Holliday junction resolution by At-HIGLE: an SLX1 lineage endonuclease from Arabidopsis thaliana with a novel in-built regulatory mechanism

Abstract Holliday junction is the key homologous recombination intermediate, resolved by structure-selective endonucleases (SSEs). SLX1 is the most promiscuous SSE of the GIY-YIG nuclease superfamily. In fungi and animals, SLX1 nuclease activity relies on a non-enzymatic partner, SLX4, but no SLX1-SLX4 like complex has ever been characterized in plants. Plants exhibit specialized DNA repair and recombination machinery. Based on sequence similarity with the GIY-YIG nuclease domain of SLX1 proteins from fungi and animals, At-HIGLE was identified to be a possible SLX1 like nuclease from plants. Here, we elucidated the crystal structure of the At-HIGLE nuclease domain from Arabidopsis thaliana, establishing it as a member of the SLX1-lineage of the GIY-YIG superfamily with structural changes in DNA interacting regions. We show that At-HIGLE can process branched-DNA molecules without an SLX4 like protein. Unlike fungal SLX1, At-HIGLE exists as a catalytically active homodimer capable of generating two coordinated nicks during HJ resolution. Truncating the extended C-terminal region of At-HIGLE increases its catalytic activity, changes the nicking pattern, and monomerizes At-HIGLE. Overall, we elucidated the first structure of a plant SLX1-lineage protein, showed its HJ resolving activity independent of any regulatory protein, and identified an in-built novel regulatory mechanism engaging its C-terminal region.


INTRODUCTION
Homologous recombination is one of the fundamental cellular processes in generating genetic diversity and repairing damaged DNA (1). One key intermediate during homologous recombination is a four-way joint DNA molecule: Holliday junction (HJ) (2)(3)(4). HJ is one of the many types of joint (branched) DNA molecules appearing during DNA metabolism (Supplementary Figure S1). An inability to process HJs results in genomic instability (5,6). HJs are processed either by dissolution or resolution (7,8). Dissolution involves the formation of a hemicatenated intermediate followed by a dissolvasome catalyzed decatenation resulting in non-crossovers. A dissolvasome comprises helicase, topoisomerase, and structural proteins (9). In contrast, HJ resolution involves structure-selective endonucleases (SSEs) that hydrolyze phosphodiester bonds near the crossover points of HJ, resulting in crossover and non-crossover products. While dissolution is a safeguarding mechanism to circumvent the loss of heterozygosity (3,10,11), resolution plays an essential role in generating genetic diversity (12).
Compared to animal and fungal resolvases, the biochemical and structural information on plant resolvases is only fragmentary (43). Although the DNA repair and recombination mechanisms are conserved among animals and plants, there are significant differences that may be unique to the plant kingdom, but have not been fully explored. Several plant resolvases have been identified based on sequence similarities with resolvases from fungi and animals. For example, plants have two homologs of GEN1: GEN1 and SEND1 (44)(45)(46). Both GEN1 and SEND1 resolve HJs by two symmetrical nicks. The MUS81-EME1 complex from Arabidopsis thaliana (At-MUS81-EME1), similar to human MUS81-EME1, prefers nicked HJ (47)(48)(49)(50), imposing a need of having another resolvase that can generate the initial nick, a function carried out by the SLX1-SLX4 complex in humans. Interestingly, neither SLX1 nor SLX4 like protein has been reported from plants. A nuclease belonging to the GIY-YIG superfamily has been identified from A. thaliana (At-HIGLE: HYL-1 Interacting GIY-YIG Like Endonuclease) (AT2G30350) (51). A BLAST (Basic Local Alignment Search Tool) search (52) identifies At-HIGLE as an excinuclease ABC (C subunit), providing clues regarding the possible function of At-HIGLE similar to fungal and mammalian SLX1. We began work on the premise that At-HIGLE may be a good candidate for plant SLX1.
In the present work, we demonstrate the HJ resolution potential of At-HIGLE, which is analogous to SLX1. We also describe a crystal structure of the nuclease domain of At-HIGLE, which in turn allowed us to reaffirm At-HIGLE as a member of the SLX1 subfamily of GIY-YIG endonucleases. Our data provide insights into the At-HIGLE substrate specificity, mode of substrate binding, as well as catalysis, advancing research in plant DNA repair and recombination. Furthermore, we find a novel regulatory mechanism involving the extended C-terminal region of At-HIGLE in delimiting its catalytic activity.

Cloning, protein expression and purification
The gene coding for At-HIGLE from Arabidopsis thaliana (1119 bp) (At2g30350) was codon-optimized (BIOMATIK, Canada) for expression in E. coli. The vector was subcloned into pET28a vector (Novagen) with N-terminal 6x His followed by a SUMO Protease cleavage site and transformed into BL21 (DE3) RIL (Novagen). At-HIGLE  and At-HIGLE 184-368 and point substitutions were introduced by Quick change Site-directed mutagenesis (Agilent) (Supplementary Table S1). For protein expression, cells were grown in LB medium at 37 • C, induced with 0.4 mM isopropyl beta-D-1-thiogalactopyranoside (IPTG) at OD 600 = 0.6-0.8, and grown overnight at 16 • C. Cells were harvested by centrifugation at 5000 rpm for 20 min at 4 • C.

Crystallization and structure elucidation
At-HIGLE 1-183 crystals were obtained by sitting drop vapor diffusion method at a final concentration of 9 mg ml -1 at 20 • C in 4.0 M Sodium formate. Crystals were cryoprotected with 30% glycerol (v/v) before data collection. X-ray diffraction data were collected at beamline ID-23-2 at European Synchrotron Radiation Facility (ESRF) at a wavelength of 0.8731Å. The crystals belonged to P2 1 2 1 2 1 space group and diffracted to a maximum resolution of 1.7Å. Diffraction data was processed and scaled using XDS (53). Phases were determined using molecular replacement in the Phase-MR module of Phenix (54). Cg-SLX1 was used as a starting model for molecular replacement (PDB: 4XM5) (40). There is one molecule per asymmetric unit. The model was refined iteratively using COOT and PHENIX to an R work /R free of 16.37% and 19.38% at 1.7Å with 10% reflections for R free calculation (55,56). In the final model, 98.0% amino acid residues resided in the allowed region of the Ramachandran plot. The structure validation was carried out using MolProbity (57). Structure analysis was performed using PyMOL (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC.). Structure-based sequence alignments were done using UCSF Chimera (58). All other sequence alignments were done using CLUSTAL W (59). The diffraction and refinement statistics are summarized in Table 1. The structure was deposited in PDB with accession code 7WME.

Nuclease assays
The synthetic DNA substrates for nuclease assays were prepared by annealing DNA oligonucleotides synthesized by Eurofins, India. The oligonucleotides sequences and various oligonucleotides used for annealing substrates are presented in Supplementary Mapping of the cleavage sites by At-HIGLE and At-HIGLE 1-183 on different DNA substrates was carried out by resolving the products (after 90 min reaction) on 12% TBE-Urea polyacrylamide gel at a constant voltage of 100 V (15 V cm -1 ) for 2 h. The samples were prepared by heating the reaction mixture for 10 min at 95 • C in formamide containing Orange G dye. The cleavages in 6-FAM and Cy5 labeled DNA strands were monitored independently by scanning the gels for 6-FAM and Cy5 signals using Typhoon scanner. The incision sites were mapped by using 6-FAM and Cy5 labeled DNA marker oligonucleotides of different lengths. The markers used for mapping the cleavage sites were chemically synthesized followed by HPLC and gel purification. Sequence of the oligonucleotides used as the marker is derived from the DNA oligos 6-FAM labeled X0-1 (X0-1 f ) and Cy5 labeled X0-4 (X0-4 c ) (Supplementary Table S2C).

Cruciform assays
Plasmid pIRbke8 mut (23,60) was transformed into TOP10 competent cells, and colonies were selected on LB plates containing 100 g ml -1 ampicillin. Plasmid DNA was isolated using Qiagen Midiprep Kit. For each experiment, plasmid DNA was diluted to 10 nM with water. Reaction was performed with 0.5 nM plasmid in a buffer containing 50 mM Tris (pH 8.5), 2.5 mM MgCl 2 , 1.0 mM DTT, 100 mM NaCl and 0.1 mg ml -1 BSA and incubated at 37 • C for 30 min to induce cruciform extrusion. Cleavage reaction was then initiated by adding 100 nM protein. Aliquots were taken out at different time points, and the reaction was quenched using 5.0 mM EDTA followed by 2 mg ml -1 proteinase K (NEB), 0.2% SDS treatment of the samples. Products were analyzed on 0.8% agarose gel stained with SYBR Gold (Invitrogen™) for 2 h and visualized under Gel Doc XR+ System (Bio-Rad).

Fluorescence anisotropy
The binding of At-HIGLE and its variants with various branched DNA substrates were studied using fluorescence anisotropy. DNA substrates were labeled with 6-FAM on F0-1. (Supplementary Table S3). Substrates were used at a concentration of 25 nM (15 nM labeled and 10 nM unlabeled), and protein concentration ranged from 0 to 600 nM. Reaction was set up in a buffer containing 20 mM HEPES (pH 7.5), 2.5 mM CaCl 2 , 100 mM NaCl, 0.5 mM DTT, 2 mM EDTA and 0.1 mg ml -1 BSA at 25 • C. Binding reactions were set in Corning 96 flat bottom black polystyrene plates. Anisotropy was measured using a POLARstar Omega microplate reader at an excitation wavelength of 485 nm and an emission wavelength of 520 nm. Fluorescence anisotropy was calculated as (I || -I ⊥ )/(I || + 2I ⊥ ), where I || and I ⊥ are intensities in parallel and perpendicular directions. Binding was studied as the change in anisotropy (A -A 0 ) versus protein concentration, where A is the observed anisotropy, and A 0 is the anisotropy of substrate alone. All experiments were done in triplicate. Dissociation constant (K d ) was calculated Nucleic Acids Research, 2022, Vol. 50, No. 8 4633

Circular dichroism
Circular dichroism (CD) experiments were conducted on a Jasco J-815 spectropolarimeter with a Peltier-type temperature controller (Jasco CDF-426 S/15). The far-UV CD spectra of At-HIGLE and its mutants were measured in the wavelength range 260-195 nm using a quartz cuvette with 0.1 cm path length. The data points were recorded using scan speed 100 nm min -1 and spectral bandwidth 1.0 nm with 25 accumulations per sample. A protein concentration of 0.3 mg ml -1 in 20 mM Tris-HCl (pH 8.5), 100 mM NaCl, 5% glycerol and 5 mM ␤-mercaptoethanol was used for all the measurements. Buffer contribution was subtracted from all protein spectra.

GIY-YIG endonuclease superfamily: plants vs. other organisms
The hallmark of the GIY-YIG endonuclease superfamily is a signature GIY-YIG hairpin followed by conserved Arg and Glu residues. The GIY-YIG nucleases are often associated with additional domains ( Figure 1A). These accessory domains regulate the catalytic activity and govern the substrate specificity of the associated GIY-YIG nuclease domain. The GIY-YIG domain of UvrC generates a 3 incision in a damaged DNA with the help of a UvrB interacting region and tandem HhH motif at the C-terminal end (61,62). I-TevI homing endonuclease from bacteriophage T4 contains a sequence-specific DNA interacting Cterminal region (63). SLX1 from animals and fungi comprises an N-terminal GIY-YIG nuclease domain and a C-terminal RING domain. The RING domain of fungal SLX1 regulates nuclease activity by participating in SLX1 homodimerization and interacting with SLX4 (40) ( Figure  1A). SLX1 in complex with SLX4 is a key HJ resolving nuclease in fungi and animals. A plant protein similar in structure and function to fungal and animal SLX1 is unknown. A yeast two-hybrid screening led to the identification of a novel GIY-YIG containing endonuclease, At-HIGLE, from A. thaliana (51).
Structure-based sequence alignment showed that the nuclease domain of At-HIGLE has conserved residues in the framework region with the signature hairpin GIY-YIG motif modified to ACY-YIG. The hairpin is followed by conserved Arginine (Arg55) and metal ion chelating Glutamate (Glu95) residues ( Figure 1B). The C-terminal region of At-HIGLE does not show similarity with any other known protein fold. Although, the N-terminal domain of At-HIGLE aligns with a high sequence similarity with the nuclease domains of SLX1 proteins, there is no sequence similarity between the extended C-terminal region of At-HIGLE and the RING domain of SLX1 from animals and fungi (Supplementary Figure S2). The other characterized GIY-YIG containing plant proteins include a mismatch-repair protein (MSH1: Muts homolog 1) (64) and a glutaredoxin (GRXS16) (65) (Figure 1). Therefore, we decided to explore At-HIGLE structurally and biochemically to scrutinize its potential function in HJ resolution in order to confirm whether At-HIGLE is indeed the hitherto unreported SLX1 from plants.

Overall structure of At-HIGLE
Full-length At-HIGLE was subjected to extensive crystallization trials to unravel its structural organization. Eventually, At-HIGLE was truncated (At-HIGLE 1-183 ) based on secondary structure predictions to remove disordered segments at the C-terminal end (Supplementary Figure S3). At-HIGLE 1-183 crystallized in P2 1 2 1 2 1 space group with one molecule in the asymmetric unit. Crystals diffracted to a resolution of 1.7Å. Phases were determined using molecular replacement with SLX1 from Candida glabrata (Cg-SLX1) as the starting model (PDB: 4XM5) (40). The structure was refined to an R free of 19.38% (Table 1, Supplementary Figure S4). The structure exhibits a mixed ␣/␤ topology. At-HIGLE 1-183 comprises of a ␤ sheet formed by five ␤ strands oriented as ␤ 2 -␤ 1 -␤ 3 -␤ 4 -␤ 5 ( Figure 1C, Supplementary Figure S5) with ␤ 2 and ␤ 3 strands present in anti-parallel orientations with respect to ␤ 1 , ␤ 4 and ␤ 5 strands. While helices ␣ 1 , ␣ 2 and ␣ 7 are present on one side of the ␤ sheet, helices ␣ 3 -␣ 6 are present on the other side. Poor electron densities did not allow the modeling of amino acids 1-26 at the N-terminal end and 180-183 at the Cterminal end. Comparison of the At-HIGLE structure with other members of the GIY-YIG endonuclease superfamily clearly showed that At-HIGLE is the most similar to SLX1 and different from other members of the GIY-YIG family. At-HIGLE 1-183 superimposes with the nuclease domain of Cg-SLX1 with a root-mean square deviation (r.m.s.d.) of 0.8Å ( Figure 1D, Supplementary Figure S6). Furthermore, a DALI (66) search using the nuclease domain of At-HIGLE, identified SLX1 proteins (PDBs: 6SEI, 6SEH, 7CQ2, 7CQ3, 4XLG and 4XM5) to be the closest hits with a Z-score of above 15.0. The Z-score with UvrC protein in the DALI search (PDB: 1YD6) is 3.5. Therefore, the structure confirms At-HIGLE to be a member of the SLX1 family of GIY-YIG nuclease superfamily.
Compared to the nuclease domain of Cg-Slx1, At-HIGLE 1-183 showed differences in the conformation of loops relevant to DNA substrate binding. Loop connecting ␣ 1 and ␤ 3 is unmodelled (and thus presumably  Figure S7). Therefore, At-HIGLE has substrate interacting loops in conformations ready for DNA binding. The average Bfactors for the loop connecting ␣ 1 and ␤ 3 , and the loop connecting ␣ 3 and ␣ 5 are 21.62 and 18.32Å 2 , respectively, whereas the average B-factors for the structure is 18.57Å 2 indicating stable conformation of the loops. A third loop, connecting ␤1 and ␤2 ( Figure 1D), is longer in the case of At-HIGLE than Cg-SLX1, and because of its proximity to one of the DNA binding sites, it may exert influence on protein-DNA interactions. In summary, At-HIGLE exists in a conformation ready for interactions with DNA substrate.

Catalytic activity of At-HIGLE
Crystal structures of Hpy188I and R.Eco29kI restriction endonucleases provide comprehensive insights into the catalytic mechanism of GIY-YIG endonuclease superfamily (68,69). A structural comparison of At-HIGLE with Hpy188I restriction endonuclease provides mechanistic insights into the catalysis by At-HIGLE (Figure 2A). At-HIGLE has a catalytic site typical of GIY-YIG endonucleases. During hydrolysis of the phosphodiester bond, a water molecule or a hydroxide performs a nucleophilic attack from one side, and a metal ion interacts with the scissile phosphate from the other side to facilitate a single substitution reaction. The metal ion destabilizes the substrate and stabilizes the transition state. Catalytically essential amino acid residues are Glu95, Arg55, Tyr31 and Tyr45. Glu95 chelates a metal ion, whereas Arg55 and Tyr31 coordinate the attacking nucleophile. Arg55 and Tyr45 interact with the scissile phosphate ( Figure 2B). In the absence of a direct experimental evidence for the catalytic mechanism of GIY-YIG nucleases, possibility of phosphodiester bond hydrolysis through a two-step mechanism involving a phosphotyrosine intermediate cannot be ruled out (70).
A highly conserved catalytic site and a close structural similarity with SLX1 encouraged us to investigate the catalytic activity of At-HIGLE on fluorescently labeled synthetic Holliday Junctions ( Figure 2C and 2D, Supplementary Figures S8 and S9). The substrate was prepared by annealing four different strands. Of the four DNA strands, two were fluorescently labeled; one with 6-FAM at the 5 end and a second with Cy5 at the 3 end (Supplementary Table S2). The rationale of using two fluorescent labels is to monitor cleavage in different DNA strands of synthetic DNA substrates. At-HIGLE and At-HIGLE 1-183 were active on HJ substrate, which showed that the enzyme does not require any additional regulatory protein for activation. A variant with a substitution of metal-chelating Glu95 to Gln (At-HIGLE E95Q ) was catalytically inactive, which confirmed the observed activity was inherent to At-HIGLE and not to any potential contaminating nucleases (Figure 2 and Supplementary Figure S9). Therefore, At-HIGLE can cleave branched DNAs (HJs), but unlike fungal and animal SLX1, it does not require any accessory protein, SLX4, for its catalytic activity.
In our activity assays, we also tested the truncated variant (At-HIGLE 1-183 ) . Remarkably, the activity of the truncated variant was much higher than the full-length protein. Moreover, the full-length protein does not immediately proceed with the secondary reaction after the initial nick (i.e. nicked or gapped DNA is not processed further). Although, a possibility of secondary reaction with full-length At-HIGLE at later time-points cannot be ruled out. In contrast, At-HIGLE 1-183 continues to cleave nicked DNA (or gapped DNA), as evident from two sets of products on native TBE-PAGE ( Figure 2C). Therefore, the C-terminal domain of At-HIGLE reduces its catalytic activity. Interestingly, fluorescence anisotropy measurements showed that At-HIGLE 1-183 and full-length protein have comparable binding affinities for HJ substrate ( Figure 2E). Collectively, these results indicate that the C-terminal region regulates the activity of At-HIGLE.
Eventually, we analyzed the reactions from the 90 min time points from the activity assays ( Figure 2C) on a denaturing PAGE to map the sites of cleavages ( Figure 2F). No differences were observed in the cleavage pattern of Holliday Junctions by the full-length and truncated At-HIGLE. Both full-length and truncated At-HIGLE cleaved 6-FAM as well as Cy5 labeled DNA strands. The cleavage sites are located three-four nucleotides from the crossover point toward the 3 side ( Figure 2F), a characteristic common to SLX1 proteins from fungi and animals (26,27,(38)(39)(40)71,72).

At-HIGLE activity on multiple branched DNA molecules
The unique feature of SLX1 among GIY-YIG superfamily endonucleases, in general, is its promiscuous substrate specificity -the ability to cleave various branched DNA substrates. A close structural similarity of At-HIGLE with SLX1 prompted us to test the catalytic activity of At-HIGLE on other branched DNA substrates. Similar to SLX1, At-HIGLE could cleave a variety of branched DNA molecules: replication fork, 5 flap, 3 flap, and splayed arm substrates ( Figure 3A, Supplementary Figure S10). Consistent with the results with HJ (Figure 2), full-length At-HIGLE was less active than At-HIGLE 1-183 on all tested branched DNA substrates ( Figure 3B). The increased catalytic activity of the truncated variant was in line with its higher affinity for the tested branched DNA substrates (Figure 3C).
We next mapped the cleavage sites for each joint DNA molecules using the full-length At-HIGLE and At-HIGLE 1-183 . At-HIGLE generates nick only in the Cy5 labeled DNA strand and no cleavage in 6-FAM labeled  strand. The cleavage occurs three nucleotides from the junction point in the 3 direction in the cases of RF, 5 flap, and 3 flap. In the case of splayed arm substrate, cleavage occurs one nucleotide from the junction point in the 3 direction. Interestingly, At-HIGLE 1-183 can generate nicks in both 6-FAM and Cy5 labeled DNA strands. The nicks made by At-HIGLE 1-183 on 6-FAM labeled strands are also toward the 3 end from the junction point. In the cases of 3 flap and splayed arm substrates, nicks in 6-FAM labeled strand are far from the junction point, implying nonspecific binding of the single-stranded portion of these two substrates near the catalytic site of At-HIGLE 1-183 ( Figure  3D-F). Both At-HIGLE and At-HIGLE 1-183 bind singlestranded DNA. However, only At-HIGLE 1-183 can generate nick in the single-stranded DNA towards the 3 edge as seen in the cleavage patterns of 3 flap and splayed-arm substrates (Supplementary Figure S11). In summary, all branched DNA substrates are cut in similar sites, and At-HIGLE 1-183 could introduce additional nicks in branched DNA substrates.

Model of DNA binding
Two DNA-bound structures of SLX1 are currently available: (1) Tt-Slx1-Slx4 CCD in complex with a hairpin-like DNA (PDB: 6SEI) (39) and (2) Sc-Slx1-Slx4 SAP+CCD in complex with 5 flap DNA substrate (PDB: 7CQ4) (67). In none of these structures, the DNA substrate is bound to the enzyme in a catalytic configuration. DNA bound structures of R.Eco29kI and Hpy188I provide insights into the alignment of DNA substrate around the catalytic site of GIY-YIG endonucleases (68,69). The insights into the catalytic interaction of DNA with GIY-YIG nuclease are provided by the structure of a substrate complex of R.Eco29kI (PDB: 3NIC). This structure provides information about the orientations of the DNA strand that undergoes cleavage and the non-cleaved strand around the active site. This enables the identification of two DNA binding sites: site I and site II ( Figure 4A). DNA-bound structure of Tt-Slx1-Slx4 CCD revealed a third interface for DNA interaction (site III) away from the active site (39) ( Figure 4A). The DNAbound structure of Sc-Slx1-Slx4 SAP+CCD contains a 5 flap substrate with a local distortion at the active site induced by the single nucleotide flap being wedged into the DNA duplex (67). Nevertheless, the structure confirmed the role of site I and site II in binding the two arms of a branched DNA substrate. If extended, the single nucleotide flap can reach and bind positively charged amino acid residues at site III. Further confirmation of DNA binding by site III comes from the interaction of a symmetry-related DNA molecule with this region in the structure of Sc-Slx1-Slx4 SAP+CCD (PDB: 7CQ4). Therefore, based on the DNA bound structures of R.Eco29kI (PDB: 3NIC), Tt-Slx1-Slx4 CCD (PDB: 6SEI) and Sc-Slx1-Slx4 SAP+CCD (PDB: 7CQ4), we generated a DNA substrate-binding model for At-HIGLE to understand the interaction of various arms of branched DNA substrates ( Figure 4A, Supplementary Figures S12  and S13).
The At-HIGLE-DNA substrate model allowed us to identify amino acid residues involved in DNA substrate binding. These included Arg57 and Arg54 from site I, Trp96 and His100 from site II, and Arg39 and Lys115 from site III ( Figure 4A). The DNA-binding residues of At-HIGLE identified based on the model are strictly conserved among plants (Supplementary Figure S5). The site I plays a vital role in orienting the DNA substrates in the correct register at the active site. Arg57 and Arg54 interact with the noncleaved DNA strand. This interaction allows the complementary cleavable strand to enter the active site with Arg55 on one side and the metal-chelating Glu95 on the other side, enabling an in-line nucleophilic attack. DNA duplex bound at site III is at a sharp angle relative to DNA bound at site I resulting in bending in the DNA substrate near the branch point. Site II is involved in interacting with the remaining arm of the DNA substrate. We validated the model by testing the effects of substituting the residues predicted to bind the DNA. The tested variants included R39A, R54A, R57A, W96A, H100A and K115A. The overall protein structure of mutants was assessed with circular dichroism spectroscopy with some deviations observed in the cases of R54A and K115A mutants (Supplementary Figure S14). All tested variants lost most of their catalytic activity and had greatly reduced affinity for DNA ( Figure 4B, Supplementary Figure S14). Moreover, our model agrees with the observed cleavage site in the DNA strand located toward the 3 side from the junction point (Figures 2 and 3). Therefore, biochemical results corroborated with the proposed model of At-HIGLE interacting with branched DNA.

Binding and processing of branched DNA substrates
The binding of branched DNA substrates and the nicking pattern is a complex interplay between the three DNAbinding sites (Sites I, II and III). At-HIGLE cleaves an HJ by generating nicks in 6-FAM and Cy5 labeled DNA strands ( Figure 2). The activity of At-HIGLE on an HJ is, therefore, identical to Cg-Slx1-Slx4 CCD . However, the main difference appears in the processing of non-symmetrical branched-DNA molecules (Replication fork, 5 flap, 3 flap, and splayed arm substrate) by At-HIGLE. Full length At-HIGLE specifically cleaves Cy5 labeled strand of nonsymmetrical branched DNA substrates with no activity on 6-FAM labeled strand (Figure 3). In contrast, Cg-Slx1-Slx4 CCD can cleave both strands (39). Multiple modes of branched DNA substrate binding utilizing the three DNA binding sites explain the activity of Cg-Slx1-Slx4 CCD on both the stands (39). Processing of only one DNA strand by At-HIGLE indicates it prefers to bind non-symmetrical branched DNA substrates only in one orientation ( Figure  4C, SI Figure 13). The processing of branched DNA substrate's arm with no discontinuity at the junction point is favored for catalysis. This distinction is absent in the case of HJs because of the molecule's symmetrical nature. The binding of asymmetrical joint DNA molecules in one particular orientation is a unique feature of At-HIGLE compared to SLX1.
A truncated At-HIGLE (At-HIGLE 1-183 ) can process both the DNA strands (6-FAM and Cy5 labeled) of symmetric and non-symmetric branched DNA substrates (Figure 2 and 3). It is, therefore, evident that the C-terminal region of At-HIGLE restricts the conformational space in which a branched DNA substrate can interact with At- HIGLE. The C-terminal region can regulate the conformational space of DNA binding either through sterically clashing with branched DNA substrates in specific orientations ( Figure 4C and Supplementary Figure S13) or by directly interacting with the DNA substrate. Attempts were made to express the C-terminal region of At-HIGLE (At-HIGLE 184-368 ) ( Supplementary Figures S15 and S16) alone and assess its DNA binding potential. However, most of the protein was produced in aggregated form, and we could not get sufficient material for DNA binding studies. Although the role of the At-HIGLE 184-368 in interaction with branched DNA substrate cannot be ruled out, the steric hindrance by the C-terminal region of At-HIGLE appears to be a more plausible scenario since truncation of the Cterminal region increases the binding affinity to all the DNA substrates, increases the catalytic activity and allows the enzyme to cleave both 6-FAM and Cy5 labeled DNA strands (Figures 2 and 3).

Holliday Junction resolution and the oligomeric state of At-HIGLE
The complete HJ resolution requires the introduction of two nicks across the Holliday Junction. The first incision is rate-limiting followed by a second coordinated incision within the life-time of the protein-DNA complex. To study the ability of At-HIGLE to perform a HJ resolution we used a cruciform assay on a supercoiled plasmid (pIRbke8 mut ) with a cruciform-like structure ( Figure 5). In a cruciform assay, a single nick on the supercoiled plasmid (pIRbke8 mut ) results in the resorption of cruciform and the formation of a nicked circular plasmid. On the other hand, two nicks within the life-time of protein-DNA complex result in a linear plasmid ( Figure 5A). A cruciform assay established the HJ resolution potential of At-GEN1 and At-SEND1 (46) similar to human GEN1 (73). GEN1 exists as a monomer in the solution and dimerizes on binding an HJ, facilitating two nicks and eventually resolving an HJ (73,74).
The full-length At-HIGLE generates both linear and nicked-circular plasmids. The appearance of a linear plasmid suggested that At-HIGLE has the potential to perform two incisions within the lifetime of the protein-DNA complex. The appearance of nicked circular plasmid provides clues about the possible non-canonical activity of At-HIGLE, similar to Yen1, which has both canonical and non-canonical activities (75) ( Figure 5B). Since, At-HIGLE has a potential to generate two incisions within the life-time of a single protein-DNA complex, there are two possible scenarios: either At-HIGLE dimerizes on interacting with an HJ as in the case of human GEN1 (73,74), or At-HIGLE exists as a homodimer in the solution as in the case of RuvC (19,76). Therefore, we determined the oligomeric state of full-length At-HIGLE and At-HIGLE  in the solution ( Figure 5C-F, Supplementary Figure S17). Full-length At-HIGLE exists as a homodimer in solution. This implies that At-HIGLE might function as a canonical resolvase--a dimeric protein introducing two nicks into an HJ by utilizing two active sites within the dimer. At-HIGLE 1-183 exists as a monomer in the solution, and generates a nickedcircular plasmid in the cruciform assay with linear plasmid appearing during the later time-point in the reaction. The  Figure S15) and has no catalytic activity (Supplementary Figure S16).

DISCUSSION
Regulation of the structure-selective endonucleases is essential for the maintenance of genome stability (77,78). For instance, G2/M arrest of HIV-infected cells results from an untimely activation of SLX4, leading to enhanced cleavage of DNA by MUS81-EME1 complex (79). Therefore, SSEs are under tight regulation: to be activated at the right time at the right place (8). SSEs regulation operates at different levels: cell cycle, nuclear localization, post-translational modification, and protein-protein interaction. Nuclear exclusion of GEN1 until nuclear envelope breaks limits its activity on HJs (80,81). Cell cycle-dependent phosphorylation of MUS81-EME1 (G2/M transition) increases its activity (81). SLX1 relies on SLX4 for its catalytic activity. Besides, SLX4 couples the catalytic activities of SLX1 and MUS81-EME1 during HJ resolution (36). SLX4 undergoes CDK1 dependent phosphorylation that plays a role in MUS81-SLX4 interaction (82).
Fungal SLX1 exists as a homodimer with inaccessible DNA binding and catalytic residues. An interaction of SLX1 with the CCD (C-terminal conserved domain) domain of SLX4 disrupts this inhibitory homodimer and exposes the catalytic site and DNA binding residues (39,40). Therefore, SLX4 is the crucial regulator of SLX1, which is otherwise a highly promiscuous endonuclease. Our structural and biochemical studies establish the At-HIGLE of A. thaliana is being equivalent to SLX1 from fungi and animals. Similar to SLX1, At-HIGLE also exists as a homodimer in solution. However, unlike fungal SLX1, the At-HIGLE homodimer is catalytically active. While catalytic activity of many structure-specific endonucleases depends upon oligomerization of the protein, e.g. RuvC (19,76), GEN1 (73,74), SLX1 (39,40) and At-HIGLE, oligomerization driven nuclease activity is also identified in the cases of other nucleases, e.g. restriction endonucleases (68), transposases (83), Drosha-DGCR8 (84), SARS2 Nsp15 nuclease (85). While in fungi, SLX4 regulates the catalytic activity of SLX1, in the case of At-HIGLE, the C-terminal region regulates its own catalytic activity. The regulation by the Cterminal domain of At-HIGLE is mediated by moderating the substrate-binding, and modulating the rate of catalysis. Interestingly, based on secondary structure predictions and Alphafold2 (86), the C-terminal domain of At-HIGLE is predicted to be largely devoid of any defined structure. However, involvement of C-terminal domain in an interaction with the branched DNA substrate cannot be ruled out. Although a protein similar to SLX4 is unknown in plants, At-HIGLE has an inbuilt mechanism to regulate its own catalytic activity. Removal of the C-terminal region releases this regulatory mechanism resulting in vigorous catalysis by the At-HIGLE. A schematic of cruciform nuclease assay using a pIRbke8 mut plasmid containing an inverted repeat sequence that adopts a conformation mimicking a Holliday Junction. Asterisks represent the sites for EcoRI restriction endonuclease. The plasmid is purified in a supercoiled state (s.c.). Single nick results in the release of supercoils resulting in a nicked circular plasmid (n.c.), whereas two simultaneous nicks result in a linear product (l.) formation. (B) Time-dependent cruciform assays using At-HIGLE (full-length) (Left gel), At-HIGLE 1-183 (middle gel), and At-HIGLE E95Q (inactive mutant, full-length) (Right gel). EcoRI serves as a positive control generating a linear product. The second lane in each gel is substrate alone. All the reactions were run on 0.8% Agarose gel. (C) Size-exclusion chromatogram (black) of At-HIGLE purified on a Sephacryl S200 gel filtration column superimposed with gel filtration markers (broken grey). The peak corresponding to At-HIGLE is demarcated by an arrow. The fractions from the demarcated peak were run on a 12% SDS-PAGE. (D) Size-exclusion chromatogram (black) of At-HIGLE 1-183 purified on a Sephacryl S100 gel filtration column superimposed with gel filtration markers (broken grey). The peak corresponding to At-HIGLE 1-183 is demarcated by an arrow. The fractions from the demarcated peak were run on a 12% SDS-PAGE. (E) Estimation of the oligomeric weight of At-HIGLE from a standard curve generated using gel filtration markers (Vitamin B12, Myoglobin, Ovalbumin, and gamma globulin (F) Estimation of the oligomeric weight of At-HIGLE 1-183 from a standard curve generated using gel filtration markers (Vitamin B12, Myoglobin, Ovalbumin, and gamma globulin). Kav was calculated as (Ve -Vo)/(Vt -Vo) where Ve, Vo and Vt are elution volume, void volume and total volume, respectively. Thyroglobulin (670 kDa) was used to determine the column's void volume.
The substrate specificity of SSEs is dependent mainly upon their specific structural attributes. Although RuvC shows preference toward a particular sequence at the crossover point (5 -A/T TT↓G/C-3 ) (18), most SSEs are specific for the structure adopted by branched DNA molecules. GEN1 dimerizes on binding an intact HJ, undergoes a transition of its active site from disordered to ordered state, and bends a DNA substrate with the help of a wedge and an H2TH motif (Helix-two-turn-helix) (23,74). Conformational changes undergone by MUS81-EME1 on binding DNA substrate followed by the formation of a 5 phosphate binding pocket dictates its substrate preference for nicked HJ over an intact HJ (33,34). SLX1 and At-HIGLE comprise surface-exposed positively charged residues organized as distinct patches ( Supplementary Figure S18). These positively charged patches interact with various arms of branched DNA substrates. Therefore, SLX1 and At-HIGLE can bind and process HJ, RF, 3 flap, 5 flap and splayed arms with varying degrees of efficiency. Furthermore, SLX1 and At-HIGLE identify the branching point in the DNA substrate by locating structural discontinuity in DNA by the spatial arrangement of DNA binding patches at an angle with each other (39,40).
At-HIGLE generates nick specifically in the doublestranded portion of the arm without any discontinuity at the junction. In contrast, SLX1 can generate incisions in all the arms of branched DNA substrates making At-HIGLE a more specific SSE than SLX1. This distinction between At-HIGLE and SLX1 disappears with the truncation of the C-terminal region of At-HIGLE. The C-terminal region of At-HIGLE is indeed guiding the conformational space in which a branched DNA substrate can interact with At-HIGLE. The conformational space of DNA substrate binding is further restricted by already ordered DNA binding loops of At-HIGLE as observed in the crystal structure of the nuclease domain of At-HIGLE. In the cases of Cg-SLX1 and Tt-SLX1, the DNA binding sites are primarily disordered and undergo a transition to an ordered state on interaction with DNA. Altogether, we determined the first structure of SLX1 lineage member of GIY-YIG nuclease superfamily from the plant kingdom and established its role in HJ resolution. Our biochemical data provide novel insights into an in-built mechanism that regulates the catalytic activity of At-HIGLE.
Among humans, Holliday Junctions are processed through dissolution involving BTR complex (BLM helicase-TOPOIII␣-RMI1-RMI2) or resolution involving SSEs (also called resolvases). Processing of HJs through BTR complex is the preferred mechanism in mammalian mitotic cells. Mutations in BLM helicase are associated with Bloom's syndrome characterized by chromosomal instability, immunodeficiency, and an early onset of cancer conditions (87). Bloom's syndrome cells exhibit increased resolvases mediated sister chromatid exchange (88,89). The phenotype can be suppressed by depleting resolvases in Bloom's syndrome cells, indicating dissolution and resolution are parallel pathways. The combined deletion of MUS81 and SLX1 have a phenotype similar to the single deletion of either MUS81 or SLX1, suggesting the two proteins work in the same pathway. However, depletion of GEN1 either with SLX1 or MUS81 has an additive effect, indicating GEN1 operates in a pathway independent of SLX1 and MUS81 (89). Furthermore, the SLX1-SLX4 complex interacts with MUS81-EME1 during the G2/M phase of the cell cycle. The catalytic sites of the two nucleases cooperate to resolve HJ through a nick-and-counter-nick mechanism. The interaction between SLX1 and MUS81-EME1 is mediated through SLX4 (36). SLX4 is a multidomain scaffolding protein interacting with several nucleases. Mutations in SLX4 are associated with a recessive human disorder, Fanconi anemia (25)(26)(27)(28)41,42).
In plants, the entire mechanism of HJ processing is not worked out as extensively as in animals and fungi, with important gaps in our understanding of this process ( Figure 6). Among plants, the processing of HJ proceeds through either dissolution or resolution. Dissolution is carried out by an RTR complex (RecQ helicase-Top3␣-RMI1) (90)(91)(92) while SSEs participate in the resolution. GEN1, SEND1 and MUS81-EME1 are the only plant SSEs known to participate in HJ resolution (44)(45)(46)(47)(48)(49)(50)93,94). Similar to animals, emerging evidence suggests that dissolution and resolution are also independent pathways in plants. A mutation in RecQ helicase increases the crossover frequency, and mutation in MUS81 affects meiotic recombination frequency. However, the mutations in both MUS81 and RecQ helicase is lethal among plants (95)(96)(97)(98). A disruption in RAD51C (a protein involved in strand invasion and formation of D-loop) can suppress the lethality caused by mutations in RecQ helicase and MUS81 (99). Silencing GEN1 or its loss of function results in male-sterility and persistent doublestranded breaks (94,100). The combined absence of SEND1 and MUS81 results in developmental defects and genome instability (101). Depletion of SEND1 does not enhance the defect caused by an absence of GEN1 (94). Therefore, these accumulated works suggest that GEN1 and SEND1 very likely work in pathways independent of MUS81-EME1.
While MUS81-EME1 from plants acts on intact HJ with reduced efficiency (47), a protein similar to animal SLX1 that can initiate HJ resolution has been unknown. An absence of SLX1 like protein in plants, therefore pose an interesting predicament. In the case of plants, At-HIGLE can be one of the SSEs responsible for generating an initial nick for MUS81-EME1. It may also be feasible that At-HIGLE might just be acting on its own to resolve HJs because of its dimeric nature. Among animals, SLX4 coordinates the activities of SLX1 and MUS81-EME1 (36). Therefore, it is imperative to determine if the actions of At-HIGLE and MUS81-EME1 are coordinate during HJ resolution. Consequently, it is interesting to explore if the C-terminal region of At-HIGLE interacts with MUS81-EME1 and its impact on the At-HIGLE dimer and catalytic activity. Furthermore, to comprehensively establish the role of At-HIGLE in plant homologous recombination and its cross-talk with different components of Holliday Junction processing machinery, the effect of a defect in At-HIGLE alone or in combination with At-MUS81, At-GEN1, At-SEND1 and At-RecQ4A need to be evaluated through in vivo studies, e.g. using homologous recombination assay (102,103).
DNA repair and recombination in plants is a neglected field, especially when many of the pathways have undergone specialization compared to animals and fungi. Deci- Figure 6. Processing of Holliday junction in plants. Dissolution by an RTR complex (RecQ Helicase/TOP3␣/RMI1) results in non-crossover product formation. Resolution of Holliday Junctions is carried out by canonical endonucleases (GEN1 and SEND1) and non-canonical endonuclease (MUS81-EME1). At-HIGLE is a GIY-YIG endonuclease similar to fungi and animal SLX1 that can resolve HJ. At-HIGLE does not require a protein similar to SLX4 for its activity.
phering the complete molecular mechanism of HR in plants has dual benefits: (a) understanding the basic biology of DNA repair and recombination in plants, and (b) improvising and innovating genome editing techniques to address the ever-increasing demand for food. One of the major hurdles in any crop improvement program is the low efficiency of crossing over. In principle, an increase in HR frequency has been demonstrated earlier by over-expressing RuvC in plants (104). In-depth knowledge of various pathways operating at the HR level and their cross-talks with other DNA repair and recombination pathways will provide insights to fine-tune the HR frequency in the existing gene-editing techniques for crop improvement in the future.