The universally-conserved transcription factor RfaH is recruited to a hairpin structure of the non-template DNA strand

Essential revisions:

- The functional importance of the observed interactions for binding affinity and binding specificity (subsection “Structural analysis of RfaH:ops contacts”) should be demonstrated experimentally with minimal effort. Although biochemical results are presented to support the inferred DNA secondary structure, no biochemical results are presented to support the inferred protein-DNA interactions. This is an important omission in view of the unusually, potentially worrisomely, small size of the inferred protein-DNA interface: "only 214 A^2.") Just 4 amino acids (K10, H20, R23, and R73) and 2 bases (T6 and G5) are inferred to be important for specificity. Accordingly, a comprehensive analysis -involving substitution of each amino acid by Ala, and quantification of affinities of wild-type and Ala-substituted RfaH derivatives for wild-type, T5-substituted, and G-6-substituted ops derivatives- would be possible.

We currently do not have an assay that would allow us to measure RfaH affinities to the EC directly and quantitatively; an EMSA with ³²P-labeled RfaH that we used in (Artsimovitch and Landick, 2002) is messy and semi-quantitative. We use an indirect way to assess defects in recruitment by measuring RNAP retention 1 and 2 nucleotides downstream from the ops site (referred to as 12 and 13 in the current manuscript). We assume that the strength of RfaH contacts to the NT strand is reflected in the delay in RNAP escape from the ops site, but this is not a true affinity measurement. Moreover, only a fraction of RNAP is delayed downstream from the ops site. Using this assay, we showed that K10A, H20A, and R73A substitutions had strong defects in delaying RNAP escape from ops, whereas R23A had a moderate effect (Belogurov et al., 2010). Based on these results, we proposed that K10, H20, R23, and R73 directly interact with the ops DNA.

In light of the reviewers’ concerns about the small size of the RfaH:ops9 interaction surface, we recalculated the interface using the PISA server. We found that the interface comprises 420Ų (average of both complexes in the asymmetric unit), a value updated in the revised manuscript. The crystal structure and the NMR data presented in the current work and a cryoEM structure of RfaH:ops EC obtained in the Darst lab support this interpretation and reveal additional contacts that were not predicted by modeling.

We have already demonstrated the importance of RfaH K10, H20, R23, and R73 (Belogurov et al., 2010) and G5 and T6 ops residues (this work) in RfaH function in vitro and in vivo and we do not yet have an assay to provide adequate quantitative analysis of RfaH binding to its real target, the EC. Thus, we cannot properly carry out a thorough analysis of these interactions in a reasonable time frame.

We agree that a systematic analysis of RfaH:EC interactions would be necessary to validate structural observations, particularly because the details of RfaH:DNA contacts are not identical. However, this cannot be achieved with “a minimal effort” because roles of all determinants, not just RfaH K10, H20, R23, R73 and G5/T6 in ops, need to be probed, and because we need to develop an equilibrium binding assay to probe contributions of RfaH and DNA residues implicated by X-ray and cryoEM structures.

At the same time, a manuscript reporting the cryoEM structure of RfaH-ops EC is expected to be accepted any day now. This manuscript was submitted five months after the results of the current study have been communicated to the authors and went through review at Cell with a lightning speed. Given these developments, we cannot afford a significant delay in publication. Notably, the two manuscripts are complementary because the Darst manuscript, on which IA and YN are coauthors, does not present any functional data on the NT DNA determinants and conformation.

To better describe what is known about RfaH:DNA interactions, we revised the text and figures to summarize the available functional data about the RfaH:DNA contacts. We indicated the demonstrated effects of RfaH substitutions on recruitment to ops (Belogurov et al., 2010) in Figure 4C and Figure 6 and discussed the observed effects of RfaH substitutions in the light of our RfaH:ops9 complex structure.

- The current description of the model for RfaH loading and action is vague (no real Figure legend for Figure 1) and it would be useful to be more concrete both in what is known and what is not known. E.g. in the modelled RfaH-TEC complex, which is very compelling, RfaH is bound to ops in the context of a TEC. The TEC is in a register where U11 would be post-translocated or possibly pre-translocated (assuming G12 has been added). It was believed that the backtracking happens at the U11 pause site (according to Figure 3, Artsimovitch and Landick, 2000) but in that case it is not easy to see how RfaH can bind to ops because the NT DNA should be shifted downstream by at least one, possibly two bases. Do the authors think that RNAP is backtracked at U11? Do the authors think that RfaH binds to this state or do they think it binds to a post-translocated TEC at U11?

Backtracking by 1 and 2-3 nt is observed (by Gre-assisted cleavage) in ECs stalled by substrate deprivation G10 and U11 ECs, respectively, in the absence of RfaH; no backtracking is seen in C9 EC. RfaH inhibits backtracking if added to the preformed ECs because it stabilizes the -10 bp at the upstream fork junction, which has to melt for backtracking to occur, as inferred from an increase in inter-strand DNA crosslinking by psoralen. These data are from a manuscript that has been just accepted for publication by Nedialkov et al.

We assume that in vivo RfaH is recruited at U11 because a strong pause at this position was observed by permanganate footprinting (Leeds and Welch, 1996) and later by NET seq, and because mutations that interfere with RNAP pausing at U11 compromise RfaH function, but we do not know whether this complex is pre-, post-, hybrid- (Guo et al., 2018 and Kang et al., 2018), or reverse-translocated. Because RfaH binds to scaffolds in which the RNA strand is present or missing similarly (Artsimovitch and Landick, 2002), it is likely that RfaH primarily senses the NT DNA conformation during the initial recruitment and binds to the EC at different positions and in different translocation states, owing to the inherent flexibility of the NT DNA. RfaH binds to the U11 ops EC which is backtracked but also to the ops EC in which both RNA and T DNA are post-translocated in the cryoEM structure of RfaH:opsEC; in the latter complex, RNA translocation is forced by the scaffold topology, and the NT DNA forms a hairpin with the C3:G8 bp.

We modified the text and the legend of Figure 1 to provide more details on what is and what is not known about the life cycle of RfaH. We also expanded the discussion of RfaH recruitment in the revised manuscript.

- The ExoIII footprints seem to argue against backtracking since the tDNA gets shorter as RNAP progresses towards U11 and the results are not consistent with the schematics at the bottom of Figure 3. It seems that the major ExoIII cleavage products in absence of RfaH are not labelled correctly: G8 should be 14, C9 should be 13, G10 should be 13/12, U11 should be 12. Likewise, in the presence of RfaH, the ExoIII footprints also get shorter (not to the same extent) – major ExoIII cleavage products seem to be 19 for C9, 19/18 for G10 and 18 for U11. These results argue against backtracking because ExoIII gains access to the tDNA strand as RNAP adds bases to the RNA 3'-end. Authors should clarify and discuss this.

We apologize for a cryptic description of the ExoIII data. We used ExoIII as a reporter of RfaH recruitment: an extension of ExoIII upstream protection is due to RfaH effects on the NT DNA and the upstream DNA duplex trajectories. RfaH induces a strong block at C9, G10, and U11 ECs, consistent with its recruitment at these positions. In Figure 3, ExoIII footprint boundaries are labeled with respect to the RNA 3’ end; in a completely post-translocated EC, RNAP protects 14 upstream bp from ExoIII (9 bp RNA:DNA hybrid + 5 extra bp), in a half- translocated (Guo et al., 2018 and Kang et al., 2018) or pre-translocated EC– 15 bp. G8 EC, in which the upstream RNA is not complementary to the T DNA, is post-translocated (the reviewer is correct and this number should have been 14). C9 is also post-translocated, the T DNA becomes 1 nt shorter, and only 1 predominant species is observed. In G10, even both 14- and 15- products are seen, whereas in U11, an additional backtracked (16) species is seen. Exo III may not be an ideal probe because once ExoIII cleaves the DNA, RNAP cannot go back. We infer the translocation register in these complexes from sensitivity to Gre cleavage (Nedialkov et al., 2018), but the Exo probing also shows that RNAP is reluctant to move forward as it approaches ops.

We expanded the section on ExoIII probing and modified Figure 3 and its legend to make it more clear.