Despite the odds: formation of the SARS-CoV-2 methylation complex

Coronaviruses protect their single-stranded RNA genome with a methylated cap during replication. The capping process is initiated by several nonstructural proteins (nsp) encoded in the viral genome. The methylation is performed by two methyltransferases, nsp14 and nsp16 where nsp10 acts as a co-factor to both. Aditionally, nsp14 carries an exonuclease domain, which operates in the proofreading system during RNA replication of the viral genome. Both nsp14 and nsp16 were reported to independently bind nsp10, but the available structural information suggests that the concomitant interaction between these three proteins should be impossible due to steric clashes. Here, we show that nsp14, nsp10, and nsp16 can form a heterotrimer complex. This interaction is expected to encourage formation of mature capped viral mRNA, modulating the nsp14’s exonuclease activity, and protecting the viral RNA. Our findings show that nsp14 is amenable to allosteric regulation and may serve as a novel target for therapeutic approaches.


Introduction
Here, we studied the interaction between the two methyltransferases encoded in the genome of the 82 coronavirus-nsp16 and nsp14. Both protteins bind to nsp10 and both are required for complete methylation.

83
Therefore, their spatial proximity mediated by nsp10 would appear as beneficial for the methylation efficacy. Prior 84 in silico analysis of structural information on nsp10/14 and nsp10/16 complexes suggested that simultaneous 85 binding of both exonucleases to nsp10 should be impossible due to steric hindrance 17 .Interestingly, both nsp14 86 and nsp16 interact with nsp10 by typical, well-defined protein-protein interfaces containing deep-reaching 87 lipophilic residues and solvent-protected hydrogen bonds. The element that causes steric overlap between nsp14 88 and nsp16 is a peculiar N-terminal "lid" domain of nsp14. This lid is mostly devoid of secondary structure and lacks 89 characteristic protein-protein interaction complementarity. Interestingly, recent structures of nsp14 without 90 nsp10 show massive rearrangement of the lid region confirming its structural flexibility 18,19 . This prompted us to 91 postulate that a local structural rearrangement of the nsp14 N-terminal lid region is therefore required and 92 possible, thus facilitating the heterotrimer complex formation.

95
Biochemical evidence of heterotrimer formation 96 Nsp10, nsp14, and nsp16 were co-expressed in E. coli and purified. A single peak was obtained in size exclusion 97 chromatography (SEC), indicating that all three proteins co-migrate (Fig 1a). Because no prior crosslinking was  Table 2).

109
The kinetics of the heterotrimer complex formation were assessed by MicroScale Thermophoresis (MST).

110
Nsp14 and nsp16 were separately expressed as histidine-tagged constructs and labeled with a high-affinity His-111 tag specific fluorophore dye. When labeled nsp14 was titrated with unlabeled nsp10, a dose-dependent increase 112 in thermophoretic signal was observed, indicating an interaction with a K d of 2.4 ± 0.2 µM, which is in the 113 agreement with the previously reported affinity with a K d of 1.1 ± 0.9 μM 20 (Fig. 1c). A comparable effect 21 was 114 observed when labeled nsp16 was titrated with unlabeled nsp10 with a K d of 0.24 ± 0.01 µM. Labeled nsp14 did 115 not directly interact with unlabeled nsp16, but when labeled nsp14 was titrated with unlabeled nsp10/16 complex, 116 a dose-dependent increase in thermophoretic signal was observed (Fig. 1c), which was interpreted as the 117 heterotrimer complex formation. Fitting the experimental data allowed the determination of the K d characterizing 118 the interaction at 0.28 ± 0.01 µM.
To further characterize the heterotrimer complex and its components, we analyzed thermal denaturation 120 profiles using nanoDSF. Each of the individual system components (nsp10, nsp14, and nsp16) was characterized 121 by a single characteristic denaturation temperature (Fig. 1d)

Functional consequences of nsp10/14/16 heterotrimer complex formation
GpppA is methylated at position N7 by nsp14 N7-methyltransferase domain, yielding N 7 MeGpppA 22 ; SAM is used 139 as a donor of the methyl group (Fig. 2a). Here, we tested whether the heterotrimer complex formation affects 140 nsp14 catalyzed methylation. To follow the N7-methyltransferase activity of nsp14, we used an indirect assay to 141 monitor changes in the level of a second reaction product, SAH, by HTRF. In the absence of one of the substrates 142 (SAM or RNA) or the tested enzyme (negative controls), there was no activity validating the experimental setup 143 (Extended Data Fig. 3a). Nsp14 showed no preference for the nascent nucleotide methylating both GpppG and 144 GpppA, as demonstrated by the production of equal levels of SAH when using either nucleotide as a substrate (Fig.   145 2b). The binary complex of nsp10/14 and the ternary complex N7-methyltransferase activities were comparable 146 to that of nsp14 alone, indicating that binding to nsp10 or the heterotrimer complex formation had no significant 147 influence on the N7-methyltransferase activity of nsp14. N7-methyltrasnferase activity was also not reported for

153
We have also tested if the nsp14 exoribonuclease (ExoN) activity affects methyltransferase activity by mutating 154 the ExoN binding site. However, no difference in N7-methyltrasnferase activity was observed, decoupling those 155 two activities from each other (Extended Data Fig. 3b). Alongside N7-methyltransferase activity, nsp14 harbors an   native and nondenaturing gel analysis) (Fig. 2c, denaturing gel). Interestingly, we observed that the nuclease 171 activity of nsp10/14 is reduced for the nsp10/14/16 heterotrimer, suggesting that nsp14 nuclease activity is 172 modulated by nsp16 (Fig. 2c, denaturing gel). To assess this, we added increasing amounts of nsp16 to preformed 173 nsp10/14 CoV-RNA1-A RNA complex and monitored the degradation of the RNA. Upon increasing the 174 concentration of nsp16, the degradation activity of nsp14 is reduced, starting at a near 1:1 ratio (Fig. 2d). This 175 effect is unlikely to be caused by the protective effect of RNA sequestering by the nsp16 as the RNA-nsp10/16 affinity is reported to be in high micromolar regime 26 (~100 µM) and yet protective functions of nsp16 appear at  with nsp16 at the surface of nsp10. The significant steric clash produced by this overlap precludes the heterotrimer complex assembly, which is mediated by the concomitant interaction of nsp14 and nsp16 with nsp10 ( Fig. 3a-c).

196
As such, the formation of the heterotrimer complex would require a significant structural rearrangement within 197 either nsp14 or nsp16. Analysis of the interactions between nsp10, nsp14 and nsp16 suggests that the former 198 binding surface contains a weakly interacting component -the N-terminal region of nsp14 (for rigorous analysis, 199 see Discussion). We hypothesize that within the nsp10/14/16 heterotrimer complex, nsp16 displaces the N-

209
To evaluate the above hypothesis, we created a "lid"-truncated mutant of nsp14 (nsp14) missing the initial

219
Guinier analysis of scattering profiles established that the largest radii of gyration (R g ) characterized the 220 heterotrimer complex (40.3±1.0 Å). Nsp14 and the nsp10/14 complexes were characterized by significantly 221 smaller R g s (28.0±0.5 and 30.1±1.8 Å, respectively); the nsp10/16 complex was characterized by the smallest (R g = 222 21.0±1.5 Å) (Extended Data Table 3). This data corresponds well with the expected molecular weights of tested 223 complexes at 1:1:1 stoichiometry, further supporting the stoichiometry of the heterotrimer complex.   The heterotrimer complex formed by nsp10, nsp14 and nsp16 was further characterized by transmission 254 electron microscopy. Negatively stained samples of a heterotrimer complex formed from full-length components 255 yielded a non-homogenous particle distribution, which precluded structural analysis (Extended Data Fig. 6).

256
However, when nsp14 methyltransferase domain was truncated out of the structure leaving nsp14 ExoN, the 257 particle distribution became more homogenous, allowing the convergent classification (Fig. 5a) and structural 258 analysis. The reconstruction obtained from the negative-stained transmission electron micrographs at 20 Å 259 resolution shows elongated particles with approximate dimensions of ~ 10 × 5 × 4.5 nm (Fig. 5b). Necking is evident   coiled-coil region connecting helix 1 and the sheet 1 and  helices 3,4 and 10; as well as -sheet 4 (Fig. 3b). In

285
Despite the fact that the interface between nsp10 and nsp14 buries a larger surface area compared to the 286 nsp10/16 interface, the affinity characterizing the components of the former complex is almost an order of 287 magnitude weaker than that of the latter. This prompted us to speculate that a significant part of the nsp10/14 288 interaction may not significantly contribute to affinity. The nsp10/14 interaction surface contains two major regions. Interactions within the N-terminal region of nsp14 involve primarily loops and other poorly structured 290 regions, such as the N-terminal coil-coiled region that interacts with the 1' helix of nsp10, -turn 1, and loops 291 between -sheet 2/3 and -sheet 7/8 (Fig. 3c). Many of those interactions are present within the region that would 292 overlap with nsp16 if the complexes were aligned. Indeed, nsp14 retains nsp10 binding properties and is 293 characterized by an affinity comparable to wild type nsp14, supporting the claim that the N-terminal region does 294 not significantly contribute to interaction. This allowed us to hypothesize that the heterotrimer complex is formed 295 when nsp16 displaces the N-terminal region ("lid") of nsp14 at the surface of nsp10 (Fig. 6). Indeed, both nsp14 296 and nsp14 readily form a heterotrimer complex with nsp16 and nsp10. stacking interactions with the exonuclease domain of nsp14 and the interaction may be described in terms of shape complementarity. In turn, the interaction of 300 the N-terminal "lid" region (amino acids 1-50) of nsp14 with nsp10 shows poor shape complementarity and is characterized only by a low number of solvent-301 exposed hydrogen bonds. b, Nsp16 binds nsp10 at the site overlapping that involved in binding of the "lid" of nsp14, but nsp10/16 interaction is characterized by

306
The "lid" hypothesis is supported by low-resolution structural data provided in this work. The gyration radii 307 and molecular weight of the heterotrimer complex derived from SEC-SAXS experiments are higher than the 308 gyration radii and molecular weights of any of the components or binary complexes (Extended Data Fig. 5a). The   Overall, in this study we provide evidence that nsp14, nsp10 and nsp16 form a heterotrimer complex 319 characterized by 1:1:1 stoichiometry, built around nsp10. The architecture of the complex follows the general 320 arrangement previously observed in nsp10/16 and nsp10/14 complexes, but nsp16 displaces the "lid" (N-terminal) 321 of nsp14 at the nsp10 surface (Fig. 6). The heterotrimer complex brings together two consecutive activities    For protein quantitation, sample separation was carried out following a simple protocol using the Prominence

404
To confirm protein content under every chromatographic peak taken for protein quantitation, mass spectrometry-405 based identification was used. The protocol for protein identification, applied with minor changes, is available 406 elsewhere 31 . Briefly, fractions acquired during protein separation were freeze-dried using CentriVap system

474
Out of 100 randomly chosen micrographs, 1'000 particles were manually picked without any structural knowledge 477 about the complex to minimize bias and assigned with 2D classes that were used in the ab initio model built. Out 478 of generated 3D classes, one was manually picked and used for the training of TOPAZ neural networks, which in 479 turn picked the next interaction of particles that were used to retrain Topaz. With this approach, approximately 480 0.5 M particles were selected to generate 50 2D classes, out of which 19 were manually picked. Collected micrographs were processed using cryoSPARC 3.1.1. Initially, 9'350 particles were picked from 497 micrographs using Blob Picker. Picked particles were subjected to a template-free 2D classification, from which 498 1'216 particles were selected and subjected to 3D reconstitution using the ab-initio reconstitution job. The 499 nsp10/14/16 complex map derived from SAXS data was used for a rigid-body fit in to 3D-reconstitution map using 500 Dock in map.