key: cord-0297457-biiv61bb
authors: Kroupova, Alena; Ackle, Fabian; Boneberg, Franziska M.; Chui, Alessia; Weitzer, Stefan; Faini, Marco; Leitner, Alexander; Aebersold, Ruedi; Martinez, Javier; Jinek, Martin
title: Molecular architecture of the human tRNA ligase complex
date: 2021-07-11
journal: bioRxiv
DOI: 10.1101/2021.07.11.451954
sha: 07849e980462f414781d15a41b531c12ecbd898e
doc_id: 297457
cord_uid: biiv61bb

RtcB enzymes are RNA ligases that play essential roles in tRNA splicing, unfolded protein response, and RNA repair. In metazoa, RtcB functions as part of a five-subunit tRNA ligase complex (tRNA-LC) along with Ddx1, Cgi-99, Fam98B and Ashwin. The human tRNA-LC or its individual subunits have been implicated in additional cellular processes including microRNA maturation, viral replication, DNA double-strand break repair and mRNA transport. Here we present a biochemical analysis of the inter-subunit interactions within the human tRNA-LC along with crystal structures of the catalytic subunit RTCB and the N-terminal domain of CGI-99. We show that the core of the human tRNA-LC is assembled from RTCB and the C-terminal alpha-helical regions of DDX1, CGI-99, and FAM98B, all of which are required for complex integrity. The N-terminal domain of CGI-99 displays structural homology to calponin-homology domains, and CGI-99 and FAM98B associate via their N-terminal domains to form a stable subcomplex. The crystal structure of GMP-bound RTCB reveals divalent metal coordination geometry in the active site, providing insights into its catalytic mechanism. Collectively, these findings shed light on the molecular architecture and mechanism of the human tRNA ligase complex, and provide a structural framework for understanding its functions in cellular RNA metabolism.

RNA ligases play critical roles in various cellular processes including tRNA splicing (Lopes et al., 2015; 17 Phizicky and Hopper, 2010; Popow et al., 2012; Yoshihisa, 2014) , unfolded protein response (UPR) 18 (Bashir et al., 2021; Filipowicz, 2014) and RNA repair (Burroughs and Aravind, 2016) . Canonical 5 '-3' 19 ligases, including bacteriophage T4 RNA ligase 1 or Saccharomyces cerevisiae tRNA ligase Trl1, are 20 ATP-or GTP-dependent enzymes that catalyze the formation of a 3',5'-phosphodiester linkage between 21 3'-hydroxyl (3'-OH) and 5'-phosphate (5'-P) RNA termini (Greer et al., 1983; Peebles et al., 1979) . In 22 tRNA splicing and other processes requiring the joining of RNA ends with a 2',3'-cyclic phosphate 23 (2',3'>P) and a 5'-hydroxyl (5'-OH), these ligases operate as part of a multi-step mechanism including 24 separate phosphatase and kinase domains (Amitsur et al., 1987; Konarska et al., 1982; Xu et al., 1990) . 25

By contrast, the RtcB-like 3'-5' ligases are GTP-dependent enzymes that catalyze the direct joining of 26 RNA strands with terminal 2',3'>P and 5'-OH (Chakravarty et al., 2012; . 27

Unlike in the 5'-3' ligases, where the source of the splice-junction phosphate group is the nucleotide 28 triphosphate cofactor, RtcB ligases incorporate the substrate-derived 2',3'>P into the resulting 3',5'-29 phosphodiester bond (Filipowicz and Shatkin, 1983; Popow et al., 2011) . 30

RtcB ligases are present throughout all domains of life with a notable absence in fungi and 31 plants (Popow et al., 2012) . Although their catalytic mechanism is conserved, their substrates and 32 consequently their cellular functions vary considerably Popow et al., 2011; Popow 33 et al., 2012) . In archaea and eukaryotes, RtcB enzymes catalyze the ligation of tRNA exon halves to 34 produce mature tRNAs upon cleavage of precursor tRNA transcripts by the tRNA splicing endonuclease 35 (Calvin and Li, 2008; Li et al., 1998; Peebles et al., 1983; Trotta et al., 1997) . Furthermore, in 36 eukaryotes, RtcB ligases also function as part of the UPR to catalyze splicing of XBP1 mRNA after its 37 stress-induced cleavage by IRE1 (Inositol-requiring enzyme 1) Kosmaczewski et 38 al., 2014; Lu et al., 2014) . In turn, bacterial RtcB is a non-essential enzyme involved in RNA repair, 39 facilitating rescue of ribosomal RNA cleaved by the MazF ribonuclease of the MazE-MazF toxin-40 antitoxin system activated upon stress (Manwar et al., 2020; Temmel et al., 2016) . 41

The RNA ligation mechanism of RtcB enzymes consists of three nucleotidyl transfer steps. First, 42 a nucleophilic attack of an invariant histidine residue in the active site (His428 in human RTCB, His404 43 in Pyrococcus horikoshii RtcB, PhRtcB, and His337 in Escherichia coli RtcB, EcRtcB) on the α-44 phosphate moiety of guanine-5'-triphosphate (GTP) results in a covalently linked RtcB-GMP 45 8 assembly (Figure 4-figure supplement 1A) . The purified complex was subsequently subjected to 195 limited proteolysis with trypsin in order to identify exposed, protease-sensitive regions such as flexibly 196 linked domains or unstructured loops. Mass-spectrometric analysis of fragments separated by size 197 exclusion chromatography revealed DDX1 cleavage resulting in the loss of a fragment spanning 198 residues Ser436-Arg694 DDX1 corresponding to the RecA2 domain (Figure 4-figure supplement 1A Ala695-Phe740 DDX1 , is sufficient to mediate tRNA-LC assembly. 202

Guided by these results, we set out to identify the minimal subunit regions sufficient for tRNA-LC 203 assembly. As with the initial deletion analysis of the full-length tRNA-LC subunits, we designed 204 polypromoter constructs for co-expression in baculovirus-infected Sf9 cells, with the following tagging 205 scheme: His6-RTCB, MBP-FLAG-DDX1, mCherry-CGI-99, and GFP-FAM98B. Each set of constructs 206 contained N-or C-terminal truncations of one of the subunits with the others remaining full-length 207 ( Figure 4A-C) . Co-precipitation experiments revealed that a minimal fragment of DDX1 corresponding 208 to the C-terminal extension (residues Ala696-Phe740 DDX1 ) was sufficient for the assembly of the four-209 subunit core tRNA-LC. In contrast, neither a shorter C-terminal DDX1 construct (residues Phe740 DDX1 ) nor a DDX1 construct containing the RecA1 and RecA2 domains but lacking the C-terminal 211 extension (residues Ala2-Ala695 DDX1 ) were able to support tRNA-LC assembly. For both CGI-99 and 212 FAM98B, we observed that their N-terminal domains, although sufficient for the CGI-99:FAM98B 213 interaction ( Figure 3C ), were dispensable for tRNA-LC assembly ( Figure 4B,C) . Instead, a CGI-99 214 fragment spanning residues Ala195-Asp232 CGI-99 was essential and sufficient for the formation of the 215 core complex. The minimal region of FAM98B that was essential and sufficient for tRNA-LC assembly 216 could be mapped to a region spanning residues Asn200-Ser239 FAM98B . Notably, these regions are 217 predicted to have a propensity for forming alpha-helical coiled coils (Lupas et al., 1991) , while the C-218 terminal extension of DDX1 is predicted to form an alpha-helix. 219

To validate the subunit features sufficient for tRNA-LC assembly, we designed a "minimal" 220 complex comprising DDX1 residues Ala696-Phe740 DDX1 , CGI-99 residues Leu102-Arg244 CGI-99 and 221 FAM98B residues Asn200-Ser239 FAM98B , along with full-length RTCB. Co-expression of the subunit 222 constructs in baculovirus-infected insect cells resulted in the assembly of a stable, catalytically active 223 complex, as determined by affinity purification, size exclusion chromatography analysis, and in vitro 9 RNA ligation assay ( Figure 4D, Figure 1-figure supplement 1B) . Of note, a fragment of CGI-99 225 comprising residues Leu102-Arg244 CGI-99 was necessary for the assembly of the minimal complex, as 226 opposed to a shorter fragment (Ala195-Asp232 CGI-99 ) sufficient for complex assembly when DDX1 and 227 FAM98B were full-length ( Figure 4B ). Taken together, these results indicate that the structural core of 228 tRNA-LC is composed of RTCB together with the C-terminal alpha-helical regions of DDX1, CGI-99 and 229 FAM98B, and suggest that DDX1, CGI-99 and FAM98B interact synergistically to form a structural 230 platform that interacts with the catalytic domain of the RTCB ligase subunit ( Figure 5) . 231

Crystal structure of human RTCB reveals conserved fold and active site with distinct metal 232 coordination geometry 233

While metazoan RtcB enzymes exist as part of the tRNA ligase complex, prokaryotic RtcB orthologs 234 function as stand-alone enzymes (Popow et al., 2012) . Although crystallographic studies of the 235 stand-alone RtcB from P. horikoshii have provided insights into the catalytic mechanism of RtcB enzyme 236 family (Banerjee et al., 2021; Desai et al., 2013; Englert et al., 2012; Okada et al., 2006) , we currently 237 lack structural information on tRNA-LC-resident RtcB enzyme. To address this, we determined, to a 238 resolution of 2.3 Å, the crystal structure of human RTCB in complex with a GMP molecule and two 239 divalent cobalt ions, Co 2+ (A) and Co 2+ (B), bound in the active site ( Figure 6A , Table 1) . 240

The conserved fold of RTCB superimposes with PhRtcB (PDB ID: 4dwq) with root mean square 241 deviation of 1.4 Å over 474 Cα atoms ( Figure 6B ). Two adjacent loops found in the vicinity of the active 242 site in RTCB are structurally distinct from PhRtcB: an N-terminal region spanning residues 243 Leu45-Pro64 RTCB , which forms an extended α-helix and a loop as compared to a shorter loop 244 (Lys37-Arg42 PhRtcB ) in PhRtcB, and a segment comprising residues Arg436-Asp444 RTCB , which is only 245 partially ordered ( Figure 6B ). To test whether the N-terminal RTCB extension is involved in the 246 inter-subunit interactions within tRNA-LC, we co-expressed a truncated RTCB containing residues 247

Pro64-Gly505 RTCB with full-length DDX1, FAM98B, and CGI-99. Affinity purification revealed that the 248 N-terminal extension of human RTCB is not essential for the formation of tRNA-LC (Figure 6-figure  249 supplement 1). 250

The structure of GMP-bound RTCB likely mimics a product-bound state ( Figure 6A ). The GMP 251 guanosine base makes hydrogen bonding interactions with the sidechains of Glu230 RTCB , Ser409 RTCB , 252 and Lys504 RTCB . The ribose 2' and 3' OH groups are hydrogen-bonded with backbone nitrogens of 253 Ala430 RTCB and Gly431 RTCB , while the phosphate oxygen atoms interact with the sidechains of 254 Asn226 RTCB and His428 RTCB ( Figure 6C ). The phosphate moiety is further coordinated by the Co 2+ (A) 255 ion. The GMP binding pocket is conserved with respect to the structure of PhRtcB in the guanylated 256 state (PDB ID: 4dwq) with the exception of Glu470 RTCB which, unlike in the PhRtcB structure, adopts a 257 conformation incompatible with hydrogen bonding with N7 of the guanosine base ( Figure 6D ) (Englert 258 et al., 2012) . The Co 2+ ions A and B adopt tetrahedral and octahedral coordination geometries, 259 respectively ( Figure 6E , left panel). Co 2+ (A) is coordinated by His259 RTCB , His353 RTCB , Cys122 RTCB , and 260 the phosphate moiety of GMP, whereas Co 2+ (B) is coordinated by Asp119 RTCB , Cys122 RTCB , His227 RTCB , 261 and three water molecules. Intriguingly, the metal coordination geometries in the RTCB-GMP structure 262 differ from those adopted by Mn 2+ ions in the PhRtcB structures ( Figure 6E , Table 2 ), in which the A-263 position ion exhibits octahedral (PDB ID: 4dwr) or trigonal bipyramidal (PDB ID: 4dwq) coordination 264 arrangements, while the B-position ion is tetrahedrally coordinated (PDB ID: 4dwr). These observations 265 indicate considerable structural plasticity of metal ion coordination and suggest that the coordination 266 geometries might dynamically vary throughout the catalytic cycle of the enzyme to facilitate each step 267 of the overall reaction mechanism. 268 11

The tRNA ligase complex is an essential factor for both tRNA biogenesis and XBP1 mRNA splicing 270 during the unfolded protein response in human cells Kosmaczewski et al., 2014; Lu 271 et al., 2014; Popow et al., 2011) . Despite considerable progress, there remain gaps in our 272 understanding of its cellular function and its molecular architecture, particularly with respect to the non-273 catalytic subunits of the complex. In this study, we provide insights into the inter-subunit interactions 274 within tRNA-LC, as well as high-resolution structures of its catalytic subunit RTCB and the N-terminal 275 CH domain of Using co-expression experiments complemented with cross-linking/mass spectrometry 277 analysis, we show that a four-subunit core complex is assembled by the C-terminal regions of DDX1, 278 CGI-99 and FAM98B together with the catalytic domain of the RTCB subunit. Each of the three non-279 catalytic subunits is required for the integrity of the complex, suggesting that they function synergistically 280 to generate an interaction platform for RTCB. As the C-terminal regions of DDX1, FAM98B and CGI-99 281 are predicted to be alpha-helical and, in the case of CGI-99 and FAM98B to form coiled-coils, we 282 hypothesize that they associate to form a helical bundle ( Figure 5 ). The C-terminal extension of DDX1 283 required for tRNA-LC assembly is located outside the helicase core comprising the tandem RecA1 and 284

RecA2 domains and the SPRY domain. This suggests that the helicase (DDX1) and ligase (RTCB) 285 modules of tRNA-LC are flexibly linked, which may serve to facilitate interactions with a diverse array 286 of RNA substrates. 287

In addition, the CGI-99 and FAM98B subunits of tRNA-LC form a stable heterodimer through 288 an independent interaction interface involving their N-terminal domains. This heterodimer further serves 289 as an interaction platform for the recruitment of ASW, the fifth tRNA-LC subunit. The existence of 290 CGI-99:FAM98B or CGI-99:FAM98B:ASW subcomplexes in vivo has not been reported to date and it 291 is currently unclear whether they have a biological role independent from tRNA-LC. CGI-99 was 292 previously shown to form a denaturation-resistant homodimer under certain conditions, although only 293 monomeric CGI-99 is associated with the other subunits of tRNA-LC (Perez-Gonzalez et al., 2014) . 294

Consistently, we have not observed CGI-99 dimerization in vitro. 295

Our crystal structure of the N-terminal domain of CGI-99 confirms that it shares similarities with 296 CH domains, which are typically found in a range of signaling and cytoskeletal proteins (Gimona et al., 297 2002; Korenbaum and Rivero, 2002; Yin et al., 2020) . The CH1-and CH2-type CH domains occur in 298 tandem in proteins such as α-actinin (Shams et al., 2016) , β-III-spectrin (Avery et al., 2017) , or 299 dystrophin (Singh and Mallela, 2012) and mediate their binding to F-actin. Contrary to the tandem 300 CH1-CH2 domains, CH3-type occur as stand-alone domains at the N-termini of proteins such as 301 calponin. Apart from actin binding, CH domains have also been found to mediate interactions with 302 microtubules (MT). The end-binding protein 1 (EB-1) regulates microtubule dynamics, with its 303 N-terminal CH domain being essential for its MT interaction (Hayashi and Ikura, 2003) . Similarly, the 304 CH domains of NDC80 and NUF2 in the NDC80 kinetochore complex mediate MT binding, although 305 unlike in EB-1 this is conditional on the formation of the NDC80-NUF2 heterodimer (Ciferri et al., 2008; 306 Wei et al., 2007) . Our structural analysis reveals that the CH domain of CGI-99 exhibits closest structural 307 homology to the microtubule binding NDC80-like family of CH domains, as proposed by an earlier study 308 (Schou et al., 2014) . There is, however, no evidence to date to indicate that CGI-99 and FAM98B, or 309 indeed tRNA-LC interacts with MTs or other cytoskeletal proteins. Nevertheless, tRNA-LC has been 310 implicated in several cytoskeleton-related processes, including RNA transport along microtubules 311 (Kanai et al., 2004 ) and regulation of cell-adhesion dynamics (Hu et al., 2008) . CGI-99 was also found 312 to interact with the tubulin binding protein PTPIP51 (protein tyrosine phosphatase interacting protein 313 51) (Brobeil et al., 2012) and the centrosomal protein ninein (Howng et al., 2004) . Further studies will 314 be necessary to understand whether the CH domain of CGI-99, perhaps together with the putative CH 315 N-terminal domain of FAM98B, facilitates a cytoskeleton-related cellular function of tRNA-LC or plays 316 another functional role. 317

Finally, the crystal structure of human RTCB determined in this study reveals a product-bound 318 state containing GMP and two divalent cobalt ions in the active site. The structure uncovers the 319 disposition of the active site residues and metal coordination geometries, contradicting previous 320 predictions based on homology modeling (Nandy et al., 2017) . Based on crystal structures of Mn 2+ -321 bound PhRtcB, metal ion A was expected to adopt an octahedral geometry, coordinated by four active 322 site residues and two oxygens of the GMP phosphate. The RTCB structure shows that the ion is instead 323 tetrahedrally coordinated and that Asn226 forms a hydrogen bond with the phosphate moiety of GMP 324 instead of directly coordinating the metal ion. The observed metal coordination geometries are distinct 325 from those found in the crystal structures of the PhRtcB ortholog, which were captured in the apo and 326 guanylated intermediate states in the presence of manganese ions. One possibility for the discrepancy 327 is that the metal ion coordination geometries are dynamic throughout the catalytic mechanism, thereby 328 13 facilitating the individual reaction steps of RtcB guanylation, 5'-exon activation and finally exon ligation. 329

Additional studies focused on structure determination of intermediate states within the catalytic 330 mechanism will thus be needed to uncover whether the observed plasticity of metal ion coordination in 331 the active sites of RtcB enzymes plays a role in the catalytic mechanism. In addition, both RTCB and 332 the entire tRNA-LC were recently shown to be redox-regulated and to undergo oxidative inactivation in 333 the presence of copper, suggesting that the binding of copper ions in the active site promotes oxidation 334 of the active site cysteine (Cys122 RTCB ), thus precluding proper divalent metal coordination (Asanović 335 et al., 2021) . The NAD(P)H-dependent interaction of RTCB with PYROXD1 counteracts this process 336 (Asanović et al., 2021) . However, the precise chemical mechanism by which PYROXD1 protects RTCB 337 is currently unknown. Furthermore, the RNA substrate binding mechanism of RTCB is likewise not fully 338 understood. Previous mutational studies of E. coli and P. horikoshii RtcB enzymes have implicated 339 conflicting sets of amino acid residues in exon RNA recognition (Englert et al., 2012; Maughan and 340 Shuman, 2016) . Although a recent crystal structure of PhRtcB bound to a 5'-OH 3'DNA oligonucleotide 341 (Banerjee et al., 2021) revealed the residues involved in 3' exon binding, additional structures of RtcB 342 enzymes bound to the 5' exon or to both exons simultaneously will be required to pinpoint the precise 343 position of the 2',3'>P end and to provide further insights into the later steps of the ligation reaction 344 beyond the initial guanylation of RtcB. 345

In conclusion, this work provides fundamental insights into the molecular architecture of the 346 human tRNA-LC, paving the way for its further structural and mechanistic studies. These will shed light 347 on specific molecular determinants of inter-subunit interactions and the ways in which they support the 348 catalytic activity as well as putative non-catalytic functions of the complex, thereby advancing our 349 understanding of tRNA-LC in metazoan RNA metabolism. His6 tag followed by a TEV protease cleavage site; DDX1 with an N-terminal mCherry tag followed by 437 a TEV protease cleavage site; FAM98B with an N-terminal GFP tag followed by a TEV protease 438 cleavage site; CGI-99 with an N-terminal (StrepII)2 tag followed by a PreScission protease cleavage 439 site; and ASW with an N-terminal MBP tag followed by a PreScission protease cleavage site. The 440 plasmids containing individual subunits were combined into a single baculovirus transfer plasmid using 441 the MacroBac protocol (Gradia et al., 2017) . The 'tRNA-LC' construct contained all five subunits 442 whereas the deletion constructs lacked each one subunit, as indicated. The CGI-99:FAM98B construct 443 contained only the two subunits CGI-99 and FAM98B. Recombinant viruses were generated using the 444 Bac-to-Bac Baculovirus expression system (Invitrogen). The proteins were expressed by infecting 50 445 ml of Sf9 insect cells at density of 1.0 x 10 6 ml -1 with the above-described viruses. Infected cells were 446 harvested by centrifugation at 500x g for 10 min at room temperature (rt). The cell pellet was frozen in 447 liquid nitrogen and stored at -20C until further use. Subsequently, the pellet was thawed and 448 formamide, 50 mM EDTA, 1 ng ml -1 bromophenol blue, 1ng ml -1 xylene cyanol, and boiled for 3 min. 486

The reaction products were separated on a denaturing 15% urea-polyacrylamide gel (SequaGel) and 487 detected by autoradiography. 

Purified tRNA-LC was diluted in cross-linking buffer (200 mM HEPES, pH 8.0, 150 mM KCl, 0.5 mM 520 TCEP) to a final concentration of 1 mg ml -1 . To initiate the cross-linking reaction, 2 µL of a 25 mM stock 521 of disuccinimidyl suberate (DSS-d0/d12, Creative Molecules) was added in independent replicates to two 522 50 µL aliquots of the complex, resulting in a final concentration of 1 mM DSS. Samples were incubated 523 for 30 min at 25 °C with mild shaking in an Eppendorf Thermomixer, and the reaction was quenched by 524 addition of ammonium bicarbonate to a final concentration of 50 mM. Quenched samples were 525 evaporated to dryness in a vacuum centrifuge and processed as described previously (Leitner et al., 526 2014) . Briefly, disulfide bonds were reduced with TCEP, free thiols were alkylated by iodoacetamide 527 and proteins were sequentially digested with endoproteinase Lys-C (Wako, 1:100 enzyme-to-substrate 528 ratio, 3 h at 37 °C) and trypsin (Promega, 1:50, overnight at 37 °C). Digested samples were desalted, 529 purified by solid-phase extraction (SepPak tC18, Waters) and fractionated by size-exclusion 530 20 chromatography (SEC; Superdex Peptide PC 3.2/300, Cytiva) (Leitner et al., 2014) . Two SEC fractions 531 enriched in cross-linked peptides were collected and evaporated to dryness. The fractions were 532 analyzed in duplicate by liquid chromatography-tandem mass spectrometry (LC-MS/MS) using a 533 Thermo Easy nLC-1000 HPLC system coupled to a Thermo Orbitrap Elite mass spectrometer. The 534 HPLC separation was performed on a Thermo Acclaim PepMap RSLC column (150 mm × 75 µm, 2 µm 535 particle size) using the mobile phases A = water/acetonitrile/formic acid (98:2:0.15, v/v/v) and 536 B = acetonitrile/water/formic acid (98:2:0.15, v/v/v), a gradient from 9 to 35% B in 60 min, and a flow 537 rate of 300 nL/min. The mass spectrometer was operated in data-dependent acquisition mode with MS1 538 detection in the Orbitrap analyzer (120 000 resolution) and MS2 detection in the linear ion trap (normal 539 resolution). For each cycle, the ten most abundant precursor ions with a charge state of +3 or higher 540 were selected for fragmentation in the linear ion trap with a normalized collision energy of 35%. 541

Previously selected precursors were put on a dynamic exclusion list for 30 s. 542

Mass spectrometry data was analyzed using xQuest (Walzthoeni et al., 2012) , version 2.1.4. MS/MS 544 spectra were searched against a custom database containing the target protein sequences and two 545 contaminant proteins and a decoy database containing the reversed sequences. The main search 546 parameters were: Enzyme = trypsin, maximum number of missed cleavages = 2, cross-linking 547 sites = Lys and N terminus, fixed modification = carbamidomethylation of Cys; variable 548 modification = oxidation of Met, MS1 mass tolerance = ± 15 ppm, MS2 mass tolerance = ± 0.2 Da for 549 "common" fragments and ± 0.3 Da for "cross-link" fragments. The scoring scheme of (Walzthoeni et al., 550 2012) was used. Post-search, the results were filtered to a mass error window of -6 to +3 ppm, a TIC 551 value of 0.1% and a deltaS value of <0.9. MS/MS spectra of all remaining candidate hits were manually 552 evaluated and the false discovery rate at the non-redundant peptide pair level was adjusted to <5%. 553

The list of cross-link peptide pair identifications from the two replicate experiments is provided in Figure  554 2 -source data 1 and 2. 555 were not heated before loading to preserve the fluorescence of the GFP and mCherry tags. The gels 586 were visualized using a Typhoon FLA 9500 fluorescence scanner (Cytiva) with the 473 nm and the 587 532 nm laser to detect GFP and mCherry, respectively, and then stained with Coomassie blue. 588

The cloning, Sf9 cell expression and purification of the minimal tRNA-LC was performed as described 590 above for the 5-subunit tRNA-LC with the following changes. DNA encoding amino acids Ala696-591

Phe740 of human DDX1 was cloned into UC Berkeley MacroLab 438C (Addgene plasmid #55220). 592

The resulting plasmid encoded for untagged RTCB(1-505); DDX1(696-740) with an N-terminal His6 tag 593 followed by an MBP tag and a TEV protease cleavage site; untagged with an N-terminal StrepII-tag followed by a GFP tag and a TEV protease cleavage 595 site. For the first affinity step, the supernatant was incubated with 9ml of Ni-NTA Superflow resin 596 in Villigen, Switzerland. The data were processed using XDS (Kabsch, 2010) and the space group was 633 determined to be P61 using POINTLESS (Evans, 2011) with two copies in the asymmetric unit. The 634 crystals diffracted to a resolution of 2.0 Å (native crystal) and 2.2 Å (S-SAD crystal). The structure was 635 determined by a single-wavelength anomalous diffraction experiment utilizing endogenous sulfur atoms 636 (S-SAD). Four 360 datasets were measured at a wavelength of 2.0173 Å and merged using XSCALE 637 (Kabsch, 2010) . Seven S sites were identified using Phenix.HySS (Grosse-Kunstleve and Adams, 638 2003) , four of which corresponded to residues Cys19 and Cys69 (in both chains) and two corresponded 639 to chloride ions coordinated by the protein. Phasing, density modification, preliminary model-building 640 and refinement were performed using Phenix.Autosol (Terwilliger et al., 2009) . The model was further 641 improved using Phenix.AutoBuild (Terwilliger et al., 2008) and the resulting structure was used as the 642 search model for molecular replacement with the higher-resolution native dataset using Phenix.Phaser 643 (Adams et al., 2010; McCoy et al., 2007) . The model building was finished in Coot (Emsley et al., 2010) 644 and refined using Phenix.Refine (Afonine et al., 2012) . The final CGI-99(2-101) model contains residues 645 2-94 in chain A and residues 2-101 in chain B with three molecules of isopropanol, one molecule of 646 hexylene glycol and one chloride bound to each chain. An inter-molecular disulfide bond was observed 647 between the Cys19 residues of the two chains in the asymmetric unit. Chain B was used as a template 648 for structural superpositions using the DALI server (Holm, 2020) . 649 Crystals of GMP-bound RTCB were obtained using the hanging drop vapor diffusion method at 20C. 671 0.5 l of protein solution containing RTCB (7.7 mg ml -1 ), GMP (0.5 mM), and an RNA substrate 672 (0.15 mM, 5'-ACGUGCAAAGGCACUC-3'p) was mixed with 0.5 l of reservoir solution (0.1 M sodium 673 acetate pH 5.5, 0.6 M sodium formate, 14% (w/v) PEG 4K, and 5 mM CoCl2). The crystals were 674 transferred to the reservoir solution supplemented with 25% (v/v) glycerol for cryoprotection and then 675 flash-cooled in liquid nitrogen. X-ray diffraction data were collected at beamline X06DA (PXIII) of the 676 Swiss Light Source at the Paul Scherrer Institute in Villigen, Switzerland. The data were processed 677 using XDS (Kabsch, 2010) and the space group was determined to be P41212 using POINTLESS 678 (Evans, 2011) . The crystal diffracted to a resolution of 2.3 Å with two copies present in the asymmetric 679 Ni-NTA anti-FLAG FLAG-DDX1 GFP mCherry 2 2 0 -2 3 9 2 -3 3 0 2 0 0 -2 3 9 2 4 0 -3 3 0 2 -1 9 9 2 2 0 -2 3 9 2 -3 3 0 2 0 0 -2 3 9 2 4 0 -3 3 0 2 -1 9 9 2 2 0 -2 

PHENIX: a comprehensive Python-based system for macromolecular structure solution

Towards automated crystallographic structure refinement with phenix.refine

Bacteriophage T4 anticodon nuclease, polynucleotide kinase and RNA ligase reprocess the host lysine tRNA

The oxidoreductase PYROXD1 uses NAD(P)+ as an antioxidant to sustain tRNA ligase activity in pre-tRNA splicing and unfolded protein response

Structural basis for high-affinity actin binding revealed by a β-III-spectrin SCA5 missense mutation

Structure of 3'-PO4/5'-OH RNA ligase RtcB in complex with a 5'-OH oligonucleotide

The molecular mechanism and functional diversity of UPR signaling sensor IRE1

Interaction of PTPIP51 with Tubulin, CGI-99 and Nuf2 During Cell Cycle Progression

RNA damage in biological conflicts and the diversity of responding RNA repair systems

RNA-splicing endonuclease structure and function

RNA ligase RtcB splices 3'-phosphate and 5'-OH ends via covalent RtcB-(histidinyl)-GMP and polynucleotide-(3')pp(5')G intermediates

Implications for kinetochore-microtubule attachment from the structure of an engineered Ndc80 complex

Rewriting the rules for end joining via enzymatic splicing of DNA 3'-PO4 and 5'-OH ends

Coevolution of RtcB and Archease created a multiple-turnover RNA ligase

Structures of the noncanonical RNA ligase RtcB reveal the mechanism of histidine guanylylation

A tRNA splicing operon: Archease endows RtcB with dual GTP/ATP cofactor specificity and accelerates RNA ligation

DDX1 is an RNA-dependent ATPase involved in HIV-1 Rev function and virus replication

Features and development of Coot

Archaeal 3'-phosphate RNA splicing ligase characterization identifies the missing component in tRNA maturation

Structural and mechanistic insights into guanylylation of RNA-splicing ligase RtcB joining RNA between 3'-terminal phosphate and 5'-OH

An introduction to data reduction: space-group determination, scaling and intensity statistics

A DEAD box protein facilitates HIV-1 replication as a cellular co-factor of Rev

Making ends meet: a role of RNA ligase RTCB in unfolded protein response

Origin of splice junction phosphate in tRNAs processed by HeLa cell extract

Functional plasticity of CH domains

MacroBac: New Technologies for Robust and Efficient Large-Scale Production of Recombinant Multiprotein Complexes

Mechanism of action of a yeast RNA ligase in tRNA splicing

Substructure search procedures for macromolecular structures

The RNA-binding protein DDX1 promotes primary microRNA maturation and inhibits ovarian tumor progression

Crystal Structure of the Amino-terminal Microtubule-binding Domain of End-binding Protein 1 (EB1)

Benchmarking fold detection by DaliLite v.5

DALI and the persistence of protein shape

A novel ninein-interaction protein, CGI-99, blocks ninein phosphorylation by GSK3beta and is highly expressed in brain tumors

FAAP, a novel murine protein, is involved in cell adhesion through regulating vinculin-paxillin association

PA subunit from influenza virus polymerase complex interacts with a cellular protein with homology to a family of transcriptional activators

The mammalian tRNA ligase complex mediates splicing of XBP1 mRNA and controls antibody secretion in plasma cells

Xds

Global analysis of protein-RNA interactions in SARS-CoV-2-infected cells reveals key regulators of infection

Kinesin transports RNA: isolation and characterization of an RNA-transporting granule

The Phyre2 web portal for protein modeling, prediction and analysis

Structure of the SPRY domain of the human RNA helicase DDX1, a putative interaction platform within a DEAD-box protein

RNA ligation via 2'-phosphomonoester, 3'5'-phosphodiester linkage: requirement of 2',3'-cyclic phosphate termini and involvement of a 5'-hydroxyl polynucleotide kinase

Calponin homology domains at a glance

The RtcB RNA ligase is an essential component of the metazoan unfolded protein response

Lysine-specific chemical cross-linking of protein complexes and identification of cross-linking sites using LC-MS/MS and the xQuest/xProphet software pipeline

Crystal structure and evolution of a transfer RNA splicing enzyme

DEAD Box 1 Facilitates Removal of RNA and Homologous Recombination at DNA Double-Strand Breaks

A role for DEAD box 1 at DNA double-strand breaks

RNA-binding protein DDX1 is responsible for fatty acid-mediated repression of insulin translation

Cutting, dicing, healing and sealing: the molecular surgery of tRNA

A Synthetic Biology Approach Identifies the Mammalian UPR RNA Ligase RtcB

Predicting coiled coils from protein sequences

The bacterial RNA ligase RtcB accelerates the repair process of fragmented rRNA upon releasing the antibiotic stress

Distinct Contributions of Enzymic Functional Groups to the 2',3'-Cyclic Phosphodiesterase, 3'-Phosphate Guanylylation, and 3'-ppG/5'-OH Ligation Steps of the Escherichia coli RtcB Nucleic Acid Splicing Pathway

Phaser crystallographic software

Homology model of the human tRNA splicing ligase RtcB

Crystal structure of an RtcB homolog protein (PH1602-extein protein) from Pyrococcus horikoshii reveals a novel fold

Novel gene ashwin functions in Xenopus cell survival and anteroposterior patterning

Precise excision of intervening sequences from precursor tRNAs by a membrane-associated yeast endonuclease

Splicing of yeast tRNA precursors: a two-stage reaction

hCLE/C14orf166 associates with DDX1-HSPC117-FAM98B in a novel transcription-dependent shuttling RNA-transporting complex

The PRIDE database and related tools and resources in 2019: improving support for quantification data

tRNA biology charges to the front

HSPC117 is the essential subunit of a human tRNA splicing ligase complex

Analysis of orthologous groups reveals archease and DDX1 as tRNA splicing factors

Diversity and roles of (t)RNA ligases

hCLE/C14orf166, a cellular protein required for viral replication, is incorporated into influenza virus particles

Cellular human CLE/C14orf166 protein interacts with influenza virus polymerase and is required for viral replication

A divergent calponin homology (NN-CH) domain defines a novel family: implications for evolution of ciliary IFT complex B proteins

Dynamic Regulation of α-Actinin's Calponin Homology Domains on F-Actin

The N-Terminal Actin-Binding Tandem Calponin-Homology (CH) Domain of Dystrophin Is in a Closed Conformation in Solution and When Bound to Factin

Novel mechanism of RNA repair by RtcB via sequential 2',3'-cyclic phosphodiesterase and 3'-Phosphate/5'-hydroxyl ligation reactions

RtcB is the RNA ligase component of an Escherichia coli RNA repair operon

The RNA ligase RtcB reverses MazF-induced ribosome heterogeneity in Escherichia coli

Decision-making in structure solution using Bayesian estimates of map quality: the PHENIX AutoSol wizard

Iterative model building, structure refinement and density modification with the PHENIX AutoBuild wizard

The yeast tRNA splicing endonuclease: a tetrameric enzyme with two active site subunits homologous to the archaeal tRNA endonucleases

False discovery rate estimation for cross-linked peptides identified by mass spectrometry

The Ndc80/HEC1 complex is a contact point for kinetochore-microtubule attachment

The cellular RNA helicase DDX1 interacts with coronavirus nonstructural protein 14 and enhances viral replication

Domain structure in yeast tRNA ligase

The DEAD-Box RNA Helicase DDX1 Interacts with the Viral Protein 3D and Inhibits Foot-and-Mouth Disease Virus Replication

Structural Characteristics, Binding Partners and Related Diseases of the Calponin Homology (CH) Domain

The input (bottom right) and bound fractions were analyzed by SDS-PAGE and 714 visualized by in-gel GFP (middle gel) or mCherry (bottom gel) fluorescence followed by Coomassie blue 715 staining (upper gel)

Figure 3 -figure supplement 1. Analysis of the N-terminal domain of CGI-99 and the 771 CGI-99:FAM98B interaction

A-B) DALI structural alignment of the N-terminal domain of CGI-99 (yellow) with (A) Chlamydomonas 773 reinhardtii IFT54 (green, PDB ID: 5fmt), and (B) Escherichia coli Lon protease (pink

101-244) with lysates of HEK293T cells transiently overexpressing 776 HA-(StrepII)2-GFP-FAM98B constructs. The input (I) and bound (B) fractions were analyzed by 777 SDS-PAGE and visualized by in-gel GFP fluorescence (upper panel) followed by Coomassie blue 778 staining (lower panel). The 'no bait' control contained HEK293T lysate incubated with GSH resin in the 779 absence of GST-CGI-99, 'no prey ctrl' contained GSH resin incubated with one of the GST-CGI-99 780 constructs, and 'CGI-99(1-244) control' is the bound fraction of

Figure 4 -figure supplement 1. Mass spectrometry analysis of RTCB:DDX1(436-783 740):FAM98B:CGI-99 subjected to limited proteolysis

Size-exclusion chromatogram of the complex before (blue) and after (orange)

Unidentified bands are indicated with an asterisk. (B) Masses identified by electrospray 787 ionization mass spectrometry analysis of the RTCB:DDX1(436-740):FAM98B:CGI-99 (upper panel), 788 peak A (middle panel) and peak B (bottom panel) from limited proteolysis of RTCB

unit. The structure was solved by molecular replacement using Phenix.Phaser (Adams et al., 2010; 680 McCoy et al., 2007) using the PhRtcB structure (PDB ID: 1uc2) as a search model. The initial model 681 building was done using Phenix.Autobuild (Terwilliger et al., 2008) and finished manually in Coot 682 (Emsley et al., 2010) . The model was refined using Phenix.Refine (Afonine et al., 2012) . The final 683 RTCB-GMP model contains residues 3-55, 60-434, and 444-505; one molecule of GMP and two Co 2+ 684 ions in chain A and residues 3-55, 59-440, 444-463, and 465-505