A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase.

Article


A Structure-Based Model for the Complete


Transcription Cycle of Influenza Polymerase


Graphical Abstract


Highlights
d Cryo-EM snapshots of the transcription elongation, termination, and recycling states d After being copied, the template 30 end rebinds the polymerase in a secondary site d Mechanism of viral mRNA poly(A) tail formation by stuttering elucidated d Efficient reformation of the promoter allows multiple transcripts from one RNP Wandzik et al., 2020, Cell 181, 1–17 May 14, 2020 ª 2020 Elsevier Inc. https://doi.org/10.1016/j.cell.2020.03.061 Authors Joanna M. Wandzik, Tomas Kouba,


Manikandan Karuppasamy, ..., Jan Provaznik, Nayara Azevedo, Stephen Cusack
Correspondence cusack@embl.fr In Brief Influenza polymerase transcribes the negative sense viral RNA genome into mRNA in the nucleus of infected cells. This work by Cusack and colleagues reports high-resolution cryo-EM structures of the polymerase at various stages of transcription providing a molecular basis for the complete transcription cycle, which should enable improved inhibitor design. Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061


Article


A Structure-Based Model for the Complete


Transcription Cycle of Influenza Polymerase


JoannaM.Wandzik,1,3 Tomas Kouba,1,3 Manikandan Karuppasamy,1 Alexander Pflug,1,4 Petra Drncova,1 Jan Provaznik,2 Nayara Azevedo,2 and Stephen Cusack1,5,*
1European Molecular Biology Laboratory, 71 Avenue des Martyrs, CS 90181, 38042 Grenoble Cedex 9, France 2European Molecular Biology Laboratory, GeneCore, Meyerhofstraße 1, 69117 Heidelberg, Germany 3These authors contributed equally 4Present address: Astra-Zeneca, Cambridge CB2 0AA, UK 5Lead Contact *Correspondence: cusack@embl.fr https://doi.org/10.1016/j.cell.2020.03.061


SUMMARY
Influenza polymerase uses unique mechanisms to synthesize capped and polyadenylated mRNAs from the genomic viral RNA (vRNA) template, which is packaged inside ribonucleoprotein particles (vRNPs). Here, we visualize by cryoelectron microscopy the conformational dynamics of the polymerase during the complete transcription cycle from pre-initiation to termination, focusing on the template trajectory. After exiting the active site cavity, the template 30 extremity rebinds into a specific site on the polymerase surface. Here, it remains sequestered during all subsequent transcription steps, forcing the template to loop out as it further translocates. At termination, the strained connection between the bound template 50 end and the active site results in polyadenylation by stuttering at uridine 17. Upon product dissociation, further conformational changes release the trapped template, allowing recycling back into the pre-initiation state. Influenza polymerase thus performs transcription while tightly binding to and protecting both template ends, allowing efficient production of multiple mRNAs from a single vRNP.


INTRODUCTION
The heterotrimeric influenza polymerase, with subunits PA, PB1, and PB2, binds to the conserved 50 and 30 termini, the ‘‘promoter,’’ of each of the eight negative sense, single-stranded viral RNA (vRNA) genome segments, which are packaged by the viral nucleoprotein (NP) into ribonucleoprotein particles (vRNPs). Following viral entry into the infected cell, vRNPs are transported to the nucleus where they associate with transcribing Pol II and the viral polymerase performs ‘‘primary’’ transcription to generate viral mRNA leading to viral protein synthesis (Eisfeld et al., 2015; Pflug et al., 2017; Te Velthuis and Fodor, 2016). Influenza polymerase employs unique mechanisms to synthesize 50 capped and 30 poly-adenylated viral mRNA. Capped oligomers of 10–15 nt long, pirated from nascent Pol II transcripts, are used to prime transcription (Lukarska et al., 2017; Plotch et al., 1981). Polyadenylation is achieved by stuttering of the polymerase on an oligo(U) stretch proximal to the template 50 end (Poon et al., 1999). The same vRNPs perform unprimed replication of vRNA into complementary cRNA, which is only stable when packaged into cRNPs using newly synthesized polymerase and NP (Vreede et al., 2004). cRNPs are the template for production of progeny vRNPs that can perform ‘‘secondary’’ transcription and/or be exported for packaging into new virions. Consistent with this, primary mRNA synthesis continues in cells infected in the presence of the translation inhibitor cycloheximide, but the stable synthesis of cRNA is inhibited due to the lack of NP and polymerase (Hay et al., 1980; Vreede et al., 2004). Furthermore, a single primary vRNP can perform transcription without additional viral components (Jorba et al., 2009), and continued accumulation of viral mRNA in the presence of cycloheximide (Vreede et al., 2004) shows that single vRNPs can perform multiple rounds of transcription. The implied efficient recycling mechanism is consistent with the need to rapidly generate large amounts of viral mRNA from the relatively small number of RNPs that initially infect the cell during a true infection (Russell et al., 2018). We have previously described how influenza polymerase accesses nascent host-capped transcripts by binding directly to the serine 5 phosphorylated CTD of Pol II (Lukarska et al., 2017), how rotation of the PB2 cap-binding domain allows the capped primer to be excised by the PA endonuclease and then directed into the polymerase active site to initiate RNA synthesis (Kouba et al., 2019; Pflug et al., 2018; Reich et al., 2014), and revealed the sequence of conformational changes that accompany the transition from the initiation to elongation phase of transcription (Kouba et al., 2019). Here, we present a detailed mechanistic model of the entire cycle of mRNA synthesis by influenza polymerase based on high resolution cryoelectron microscopy (cryo-EM) structures of actively transcribing bat influenza A (FluA) polymerase that cover the complete transcription cycle from pre-initiation, through elongation and termination, to product dissociation and template recycling. These structures give new insight into the template trajectory after passing through the active site, the mechanics Cell 181, 1–17, May 14, 2020 ª 2020 Elsevier Inc. 1 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 of poly-adenylation, and how template recycling likely occurs. In particular, we find that after translocation through the active site, the 30 extremity of the template docks into a partially buried binding site on the polymerase surface and remains sequestered there through termination and product dissociation. By tightly holding on to both extremities of the template throughout mRNA synthesis, the polymerase both protects the template from degradation and can efficiently recycle it, reform the promoter, and initiate the next cycle of transcription. The proposed model is fully compatible with transcription occurring, physiologically, in the RNP context, which necessitates NP uncoating and recoating of the incoming and outgoing template strands, respectively.


RESULTS


Transcription Elongation State of Bat FluA Polymerase in the Pre-catalytic State
To obtain snapshots of successive states of actively transcribing polymerase, we developed strategies to stall the polymerase after different amounts of template translocation (Figures 1A and 1B). Previously we used an extended 18+3-mer template, capped 15-mer primer, and 14-mer 50 vRNA activating ‘‘hook,’’ together with ATP and GTP, to stall influenza B (FluB) polymerase, through lack of CTP, after the addition of only five nucleotides to the primer (Kouba et al., 2019). Here, we used bat FluA polymerase with the same RNAs but added first ATP and GTP, and subsequently CTP together with a non-hydrolysable UTP analog, UpNHpp (Figure 1D). The template then translocates by 12 nt from the initiation state before stalling at the first adenine in the template sequence at position 15 from the native 30 end of the template. At this point, the principle product is a capped 26-mer, as confirmed by next-generation sequencing of the transcription reaction products (Figures 1D and S6). Cryo-EM analysis of this early elongation state confirms the biochemical characterization with the two major 3D classes clearly showing a template-product duplex with UpNHpp bound in a pre-catalytic configuration base-paired to A15 of the template at the +1 position (Figures 1B and S1; Table S1; Method S1B). In one 3D class, at 3.02 Å resolution (PDB: 6T0V), the cap-binding domain still associates with the polymerase core with the cap-proximal part of the primer bound to it. There is smeared density for the bulged-out part of the product mRNA connecting to the duplex. The second major 3D class is at 2.50 Å resolution (PDB: 6SZV) (Figure 1B;Method S1B), enabling ions, ribose hydroxyls, base substituents, and >210 water molecules to be resolved. In this structure, there is no density for the cap-binding domain, and only the part of the product in the product-template duplex within the active site cavity is visible. Comparison of this elongation state with the structure of the bat FluA pre-initiation state (PDB: 6T0N) (Figure 1B; Method S1A) shows that the initiation-elongation transition of bat FluA polymerase exhibits the same characteristic conformational changes as previously described for FluB polymerase (Kouba et al., 2019), confirming that the mechanism is likely general for all influenza polymerases (Figure S2). These changes include (1) complete extrusion of the priming loop and opening of the template exit channel, (2) promoter collapse involving 28 rota- 2 Cell 181, 1–17, May 14, 2020 tion of the PB1 b-ribbon, (3) refolding of PB1/667-681 and formation of a three-strand sheet with the b-ribbon (Figure S2A), (4) outward rotation of the PB1 thumb domain and associated PB2-N domains by 5 , (5) displacement of PB2/37-44 to allow duplex growth, and (6) duplex strand-separation by stacking of PB2 lid domain Tyr205 and PB1/Arg706 on the last duplex base pair (Figure S2B). The only difference with FluB polymerase is that in FluA, promoter collapse is correlated with a refolding and shift of the PA arch residues 372–385, bringing a cluster of acidic residues PA/D375, E377, and D378 into contact with PB1/K188 and R203 on the PB1 b-ribbon (all cited residues conserved in FluA) (Figure S2C). The FluA elongation structure allows visualization at high resolution of the pre-catalytic state with incoming non-hydrolysable UpNHpp with both magnesium ions, Mg(A) and Mg(B), configured for catalysis (Figure S3A). Compared to the inactive configuration (Kouba et al., 2019), Mg(A) shifts by 5 Å from the Mg(A0) position to coordinate the 30 hydroxyl of the priming nucleotide and the a-phosphate of the incoming NTP, with concomitant rotamer changes of the motif A and C aspartates, but otherwise no major structural changes occur (Figure S3A, right). This structure, together with the high resolution post-catalytic state (see below, Figure S3B), provides details that will be useful for drug development targeting the RNA synthesis active site but will not be described further here.


Template Exit Pathway
Compared to the previously described FluB early elongation state, in the FluA UpNHpp structure, the template has translocated an additional 7 nt beyond the end of the product-template duplex. Four of these are reasonably well-resolved in the cryoEM map, thus revealing the template exit path (Figures 2A and 2B). The exiting template runs in a narrow channel between residues 680-697 of PB1 helix a20 on one side, and the base of the extruded priming loop (PB1/632-634) and the PB2-N1 domain (PB2/54-110) on the other side. The floor of the groove is formed by extended peptide PB1/671-674 with basic residues K669, R670, and R672 contributing to its electro-positivity. PB1/ His633 stacks on U4, PB2/Arg88 is close to the phosphates of U4 and G3, and PB1/Arg687 is in proximity to the phosphates of C2 and U1. Prior to the transition to elongation, the exit groove does not exist (Figures 2A–2C). It is blocked first by the unextruded priming loop and second by the PB2/80-90 loop of the PB2-N1 domain, which is sandwiched between the PB1/ Arg670-Glu686 salt-bridge and PB1/Arg687 (Figures 2B and 2C). Its opening requires the concerted rigid-body movement of the thumb and PB2-N1 domains that accompanies the transition to elongation as well as an additional outward movement of the tip of the PB2/80-90 loop. This correlates with a reconfiguration of the PB1/671-674 peptide with change of salt-bridge partner of Glu686 from Arg670 to Arg672 (Figure 2B).


Late Elongation and Termination States
To determine the subsequent path of the template and to give insight into the mechanism of polyadenylation, we determined cryo-EM structures after initiating mRNA synthesis with a single-chain 44-mer mini-vRNA template in which the native 30 and 50 ends are connected by a loop. The template either had A B C D E Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 a 6U poly-adenylation (poly(A)) signal starting at the natural position 17 nt from the 50 end (v44-6U template), which is expected to promote poly(A) tail formation by stuttering, or the 6U sequence was replaced by UGUGUA (v44 template) to produce a defined, non-polyadenylated product (Figure 1A). The template sequences were chosen so that the first adenine occurred at position 16 (v44-6U) or 17 (v44) from the 50 end. For the v44-6U template, addition of all NTPs produced, as expected, a smear of polyadenylated products ranging in apparent size from around 50–200 nt (Figure 1E). In the case of the v44 template, ATP, GTP, CTP, and non-hydrolysable UpNHpp were added with the expectation that transcription would stall with UpNHpp base pairing with A17 at the +1 position, resulting in a capped 37-mer as main product (Figure 1A). Addition of all NTPs should increase the product size to a 38- mer, resulting from incorporation of U opposite template position A17. In fact, the major products appear to migrate at a size of around 55–60 nt (Figure 1E). Loading and running of control RNAs or transcription products on denaturing PAGE at different temperatures (Figure S4B) clearly shows that the aberrant mobility is due to formation of a high-melting point product-template duplex (Figure S4C). To further confirm the identity of the capped transcription products, we performed, after decapping, next-generation sequencing of the main product bands of the 18+3-mer and two v44 template transcription reactions (Figures 1D, 1E, boxed, and S6; Table S3). We find that of the reads that start with the primer sequence, around 60% end at the expected position with a minority of these exhibiting primer realignment resulting in one or two repeats of the AGC at the primer 30 end (Figure S6D). The isolated band from the v44 reactions also contains approximately 1:1 ratios of the template RNA (Table S3) supporting that it is hybridized to the product in the gel. The combined evidence of the PAGE migration analysis, next-generation sequencing and the sequence register deduced from the cryo-EM density (Figures S6A–S6C), thus unambiguously identify the observed transcription products. Interestingly, an apparent larger size product was previously observed for related viral transcription/replication systems (Pyle and Whelan, 2019; Vogel et al., 2019) and our observations now provide a plausible explanation of this phenomenon.


Transcription of the v44 Template Reveals a Secondary


30 End Binding Site
The sample with the v44 template reacted with ATP, GTP, CTP, and non-hydrolysable UpNHpp produced a major 3D class at 2.41 Å resolution, denoted ‘‘pre-termination’’ state (PDB: 6SZU) (Figure S1; Table S1; Methods S1C). This is a pre-translo- (D) Elongation products formed using bat influenza A polymerase with the cappe drolysable UTP analog (UpNHpp). Standards (RNAswith cap1) are indicated on th with or without addition of UpNHpp, as expected, but longer products, generated indicated. The sample for cryo-EM was pre-incubated with ATP and GTP, subseq freezing. This minimized unwanted products, which accumulate over longer time for RNA sequencing. (E) In vitro transcription assay using the 44-mermini-vRNA template with disrupted the capped 13-mer primer, endonuclease inhibitor, and NTPs as indicated, sho heterogeneous length (v44-6U). The products migrate slower than expected due boxes indicate bands whose equivalents were used for RNA sequencing. See also Figures 7, S4B, and S4C and Table S3. 4 Cell 181, 1–17, May 14, 2020 cation product complex with U18 at the +1 position base-pairing with A37 incorporated into the capped 37-mer product, together with pyrophosphate (Figures 1A, 1B, S3B, S5, and S6A; Video S1, part 1). However, we expected to observe UpNHpp opposite template A17 at the +1 position. This suggests that placing A17 in the +1 position is unfavorable, probably due to strain in the connection to the firmly bound 50 hook, but it can occur transiently, because when UTP is used instead of UpNHpp, an extra base is incorporated to form the capped 38-mer product (see above, Figures 1E, S4A, and S4B). Indeed, cryo-EM analysis of the v44 template incubated with all NTPs, visualized at 2.70 Å resolution, unambiguously shows U38 of the product at the +1 position opposite A17 in a complex we denote ‘‘termination’’ state (PDB: 6TW1) (Figure S6B; Methods S1D). These observations are discussed below in the context of the mechanism of polyadenylation. In the pre-termination state structure, there is excellent density for almost the entire 44-mer template RNA, allowing unambiguous assignment of the complete sequence (Figures S5 and S6A). Furthermore, the catalytic metal ions Mg(A) and Mg(B) are clearly visible (Figures S3B and S5C) as well as 275 ordered water molecules, mainly hydrating the proteinRNA interface (Figure S5D). The pre-termination state corresponds to translocation of the template by 24 nt (i.e., capped 13-mer primer extended to a 37-mer), with the template 30 end 27 nt from the +1 active site position. The important new features of this structure are, first, the well-ordered direct connection of the template from the 50 hook to the active site, and second, the outgoing template is observed to progress along the exit channel and bend around the PB2-N1 domain, with the five 30 terminal nucleotides binding into a partially buried groove between the PB1 thumb and PA-C domains, which we call the ‘‘secondary 30 end binding site.’’


Template EntranceConformation in the Pre-termination and Termination States
In the v44 pre-termination state, the template makes an almost direct connection between the 50 hook and the active site rather than the sinuous route taken by the 30 end of the promoter at the start of transcription (Figure 3A). The connector RNA has an irregular but highly ordered conformation that is stabilized by several highly conserved arginines, which both stack on bases and interact with phosphates (Figures 3B and S5B). The bases of A13 and C15 stack together and are sandwiched between PB1/Q186 and PB1/R203 of the collapsed b-ribbon on the A13 side and PB1/R353 on the C15 side. PB1/R353 also interacts with the phosphate of C15. The base of A16 stacks on the other d 15-mer primer with different combinations of ATP, GTP, CTP, and non-hye left and expected products on the right. The capped 26-mer product is formed by misincorporation (read-through) or primer realignment, are also present as uently CTP and UpNHpp were added to the reaction for 2 min (lane 13) before (lanes 14–17). The yellow box indicates the band whose equivalents were used (v44, left) or native 6xUpolyadenylation signal (v44-6U, right) in the presence of ws synthesis of a single main product (v44) and polyadenylated products of to robust hybridization with the complementary template (see text). The yellow A B C Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 side of PB1/R353 and also partially on the backbone of PB1/G33 as well as hydrogen bonding with PA/N508. The base of G14 is bulged out into a distinct pocket, stacking between the aliphatic part of PA/R507 and PB2/R38 and with the N7 hydrogen bonding to the main-chain of PB1/L675. PA/R507 also makes multivalent interactions with the phosphate of G14 and base and ribose of A11. PB1/R249 and R126 interact with the phosphates of A17 and U18, respectively, although the base of A17 is less well-ordered having no stacking interactions. The position of nt 1–14 from the 50 end are the same as in the elongation structure, and similarly for nt 16–18. However, in the elongation structure, the 50 end hook and the incoming template are not connected (mimicking the indirect linkage by a long loop of RNA for a physiological vRNA template), and in this case, the two segments of RNA are kept apart by a different position of the side chain of PB1/R353 (Figure 3C). In the termination state, base A17 moves into the +1 position, maximally straightening the connection to the 50 hook (Figure 3D). C15 pushes PB1/R353 to a position similar to the elongation state and A16 relocates to pre-termination position of A17.


The Secondary 30 End Binding Site
In the pre-termination and termination states, the outgoing template is guided along a positively charged groove around the PB2-N1 domain and into a partially buried extended pocket between the PA-C and PB1 thumb domains where the 30 extremity binds in a sequence-specific manner (Figure 4). For longer templates, this implies that as transcription proceeds, the emerging template would have to bulge away from the protein while the 30 extremity remains tightly bound in the secondary binding site. Indeed, in the termination state, there is no clear density for the nucleotides along the exit channel, probably due to disorder resulting from the extra base translocation of the template. Comparison of the UpNHpp elongation and pre-termination structures shows that in the first section of the exit groove, bounded by the extruded priming loop, PB2-N1 domain, and PB1 helix a20, different sequences and conformations of RNA can be accommodated as expected for a non-specific translocation channel (Figure 2B). The trajectory of the template follows the positively charged electrostatic surface of the groove that naturally guides the 30 extremity into the secondary binding site (Figure 4A). In this site, identical for both pre- and termination states, there are numerous interactions of PB1/553-571, PA/ 302, 351-352, 459-491, 577, and PB2/53-54 with the backbone and bases of the 5 terminal nucleotides (30-UCGUC) (Figure 4B). For example, the phosphates of nt 1–4 are contacted, respectively, by PB1/K553, PA/K483 and Y459, PB1/Y557 and R560, (B) Comparison of the template exit channel in the pre-initiation (left), elongation ( with positively charged residues (PB1/R670, R672, and R687), which together w initiation to elongation transition, PB1/E686 exchanges its partner from PB1/R67 template. In the pre-termination state, the RNA bases are disposed differently in th secondary 30 end binding site. (C) Surface representation showing changes in the exit channel occurring from th template exit channel is blocked by the stem of the priming loop and close inter elongation, the exit channel opens allowing the 30 end of the template to emerge the partially buried secondary binding pocket (black box). In the product dissoci (white arrow) but is disordered except in the secondary 30 end binding pocket. 6 Cell 181, 1–17, May 14, 2020 and PB1/Arg571 (Figure 4B). Aromatics PB1/Y557, PA/F352, and PA/Y459 each undergo induced fit rotamer flips upon RNA binding with PB1/Y557 and PA/F352 stacking on bases 2 and 3 (Figure 4C). Base-specific interactions are made, for example, by PA/R577 and T462 to C2 and PA/Q476 to U4. Most of the 30 end interacting residues are absolutely conserved among all FluA strains (except bat FluA-specific PA/F352), and many residues are functionally conserved in other influenza types (Figure 4D). Using fluorescence polarization to measure binding of labeled RNA 6-mers to bat FluA polymerase, we showed first that the wild-type 30-UCGUCU sequence has a Kd of 20 nM whereas swapping purines for pyrimidines in the first four positions (30-AGCACU) or an A6 sequence had 100-fold less affinity (Figures 4E and S7A). To confirm the site of binding and the importance of various interacting residues, we showed that single mutations of key residues PA/K483, PB1/K553 and R560, and PB2/K54 reduced binding by 15- to 100-fold and double and triple mutants by 200-fold (Figure S7B), with no effect of the mutations on protein integrity as judged by the intrinsic RNA synthesis activity of the polymerase (Figure S7C) (Reich et al., 2017). By cryo-EM, we confirmed that the site and mode of binding is conserved in FluB polymerase (PDB: 6T0W) (Figure S7E; Table S1; Methods S1G). Importantly, our high resolution maps show unambiguous density for vRNA 30 end nt 1–5 bound specifically in the secondary binding site (Figure 4F). In contrast, in a lower resolution structure of influenza D polymerase (Peng et al., 2019), vRNA nucleotides positions 2–6 were assigned to the binding site, even though in the correspondingmap there is no density for the implied overhanging first nucleotide and the last visible ribose is orientated such that no prolongation is possible (as in the Bat FluA structure). Interestingly, other recent cryo-EM structures for influenza D (Peng et al., 2019), FluB, and A/H3N2 (Fan et al., 2019) polymerases, show that the cRNA 30 end can also bind in this site. In all these cRNA bound structures, the secondary binding site nucleotides are assigned to positions 4–8 of the cRNA. However, we have high resolution cryo-EM maps showing unambiguous binding of cRNA nt 1–5, 30-UCAUC, in the secondary binding site of bat A polymerase (unpublished data). In summary, our structural and biochemical results are consistent with the secondary binding site being specific for the conserved 30 extremity of vRNA and cRNA for all influenza strains with consensus sequence 30-U1C2A/G3U4C5 (exceptionally FluD has an adenine at position 5 from the vRNA 30 end) (Figure S7D). Finally, we note that the influenza polymerase secondary 30 end binding site corresponds precisely to that observed for 30 end vRNA binding to La Crosse bunyavirus (LACV) polymerase (Gerlach et al., 2015) (Figure S7F, left). For LACV, the binding is also both sequence-specific middle), and pre-termination states (right). The floor of the exit channel is lined ith PB2/K82 and R88 accommodate the emerging RNA template. In the pre0 to R672. PB1/H633 changes orientation to stack on a base of the emerging e exit groove and the template curves around the PB2-N1 domain to bind in the e pre-initiation to the product dissociation states. In the pre-initiation state, the action of PB2-N1 and PB1/C-ter domains (black box). During the transition to on the polymerase surface. As elongation progresses the 30 end navigates into ation state, the template (dotted) remains in place in the widened exit channel A B C D Figure 3. Structure of the Incoming Template in the Pre-termination and Termination States (A) Superposition of incoming RNA moieties in the pre-initiation complex (50 vRNA in pink, 30 vRNA in orange) with the pre-termination state (vRNA in yellow, mRNA product in blue, magnesium ions in green) shows the direct connection of the 50 hook to the active site at pre-termination compared to the indirect path taken by 30 vRNA in the pre-initiation state. (B) Protein-RNA contacts stabilizing vRNA residues A13-G19 in the pre-termination state by interacting with the phosphate backbone (e.g., PA/R507, PB1/ R126, R203, R249, R353, and R365), bases (e.g., PA/H505, N508 and PB1/Q186, and K229) or ribose moieties (e.g., PA/R507 and PB1/R353). (C) Superposition of incoming RNA moieties in the pre-termination state (vRNA in yellow except for 50 hook in pink, product mRNA in blue) with the elongation state (gray). Whereas in pre-termination the RNA is continuous, throughout elongation, the 50 and 30 ends are not directly connected (gray ends in inset). A switch in the orientation of PB1/R353 either stabilizes the direct connection in pre-termination (see C) or allows the RNA to loop out between the stably bound 50 hook and template translocating into the active site. (D) Comparison of the incoming template path in the pre-termination state (vRNA in yellow except for 50 hook in pink, product mRNA in blue), with U18 at the +1 position, and the superposed termination state (gray), with A17 in the +1 position. The RNA conformation between G14 and A16 is straightened to allow A17 to reach the active site with a concomitant adjustment of PB1/R353 to allow C15 to reposition. See also Figures S5 and S6. Cell 181, 1–17, May 14, 2020 7 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 A B C D E F Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 and with the extreme 30 end ribose bound to preclude continuation. As discussed below, the sequestration of the 30 end during active transcription now provides a rationale for the existence of this site. The ‘‘Stuttering’’ State with the v44-6U Template Gives Insight into Polyadenylation A sample of the v44-6U template incubated with all NTPs was analyzed by cryo-EM. In this poly-adenylating sample, one of the 3D classes, at 3.04 Å resolution (PDB: 6T0S) (Table S1; Methods S1E) resembles the v44 template termination state, but, according to the base assignment in the duplex (Figure S6C), the polymerase has stalled after transcription of the U17-U22 polyadenylation signal, but prior to poly(A)tail elongation (Figures 1A, 1B, 5A, and 5B). The register of the product strand is the same as in the termination state but the template strand is distorted and the connection between 50 end nt 13 and 17 is not visible. We refer to this as the ‘‘stuttering’’ state. There is weak density for A38 incorporated into the product at the +1 position. However, the U17A38 base pair is not formed. Instead, template base U17 is at the +2 position and U18 is at the +1 site forming a tilted base pair with A37 of the product (Figure 5B). Indeed, the proximal part of the template-product duplex is stretched and tilted compared to that in the v44 termination state, before regaining the same geometry at the top of the duplex (Figures 5A and 5B). As in the termination state, the template in the exit channel is poorly ordered, whereas its extreme 30 end is clearly bound in the secondary 30 end binding site. We interpret the v44-6U stuttering structure as partially translocated. The distortion in the duplex backbone structure during translocation is accompanied by a concerted longitudinal shift in PB1 helix a8 276–292 and the preceding turn (PB1/ 271-275) as well as PB1/124-128, both of which abut the template backbone (Figure 5C). Moreover, Mg(A) is observed in the shifted position, Mg(A0), coordinated by PB1/D446 and E491 (Figure 5B). Based on these structures, a model for the mechanism of polyadenylation is proposed in the discussion (Figure 5D).


Product Dissociation State
The v44 pre-termination and v44-6U samples each have a second 3D class of interest at, respectively, 3.12 (B) Protein-RNA interactions contributing to the specific binding of the five 30 term formed at the interface of PA (green), PB1 (blue), and PB2 subunits (red) in bat in K483, PB1/K553, R560, R571, and PB2/K54) or interact specifically with RNA ba (C) To accommodate the 30 end RNA, three aromatic residues (PA/F352, Y459, an gray to colors) to stack with RNA bases (PA/F352 and PB1/Y557 with U4 and C2 ribose of C2 and PB1/Y557 with the phosphate of G2). (D) Sequence alignment of selected regions of the PA, PB1, and PB2 subunits that Interacting residues that are strictly conserved in influenza A–D strains are highlig F352 (green), stacking on vRNA base U4, is specific for bat polymerase. Residue (E) The 30 vRNA terminal nucleotide positions one to five bind sequence specificall swap at positions one to four of the conserved 30-U1C2Py3U4Pu5 sequence con binding. The graph shows binding curves of FAM-labeled WT (30-UCGUCU, g measured by fluorescence polarization. Data shown is for n = 4 ± SD. (F) Cryo-EM density for the 30 end of the vRNA template in the pre-termination s See also Figure S7. (PDB: 6T0U) and 2.82 Å (PDB: 6T0R) resolution (Table S1; Methods S1C and S1E). A very similar class is also observed for the v44 termination state (data not shown). All these structures clearly show the 50 proximal part of the template entering the active site (e.g., up to U17 in the v44-6U case) and the extreme 30 end of the template bound in the secondary binding site as in the termination and stuttering structures, but lack density for the product-template duplex in the active site cavity (Figures 6A and 6B). We interpret this as the ‘‘product-dissociation’’ state in which the three-stranded beta-sheet together with the clamped down b-ribbon remains formed, the priming loop is fully extruded, and the template stays threaded through the active site cavity, but the product mRNA has dissociated (Figures 1A, 1B, and 6A). Due to the lack of a well-defined duplex imposing an ordered structure, there is only low resolution density for the template passing through the active site cavity and exit channel, but that in the secondary binding site is unambiguous (Figure 6B). A major difference between the product-dissociation state and the preceding elongation and termination states (which are globally similar) is a further rotation by 6 , but about a different axis, of the entire thumb domain (PB1/520-669) and rigidly associated regions of PA-C and PB2-N (PB2/43-250), while preserving template binding through the exit channel into the secondary 30 end site (Figure 6C). This we call the ‘‘dislocated’’ state of the polymerase, which is characterized by refolding of the PB1/500-513 peptide immediately after conserved polymerase motif E at the start of the thumb domain (Figures 6C and S8A–S8C). This peptide (500- GFVANFSMELPSFG-513) is highly conserved in all influenza polymerases (Figure S8B). In the dislocated state, a characteristic third beta-strand is formed by PB1/501-503 with the preceding motif E hairpin loop (Figures 6C and S8C). Furthermore, the new position of PB1/S506-M507-E508 would clash with the product strand in the duplex (Figure 6C) and PB1/ R126 and the PB1/521-536 helix a16 with the template strand, emphasizing that this state is unable to bind a duplex. The two major transitions that occur during transcription, from (pre-) initiation to elongation and from termination to product dissociation (i.e., dislocation), are complex, having distinct local and more global components, as shown in Figure S8D and Video S1 (part 2). inal residues (yellow, numbering from 30 end) in the secondary binding pocket fluenza A polymerase. Multiple residues coordinate RNA backbone (e.g., PA/ ses (e.g., PA/K351, T462, Q476, R485, R577, and PB2/M53). d PB1/Y557) undergo induced fit rotations (as visualized by the transition from , respectively) and hydrogen bond with the RNA backbone (PA/Y459 with the contribute to the RNA-protein interactions in the secondary 30 end binding site. hted in red, and those with similar chemical properties, in blue. Note that PA/ numbering is according to bat FluA/H17N10 strain. y to the 30 secondary binding site with high affinity. A pyrimidine (Py)-purine (Pu) sensus for vRNA/cRNA (see Figure S7D), or oligo(A) completely abolishes the ray), mutant (30-AGCACU, blue) and control (30-AAAAAA, red) 6-mer RNAs tructure at 2.41 Å resolution showing clear definition of each base. Cell 181, 1–17, May 14, 2020 9 A C B D (legend on next page) 10 Cell 181, 1–17, May 14, 2020 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061


Recycling State
Interestingly, we and others have observed the dislocated state of the polymerase together with the vRNA template 30 end bound in the secondary binding site in situations where the polymerase has never been actively transcribing. When either bat FluA or FluB resting polymerase-promoter complexes (i.e., in the absence of NTPs), with or without capped primer, are examined by cryo-EM, in addition to majority 3D classes corresponding to the un-dislocated, promoter-bound, pre-initiation state, there is always a small fraction of particles in the dislocated state, with the 30 end of the template bound in the secondary binding site (PDB: 6T2C) (Figures 6D and 6E; Table S1; Methods S1F). In these open and flexible structures, there is disorder in the C-terminal region of PB1 beyond 670 and the N-terminal region of PB2 up to residue 40 with the PB1-PB2 interface and endonuclease not being visible (this is also the case in recent A/H3N2 polymerase cryo-EM structures, e.g., PDB: 6QWL, 6QX3, and 6QX8) (Fan et al., 2019). Furthermore, whereas the 50 end side of the promoter duplex is reasonably well-defined, the 30 end side is less so, and the connection to the secondary site is also poorly ordered. An exception is the recent FluD polymerase structure (PDB: 6KUR), which is also dislocated (Figure S8C), but where the connection between the template extremity bound in the secondary site to the promoter duplex region is clearly seen (Figure S7F, right). The authors refer to this promoter configuration as mode B, distinct from mode A, when the 30 end is in the active site as in the pre-initiation state (Peng et al., 2019). The RNA bound LACV structure (Gerlach et al., 2015) also exhibits mode B promoter binding (Figure S7F, left). Recent cryo-EM structures of dimeric (PDB: 6QX8) and monomeric (PDB: 6QX3) A/H3N2 polymerase, with or without bound cRNA promoter, are also in the dislocated state (Figure S8C). When polymerase is bound to the cRNA promoter, mode B exclusively predominates over mode A (Fan et al., 2019; Peng et al., 2019), probably because of the two nucleotide Figure 5. The Polymerase Stutters on U17 to Produce Poly(A) Tails (A) Sequences of the disrupted (v44) and native polyadenylation signal (v44-6U) termination/termination and stuttering structures, respectively. (B) Structure of the template-product duplex in the pre-termination (left), terminatio the bottomof the helix in the latter. In the pre-termination complex, regular base pa the +1 position base pairing with template U18. In the termination structure, the r position. In the stuttering complex, the product strand is partially translocated an pulled back to the +1 active site position while maintaining a base pair with A37 panels). (C) Superposition of the RNA duplexes in the stuttering (color) and termination st PB1/124-128 on the opposite side of the duplex. (D) The mechanism of polyadenylation with states visualized by structures as ind blue) without tension on the template linker (A13-U17) connecting to the 50 hook. T the interaction with PB1/R353 (Figure 3C). (2) The next translocation (visualized by U17 (pink) into the +1 position, forcing the A13-U17 linker to partially unfold fr maximally stretched A13-U17 linker (as seen in termination structure, Figure 3D) complementary with the template. (4) Motif B PB1/Met410 holds the product in t U17 backward (red arrow), from the +1 to the +2 position causing the lower half of stabilized by nine regular base pairs, nevertheless the A13-U17 linker is disordere duplex rupture the duplex with the product slipping by one base, extruding t from 9 to 10). Slippage is facilitated by the maintenance of five A-U base pairs The product +1 position is then vacant for another round of ATP incorporation. (6 active site +1 position, and upon incorporation of another adenine (7) the system See also Figures S4D–S4F. longer single-stranded region of the cRNA 30 end, whereas for vRNA binding, the majority are in mode A (Peng et al., 2019), as also observed for bat influenza polymerase. We interpret the dislocated, mode B promoter bound, vRNA structure as a critical recycling intermediate between the product disassociation state (also dislocated and highly flexible) and the mode A pre-initiation state in which the promoter is stably reformed (see Discussion).


DISCUSSION


Mechanism of Polyadenylation
Polyadenylation of influenza mRNAs occurs by ‘‘stuttering’’ at the 50 proximal oligo(U) motif (Li and Palese, 1994; Poon et al., 1999) and the pre-termination, termination, and stuttering structures described above allow us to propose a mechanism for exactly how this takes place, consistent with the known biochemistry (Figure 5D). The polyadenylation signal comprises a track of five or six, exceptionally seven, uridines always starting at position 17 from the 50 end of the vRNA template. The exact position from the 50 end is critical, because displacement of the oligo(U) tract forward by only one position dramatically decreases polyadenylation levels (Li and Palese, 1994; Poon et al., 1999). This we confirmed by using a v45-6U template, in which C17 was inserted so that the oligo(U) motif starts at position 18, in which case polyadenylation is eliminated (Figures S4D–S4F). The v44 pre-termination state structure reveals the path of the template RNA from the tightly bound 50 hook structure (A1–A10), via the sandwiched dinucleotide A11-G12 to the irregular but highly ordered 13–17 linker, allowing U18 to enter the +1 position in the active site (Figure 3C). This structure also shows that even in the presence of a non-hydrolysable NTP complementary to nucleotide 17, the corresponding pre-catalytic base pair is not observed at the +1 position (Figure 5D, 1). This suggests that it at position U17-U22 of the two 44-mer mini-vRNA templates used in the pre- n (middle), and stuttering (right) complexes showing non-standard geometry at irs are formed between template (yellow) and product strands (blue) with A37 at egister shifts by one position forward placing the U38-A17 base pair at the +1 d A38 occupies the +1 site without U17 opposite to it. At the same time U18 is , resulting in a tilted RNA duplex (compare green lines in left/middle and right ates (gray) showing a concerted longitudinal shift in helix PB1/a8 and the loop icated. (1) U18 is the last template position that can be transcribed (product in he specific linker conformation is stabilized bymultiple protein contacts such as movement of the penultimate brown base pair in the duplex) pulls (green arrow) om its highly structured configuration (dashed lines and pale colors). (3) The allows incorporation of ATP (black), which prolongs the oligo(A) stretch to 6 nt, he +1 position (green arrow) but the strain in the A13-U17 template linker pulls the RNA duplex to tilt. The stuttering structure visualizes this state which is still d (dashed lines and pale colors). (5) The opposing forces on each strand of the he base in position 9 out of the active site chamber (brown base moves but three mismatches are created in the upper half of the duplex (red crosses). ) Base-pairing with incoming ATP (gray) pulls U17 transiently back into to the then recycles to state (4). Cell 181, 1–17, May 14, 2020 11 A B C D E (legend on next page) 12 Cell 181, 1–17, May 14, 2020 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 is unfavorable for the template nucleotide at position 17 to enter the +1 position, because this requires disruption of the network of interactions that maintain the conformation of nt 13–16, in particular, PB1/R353 intercalating between A16 and C15 (Figures 3B and 3C). However, consistent with the observation that incorporation does occur at position 17 of the v44 template (Figure 1E), and a uridine at position 17 is essential for polyadenylation (Figures S4D and S4F), the termination structure shows that the position 17 base is able to dock into the +1 position despite the necessary stretching of nt 13–16 (Figure 3D). However, only if an incoming NTP can base pair with it, immediately proceeding to catalysis with incorporation into the product, is the energy cost compensated (Figure 5D, 2 and 3). Importantly, we could only trap the termination structure because the v44 template sequence results in a strong, G-C rich template-product duplex, which compensates for the maximal stretching of the linker to the 50 hook. In contrast, in the true polyadenylation situation, with the v44-6U template, there is a considerably weaker, A-U rich duplex, and after incorporation of the adenine opposite U17, the tension in the stretched template causes template backtracking. U17 is pulled out of the active site and U18 moves back into the +1 position, tilting its base pair with the adenine in the 1 position as well as the U:A base pairs involving U19-21 higher up the duplex (Figure 5D, 4). At this stage, the template-product duplex is shortened to nine base pairs and contains five, tilted A-U base pairs. The unpaired +1A product nucleotide stacks on PB1 motif B M410 but has yet to be translocated. This state corresponds to the partially translocated v44-6U ‘‘stuttering’’ structure that we observe (Figure 5B). The overall weakening of the duplex, together with the backward pull on the template and forward force on the product, promotes inter-strand slippage and product translocation by one base, leaving the +1 position on the product site vacant (Figure 5D, 5). The previously described conformational change in the motif B loop (Kouba et al., 2019) may assist this step. The next incoming ATP can then engage with U17, which is flipped transiently back into the active site for another round of incorporation into the growing poly(A) tail (Figure 5C, 6 and 7). Initial product slippage, however, introduces three mismatches in the upper half of the duplex (Figure 5C, 5–7), reducing significantly the number of regular base pairs. This facilitates template-product slippage in the next rounds of ATP incorporation but also increases the probability of product dissociation. Furthermore, as Figure 6. Product Dissociation and Recycling States (A) Superposition of RNAmoieties in the product dissociation (template in yellow w product in light blue). The position of the 50 hook is unaltered, but the template p bound in the secondary binding site (bracketed residue numbering indicates the (B) In the product dissociation state, densities for 30 (yellow) and 50 (pink) termini of bound by the polymerase, but residues U18 to U38 (indicated by dashed line) are formation of a stable duplex with the product strand. (C) The transition from the termination (gray and light blue for product strand) to t PA-C domain (e.g., a14 and a17), the PB1 thumb domain (e.g., a16, PB1/520-5 (inset). The PB1/504-508 peptide is repositioned and becomes incompatible with and residues PB1/501-504 form a characteristic third strand with the motif E bet (D) Cryo-EM structure of the recycling state for bat FluA (PDB: 6T2C) correspondin remains bound in the secondary binding pocket, but the viral promoter duplex is (E) As (D) but for human FluB polymerase (PDB: 6T0W). See also Figure S8. observed, if the oligo(U) tract is six nucleotides long, polyadenylation is more efficient, because there will be fewer mismatches in the duplex, favoring slippage and disfavoring product release. In summary, we propose a mechanism of stuttering and slippage in which U17 flips in and out of the active site +1 position, with the template in the active site cavity alternating between 9 and 10 nt in length between every ATP incorporation round without any net translocation. Thus, in contrast to regular translocation during elongation, at polyadenylation, only the product translocates out of the active site chamber.


Overall Model of Transcription
Based on our extensive, high resolution structural data showing successive snapshots of actively transcribing influenza polymerase, we propose a comprehensive and coherent model for the full transcription cycle of influenza polymerase (Figure 7; Video S2). The model provides a strong rationale for the existence of the secondary 30 end binding site, first observed in LACV polymerase (Gerlach et al., 2015) and subsequently for influenza polymerase (Fan et al., 2019; Peng et al., 2019). The initiation and transition to elongation steps have been described previously (Kouba et al., 2019) for FluB and confirmed here for FluA (Figure S2). The key elements are the growth of the template-product duplex, which correlates with disruption of the promoter, clamping down of the b-ribbon, opening of the PB1 thumb domain, unblocking of the template exit channel via extrusion of the priming loop and strand separation by the PB2 lid domain (Figures 7A–7C). Further elongation drives the template along a positively charged groove and after translocation of 24 nt, the leading extremity of the template docks into the 30 end secondary binding site in a sequence-dependent manner (Figure 7D). The template 30 end is sequestered by the polymerase during all subsequent steps of transcription, protecting it from degradation by cellular exonucleases. Thereafter, the exiting template is forced to bulge out in a loop that increases in size as transcription proceeds to termination (Figure 7E). Correspondingly, the incoming template loop between the bound 50 hook and the active site decreases in size until nucleotide 18 is at the +1 position and further translocation cannot occur without introducing strain in the linkage to the 50 hook (Figure 7F). Poly(A) tail synthesis then occurs as described above (Figure 5D). The exact determinants of poly(A) length and the potential involvement of host-factors in polyadenylation ith the 50 hook in pink) and the pre-termination structures (template in gray and ath toward the active site cavity changes. The 30 end of the template remains count from the 30 end). vRNA template are well defined indicating that the template extremities are still flexibly threaded through the active site cavity and disordered due to the lack of he product dissociation/dislocated state (bright colors) entails a rotation of the 36) and refolding of the PB1 peptide 500-513 adjacent to polymerase motif E the presence of RNA duplex (e.g., Met507 would clash with the product strand) a-hairpin (PB1/489-498). g to mode B vRNA promoter binding (Peng et al., 2019). In this state, the 30 end reformed. Cell 181, 1–17, May 14, 2020 13 A B C D E F G H I J K Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 remain unclear (Landeras-Bueno et al., 2011). At some point, the relatively weak, U-A rich, and mismatched product-template duplex dissociates, and the capped and polyadenylated viral mRNA is released. Duplex destabilization and product-dissociation is likely facilitated by coupling to a conformational switch of the polymerase into the dislocated state. This state is incompatible with duplex binding due to the reconfiguration of the postmotif E peptide and repositioning of PB1/521-536 helix such that they would clash with the product and template strands, respectively (Figure 7G). In the product-dissociation state, the 50 proximal part of the template forms a loop that is trapped inside a tunnel formed by the active site cavity and exit channel. The tunnel walls comprise the priming loop and PB2-N1 domain on one side and PB1/680700 and PB2/26-37 on the other side, both of which are connected to the PB2-Nter/PB1-Cter helical bundle that packs against the 170-loop of the PB2 helical lid domain, burying the template (Figures 7G and S8E). The template cannot be extracted without a dramatic conformational change of the polymerase. One way this may occur is for the PB2-Nter/PB1-Cter helical bundle to peel away, opening the tunnel and allowing the template loop to be released (Figure 7H). Indication that this is feasible comes from a crystal structure of dimeric A/ H7N9 polymerase core (Table S2) (PDB: 6TU5) in which the PB2-Nter/PB1-Cter helical bundle has rotated down as envisaged (Figure S8F). This would need to be followed by the PB2Nter/PB1-Cter helical bundle threading back through the template loop in a ‘‘skipping rope’’ fashion. The polymerase could then reconfigure into the still dislocated ‘‘recycling’’ state, with the 30 and 50 ends remaining bound in their respective sites, corresponding to mode B promoter binding, but with the observed weak promoter duplex (Figure 7I). Finally, the polymerase can switch back into the preferred pre-initiation state with mode A promoter binding. This is probably driven by polymerase closure (i.e., reversing the dislocation) and robust promoter duplex formation resulting in release of the 30 end from its secondary binding site. In the case of cRNA, which binds preferentially in mode B (Peng et al., 2019), dimer formation may be required to release the 30 end from the secondary binding site to facilitate cRNA to vRNA replication (Fan et al., 2019; Peng et al., 2019). (C) Template translocation initially occurs by straightening of the template entra Thumb rotation widens the active site cavity allowing growth of the product/templ helical lid, forcing strand separation. (D) The exiting template is guided along the positively charged exit groove and af remains bound there until the end of the transcription cycle. (E) Further translocation forces the exiting template RNA to loop out. (F) The polymerase processively transcribes the template until the incoming vRN and out of the active site, producing the poly(A) tail (see Figure 5D). (G) Due to mismatches in the A-U rich product-template duplex, the polyadenyla merase to the open, dislocated state. However, the template RNA is still threade (H) The template can be released, for instance, by an outward flip of the PB1-C/ skipping rope held at each end. (I) Once the vRNA is completely outside, the promoter can reform with the templa promoter formation, the polymerasemust transition back to the non-dislocated, p site and preferential binding in the active site. Another round of transcription can (J) In the vRNP context, NP molecules must be successively stripped off the inc (K) Because the transcribing polymerase sequesters roughly the same length of R (light brown NP(n)) before re-docking onto the growing loop of the exiting templa


Transcription in the vRNP Context
In the viral RNP, the vRNA genome is pseudo-circularized by promoter binding to the viral polymerase with the rest of the RNA forming a supercoiled helical loop coated by NP (Arranz et al., 2012). Influenza NP is believed to bind to 24–26 nt RNA (Hutchinson et al., 2014; Ortega et al., 2000), although recent studies show that the coating is not uniform andRNA stem-loops may extrude between the NPs (Le Sage et al., 2018; Lee et al., 2017). During transcription, as the template is progressively translocated through the polymerase active site, the NPs must successively peel off as the pre-transcription template RNA loop shrinks. We have now established that the outgoing, transcribed template gradually bulges out on the other side of the polymerase, between the exit channel and the secondary 30 end binding site, forming a correspondingly growing loop as transcription proceeds. It is therefore plausible that each successive RNA-stripped NP is transferred the short distance from the incoming to outgoing template strands where it can rebind to the growing loop of post-transcribed RNA (Figures 7J and 7K). This ensures that the transcribed RNA remains protected by bound NP, and no naked RNA is exposed to the nuclear environment. Interestingly, the amount of template RNA that is sequestered within the polymerase active site cavity, exit channel, and 30 end binding site, before the outgoing template begins to bulge out and is able to be rebound by NP, is around 25 nt, the same amount of RNA that can bind to one NP. This suggests that there is a docking site for at least one RNA-free NP before it can rebind the outgoing template strand (Figures 7J and 7K). This NP would be held in place by transient interactions with the polymerase as well as daisy-chain like NP-NP interactions with its neighbors (Ye et al., 2006). The growing loop of NP-coated, post-transcribed vRNA will presumably acquire a similar structural topology as the original super-helical genome segment. Therefore, images of a transcribing vRNP will likely show only a perturbation at the position of the polymerase as it appears to move from one end to the other, as very recently reported (Coloma et al., 2020). Finally, the conservation of the secondary 30 end binding site in both influenza and La Crosse virus suggests that its role to sequester the template extremity during transcription is nce pathway, but eventually the promoter melts and the b-ribbon collapses. ate duplex to the steady-state nine base pairs, before abutting against the PB2 ter translocating 24 nt, its extremity docks into the secondary 30 end site and A loop fully tightens up, whereupon the polymerase flips template base U17 in ted viral mRNA is eventually released, concomitant with a switch of the polyd through the polymerase, trapped inside a tunnel. PB2-N helical bundle that allows it to swish out of the active site cavity like a te 30 end still bound in the secondary site (recycling state). However, for robust re-initiation configuration, which forces release of the 30 end from the secondary then start. oming template (dotted) as it translocates into the polymerase. NA as binds to a single NP ( 25 nts), at least one NP must remain free of RNA te. Cell 181, 1–17, May 14, 2020 15 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 maintained in all segmented negative strand polymerases. Moreover, given that the cRNA and vRNA can both bind in the secondary 30 site and the template exit pathway is likely independent of RNA synthesis mode, it is plausible that the model proposed here for the trajectory of the template is the same for replication and transcription. STAR+METHODS Detailed methods are provided in the online version of this paper and include the following: d KEY RESOURCES TABLE d LEAD CONTACT AND MATERIALS AVAILABILITY d EXPERIMENTAL MODEL AND SUBJECT DETAILS d METHOD DETAILS B Site-directed mutagenesis B Bat FluA polymerase expression and purification B A/H7N9 polymerase core expression, purification, crystallization and structure determination B Endonuclease inhibitor B Cap-dependent polymerase transcription assays B Transcription product sequencing B Urea-PAGE migration analysis B Fluorescence polarization-based RNA synthesis assay B Fluorescence polarization-based RNA binding assay B CryoEM sample preparation and data collections B CryoEM data processing and structure determination B Oligonucleotides d QUANTIFICATION AND STATISTICAL ANALYSIS d DATA AND CODE AVAILABILITY SUPPLEMENTAL INFORMATION Supplemental Information can be found online at https://doi.org/10.1016/j. cell.2020.03.061.


ACKNOWLEDGMENTS
We thank FelixWeis andWimHagen for access to the Titan Krios at EMBLHeidelberg;Michael Hons,Wojtek Galej, and Erika Pellegrini for access to theGlacios at EMBLGrenoble; and Aymeric Peuch for help using the EMBL-IBS computer cluster. We thank the EMBL-ESRF Joint Structural Biology group for access to beamline ID30-A3. We thank the EMBL Heidelberg GeneCore (Vladimir Benes), the EMBL Grenoble eukaryotic expression facility (Alice Aubert), and Stefan Reich for help with fluorescence data analysis. This work was supported by ANR (ANR-18-CE11-0028 to S.C.). T.K. holds an EMBL Interdisciplinary Postdoc (EI3POD) fellowship, co-hosted byMartin Beck (EMBL Heidelberg) and co-funded by the Marie Sk1odowska-Curie Actions (664726, European Commission). This work used the platforms of the Grenoble Instruct-ERIC center (ISBG; UMS 3518 CNRS-CEA-UGA-EMBL) within the Grenoble Partnership for Structural Biology (PSB), supported by FRISBI (ANR-10-INBS-05-02) and GRAL, financed by the University Grenoble Alpes Ecoles Universitaires de Recherche CBH-EUR-GS (ANR-17-EURE-0003).


AUTHOR CONTRIBUTIONS
J.M.W. and P.D. expressed and purified bat FluA and FluB polymerase. J.M.W. performed and analyzed the FP-based assays. J.M.W. and T.K. performed transcription assays. M.K. pioneered cryo-EM structure determination of influenza polymerase and collected and analyzed the data on FluB polymerase. A.P. cloned, produced, crystallized, and solved the structure of the A/ 16 Cell 181, 1–17, May 14, 2020 H7N9 polymerase core. N.A. performed RNA sequencing. J.P. analyzed the sequencing results. J.M.W. and T.K. prepared cryo-EM grids, collected cryo-EM data, performed image processing and 3D reconstruction, and built initial models. S.C. conceived and supervised the project and refined atomic models. J.M.W., T.K., and S.C. prepared the manuscript and figures. P.D. made the schematic animation.


DECLARATION OF INTERESTS
S.C., T.K., and P.D. have a patent application related to this work. Received: November 4, 2019 Revised: February 18, 2020 Accepted: March 26, 2020 Published: April 17, 2020


KEY RESOURCES TABLE
REAGENT or RESOURCE SOURCE IDENTIFIER Bacterial and Virus Strains MultiBac baculovirus expression system Nie et al., 2014 N/A Chemicals, Peptides, and Recombinant Proteins GIBCO Express Five SFM GIBCO Cat.# 10486025 GIBCO L-Glutamine (200mM) GIBCO Cat.# 25030149 His60 NiNTA Superflow resin Clontech / Takara Bio Cat.# 635660 Strep-Tactin Superflow high capacity resin IBA Lifesciences Cat.# 2-1208-025 cOmplete, Mini, EDTA-free Protease Inhibitor Cocktail Roche Cat.# 11836170001 UpNHpp Jena Bioscience Cat.# NU-930S NTP set Thermo Fisher Cat.# R0481 Decade Markers System Thermo Fisher Cat.# AM7778 EasyTides [gamma-32P]-ATP Perkin Elmer Cat.# NEG502H-250UC EasyTides [alpha-32P]-ATP Perkin Elmer Cat.# NEG503H-250UC EasyTides [alpha-32P]-GTP Perkin Elmer Cat.# NEG506H-250UC RNA 50 Pyrophosphohydrolase (RppH) NEB Cat.# M0356S Monarch RNA Cleanup Kit (10 mg) NEB Cat.# T2030L Low Range ssRNA Ladder NEB Cat.# N0364S SYBR Gold Nucleic Acid Gel Stain Thermo Fisher Cat.# S11494 Influenza A/little yellow-shouldered bat/2010/H17N10 polymerase (wild-type) Pflug et al., 2014 N/A Influenza A/little yellow-shouldered bat/2010/H17N10 polymerase (PA/K134A) This study N/A Influenza A/little yellow-shouldered bat/2010/H17N10 polymerase (PA/K483E) This study N/A Influenza A/little yellow-shouldered bat/2010/H17N10 polymerase (PB1/K553E) This study N/A Influenza A/little yellow-shouldered bat/2010/H17N10 polymerase (PB1/Y557A) This study N/A Influenza A/little yellow-shouldered bat/2010/H17N10 polymerase (PB1/R560E) This study N/A Influenza A/little yellow-shouldered bat/2010/H17N10 polymerase (PB2/K54E) This study N/A Influenza A/little yellow-shouldered bat/2010/H17N10 polymerase (PA/K483E, PB1/K553E) This study N/A Influenza A/little yellow-shouldered bat/2010/H17N10 polymerase (PA/K483E, PB1/K553E, PB1/Y557A) This study N/A Influenza B/Memphis/13/03 polymerase (wild-type) Reich et al., 2014 N/A Endonuclease inhibitor 5-Hydroxy-4-oxo-1-{[1-(1Hpyrrolo[2,3-b]pyridin-1-yl)cyclopentyl]methyl}-1, 4-dihydro-3-pyridinecarboxylic acid. SMILES: N1(C2(CN3C = C(C( = O)C( = C3)O)C( = O)O)CCCC2) c2c(cccn2)C = C1 Miyagawa et al., 2016; Gift from Savira pharmaceuticals. Cat.# ENDO483 Critical Commercial Assays SMARTer smRNA-Seq kit Takara Bio USA Cat.# 635029 Agilent DNA High Sensitivity Bioanalyzer Kit Agilent Technologies Cat.# 5067-4626 Agilent RNA Pico 6000 Bioanalyzer Kit Agilent Technologies Cat.# 5067-1513 Qubit RNA HS Assay Kit Life Technologies Cat.# Q32852 (Continued on next page) e1 Cell 181, 1–17.e1–e9, May 14, 2020


Continued
REAGENT or RESOURCE SOURCE IDENTIFIER Qubit dsDNA HS Kit Life Technologies Cat.# Q32854 SPRIselect Reagent Kit Beckman Coulter Cat.# B23319 MiSeq Reagent Kit v3 (150-cycle) Illumina Cat.# MS-102-3001 Deposited Data Cryo-EM map and model of Bat Influenza A polymerase pre-initiation complex This study 6T0N (PDB); 10356 (EMDB) Cryo-EM map and model of Bat Influenza A polymerase elongation complex with incoming UTP analog (core + endonuclease only) This study 6SZV (PDB); 10355 (EMDB) Cryo-EM map and model of Bat Influenza A polymerase elongation complex with incoming UTP analog (complete polymerase) This study 6T0V (PDB); 10360 (EMDB) Cryo-EM map and model of Bat Influenza A polymerase pre-termination complex with pyrophosphate using 44-mer vRNA template with mutated oligo(U) sequence This study 6SZU (PDB); 10354 (EMDB) Cryo-EM map and model of Bat Influenza A polymerase product dissociation complex using 44-mer vRNA template with mutated oligo(U) sequence This study 6T0R (PDB); 10357 (EMDB) Cryo-EM map and model of Bat Influenza A polymerase termination complex with pyrophosphate using 44-mer vRNA template with mutated oligo(U) sequence This study 6TW1 (PDB); 10603 (EMDB) Cryo-EM map and model of Bat Influenza A polymerase stuttering complex using 44-mer vRNA template with intact oligo(U) sequence This study 6T0S (PDB); 10358 (EMDB) Cryo-EM map and model of Bat Influenza A polymerase product dissociation complex using 44-mer vRNA template with intact oligo(U) sequence This study 6T0U (PDB); 10359 (EMDB) Cryo-EM map and model of Bat Influenza A polymerase recycling complex This study 6T2C (PDB); 10368 (EMDB) Cryo-EM map and model of Human Influenza B polymerase recycling complex This study 6T0W (PDB); 10361 (EMDB) Crystal structure and model of Influenza A/H7N9 polymerase core (apo) This study 6TU5 (PDB) RNA sequencing data, sample v44 (+AGC) This study SAMEA6563093 (ENA) RNA sequencing data, sample v44 (+AGCU) This study SAMEA6563094 (ENA) RNA sequencing data, sample v30 RNA 18+3 (+AGC) This study SAMEA6563095 (ENA) Experimental Models: Cell Lines Trichoplusia ni High Five Cells Invitrogen Cat.# B85502 Oligonucleotides (see the STAR Methods for all oligonucleotides used) N/A N/A Recombinant DNA Expression plasmid for bat Influenza A polymerase (wild-type) Pflug et al., 2014 pKL-PBac-BatPol Expression plasmid for bat Influenza A polymerase (PA/K134A) This study pKL-PBac-BatPol-PA-K134A Expression plasmid for bat Influenza A polymerase (PA/K483E) This study pKL-PBac-BatPol-PA-K483E Expression plasmid for bat Influenza A polymerase (PB1/K553E) This study pKL-PBac-BatPol-PB1-K557E (Continued on next page) Cell 181, 1–17.e1–e9, May 14, 2020 e2 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061


Continued
REAGENT or RESOURCE SOURCE IDENTIFIER Expression plasmid for bat Influenza A polymerase (PB1/Y557A) This study pKL-PBac-BatPol-PB1-Y557A Expression plasmid for bat Influenza A polymerase (PB1/R560E) This study pKL-PBac-BatPol-PB1-R560E Expression plasmid for bat Influenza A polymerase (PB2/K54E) This study pKL-PBac-BatPol-PB2-K54E Expression plasmid for bat Influenza A polymerase (PA/K483E, PB1/K553E) This study pKL-PBac-BatPol-2xmut Expression plasmid for bat Influenza A polymerase (PA/K483E, PB1/K553E, PB1/Y557A) This study pKL-PBac-BatPol-3xmut Expression plasmid for human Influenza B polymerase (wild-type) Reich et al., 2014 pKL-PBac-FluBPol Expression plasmid for Influenza A/H7N9 polymerase core (PA 201-716, PB1 full-length, PB2 1-127) This study pFastBacDual-H7N9-core Software and Algorithms RELION 3.0 Scheres, 2012 N/A MotionCorr Zheng et al., 2017 N/A Gctf Zhang, 2016 https://www.mrc-lmb.cam.ac.uk/kzhang Gautomatch N/A https://www.mrc-lmb.cam.ac.uk/kzhang WARP Tegunov and Cramer, 2019 http://www.warpem.com/warp/# Coot Emsley et al., 2010 https://www2.mrc-lmb.cam.ac.uk/ personal/pemsley/coot PHENIX Afonine et al., 2018 https://www.phenix-online.org CCP-EM Burnley et al., 2017 https://www.ccpem.ac.uk LocScale Jakobi et al., 2017 N/A AUTOPROC/STARANISO Tickle et al., 2018 N/A PHASER McCoy et al., 2007 N/A Refmac5 Murshudov et al., 1997 N/A UCSF Chimera Pettersen et al., 2004 https://www.cgl.ucsf.edu/chimera PyMOL Schrödinger https://pymol.org/2/ CCP4MG McNicholas et al., 2011 https://www.ccp4.ac.uk/MG APBS Jurrus et al., 2018 N/A LigPlot+ Laskowski and Swindells, 2011 N/A SnapGene SnapGene https://www.snapgene.com:443/ Cutadapt v 2.3 Martin, 2011 N/A NR-grep v 1.1.2 Navarro, 2001 https://users.dcc.uchile.cl/ ~gnavarro/software/ GNU grep GNU https://www.gnu.org/software/grep/ manual/grep.html Je demultiplexer Girardot et al., 2016 https://gbcs.embl.de/portal/tiki- index.php?page=Je R version 3.5.1 R Foundation for Statistical Computing https://www.R-project.org cowplot 0.9.4 R package https://cran.r-project.org/web/ packages/cowplot/index.html ggplot2 3.2.0 R package https://cran.r-project.org/web/ packages/ggplot2/index.html readr 1.3.1 R package https://cran.r-project.org/web/ packages/readr/index.html (Continued on next page) e3 Cell 181, 1–17.e1–e9, May 14, 2020 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061


Continued
REAGENT or RESOURCE SOURCE IDENTIFIER GraphPad Prism GraphPad https://www.graphpad.com UNAFold Markham and Zuker, 2008 http://unafold.rna.albany.edu WebLogo Crooks et al., 2004 https://weblogo.berkeley.edu/ Other Quantifoil R 1.2/1.3 Au 300 Quantifoil N/A HiTrap Heparin HP 5 mL column GE Healthcare Cat.# 17040703 Superdex 200 Increase 10/300 GL GE Healthcare Cat.# 28990944 Amicon Ultra-15 (50 kDa cut-off) Merck Millipore Cat.# UFC905024 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061


LEAD CONTACT AND MATERIALS AVAILABILITY
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Stephen Cusack (cusack@embl.fr). All unique/stable reagents generated in this study are available from the Lead Contact with a completed Materials Transfer Agreement.


EXPERIMENTAL MODEL AND SUBJECT DETAILS
Trichoplusia ni HighFive cells (Invitrogen) cultured in GIBCO Express Five TM SFM media supplemented with GIBCO L-glutamine were used for recombinant protein expression.


METHOD DETAILS


Site-directed mutagenesis
Mutations were introduced in influenza A/little yellow-shouldered bat/2010/H17N10 polymerase (bat FluA), including single mutants (PA/K134A, PA/K483E, PB1/K553E, PB1/Y557A, PB1/R560E and PB2/K54E), a double (PA/K483E + PB1/Y553E) and a triple mutant (PA/K483E + PB1/K553E + PB1/Y557A), by site-directed mutagenesis. Introduction of desired mutations was confirmed by sequencing of the entire polymerase gene. Bat FluA polymerase expression and purification The bat FluA polymerase self-cleaving polyprotein construct was expressed and purified as previously described (Nie et al., 2014; Pflug et al., 2014) unless stated otherwise. Briefly, wild-type and mutant polymerase were expressed in HighFive insect cells using the baculovirus expression system. Cells were collected by centrifugation, re-suspended in lysis buffer (50 mM Tris-HCl pH 8, 500 mM NaCl, 10% glycerol, 2 mM TCEP) supplemented with protease inhibitor (Roche, complete mini, EDTA-free) and disrupted by sonication followed by centrifugation (35 000 rpm, 35 min, 4 C) to yield cleared supernatant. The latter was subjected to ammonium sulfate precipitation (final concentration 0.5 g/mL) and the pellet was recovered by centrifugation (30 000 rpm, 30min, 4 C, rotor 45Ti, Beckman Coulter). The pellet was dissolved in lysis buffer and subjected to a final centrifugation step (35 000 rpm, 35min, 4 C). Cleared supernatant was incubated with nickel resin (His60 NiNTA, Clontech) and protein was eluted with lysis buffer supplemented with 400 mM imidazole. The sample was diluted to decrease imidazole concentration to 250 mM and subjected to affinity chromatography on Strep-Tactin resin (Superflow, IBA). Protein was eluted with lysis buffer supplemented with 2.5 mM d-desthiobiotin, diluted to 250 mM salt concentration and loaded on heparin column (HiTrap Heparin HP, 5 mL, GE Healthcare). Elution was performed with a 25% to 100% gradient of buffers A (50 mM HEPES pH 7.5, 5% glycerol, 2 mM TCEP) and B (buffer A supplemented with 1 M NaCl). At this stage, monomeric and RNA-free wild-type polymerase fractions were pooled, concentrated (50 kDa cut-off, Amicon), flash-frozen and stored at 80 C (Figure S2D). In the case of mutant polymerases an additional step was introduced. Pooled fractions were concentrated and injected onto size exclusion column (Superdex 200 Increase 10/300, GE Healthcare) equilibrated in buffer C (50 mM Tris-HCl pH 7.5, 500mMNaCl, 5% glycerol, 2 mM TCEP). Monomeric and RNA-free polymerase fractions were pooled, concentrated (50 kDa cut-off, Amicon), flash-frozen and stored at 80 C. A/H7N9 polymerase core expression, purification, crystallization and structure determination PA, PB1 and PB2 of influenza A/Zhejiang/DTID-ZJU01/2013(H7N9) were expressed from synthetic codon-optimized reading frames (DNA2.0) corresponding to GenBank (Uniprot) entries AGJ51952.1 (M9TI86), AGJ51960.1 (M9TLW3) and KJ633805.1 (X5F427), respectively. A co-expression construct for the A/H7N9 core, was generated using the baculovirus expression vector pFastBacDual Cell 181, 1–17.e1–e9, May 14, 2020 e4 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 (Thermo Fisher) as follows. The complete PB1 protein coding sequence was inserted into the PolH-MCS using the restriction sites BamHI and RsrII. The sequence encoding residues 1-127 of PB2, supplemented with a C-terminal TEV-cleavable poly-histidine tag (GSGSENLYFQGSHHHHHHHH) was inserted into the P10-MCS using the restriction sites BbsI and XhoI. The sequence encoding for residues 201-716 of PA was cloned first into the vector pACEBac via the restriction sites BamHI and EcoRI, then amplified from this construct including the PolH promoter and SV40 polyA signal and subcloned into the pFastBacDual_PB1_PB2 construct using a unique AvrII restriction site and SLIC cloning technology. The core of A/H7N9 polymerase (PA 201-716, PB1 full-length, PB2 1-127) was produced in HighFive insect cells using the baculovirus expression system. Cells were lysed by sonication in buffer A (50 mM Tris-HCl pH 8.0, 500 mM NaCl, 10% (v/v) glycerol), cell debris was spun off (30 min, 4 C, 35000 g) and ammonium sulfate added to the supernatant (final concentration 0.5 g/ml) to precipitate the protein. The precipitated protein was collected by centrifugation (30 min, 4 C, 35000 g), re-suspended in buffer A and a cleared by centrifugation (30 min, 4 C, 35000 g) before subjecting it to immobilized metal ion affinity chromatography (IMAC). Elution fractions containing polymerase protein were pooled und subjected to a digestion with TEV protease (in buffer A supplemented with 5 mM b-mercaptoethanol). After dialyzing the digested protein sample back into buffer A it was passed through an IMAC column a second time. The sample was diluted to a salt concentration of 250 mMNaCl prior to loading it on a heparin column (HiPrep Heparin HP, GE Healthcare) pre-washed with buffer B (50 mMHEPES pH 7.5, 5% (v/v) glycerol, 2 mM TCEP, 150 mMNaCl). The protein was eluted via a salt gradient plateauing at 1MNaCl. Monodisperse and RNA-free polymerase was concentrated to about 9mg/ml, flashfrozen and stored at 80 C. A/H7N9 polymerase core crystallized within 4-5 days at 4 C in sitting drops in conditions of 0.1 M Tris pH 7.0, 8%–13% PEG 8K, 0.2 M MgCl2, 0.1 M guanidine hydrochloride with drop mixing ratios of 1:2 - 1:3 (protein:well) (Table S2). Diffraction data were collected at 100 K on beamline ID30A3 (MASSIF 3) at the European Synchrotron Radiation Facility (ESRF) and integrated and scaled with AUTOPROC with STARANISO (Tickle et al., 2018). Initial phases were obtained by molecular replacement using PHASER (McCoy et al., 2007) with the bat FluA polymerase structure (PDB: 4WSB). Structural adjustments were performed with COOT (Emsley et al., 2010) and refinement done with Refmac5 (Murshudov et al., 1997).


Endonuclease inhibitor
A potent endonuclease inhibitor, denoted ENDO483, was used as indicated below. ENDO483 is 5-Hydroxy-4-oxo-1-{[1-(1H-pyrrolo [2,3-b]pyridin-1-yl)cyclopentyl]methyl}-1,4-dihydro-3-pyridinecarboxylic acid (Miyagawa et al., 2016). The sample used was a gift from Savira pharmaceuticals. Cap-dependent polymerase transcription assays To test the transcriptional activity of polymerase toward synthetic vRNA templates used in this study, an in vitro transcription assay based on incorporation of radiolabeled nucleotides (a32P-ATP, Perkin Elmer) was performed. 0.8 mM bat FluA polymerase pre-bound to 35 mM endonuclease inhibitor ENDO483, 0.8 mM template RNA v44, v44-6U or v45-6U, 2 mM capped RNA primer, 100 mMNTPs and 12 nM a32P-ATP were incubated in cryoEM buffer (see below) at 28 C for 10, 30 min and 2 h. Samples were separated by 7 M urea, 20% polyacrylamide gel electrophoresis (urea-PAGE) in Tris-borate-EDTA (TBE) buffer, exposed to a phosphor storage screen and imaged with a Typhoon scanner. Decade marker system (Ambion) was used as RNA ladder. To test elongation stalling reaction conditions, 0.9 mMbat FluA polymerase pre-bound to 35 mMendonuclease inhibitor ENDO483, 0.81 mM template 18+3-mer 30 vRNA, 0.81 mM 14-mer 50 vRNA, 1.8 mM capped RNA primer and 100 mM ATP, GTP (and CTP or CTP and UpNHpp) and 12 nM a-32P-ATP were incubated in cryoEM buffer at 28 C for 5 h. Additionally, CTP or CTP and UpNHpp were added for 2, 4 or 10 minutes. Samples were separated and visualized by the later procedure. Decade marker system (Ambion) and 20-mer, 23-mer and 28-mer RNAs, 20-O-methylated at first ribose adjacent to the cap and radio-labeled at the 50 end were used as markers.


Transcription product sequencing
To confirm the identity of transcription products and address the discrepancy between expected size and urea-PAGE migration, sequencing of the main product was performed. First in vitro transcription assay was performed by incubating 0.8 mM bat FluA polymerase pre-bound to 35 mMendonuclease inhibitor ENDO483, 2 mM13-mer capped primer, 0.8 mM template RNA v44 and 100 mM ATP, GTP, CTP and without or with UTP in cryoEM buffer for 3h at 28 C. Analogous reaction was set up with 0.8 mM bat FluA polymerase pre-bound to 35 mMendonuclease inhibitor ENDO483, 2 mM15-mer capped primer, 0.8 mM template 18+3-mer 30 vRNA and 14-mer 50 vRNA and 100 mMATP, GTP andCTP for 3h at 28 C. After completion of the reaction, proteins were removed byMonarch RNACleanup Kit (NEB) and RNAs were then decapped using RNA 50 pyrophosphohydrolase (RppH, NEB) and subjected again to the Monarch RNA Cleanup Kit (NEB). Purified RNA were subjected to the urea-PAGE, stained with SYBR Gold (Thermo Fisher) and RNA was extracted from the bands of interest (shown on Figures 1D and 1E). RNA integrity was checked using the RNA Pico 6000 Assay Kit of the Bioanalyzer 2100 system (Agilent Technologies, Santa Clara, CA), and concentration was measured with Qubit RNA Assay Kit in Qubit 2.0 Fluorimeter (Life Technologies). Small RNA-Seq libraries were prepared manually from 2 ng of RNA as input using the SMARTer smRNA-Seq kit for Illumina Platforms (Takara Bio). Library preparation was done following the manufacturer’s instructions. The size distribution of the libraries was assessed on Bioanalyzer with a DNA High Sensitivity kit (Agilent Technologies), and concentration was measured with Qubit DNA High Sensitivity e5 Cell 181, 1–17.e1–e9, May 14, 2020 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 kit in Qubit 2.0 Flurometer (Life Technologies). Libraries that passed the QC step were pooled in equimolar amounts and final pool was purified with SPRI select beads (Beckman Coulter) in a 1.3x ratio. 8 pM solution of the pool of libraries was loaded on the Illumina sequencer MiSeq and sequenced uni-directionally, generating 18 million reads, each 150 bases long. Base-calling files were processed using Illumina bcl2fastq version 2.20.422 (Illumina, San Diego, USA). Resulting fastq reads were trimmed using Cutadapt v 2.3 (Martin, 2011) with settings recommended by the library preparation kit (-u 3 -a ‘AAAAAAAAAAAAAA’) omitting the filter for only reads longer than 25 bases. Trimmed reads were subsequently used as an input for pattern lookup performed by combination of NR-grep (Navarro, 2001) and GNU grep (https://www.gnu.org/software/grep/manual/grep.html). NRgrep allows for fuzzy search including substitutions, deletions and insertions. The list of search regular expression patterns and allowed substitutions are in Table S3. After the filtering, reads were sorted and unique occurrences were counted.


Urea-PAGE migration analysis
To investigate migration and level of denaturation of dsRNA, 2 mM 38-mer RNA (RNA38) and equimolar amount of its reverse complementary sequence (RNA38-RC) were incubated separately and together in cryoEM buffer for 15 min at room temperature. 2xRNA loading buffer was added to all samples (95% formamide, 0.025% SDS, 0.025% bromophenol blue, 0.025% xylene cyanol FF, 0.5 mM EDTA) and samples were boiled at 95 C and analyzed by 7 M urea 20% polyacrylamide electrophoresis at three different temperatures. Three gels were loaded and run in parallel either on ice ( 4 C), at room temperature ( 25 C) or with hot buffer ( 95 C) to comparemigration of RNA samples. Gelswere stainedwith SYBRGold stain (Thermo Fisher) and imaged after 15minutes. Decade marker system (Thermo Fisher) and low range ssRNA ladder (NEB) were used as markers. Fluorescence polarization-based RNA synthesis assay In order to test activity of newly produced polymerasemutants, an in vitro transcription assay based on fluorescence polarization (FP) and using short fluorescently labeled template vRNA was performed as previously described (Reich et al., 2017). Briefly, 0.25 mM wild-type or mutant polymerase was prebound to 1.2x excess of 50-vRNA (50-pAGUAGUAACAAGAG-30) and mixed with 1.5 mM capped RNA primer (50-m7GpppAAUCUAUAAUAGC-30), 0.3 mM FAM-labeled vRNA template (18-mer, 50-FAM-Ex-5-UAUACCU CUGCUUCUGCU-30) and 150 mM NTPs in assay buffer (50 mM HEPES pH 7.5, 200 mM NaCl, 10% (v/v) glycerol, 2 mM TCEP, 0.05% Tween20) supplemented with 5 mMMgCl2. Samples were incubated at 25 C and reaction was quenched by 4.5 M NaCl after 1, 1.5, 4, 7, 16, 20, 31, 45 min, 1h 30, 2h 15, 3h 30, 5h 30, 19h 30, 21h 30. After reaching equilibrium, fluorescence polarization was measured with a CLARIOstar microplate reader (BMG Labtech) at 25 C using excitation and emission filters of 485 and 520 nm, respectively. Progress curves were fitted into double exponential equation with bi-phasic reaction kinetics: fðtÞ = A$e k1t B$e k2t +C where t is reaction time,A andB are observed signal amplitudes, k 1 and k2 are observed rate constants for corresponding phases and C is final FP signal. Data were analyzed and fitted using GraphPad Prism. Fluorescence polarization-based RNA binding assay Binding assays were performed with short fluorescently labeled wild-type (50-UCUGCU-30), mutant (50-UCACGA-30) and negative control (50-AAAAAA-30) RNA (IBA Lifesciences). Binding reactions were carried with 0.02 mM FAM-labeled RNA and increasing concentration of wild-type or mutant polymerase in assay buffer (50 mM HEPES pH 7.5, 200 mM NaCl, 10% (v/v) glycerol, 2 mM TCEP, 0.05% Tween20). The binding reactions were set-up in 20 mL volume and incubated at room temperature for about 30 min. After reaching equilibrium, fluorescence polarization was measured with a CLARIOstar microplate reader (BMG Labtech) at 25 C using excitation and emission filters of 485 and 520 nm, respectively. Recorded fluorescence polarization signal was corrected for changes in fluorescence intensity according to the equation: fb = ðFP FPfreeÞ Ibound Ifree $ðFPbound FPÞ+FP FPfree where fb represents the fractional concentration of bound RNA , FP the observed fluorescence polarization, FPfree and FPbound the fluorescence polarization of free and bound RNA, and Ifree and Ibound the fluorescence intensity of free and bound RNA, respectively. Dissociation constants (KD) were obtained using GraphPad Prism software by fitting fluorescence polarization data as a function of total concentration of protein (P) using the following equation: fb = L+P+KD ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðL+P+KDÞ2 4LP r 2L Cell 181, 1–17.e1–e9, May 14, 2020 e6 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061 where f represents the fractional concentration of bound RNA, L b the total concentration of labeled RNA, P the concentration of protein and KD the dissociation constant. For the competition assay, 0.04 mM wild-type polymerase was pre-bound with an equimolar quantity of FAM-labeled short RNA (50-UCUGCU-30) and increasing amounts of unlabeled RNA competitor were added. Data were recorded and fitted in the same manner as for direct binding experiments. CryoEM sample preparation and data collections To capture distinct states of transcription cycle, state specific vRNA templates and sample preparation protocols (see below) were applied. Either wild-type polymerase with endonuclease inhibitor or an endonuclease-inactive polymerase (PA/K134A) were used to prevent possible endonuclease mediated vRNA template or product degradation. All complexes were assembled in cryoEM buffer (50 mM HEPES pH 7.5, 150 mM NaCl, 5 mM MgCl2, 2 mM TCEP). Sample 1: For the pre-initiation complex, equimolar amounts of bat FluA polymerase at 0.9 mM (prebound to 35 mMendonuclease inhibitor (ENDO-483)), 30 vRNA (16-mer) and 50 vRNA (14-mer) in cryoEM buffer were incubated on ice for 15 min. Sample 2: For the elongation complex, 0.9 mM bat FluA polymerase (pre-bound to 35 mM endonuclease inhibitor ENDO483), 0.81 mM 30 vRNA (18+3-mer) and 50 vRNA (14-mer) and 1.8 mM 15-mer capped were mixed in cryoEM buffer and incubated for 5 hours at 28 C with 100 mM ATP and GTP. Additionally, 100 mM CTP and 300 mM UpNHpp were added for 2 minutes at 28 C and immediately cooled to 4 C. Sample 3 and 4: For the pre-termination and termination complexes, equimolar amounts of bat FluA polymerase at 0.85 mM (pre-bound to 35 mMendonuclease inhibitor ENDO483) and the v44 template weremixedwith 2.5molar excess of 13-mer capped RNA primer, 100 mM ATP, CTP, GTP and 300 mM UpNHpp in cryoEM buffer and incubated at 28 C for 4 h. For the termination complex, 100 mM UTP instead of UpNHpp was used. Sample 5: For the stuttering complex, equimolar amounts of bat FluA PA/K134A polymerase at 0.7 mM and the template v44-6U (44-mer template RNAwith 6xU polyadenylation signal) were mixed with 2.5 molar excess of 13-mer capped RNA primer, 170 mM ATP and 17 mM CTP, GTP and UTP in cryoEM buffer and incubated at 30 C for 1.5 h. Sample 6: For the recycling complex, equimolar amounts of bat FluA polymerase at 0.9 mM (pre-bound to 35 mM endonuclease inhibitor ENDO483) and template vRNA7L (34-mer) in cryoEM buffer were incubated on ice for 20 min. Sample 7: For the FluB recycling complex, wild-type human FluB polymerase at 1.2 mMwas incubated with 1.25 molar excess of template vRNA7L (34-mer) and 12-mer capped RNA primer in cryoEM buffer. Aliquots of 3 ml were applied to glow discharged grids (R1.2/1.3, Au 300, Quantifoil), blotted for 2 s and immediately plungefrozen in liquid ethane using an FEI Vitrobot IV at 4 C and 100 % humidity. Grids were loaded onto 300 kV FEI Titan Krios or 200 kV FEI Talos Glacios electron microscopes and data were acquired in electron counting mode. Further details regarding each data collection are presented in Table S1. CryoEM data processing and structure determination CryoEM data processing was performed according to previously published strategy (Kouba et al., 2019) and additional details regarding each dataset can be found in Table S1 and Methods S1. Briefly, in order to limit radiation damage, only selected initial frames from the collected movies were aligned using MotionCorr2 (Zheng et al., 2017) and then used for contrast transfer function parameter calculation byGctf (Zhang, 2016). Particle were picked using eitherWARP (Tegunov andCramer, 2019) or Gautomatch (by K. Zhang) and were next subjected to reference-free two-dimensional (2D) classification using RELION 3.0 (Scheres, 2012). After several rounds of 2D classifications on binned images, particles in classes with poor structural features were discarded. Classes containing good quality particles were subjected to 3D classification with alignment. Next, the classes representing similar conformational stateswere pooled together for global 3D auto-refinement andwere subjected to another round of 3D classification with fine angular searches. This procedure was repeated until conformationally homogeneous classes were obtained. The resulting maps were improved by CTF refinement and Bayesian polishing procedures in RELION and further 3D auto-refined. The final cryoEM density maps were sharpened using post-processing procedure in RELION and the resolution was estimated using 0.143 gold standard Fourier Schell Correlation (FSC) cutoff. Local resolution maps were generated using RELION’s own implementation. Final maps were sharpened and blurred into MTZ format using CCP-EM (Burnley et al., 2017) and local amplitude scaling was performed using LocScale (Jakobi et al., 2017). The previously solved structures of bat influenza A polymerases (4WSB, but also models generated in this study) were rigid-body fitted into the EM densities using Molrep in CCP-EM and further model building of RNA and flexible domains was performed manually in COOT (Emsley et al., 2010). Real-space refinement was done in PHENIX (Afonine et al., 2018). For the highest resolution structures, Ramachandran and side-chain restraints were not used. Structures were analyzed and figures were prepared using following software packages: PyMOL (Schrödinger, Inc.) with APBS plugin (Jurrus et al., 2018), USCF Chimera (Pettersen et al., 2004), CCP4MG (McNicholas et al., 2011) and LigPlot+ (Laskowski and Swindells, 2011). e7 Cell 181, 1–17.e1–e9, May 14, 2020 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061


Oligonucleotides
Sequence Source Name (m7Gppp)AAUCUAUAAUAG TriLink Biotechnologies Capped primer (12-mer) (m7Gppp)AAUCUAUAAUAGC TriLink Biotechnologies Capped primer (13-mer) (m7Gppp)GAAUGCUAUAAUAGC TriLink Biotechnologies Capped primer (15-mer) pAGUAGUAACAAGAG IBA Lifesciences 50 vRNA (14-mer) UACCUCUGCUUCUGCU IBA Lifesciences 30 vRNA (16-mer) UAUACCUCUGCUUCUGCU IBA Lifesciences 30 vRNA (18-mer) UAUACCUCUGCUUCUGCUAUU IBA Lifesciences 30 vRNA (18+3-mer) pAGUAGUAACAAGAGCAAUGUGUCCGUCUCG CCUCUGCUUCUGCU IBA Lifesciences v44 (44-mer) pAGUAGUAACAAGAGCAUUUUUUCCGUCUCG CCUCUGCUUCUGCU IBA Lifesciences v44-6U (44-mer) pAGUAGUAACAAGAGCACUUUUUUCCGUCU CGCCUCUGCUUCUGCU IBA Lifesciences v45-6U (45-mer) pAGUAGUAACAAGAGGUAUUACCUCUGC UUCUGCU IBA Lifesciences vRNA7L (34-mer) UCUGCU IBA Lifesciences 30 vRNA (6-mer) UCACGA IBA Lifesciences 30 vRNA (mut) AAAAAA IBA Lifesciences 6A (FAM-Ex-5)-UCUGCU IBA Lifesciences 30 vRNA (6-mer), FAM-labeled (FAM-Ex-5)-AAAAAA IBA Lifesciences 6A, FAM-labeled (FAM-Ex-5)-UAUACCUCUGCUUCUGCU IBA Lifesciences 30 vRNA (18-mer), FAM-labeled (m7Gppp)-mO2GAAUCACAUAAAGACCAGGC (Kouba et al., 2019) 20-mer marker (m7Gppp)-mO2GAAUCUAUACAUAAAGACCAGGC (Kouba et al., 2019) 23-mer marker (m7Gppp)-mO2GAAUCUAUAAUAAACAUAAA GACCAGGC (Kouba et al., 2019) 28-mer marker pAGUAGUAACAAGAGCAUUUUUCA GACUCUGCUUCUGCU IBA Lifesciences RNA38 (38-mer)


AGCAGAAGCAGAGUCUGAAAAAUGC
UCUUGUUACUACU IBA Lifesciences RNA38-RC (38-mer)


GACTATTACTATGAAGCGGCTTCTAA
GCTCAAAGGAGAGAACG Thermo Fisher PA_K134A_fwd


CTTAGAAGCCGCTTCATAGTAATAGTCC
TCAACCTTCCGTCTGG Thermo Fisher PA_K134A_rev AATCATAACCGAATGTCGGGACACCTCAGG Thermo Fisher PA_K483E_fwd


GTCCCGACATTCGGTTATGATTGGGATTAC
TTGGTAC Thermo Fisher PA_K483E_rev


CAGCTGTTCATAGAGGACTACAGATATAC
CTATAGGTGC Thermo Fisher PB1_K553E_fwd TCTGTAGTCCTCTATGAACAGCTGTATGGCCATC Thermo Fisher PB1_K553E_rev TATACCTATGAGTGCCATCGCGGCGATACCAAC Thermo Fisher PB1_R560E_fwd


GCGATGGCACTCATAGGTATATCTGTAGTCCTTT
ATGAACAGC Thermo Fisher PB1_R560E_rev GACTACAGAGCTACCTATAGGTGCCATCGCGGCG Thermo Fisher PB1_Y557A_fwd


CCTATAGGTAGCTCTGTAGTCCTTTATG
AACAGCTGTATGGC Thermo Fisher PB1_Y557A_rev


CCTATAGGTAGCTCTGTAGTCCTCTATGAACAG
CTGTATGGC Thermo Fisher PB1_Y557A_rev2 ATGGCCATGGAGTATCCAATAAGCGCC Thermo Fisher PB2_K54E_fwd ATTGGATACTCCATGGCCATCATCCATTTCATCC Thermo Fisher PB2_K54E_rev Cell 181, 1–17.e1–e9, May 14, 2020 e8 Please cite this article in press as: Wandzik et al., A Structure-Based Model for the Complete Transcription Cycle of Influenza Polymerase, Cell (2020), https://doi.org/10.1016/j.cell.2020.03.061


QUANTIFICATION AND STATISTICAL ANALYSIS
Enzymatic and binding assays represent a mean of n independent experiments with standard deviation (SD) and the number of experiments (n) is indicated in legends to Figures 4E and S7A–S7C. Structural data were validated by Phenix validation tools and are presented in Tables S1 and S2.


DATA AND CODE AVAILABILITY
Co-ordinates and cryoEMmaps generated during this study are available at Protein Data Bank (PDB) and Electron Microscopy Data Bank (EMDB) under following accession numbers: d Bat Influenza A polymerase pre-initiation complex, PDB ID: 6T0N, EMDB ID: 10356 d Bat Influenza A polymerase elongation complex with incoming UTP analog (core + endonuclease only), PDB ID: 6SZV, EMDB ID: 10355 d Bat Influenza A polymerase elongation complex with incoming UTP analog (complete polymerase), PDB ID: 6T0V; EMDB ID: 10360 d Bat Influenza A polymerase pre-termination complex with pyrophosphate using 44-mer vRNA template with mutated oligo(U) sequence, PDB ID: 6SZU, EMDB ID: 10354 d Bat Influenza A polymerase product dissociation complex using 44-mer vRNA template with mutated oligo(U) sequence, PDB ID: 6T0R, EMDB ID: 10357 d Bat Influenza A polymerase termination complex with pyrophosphate using 44-mer vRNA template with mutated oligo(U) sequence, PDB ID: 6TW1, EMDB ID: 10603 d Bat Influenza A polymerase stuttering complex using 44-mer vRNA template with intact oligo(U) sequence, PDB ID: 6T0S, EMDB ID: 10358 d Bat Influenza A polymerase product dissociation complex using 44-mer vRNA template with intact oligo(U) sequence, PDB ID: 6T0U, EMDB ID: 10359 d Bat Influenza A polymerase recycling complex, PDB ID: 6T2C, EMDB ID: 10368 d Human Influenza B polymerase recycling complex, PDB ID: 6T0W, EMDB ID: 10361 Co-ordinates and structure factors generated during this study are available at Protein Data Bank (PDB) under following accession number: d Influenza A/H7N9 polymerase core (apo), PDB ID: 6TU5 RNA sequencing data generated during this study are deposited at the European Nucleotide Archive (ENA) under study accession number PRJEB36795 with following accession numbers for each sample: d Sample v44 (+AGC), ENA ID: SAMEA6563093 d Sample v44 (+AGCU), ENA ID: SAMEA6563094 d Sample v3018+3 (+AGC), ENA ID: SAMEA6563095 e9 Cell 181, 1–17.e1–e9, May 14, 2020


Supplemental Figures
Figure S1. Cryo-Electron Microscopy of Influenza Polymerase Conformations from Pre-initiation to Recycling States, Related to Figure 1 and Methods S1 A. Representativemicrograph at 165,000 nominalmagnification with defocus of 2.0 mm (left) and two-dimensional (2D) classes of influenza polymerase complex (right). Data are from the pre-termination state structure determination. (legend continued on next page) B. Distribution of local resolution of the maps for pre-initiation (left), elongation (middle) and pre-termination complexes (right) and a slice through the polymerase core for each map. Local resolution maps were calculated with RELION and resolution range is indicated according to the color bar. C. Fourier shell correlation (FSC) curves for all complexes described in this study. The plot of the FSC between two independently refined half-maps shows the overall resolution of the two maps as indicated by the gold standard FSC 0.143 cut-off criteria. Figure S2. Conformational Changes Underlying the Transition from the Pre-initiation to the Elongation States in Bat FluA Polymerase, Related to Figure 2 A. The transition from pre-initiation (gray) to elongation complex (color) is marked by the collapse of the PB1/b-ribbon with formation of a three-stranded b sheet together with refolded peptide PB1/667-681; extrusion of the priming loop (ruby) and shift of PB2/37-44 peptide (red) as previously observed in the FluB elongation complex (Kouba et al., 2019). B. The collapse of the PB1/b-ribbon is accompanied by shift of PA/arch that forms new interactions between PB1/K188, R203 with PA/D378 and D375, respectively. C. Residues involved in the RNA duplex strand separation are PB2/helical lid Tyr205 and PB1/Arg706, which stack on template (yellow) and product (blue) strands, respectively. (D). SDS-PAGE of purification of trimeric bat FluA polymerase.


Figure S3. Active Site in the Pre- and Post-catalytic States, Related to Figure 1
A. CryoEM density for UpNHpp moiety (red sticks) stalled in the pre-catalytic state in the elongation complex (left) and RNA-protein contacts in the catalytic site (right) with conserved RNA-dependent RNA polymerasemotifs colored accordingly: motif A (green), B (dark gray), C (pink), D (cyan), E (orange) and F (yellow) and RNA duplex between vRNA (yellow) and product strand (blue) with magnesium ions (green spheres) and water molecules (red spheres). Compared to the inactive configuration previously described (Kouba et al., 2019), Mg(A’) (gray sphere coordinated by gray residues) shifts by 5 Å (arrow) with repositioning of D446. B. CryoEM density for A37 incorporated into the product strand (blue) and leaving pyrophosphate (orange) in the post-catalytic state in the pre-termination complex (left) and RNA-protein contacts color coded as described in A (right). Figure S4. In Vitro Transcription by Bat FluA Polymerase Mini-vRNA Templates, Related to Figures 1D, 1E, and S6 A. Analysis of the transcription products formed using the v44 template, with disrupted polyadenylation signal, in the presence of endonuclease inhibitor ENDO483 and with addition of selected nucleotides as marked. Misincorporation and run-off transcription products are indicated. B. Urea-PAGEmigration pattern analysis. Three identical gels were loaded and run at different temperatures: on ice ( 4 C, left), at room temperature (middle) and with hot running buffer ( 95 C, right). Synthetic 38-mer (RNA 38, lane 1) and its reverse complement 38-mer (RNA 38-RC, lane 2) individually migrate at the expected size. Their mixture makes a stable RNA duplex (dsRNA-38) with high melting temperature (Tm = 92.8 C, see C). At temperatures low enough that dsRNA-38 is a stable duplex, an apparent double sized product ( 80-mer) is observed (left panel, lane 3, red box) even in conditions of 7 M urea PAGE. In contrast, at medium and high temperatures, dsRNA-38 is denatured and migrates as the individual ssRNA components (middle/right panels, lane 3, black box). The 44-mer template RNA (v44) alone migrates at the expected size (all panels, lane 4). The expected 37-mer (without UTP) or 38-mer (with UTP) capped transcription products can form fully complementary RNA duplexes of 27 or 28 base-pairs with the v44 template, with 10 nts primer and 17 or 16 nts template 50 overhangs, respectively (see C). As for themixture of the complementary 38-mer control RNA, the product-template hybrids have highmelting temperatures (97.8 and 98.3 C, respectively) and are retarded at apparent double the expected size at low and medium temperatures (left/middle panels, compare lane 3, red box and 5 and 7, red star). There is a shift inmigration of the product-template hybrid when the product is de-capped (all panels, compare lane 5with 6 and 7with 8). At high temperature, the product-template hybrid is melted (right panel, lanes 5-8) and the 37- and 38-mer transcription products migrate in accordance with their sizes. A minor run-off transcription product (54-mer) is present only for the sample where all NTPs were included (right panel, lane 7-8, green star). The capped and de-capped transcription products of the v30 18+3 RNA template were also visualized (middle panel, lanes A-B), the main product is the expected 26-mer (red circle) with minor products due to misincorporation/runoff (30-mer, black circle) and probable re-alignment products (green circle). C. Melting temperatures for dsRNAs calculated using the UNAFold server (Markham and Zuker, 2008) show strong hybridization for RNA-38 and its reverse complement and for v44 template RNA and its respective transcription product. D. Comparison of the polyadenylation signal sequence and position in the three mini-vRNA templates used in this study. E., F. Comparison of in vitro transcription products using the v44-6U (native polyadenylation signal) or v45-6U (polyadenylation signal shifted one nucleotide away from the 50 end by insertion of C17) templates shows that U17 is essential for polyadenylation.


Figure S5. Cryo-EM Density for the Pre-termination State, Related to Figures 3
A. Fit of the entire 44-mer template (v44) into the amplitude scaled (LocScale) cryoEM map of the pre-termination state. B. Fit of the nts A11-A17 connecting the 50-hook with the active site +1 position into the cryoEM density in the pre-termination state filtered at 2.41 Å resolution. C. Fit of the RNA duplex between template (yellow) and product (blue) strands into the cryoEM density in the pre-termination state filtered at 2.41 Å resolution. D. Comparison of water molecules (red spheres) contacting vRNA nucleotide U6 in the pre-initiation (left), elongation (middle) and pre-termination (right) state in cryoEM maps (filtered to 2.54 Å, 2.50 Å and 2.41 Å, respectively) show their conserved positions in independent structure determinations. (legend on next page) Figure S6. Verification of the Sequences in the In Vitro Transcription Products by Cryo-EM and RNA Sequencing, Related to Figure 5 and Table S3 A–C. CryoEM density for the template-product duplex in the pre-termination (A), the termination (B) and the stuttering (C) structures shows the expected sequences. D. RNA sequencing results for the products of the in vitro transcription reaction with the v44 and v30 RNA 18+3 templates on gel extractedmaterial as indicated on Figures 1D and 1E (yellow box). Bioinformatics analysis was performed on all reads starting with the primer sequence and dividing all products into mutually exclusive categories by applying the appropriate search patterns (Table S3). The table (top) summarizes the number of reads in each of the six indicated categories, which include full-length expected product (with and without re-alignment), aberrant products (with and without re-alignment), primer sequence and fulllength run-off product. Of note is that ‘full-length run-off product’ was not observed in the v44 + ATP/GTP/CTP sample and as little as 38 readswere found in v44 + ATP/GTP/CTP/UTP sample indicating that 50 hook release and run-off transcription is a negligible event under these conditions. A plot (bottom) shows a majority of products with the expected sequence including those that underwent re-alignment during initiation. Remaining products correspond to almost exclusively to aberrant transcription products. (legend on next page) Figure S7. Validation of Specific Interactions in the vRNA 30 End Secondary Binding Site and Conservation of Interacting Residues in Other Influenza Strains, Related to Figure 4 A. Fluorescence polarization competition assay. The WT 30 end 6-mer vRNA, but not a control oligo(A) 6-mer can compete FAM-labeled 6-mer vRNA from the 30 secondary binding site, as shown by the decrease of fluorescent polarization (FP) signal. Data shown is for n = 2 ± SD. B. Mutations in key residues interacting with the 30 end in the secondary binding site (PA/K483E, PB1/K553E, Y557A, R560E and PB2/K54E) and their combinations reduce the binding of the WT FAM-labeled 30 RNA (30-UCGUCU-50) to the apo FluA polymerase, as shown by the decrease of fluorescent polarization (FP) signal and measured KD’s. Data shown is for n = 4 ± SD. C. Transcription activity test (Reich et al., 2017) on the purified secondary binding site mutants using an 18-mer vRNA template shows that the mutations do not affect the RNA synthesis activity on a short templates that does not reach the secondary site. Data shown is for n = 4 ± SD. D. Conservation logos representing 30 ends of vRNA (left) and cRNA (right) comparing all influenza genera. Data were generated with WebLogo online server (Crooks et al., 2004) and available sequences for the following strains: A/WSN/1933, A/little yellow-shouldered bat/Guatemala/060/2010, B/Lee/1940, C/Johannesburg/1/66 and D/swine/Oklahoma/1334/2011. E. RNA-protein contacts in the secondary binding pocket in human FluB polymerase recycling complex (left) and overall view on the binding pocket (right). Note U1 is differently orientated than in the bat FluA complex (Figure 4B). Color code: PA (green), PB1 (cyan) and PB2 (red). F. Overall view of the 30 end sequence specific secondary binding pocket in La Crosse virus L protein (Gerlach et al., 2015), the recycling bat FluA polymerase complex (this work) and mode B promoter binding of FluD polymerase (Peng et al., 2019). The LACV L protein is monomeric but colored as for influenza polymerase subunits to show the structural similarity. LACV has an additional structural element, the clamp (slate), which partially buries the 30 end. (legend on next page) Figure S8. Refolding of the Peptide PB1/500-513 andConformational Transitions in the Polymerase Core as Transcription Proceeds, Related to Figures 6 and 7, Table S2, and Video S1 (part 2) A. Fits of peptide PB1/500-513 in the cryoEM density in the pre-termination state (left) and product dissociation/dislocated state (right) shows different folding of this region. B. Sequence alignment of PB1/500-513 region showing high conservation of this peptide in all influenza strains. Numbering according to the strain A/little yellowshouldered bat/Guatemala/060/2010(H17N10). C. Superposition of the RNA-dependent RNA polymerase motif E hairpin loop and peptide PB1/500-513 in the pre-termination (gray) and product dissociation (green) states with human H3N2 (active in blue and dislocated in beige) and FluD polymerase (red). Arrows indicate folds which correspond to closed-active and dislocated-inactive (with extra beta-strand) conformations of the polymerase. The FluD conformation is intermediate. D. (Left) Superposition of the pre-initiation (light gray) and pre-termination (colors, PA: green, PB1: cyan, PB2: red) structures showing global changes in the protein conformation including the collapse of PB1/b-ribbon accompanied by rotation of PA/arch, extrusion of the priming loop and rotation of PB1/thumb and PB2-N1 domains. PB1/1-551 were superposed, excluding the b-ribbon, i.e., 476 residues aligned with root-mean-square deviation (RMSD) on Ca positons of 0.45 Å. The b-ribbon (PB1/182-210) rotated by 28 , PB2-N1 (44-101, excluding the loop 80-87) by 3.8 and the thumb (PB1/560-670, excluding the priming loop 634-660), by 5.7 . (Middle) Superposition of the pre-termination (colors) and product dissociation/dislocated (dark gray) structures showing further conformational changes including correlated rotation of PA-C, PB1/thumb, PB1-PB2 interface and endonuclease domains. PB1/1-498 were superposed (excluding 274-300) i.e., 451 residues aligned with RMSD on Ca positons of 0.37 Å. The thumb rotates by 6.8 and PB2/93-250 by 5.5 . (Right) Superposition of the pre-initiation (color) and product dissociation structures (dark gray) with modeled template (yellow) showing overall changes during the transcription cycle. PB1/1-496 were superposed, excluding the b-ribbon, i.e., 388 matched residues with RMSD on Ca positons of 0.41 Å. The thumb rotates by 10.8 . For each structure, the complete pre-termination state template is superposed (yellow). E. Surface representation of the bat FluA product-dissociation state showing template path modeled from the pre-termination state structure. The template is trapped inside a tunnel formed by the PB2-Nter/PB1-Cter helical bundle packing on the 170-loop of the PB2 helical lid domain. F. Model based on the crystal structure of dimeric apo-H7N9 core showing one observed position of the PB2-Nter/PB1-Cter helical bundle consistent with release of template. The helical bundle swings out with hinges at PB1/670 and PB2/40 (consistent with the limits of visible density in the recycling structures in Figures 6D and 6E) and would need to thread back through the template loop in ‘skipping rope’ fashion to fully free the template. The observed position of the PB2-Nter/PB1-Cter helical bundle is likely a crystallization artifact (for the second monomer in the dimer, it is in a different orientation) but serves to show that substantial relocation of the PB2-Nter/PB1-Cter helical bundle is possible. In several recently published cryoEM structures of the apo-dimer ormode B (recycling) state of A/H3N2 polymerase (PDB: 6QWL, 6QX3 and 6QX8) (Fan et al., 2019), PB1 beyond 670 and PB2 before 40 are not present in the cryoEM density, also indicating flexibility at these hinge points.


Metadata
Authors
Joanna M Wandzik, Tomas Kouba, Manikandan Karuppasamy, Alexander Pflug, Petra Drncova, Jan Provaznik, Nayara Azevedo, Stephen Cusack
Journal
Cell
Publisher
Date
pm32304664
PM Id
32304664
PMC Id
Images