The human gut Firmicute Roseburia intestinalis is a primary degrader of dietary β-mannans

β-Mannans are plant cell wall polysaccharides that are commonly found in human diets. However, a mechanistic understanding into the key populations that degrade this glycan is absent, especially for the dominant Firmicutes phylum. Here, we show that the prominent butyrate-producing Firmicute Roseburia intestinalis expresses two loci conferring metabolism of β-mannans. We combine multi-“omic” analyses and detailed biochemical studies to comprehensively characterize loci-encoded proteins that are involved in β-mannan capturing, importation, de-branching and degradation into monosaccharides. In mixed cultures, R. intestinalis shares the available β-mannan with Bacteroides ovatus, demonstrating that the apparatus allows coexistence in a competitive environment. In murine experiments, β-mannan selectively promotes beneficial gut bacteria, exemplified by increased R. intestinalis, and reduction of mucus-degraders. Our findings highlight that R. intestinalis is a primary degrader of this dietary fiber and that this metabolic capacity could be exploited to selectively promote key members of the healthy microbiota using β-mannan-based therapeutic interventions.

This paper is of significant interest to the field as it details the degradation of complex polysaccharide by the firmicute R. intestinalis, most studies of polysaccharide degradation in the human gut have centred on the bacteroides. It provides new insights and adds depth to the body of knowledge which could influence the development of targeted prebiotics. The paper has a significant amount of data including bacterial growths, regulation data, enzyme activities, competition experiments and in vivo experiments. It is clear to see that the paper is a considerable amount of work by the authors.
Major points to address. Fig.2a The authors show growth curves for R. intestinalis on different carbon sources. The figure shown only displays growth curves over 8hrs which only covers growth into exponential phase. Usually, I would expect to see a full growth curve right through to stationary phase. Later in the manuscript the authors show HPLC traces of culture supernatant from stationary phase growths. It would be beneficial to show the full growth curve in Fig2a. Supplementary Fig 1.a. The chromatogram is shown from 2.5 min. There appears to be peaks that would align with the mannose and oligo standards. The Authors state that there was no observed build-up of oligos or mannose, what are the peaks between 2.5 and 10 min?
Page 10 line 201 onwards, the authors claim that the frequency of galactose substitutions does not affect RiGH36 catalytic efficiency. From the data shown it is hard to come to that conclusion, the authors show end point assays displaying RiGh36 is capable of fully hydrolysing oligos with one or two internal substitutions, this does not show how quickly or efficiently the oligos are hydrolysed. To assess the catalytic activity the authors need to either measure the catalytic parameters of the enzyme for each substrate or a time course showing the oligos are degraded in a similar fashion/speed using the same conditions for both oligos.
Finally the authors show a model of β-mannan degradation based on the assumption RiGH26 is located on the cell surface and that is the key enzyme which breaks down β-mannan to oligos for transport by the ABC transporter. I think if this is the case the authors need to show the cellular location for RiGH26 experimentally.
Minor points.
The supplemental figures need to be re-ordered so that they appear in the order they are cited in the text.
Also there is no consistency in how the HPLC traces are displayed and labelled. Some have different colours, some are labelled beneath the trace or above. It would be helpful if the authors have a uniform way to display the HPLC traces. Also, some of the traces were hard to see as the plots are too close together and should be spaced out more. For example Supplemental fig 7.b it is very difficult to see the gal peak as it overlays with another plot and they both have a similar colour.
We thank the editor and the reviewers for the valuable comments to our manuscript.
We now submit a revised version of NCOMMS-18-21943-T that addresses their concerns. Our point-by-point answers to the comments are dealt with chronologically. Our responses are in blue lettering.

Reviewer #1
The authors have demonstrated through a combination of multi "omic" analyses and biochemical studies that R. intestinalis can break down plant cell wall polysaccharides β-Mannans, a common component of the human diet. Their detailed biochemical studies of the encoded enzymes have resulted in a model of sequential action for the mannan utilization system encoded by R. intestinalis. Additionally, their in vivo experiment suggests that the use of β-mannan-based prebiotics could promote SCFA-producers and provide a competitive advantage over colonic mucindegrading bacteria.
These findings are novel, the work is thorough and convincing, and the paper should be of interest to a wide audience.
We appreciate the reviewer's interest in our data, his/her recognition that the work provides novel insights into poorly characterized β-mannan utilization in Firmicutes and the comments to further improve the manuscript. 1) I was somewhat surprised to see that R. intestinalis relative abundance is only significantly increased on the first day of AcGGM treatment and returns to near baseline by day 7. M. formatexigens must be very efficient once it had a start on the AcGGM to then be able to outcompete/suppress B. ovatus and R.
intestinalis by the end of the week trial. Perhaps the authors could consider competition experiments with this acetogen as well.
Re: Although we agree with the reviewer's comment that would be of interest to conduct in vivo competition experiments to show that M. formatexigens is able to outcompete B. ovatus and R. intestinalis, we are currently unable of performing additional murine experiments. Unfortunately, setting up a new biassociated mice feeding trial is not feasible due to very limited availability of gnotobiotic mice in Prof. Martens lab, costs and time constrains. Perhaps most importantly, we are truly reluctant to try to justify sacrificing more animals for this purpose. Due to the above points, we have conducted in vitro experiments and results are shown in the Supplementary Fig. 17. Overall, our findings show that M. formatexigens possesses a degradation apparatus for β-manno-oligosaccharides ( Supplementary Fig.17a), is able to grow ( Supplementary Fig.17b) and utilizes the manno-oligosaccharides contained in the AcGGM preparation ( Supplementary Fig.17c-d) in vitro. However, we were surprised to find that M. formatexigens did not outcompete either B. ovatus or R. intestinalis in an in vitro setting (Supplementary Fig.17e-f). Taken together with the in vivo experiment results shown in Fig. 6e, it is likely that, when present as part of a microbial community, M. formatexigens may be indirectly benefiting by either feeding/cross-feeding on mannooligosaccharides or another mechanism. Even though we do not know the biological significance of this finding without additional experimental evidence which is outside the scope of this manuscript, we have added the following hypotheses to the discussion lines 419-439: "Intriguingly, R. intestinalis' response did not last over the 7 day feeding treatment and the acetogen M. formatexigens seemed to replace it. A cluster of genes with predicted functions in β-manno-oligosaccharide utilization (BRYFOR_07194-BRYFOR_07206) was identified in the genome of M.
formatexigens ( Supplementary Fig.17a). The results shown in Supplementary 2) Line 52: bifidobacteria and lactobacilli to be written in lower case, and 'strains' can be removed (and perhaps it is better to say: 'certain bifidobacterial and lactobacilli'. Re: Done as suggested. The growth experiments with a simplified microbial community showed in Fig. 6a-d were performed in a buffered system and the pH of the stationary phase cultures after growth on AcGGM was 5.8 ± 0.16. If the findings of the competition experiments on AcGGM were the results of differences in acid sensitivities between the two strains, we believe that we should have observed the same outcome when R. intestinalis and B. ovatus were cocultivated on mannose, where the pH of the stationary phase culture was 5.6 ± 0.11. Additionally, using the same culturing conditions, Leth et al. has shown that R. intestinalis is able to outcompete B. ovatus during the exponential-phase of growth on xylotetraose and xylans (Leth et al., Nature Microbiol, 2018), whereas the proportional representation of B. ovatus increased in the stationary phase. Due to the above points, we hope that the reviewer will agree with us that the findings of the in vitro experiment are not the result of a pH-effect.
We have now added this information in the legend of Fig. 6 lines 1003-1005: "The pH of the stationary phase cultures after growth on either AcGGM or mannose was 5.8 ± 0.16 and 5.6 ± 0.11, respectively, thus showing that the results are not due to differences in acid sensitivity between the two strains".
Re: Done as requested.
Re: Done as requested.

Reviewer #2
The manuscript by La Rosa et al is concerned with the breakdown of β-mannan by Roseburia intestinalis. They show that two loci are expressed in response to βmannan and they describe detailed activities for the enzymes involved in β-mannan degradation. The authors describe the endo-mannanase activity of RiGH26 and hypothesise on the role of two CBM modules to anchor manno-oligosaccharides to the surface of the cell. The authors also report a new specificity for RiGH113 as a reducing end mannose-releasing exo-oligomannosidase. The authors also describe a number of enzyme activities involved in β-mannan degradation and go on to predict a model for β-mannan degradation by R. intestinalis. The authors also test the ability of R. intestinalis to compete with another β-mannan degrader Bacteroides ovatus showing that the two can co-exist whilst using β-mannan as a carbon source.
Notably, they observed R. intestinalis was able to out compete B. ovatus during stationary phase suggesting R. intestinalis has a selective advantage during nutrient limitation. Finally the authors showed that in a minimal microbiota grown on βmannan members capable of breaking down β-mannan are enriched and that mucin degrader levels decreased. This paper is of significant interest to the field as it details the degradation of complex polysaccharide by the firmicute R. intestinalis, most studies of polysaccharide degradation in the human gut have centred on the bacteroides. It provides new insights and adds depth to the body of knowledge which could influence the development of targeted prebiotics. The paper has a significant amount of data including bacterial growths, regulation data, enzyme activities, competition experiments and in vivo experiments. It is clear to see that the paper is a considerable amount of work by the authors.
We thank the referee for positive comments, overall interest in data, and valuable suggestions to improve the manuscript.
Major points to address:  Supplementary Fig 1a. The chromatogram is shown from 2.5 min. There appears to be peaks that would align with the mannose and oligo standards. Moreover, info about the concentration of the enzymes and substrates used for the assays have been added to the legend of the Supplementary Fig. 8.

4) Finally, the authors show a model of β-mannan degradation based on the
assumption RiGH26 is located on the cell surface and that is the key enzyme which breaks down β-mannan to oligos for transport by the ABC transporter. I think if this is the case the authors need to show the cellular location for RiGH26 experimentally.

Re:
The reviewer brings up a good point and following his/her suggestion we have obtained the polyclonal antibody raised against the recombinant RiGH26 and conducted immunofluorescence microscopy. The results, shown in Fig. 3 in the revised manuscript, demonstrate that RiGH26 is located on the surface of the R. intestinalis cells, similarly to the previously shown surface enzyme RiXyn10A from the same bacterium (Leth et al., Nat Microbiol, 2018).
We have now added the following text in the results section lines 132-133: "The extracellular localization of RiGH26 was corroborated experimentally by immunofluorescence microscopy (Fig. 3)" In the methods section lines 574-594 and in the figure 3 legend lines 942-946.
Minor points:

4)
The supplemental figures need to be re-ordered so that they appear in the order they are cited in the text.
Re: We thank the reviewer for this constructive suggestion. We have rearranged the panels depicted in the supplemental figures and reordered the supplemental figure numbers to match the sequence they are cited in the text.

Reviewers' comments:
Reviewer #1 (Remarks to the Author): In the present revised manuscript the authors present an extensive study of the enzymatic pathway responsible for beta-mannan metabolism in Roseburia intestinalis, while also showing how this group of complex polysaccharides allows this gut commensal to compete in its natural environment. The topic is exciting and novel and the work presented in this manuscript is highly impressive and of excellent quality. The revision has significantly improved from theo riginal submission, though there are still a couple of outstanding and relatively minor issues.
The manuscript however is in places difficult to follow, especially in the Results section, probably due to the high density of the provided experimental data. The authors can easily solve this by explaining "what" the authors wanted to show with their experiments and "how" they decided to do it.
Below, some further issues are highlighted: I could not find any indication of the source of the strain being the main study object of this study (I imagine is a sequenced strain because is publicly available, but was it sequenced within this study or where it was sourced?).
Line 110; rather than listing GH families, it may be better to say 'predicted glycosyl hydrolases belonging to GH113, GH36 and GH1, Line 120; 'may be' can be written as 'is' since that part of the sentence already contains the word 'suggesting'. Line 121/122; the sentence may contain a reference to the next paragraph when the function of the GH26 endomannanase will be revealed.

Line 131; 'regions' should be 'region'
Lines 132-134 and Fig 3; the figure is not particularly insightful due to the absence of a negative control (being the strain grown on glucose and/or a Roseburia strain which does not encode this protein). A negative control should therefore be provided or mentioned. Line 141; 'as much as 50 mg/mL Spruce AcGGM', it may be useful to indicate the amount of time and enzyme it takes to do this. Line 148; the description 'mannanase from Firmicutes mostly in other members of the' is a bit ambiguous, should it be 'mannanases encoded by Firmicutes belonging to various other members of the'? Line 154; 'to' should be 'for', and 'with highest affinity on' should be 'with its highest affinity for' Line 168; 'with degree' should be 'with a degree' Line 406; nutrients should be nutrient Line 471; spectro-photochemically should be spectrophotometrically? Line 498; transcription should be translation M&M; The author may provide more information on bioinformatic analysis; I could not find details of the procedures and parameters used (other than stating the source of the program or online tools used).
M&M; It was not clear to me why the authors in some cases used His-tag purification and in other cases not (in methods section they state that "where appropriate His-tagged transcription [should be translation!] was prevented by the introduction of one of two stop codons"). Also, if non-Histagged proteins were purified, how was this performed?
Reviewer #2 (Remarks to the Author): The authors have included experimental data to show the cellular location of the RiGh26 and additional experiments (to the reviewers) to demonstrate that mannose or manno-ologos do not build up in the culture supernatant. The authors have addressed the comments from the reviewers and have edited the manuscript accordingly.
We appreciate the valuable comments to our manuscript. We now submit a revised version of NCOMMS-18-21943-A that addresses reviewer #1 concerns. Our pointby-point answers to the comments are dealt with chronologically. Our responses to the comments are in red lettering.

Reviewer #1
In the present revised manuscript the authors present an extensive study of the enzymatic pathway responsible for beta-mannan metabolism in Roseburia intestinalis, while also showing how this group of complex polysaccharides allows this gut commensal to compete in its natural environment. The topic is exciting and novel and the work presented in this manuscript is highly impressive and of excellent quality. The revision has significantly improved from the original submission, though there are still a couple of outstanding and relatively minor issues.

Re:
We appreciate the reviewer's interest in our data and the very positive comments.
The manuscript however is in places difficult to follow, especially in the Results section, probably due to the high density of the provided experimental data. The authors can easily solve this by explaining "what" the authors wanted to show with their experiments and "how" they decided to do it.

Re:
We realize that, as referee 1 points out, due to the high concentration of data, the rationale of analysis may be perceived as a little 'hidden' in some places. In most cases, each part of the results section start with a short sentence indicating the hypothesis we wanted to prove with the experiment and how this was performed.
Additionally, the title of the each results subsection summarizes the major results.
We have identified some less clear parts and added some sentences to improve clarity in the results section.
To clarify the scope of the comparative genomic analysis, binding experiments with the carbohydrate binding modules and the solute binding protein, we have added the following text at lines 119-121: "We carried out a comparative genomic analysis to establish the distribution of βmannans utilization loci equivalent to the identified MULL and MULS in other representative Roseburia spp. and Clostridium cluster XIVa members". and lines 154-158: "To investigate the biochemical properties of the two modules, RiCBM27 and RiCBM23 were expressed in E. coli and their capacities to bind to a range of different soluble cello-and manno-oligosaccharides were evaluated using surface plasmon resonance (SPR)." and lines 170-173: "The thermodynamic binding parameters of the ABC-transporter associated solute binding protein, RiMnBP, to linear and substituted manno-oligosaccharides were determined using isothermal titration calorimetry (ITC)." Below, some further issues are highlighted: 1) I could not find any indication of the source of the strain being the main study object of this study (I imagine is a sequenced strain because is publicly available, but was it sequenced within this study or where it was sourced?).

Re:
We apologize for this oversight and we thank the reviewer for catching this. This info was provided quite late in the results text, lines 90-91. We have now made clear that our study was conducted using the R. intestinalis L1-82 and added the strain name at the introduction line 75. In addition, we added the reference describing the isolation of this bacterium at line 473 and the BioProject accession number at line 497.
2) Line 110; rather than listing GH families, it may be better to say 'predicted glycosyl hydrolases belonging to GH113, GH36 and GH1.
Re: Done as suggested.
3) Line 120; 'may be' can be written as 'is' since that part of the sentence already contains the word 'suggesting'.
Re: Done as suggested.

4)
Line 121/122; the sentence may contain a reference to the next paragraph when the function of the GH26 endomannanase will be revealed.

Re:
We thank the reviewer for this suggestion. We have added the following text at lines 125-126: "(see later results for R. intestinalis β-mannanase RiGH26)" 5) Line 131; 'regions' should be 'region'.
Re: Done as suggested. Fig 3; the figure is not particularly insightful due to the absence of a negative control (being the strain grown on glucose and/or a Roseburia strain which does not encode this protein). A negative control should therefore be provided or mentioned.

Re:
We thank the reviewer for the constructive suggestion. As a negative control, we have now included a picture of R. intestinalis L1-82 cells cultured on YCFA supplemented with 0.5% w/v glucose and incubated with polyclonal antibodies raised against RiGH26. Cells grown on glucose display a minor fluorescence signal, consistent with the proteomics data indicating that RiGH26 is expressed at basal levels. We believe now that Figure 3 shows the results that would be expected.
Legend (lines 958-967) has been modified as follows: Sequences with coverage <60% and amino acid similarity <45% were excluded". 15) M&M; It was not clear to me why the authors in some cases used His-tag purification and in other cases not (in methods section they state that "where appropriate His-tagged transcription [should be translation!] was prevented by the introduction of one of two stop codons"). Also, if non-His-tagged proteins were purified, how was this performed? Re: We thank the reviewer for this insight. When we started this study, we conducted protein sequence analyses with previously characterized enzymes to identify catalytic and active residues and determine the most suitable cloning strategy. The reasons for having four non-His tagged proteins -RiGH113, RiGH36, RiMep and RiPgm -are as follows. Based on the sequence similarities with the Ruminococcus albus epimerase RaCE (PDB ID: 3VW5), we observed that RiMep has a catalytic histidine (His 395) close to the C-terminus that could have interaction with a potential C-terminal His-tag. Based on the sequence similarity with the AaManA sequence from Alicyclobacillus acidocaldarius (Zhang Y et al, J Biol Chem 2008), RiGH113 has an active residue (Trp-273) close to the C-terminus that may interact with a potential C-terminal His-tag. Additionally, the His-tagged version of RiGH36 and RiPgm were inactive.
The four non-His tagged proteins (RiGH113, RiGH36, RiMep and RiPgm), were purified using hydrophobic interaction chromatography (HIC) followed by Size Exclusion Chromatography (SEC), as indicated in the Supplementary table 9. The purification methods were already described at methods lines 520-531. We have now made a few minor changes in the methods section for clarification: a) lines 508-511: "Recombinant proteins generally contained a C-terminal His 6 -tag, although, in some cases, His-tag translation was prevented by the introduction of one or two stop codons at the end of the open-reading frame (RiMep, RiGH36, RiPgm and RiGH113)." b) added the following text at lines 511-512: "The His 6 -tag was excluded to prevent interaction with putative C-terminal active or catalytic residues that could be detrimental to the enzymes' activity" c) and specified that the proteins RiMep, RiGH36, RiPgm and RiGH113 were purified by HIC at line 524.