Single-peptide DNA-dependent RNA polymerase homologous to multi-subunit RNA polymerase

Transcription in all living organisms is accomplished by multi-subunit RNA polymerases (msRNAPs). msRNAPs are highly conserved in evolution and invariably share a ∼400 kDa five-subunit catalytic core. Here we characterize a hypothetical ∼100 kDa single-chain protein, YonO, encoded by the SPβ prophage of Bacillus subtilis. YonO shares very distant homology with msRNAPs, but no homology with single-subunit polymerases. We show that despite homology to only a few amino acids of msRNAP, and the absence of most of the conserved domains, YonO is a highly processive DNA-dependent RNA polymerase. We demonstrate that YonO is a bona fide RNAP of the SPβ bacteriophage that specifically transcribes its late genes, and thus represents a novel type of bacteriophage RNAPs. YonO and related proteins present in various bacteria and bacteriophages have diverged from msRNAPs before the Last Universal Common Ancestor, and, thus, may resemble the single-subunit ancestor of all msRNAPs.

3) line 84 and Supp. Fig 1: Is YonO prone to non-templated 3'-end addition? It looks like that might be the case. Also, if the effective Kd for NTPs is less than 1 microM (as it might be in view of Fig. 1d) are there concerns about the purity of the CTP and UTP, used at high concentrations for the misincorporation test? 4) line 89 and Fig. 1e: I don't think that this expt. at a single pH justifies the absolute statement. The characteristic pKs of the hydrolytic process of the 2 enzymes might be different. A pH titration is indicated. 5) line 132 and Fig. 2e: The duplex DNA transcription expt. is inadequate (besides being a bit of a mess): What establishes that the thick band in the left-hand lane is a single run-off transcript (cf Fig. 2d) and, more importantly, that it is not an end-to-unknown-terminator transcript? And is it released from the template or is it produced as an RNA-DNA duplex? Hasn't the 5' end of the in vitro transcript already been mapped along with the in vivo RNA? The implication of this experiment that the authors seem to convey (see the cartoon at the top of Fig. 2e and "run-off transcript" at the side) is that the YonO protein, like the unrelated T7 single-subunit RNA polymerase, finds its transcriptional start site without the aid of a sigma-like (or archeal TBP/TFBlike) initiation factor. A better and better-executed experiment is called for to even suggest this important implication. The last sentence of Discussion states that work on initiation of transcription is continuing. In Fig. 2e, the authors imply they believe they already know the answer, but the evidence is just not there. I hope it can be provided in this publication. If time constraints do not allow that, Fig. 2e needs to be omitted (which would be regrettable).
(signed) E. Peter Geiduschek Reviewer #3 (Remarks to the Author): This paper reports a significant discovery, an RNA polymerase with apparent homology to the universal family of multisubunit RNA polymerases, but of a highly limited and restricted sort: not much more than a handful of essential residues in the active center, in a single polypeptide. Nonetheless, the relation is persuasive and evidence of the activity and function of the RNA polymerase are convincing. The data are generally good and technically well produced. The discovery is important enough to deserve a publication of some prominence.
Several questions arise.
1. No comment is made about the +/-NT components of Fig. 1B. 2. It is argued that the YonO polymerase pauses much less than the E. coli multisubunit RNA polymerase. However, the experiment was performed at only one (low) concentration of substrate. Wouldn't this be the result if the Km were simply quite different for the two enzymes, but the pausing behavior not significantly different? 3. The identification of the promoter by runoff synthesis in Fig. 2E does show a (fuzzy) band of the expected size, but there is other RNA as well. It would have been useful to show that random fragments make no bands of this size, or that different fragments with the sequence make runoffs of predicted size. This comment has partly to do with the absence of more precise description of how the triphosphate end analysis was done to make it the expectation that there is only one 5' end that should map in this fragment. Probably the authors can satisfy this with a few more words. Concerns 1-Those results raised the following question: "It walks as a duck, quacks as a duck but is it really a dusk?" The authors point out to the residues with homology to ß and ß' of msRNAPs required for catalysis. Those should be mutated, the proteins purified and tested for activity.
As suggested by the Reviewer, we mutated YonO's aspartate triad that chelates the first Mg 2+ ion in msRNAPs, and that is among the few amino acids conserved between YonO and msRNAPs. We purified and characterised the mutant enzyme, which was absolutely inactive in RNA synthesis, despite forming stable ECs (Fig. 1c ). The result indicates that the aspartate triad are indeed the catalytic residues of YonO.

2-It is surprising that, although not sequence related, the authors do not refer to the simple single subunit RNA polymerases, which lack proofreading and back-tracking activity, show reduced pausing, are faster than the ms RNAPs and quite processive.A priori, a trigger loop is not essential
We agree with the Reviewer that the higher processivity, lack of proofreading, etc., could be a result of functional convergence between evolutionary unrelated bacteriophage RNAPs (single-subunit RNA polymerases and YonO), in contrast to msRNAPs, because of their similar tasks during phage development. We have added this discussion to the revised version.
However, we refrain from discussing the essentiality of the Trigger Loop for the following reason. As mentioned by the Reviewer, the single-subunit RNA polymerases (ssRNAPs) are not related to msRNAPs or YonO. Therefore, while the convergence of the functions of RNA synthesis between ssRNAPs and msRNAPs is highly remarkable indeed, the structural and/or chemical aspects of these functions cannot be compared directly. The O-helix of ssRNAP was proposed to be functionally analogous to the Trigger Loop in delivering the substrate to the active conformation. However, although it suggests the importance of the rearrangement of the active centre during NMP addition, this analogy is limited. For example, the catalysis of phosphodiester bond formation by ssRNAP involves acid-base catalysis, which is not the case for msRNAP, whose Trigger Loop participates in stabilisation of the transition state of the reaction. Thus one cannot argue on the essentiality or nonessentiality of the Trigger Loop based on comparisons between msRNAP and ssRNAP.
In contrast, the properties of YonO and msRNAPs, such as processivity, proofreading, etc., can be directly compared as these enzymes share a common ancestor. Hence, it is surprising that an enzyme apparently lacking the part of the active centre conserved among all its relatives (msRNAPs) is as active as them.
3-One outstanding question is the mechanism of promoter recognition. The authors indicate that those are questions for the future. Having a single promoter, footprinting, template base changes and crosslinking to identify protein and template residues recognized upon binding is very straightforward.
As suggested by Reviewer 2, because of the lack of clarity on the mechanisms involved, we have removed panel 2e and accompanying discussions. Experiments on initiation that we performed during revision (footprinting, abortive initiation, etc.) have raised more questions about promoter properties and the requirements for initiation than provided answers. While we can obtain initiation on double-stranded DNA, we do not see formation of the promoter open complex in vitro and we cannot explain why in vitro transcription by YonO is much weaker than we observe in vivo, or why we observe substantial non-specific initiation in vitro but not in vivo. This may imply a requirement for a specific DNA structure, superhelicity or additional factors to increase specificity and/or efficiency of initiation. Investigation of these mechanisms is not feasible within the revision process, and is a separate, possibly several-year, project that we are starting in the lab.

Reviewer #2 (Remarks to the Author):
The cellular DNA-dependent RNA polymerases are universally multi-subunit enzymes. Thus, the conjecture, more than 10 years ago, that an ORF of a B. subtilis inducible prophage might point to a very ancient (pre-LUCA) single-subunit origin of the contemporary multi-subunit enzymes should have aroused wider interest and follow-up than was the case. This interesting initial communication takes up the challenge of that conjecture by showing that the YonO protein of the B.subtilis 168 SPbeta prophage is a DNA-dependent RNA polymerase and that its function is essential for expression of viral late genes after induction of the lytic viral multiplication cycle. In my judgment, this brief note and the promise of further work on this very special single-subunit RNAP will elicit sufficiently wide interest to merit publication in Nature Communications.
I do have specific concerns to convey to the authors, but all of them involve further experiments that will require little time. In text order (item 5 is my principal concern): 1) line 75 and Fig 1c: Is the YonO protein "strict" with Mn(+2) in place of Mg(+2)? I suspect that perhaps not.
We have repeated the substrate specificity experiment (RNA vs DNA, NTPs vs dNTPs) in the presence of Mg 2+ or Mn 2+ . YonO, as well as msRNAPs, can use Mn 2+ as efficiently as Mg 2+ . Mn 2+ had little effect on YonO specificity, apart from the DNA-DNA "hybrid" in the presence of NTPs, where it increased the otherwise poor extension. The results are presented in Supplementary Fig. 1d of the revised version, and are discussed in the text. Fig 1d (also line 173 and Fig. 2e): Is the transcript released or is an RNA-DNA hybrid formed (tested with RNases H,A)?

2) line 80 and
We performed the suggested analysis, and showed that YonO does facilitate RNA disengagement from the template DNA behind the elongation complex, as do msRNAPs ( Supplementary Fig. 1c)

3) line 84 and Supp. Fig 1: Is YonO prone to non-templated 3'-end addition? It looks like that might be the case. Also, if the effective Kd for NTPs is less than 1 microM (as it might be in view of Fig. 1d) are there concerns about the purity of the CTP and UTP, used at high concentrations for the misincorporation test?
Indeed, during prolonged incubation in a high NTP concentration, YonO adds non-templated NMPs to the 3' end of the run-off transcript (we found that RNA stays in the complex at the end of template, as in the case of E. coli RNAP; Supplementary Fig. 1c). E.coli RNAP also added nontemplated NMPs but less efficiently. We found that, although the K m for correct NTPs of YonO was similar or slightly higher than that of E. coli RNAP, K m for incorrect ones was much lower, suggesting a possible explanation for the more efficient 3'-end additions by YonO.
Similarity of K m s to NTP between two enzymes also suggests that the "faster" transcription by YonO in Fig. 1d (Fig. 1e in revised manuscript) is rather explained by a higher processivity of YonO. To confirm that we performed extension at high (100 μM) NTPs concentration, which showed that E. coli msRNAP had still lower processivity as compared to YonO (right panel in Fig. 1e in revised manuscript).
As we found K m s for the correct NTPs to be similar for YonO and E. coli RNAP, the observation of the extension with the wrong NTPs cannot be explained by contamination with correct ones. Furthermore, we have no concerns with the NTPs' purity because in most cases our Urea-PAGE system clearly separates RNAs with different 3' NMPs (this can be seen in the misincorporation gels; Fig. 1f and Supplementary Fig. 1e in the revised version), and, in many cases, transcripts of the same length but with different sequences. Some years ago we did perform additional purification of commercial NTPs (GE Healthcare), but found it be redundant. Fig. 1e: I don't think that this expt. at a single pH justifies the absolute statement. The characteristic pKs of the hydrolytic process of the 2 enzymes might be different. A pH titration is indicated.

4) line 89 and
As suggested by the Reviewer, we performed proofreading reactions at different pHs. As can be seen from Supplementary Fig. 1f of the revised version, while cleavage by E. coli RNAP increased significantly with the increase of pH, YonO remained completely inactive even during prolonged incubation. Though we cannot exclude that at an even higher pH YonO may show some activity in hydrolysis, this can hardly be observed experimentally since at these pHs/times RNA begins to be non-enzymatically hydrolysed in the presence of Mg 2+ . Similarly, activity with a much higher pKa is unlikely to have biological relevance. Fig.  2d) and, more importantly, that it is not an end-to-unknown-terminator transcript? And is it released from the template or is it produced as an RNA-DNA duplex? Hasn't the 5' end of the in vitro transcript already been mapped along with the in vivo RNA? The implication of this experiment that the authors seem to convey (see the cartoon at the top of Fig. 2e and "run-off transcript" at the side) is that the YonO protein, like the unrelated T7 single-subunit RNA polymerase, finds its transcriptional start site without the aid of a sigma-like (or archeal TBP/TFB-like) initiation factor. A better and betterexecuted experiment is called for to even suggest this important implication. The last sentence of Discussion states that work on initiation of transcription is continuing. In Fig. 2e, the authors imply they believe they already know the answer, but the evidence is just not there. I hope it can be provided in this publication. If time constraints do not allow that, Fig. 2e needs to be omitted (which would be regrettable).

5) line 132 and Fig. 2e: The duplex DNA transcription expt. is inadequate (besides being a bit of a mess): What establishes that the thick band in the left-hand lane is a single run-off transcript (cf
We performed additional experiments to understand the mechanism of initiation, but these raised more questions than answers. As mentioned in the reply to Reviewer 1, although we can observe run-off transcription, we cannot exclude several nearby starts as the band is usually fuzzy; we do not see formation of the promoter open complex; and cannot explain the lower efficiency and specificity of transcription as compared to our in vivo observations. It will not be feasible to address these problems during the revision, and we have removed Fig. 2e and corresponding discussions as suggested by the Reviewer.

Reviewer #3 (Remarks to the Author):
This paper reports a significant discovery, an RNA polymerase with apparent homology to the universal family of multisubunit RNA polymerases, but of a highly limited and restricted sort: not much more than a handful of essential residues in the active center, in a single polypeptide. Nonetheless, the relation is persuasive and evidence of the activity and function of the RNA polymerase are convincing. The data are generally good and technically well produced. The discovery is important enough to deserve a publication of some prominence.

No comment is made about the +/-NT components of
We thank the Reviewer for spotting this miss. We have added a discussion of the experiment. Briefly, while msRNAP elongation complex (EC) without NT is stable at normal ionic strength, it cannot withstand high (1M) salt. This property is explained by the critical role of the downstream DNA duplex in EC stabilisation. In the mentioned experiment, we tried to show similarities of the requirements for EC stability and, thus possibly, their structures for YonO and msRNAP.
2. It is argued that the YonO polymerase pauses much less than the E. coli multisubunit RNA polymerase. However, the experiment was performed at only one (low) concentration of substrate. Wouldn't this be the result if the Km were simply quite different for the two enzymes, but the pausing behavior not significantly different?
We have repeated the experiment at a high (100μM) NTP concentration and observed similar behaviour; YonO was more processive than E. coli msRNAP (Fig. 1e of the revised version). Furthermore, we measured the K m for NTP of YonO and E. coli msRNAP in elongation complex and found them to be similar. Fig. 2E does show a (fuzzy) band of the expected size, but there is other RNA as well. It would have been useful to show that random fragments make no bands of this size, or that different fragments with the sequence make runoffs of predicted size. This comment has partly to do with the absence of more precise description of how the triphosphate end analysis was done to make it the expectation that there is only one 5' end that should map in this fragment. Probably the authors can satisfy this with a few more words.

The identification of the promoter by runoff synthesis in
Triphosphorylated 5' end of the transcript was determined by RNA-seq. In order to enrich for 5' triphosphorylated transcripts during RNA-seq, the RNA sample was treated with 5′ terminator exonuclease (TEX), which degrades RNAs containing a 5′-monophosphate, thereby enriching for transcripts containing 5′-triphosphates. We have added this description to the text. However, as suggested by Reviewer 2, we decided to remove figure 2e and the corresponding discussion. The reason for this was that the additional experiments which we have performed (footprinting, abortive synthesis, etc.) raised more questions than delivered answers. We cannot exclude several nearby starts as the band is usually fuzzy (which though is not observed on different DNA templates). We do not yet understand the properties of the promoter, do not observe open complex formation, and cannot explain lower efficiency and specificity of transcription in vitro as compared to in vivo. Answering these questions is a separate study in the lab, and is not possible within this revision.
4. Could the authors reference the non-essentiality of the various domains of multisubunit RNA polymerases, as asserted in the discussion (p. 8). Or would these be the references already given for the functions of these domains?
The references provided are the references about the functions of these domains. These works showed non-essentiality of these domains in robust RNA synthesis in vitro, but their requirements for response to some specific regulatory signals or in some specific conditions.