in the VMA inteins on DNA recognition. (A) Primary structure comparison of ACDs in Tli and Pho VMA inteins. Regions with high similarity are highlighted with red rectangles. (B) The ACD structures of Tli and Pho VMA inteins and a comparison with the bacteriophage 434 repressor (434R, PDB: 2or1) (Aggarwal et al., 1988) and the ACDs of PI-TkoII (PDB: 2cw7), PI-PfuI (PDB: 1dq3), and PI-SceI (PDB: 1lws). The regions highlighted with red rectangles in (B) are colored in red. (C) Schematic illustration of the swapping experiments using different ACDs. The ACD in the TliVMA intein was replaced with the respective ACDs from the VMA intein of P. furiosus (Pfu), P. abyssi (Pab), or the 434-bacteriophage repressor domain. (D) DNA cleavages of λ-phage DNA using Tli and Pho VMA intein variants carrying different ACDs. Cleavage of λ-phage DNA was tested by overnight incubation with the indicated intein variants at 80°C. Agarose gel analysis of the digestion reactions. Lane 1, λ-phage DNA (48k bp) without intein; lane 2, the wild-type TliVMA intein; lane 3, TliVMA intein with deletion of ACD (TliVMA^ΔACD); lane 4, TliVMA intein with phage 434 repressor as ACD (TliVMA₄₃₄); lane 5, reactivated PhoVMA_Act intein; lane 6, the reactivated PhoVMA intein with ACD from TliVMA intein (PhoVMA_Act-ACD(Tli)); lane 7, TliVMA with ACD from PfuVMA (TliVMA_ACD(Pfu)); lane 8, TliVMA intein with ACD from PabVMA (TliVMA_ACD(Pab)). M stands for the DNA size ladder.

FIGURE 6

(A) The intein homing cycle model. The intein homing cycle model starts with the (1) Invasion of a HEN-containing intein via horizontal gene transfer followed by (2) Fixation into vacant alleles within the population. Depletion of HEN homing sites causes (3) Degeneration of the HEN due to accumulation of mutations tolerated by the lack of selection. Degenerated HENs are prone to (4) Deletion, rendering the intein incapable of competing with intein-free alleles, which might cause (5) Intein-loss upon interbreeding with strains providing an intein-free allele. (B) Intein spread model. Inteins might obtain an ACD modulating the HEN specificity, e.g., via changes in the ACD which can lower the HEN specificity to find novel insertion sites. Thus, the acquisition of a diverse ACD provides a spreading mechanism to prevent degeneration and extinction.