Author response:
The following is the authors’ response to the original reviews.
Public Reviews:
Reviewer #1 (Public review):
Summary:
Fecal virome transfer (FVT) has the potential to take advantage of microbiome associated phages to treat diseases such as NEC. However, FVT is also associated with toxicity due to the presence of eukaryotic viruses in the mixture, which are difficult to filter out. The authors use a chemostat propagation system to reduce the presence of eukaryotic viruses (these become lost over time during culture). They show in pig models of NEC that chemostat propagation reduce the incidence of diarrhea induced by FVTs.
Strengths:
The authors report an innovative yet simple approach that has the potential to be useful for future applications. Most of the experiments are easy to follow and performed well.
Weaknesses:
The biggest weakness is that the authors show that their technique addresses safety, but they are unable to demonstrate that they retain efficacy in their NEC model. This could be due to technical issues or perhaps the efficacy of FVT reported in the literature is not robust. If they cannot demonstrate efficacy of the chemostat propagated virome mixture, the value of the study is compromised.
We appreciate the reviewer’s assessment and fully acknowledge that our inability to demonstrate NEC protection by FVT is a limitation to the study. If technical issues cover the variability in disease phenotype in our animal model, which is of a spontaneous nature, then yes we fully agree. Issues with FVT preparation are however unlikely, as this is performed per protocol. The effect of FVT on NEC has hitherto only been demonstrated by our research group in two individual studies using separate donor fecal material, so it is indeed too early to speculate about robustness in FVT response. We have briefly mentioned this in the results (lines 563-565) and discussion (lines 777-779), but agree that it needs further elaboration. We have now revised the discussion and conclusion to better emphasize the extent and consequences of this limitation (lines 793-797 + lines 817-818). Importantly, we show that inclusion of specific nutrients, such as milk oligosaccharides, impacts the resulting propagated fecal-derived virome. One can argue that this is not surprising, but it has nevertheless not been shown before – and it opens up possibilities for future “tailor-made” fecal-derived viromes with predictable profiles and effects.
Even though we do not demonstrate an effect of the chemostat-propagated virome, we still believe that the study provides valuable insights as a proof-of-concept. Specifically, we demonstrate that in vitro chemostat propagation can significantly modulate the safety profile of FVT, while still driving changes in the microbiome, e.g., by decreasing C. perfringens.
The above issue is especially concerning because the chemostat propagation selected for bacteria that may not necessarily be the ones that harbor the beneficial phages. Without an understanding of exactly how FVT works, is it possible to make any conclusion about the usefulness of the chemostat approach?
The chemostat work was based on the idea that if we culture a fecal inoculum under suitable conditions, then the phageome would propagate alongside and allow for a scalable production method for standardized donor-independent FVT. We are cognizant that the chemostat end-culture diverged quite markedly from the fecal inoculum. In reality, such divergence is unavoidable when performing in vitro simulation of intestinal growth conditions. On the positive side, we showed that we could drive an expansion of Bacteroides spp. by supplementing the media with human milk oligosaccharides. We have previously shown that Bacteroides spp. engraft FMT recipients that are in turn protected from NEC. However, there is much room for refinement of the chemostat culture condition; i.e. to preserve the rich repertoire of lactobacilli from the inoculum e.g. by means of lowering the pH. Moreover, the loss of viral diversity in the chemostat end-culture also needs to be addressed, potentially by lowering the chemostat dilution-rate to allow the time for phage propagation. Based on these insights, we will in the near future invest heavily in improving the chemostat procedure to end up with a propagated fecal virome with better resemblance to the fecal inoculum.
Finally, can the authors rule out that their observations in THP-1 cells are driven by LPS or some other bacterial product in the media?
We thank the reviewer for raising this point. To minimize the influence of bacterial contaminants such as LPS or other small bacterial products, we implemented several steps during sample preparation. Specifically, we performed ultrafiltration using a 300 kDa molecular weight cut-off, which should remove small molecules, including LPS, bacterial metabolites, and other potential soluble immunomodulators. Hereafter, all viral preparations underwent endotoxin removal procedures prior to cell exposure. These precautions reduce the likelihood that our observed effects in THP-1 cells are attributable to bacterial products rather than viral components. This is explained in the referenced article (20), but we have now added the clarification to the Methods section of the revised manuscript (lines 222 and 227). The immune expression profile differs markedly between the viral preparations and the E. coli control, e.g. IFNG, TLR3, TLR8, making it highly likely that viral epitopes are the major drivers of the viral preparations with less impact by any potential bacterial epitope contaminant. This is now mentioned in the results section (line 541-543):
Reviewer #2 (Public review):
Major revision
(1) As authors state that the aim of the research is 'We hypothesized that chemostat propagated viromes could modulate the GM and reduce NEC lesions while avoiding potential side effects, such as earlier onset of diarrhea'.
(a) For the efficacy, in Fig 5, there are no significance in stomach pathology and enterocolitis between groups, even between control group and experimental groups, is it because of the low incidence of NEC? This may affect the statistical power of the conclusions. Therefore, it is unclear how one can draw the conclusion that chemostat can reduce NEC lesions?
Thank you for highlighting this important point. We fully agree and would like to clarify that it is not our intention to conclude that chemostat propagation reduces NEC lesions under the experimental settings within this paper. Rather, this was our initial hypothesis, which could not be confirmed. The unexpectedly low incidence of NEC across groups in Piglet Experiment 1 did not allow for a clear conclusion, but the second Piglet Experiment 2 failed to show a NEC-reducing effect. We have stated this important point in the following sections:
- Abstract (line 42-44): “However, these signatures were lost in recipients of chemostat-propagated viromes, and only minor microbiome effects and no NEC prevention were observed.”
- Results (line 699): “This highlights that while chemostat propagation effectively mitigates virus-associated diarrhea, the method needs further optimization to targt NEC.”
- Discussion (lines 773–775): “However, the MO-propagated chemostat virome did not increase Bacteroides or Parabacteroides spp. in the recipient’s gut, nor did it provide NEC protection.”
- We have rephrased this to emphasize the importance of Experiment 2.
- To avoid any potential misinterpretation, we have rephrased line 598 to reflect that we observed “a difference in the clinical side effect pattern” rather than implying efficacy.
- Furthermore, we have updated the summary title for Figure 8 (line 704) to clearly state: “MO-propagated virome modestly exacerbates gastric injury and fails to improve NEC.”
- Also, we have added the following section to the discussion (lines 793-797): “However, we acknowledge that the absence of demonstrated NEC prevention by the native donor virome is a significant limitation to conclusions regarding efficacy. Without a protective baseline, we cannot assess whether the virome efficacy was lost during chemostat propagation. Consequently, we cannot confirm or dismiss the hypothesis that chemostats can preserve a phage community capable of preventing NEC.”
- Lastly, we have updated the conclusion (lines 817-818): “However, as neither the chemostat-propagated viromes nor the native donor virome demonstrated NEC prevention, the efficacy of the chemostat approach remains inconclusive.”
- These changes should clarify that while the study demonstrates improved safety via reduced diarrhea, NEC efficacy was not obtained.
(b) More convincing pathology images would be helpful.
Since we did not observe a protective effect against NEC with either of the treatments, we opted not to include pathology images. However, extensive examples can be found in the cited paper (reference 37), which describes our NEC scoring methodology in the Methods section (lines 268-271): https://doi.org/10.1016/j.yexmp.2024.104936.
(c) For the safety, such as body weight development, FVT had no statistical significance difference from control, CVT, and CVT-MO, so how can you drawn the conclusion that chemostat can avoid potential side effects?
We appreciate the reviewer’s observation. To clarify, we do not claim that chemostat propagation completely avoids all potential side effects, but rather that it mitigates them. As shown in Fig. 5G, FVT recipients exhibited significantly reduced body weight gain compared to controls, CVT, and CVT-MO specifically on day 4, but not on day 5. This transient effect suggests that side effects such as reduced growth and early-onset diarrhea are delayed, not entirely prevented, by chemostat propagation. This is stated in the results section in lines 593-595. We also believe that this is consistent with the paper title and the conclusion that the chemostat process minimizes the adverse effects associated with native FVT (line 813).
(d) There is lack of evidence to convince the reader that there is a decrease of eukaryotic viruses. More quantitative data here would be useful.
Apart from the fact that it is impossible for eukaryotic viruses to shed in a system devoid of eukaryotic cells, and that the chemostat runs continuously exchanges the culture, thereby diluting any substance incapable of propagation, we agree that quantitative data to demonstrate a reduction of eukaryotic virus load is lacking.
However, in this case we believe the relative viral abundance data are almost as convincing. To make this even clearer, we have produced new graphs showing 1) the eukaryotic viral abundance relative to total viral abundance and 2) observed eukaryotic viral species, both after medium subtraction. Eukaryotic viral relative abundances decrease from around 0.4% to approach zero already in the batch phase, and similarly number of eukaryotic viral species decrease from around 10 in the fecal inoculum to zero midway through the chemostat phase. These new graphs are now part of Supplementary figure S3 B-C. Moreover, an error in the eukaryotic viral heatmaps presented in Figure 3F now means that the relative abundance of each sample (column) now sums up to 100%. Please also notice from the lower heatmap (where the virome signature of the medium is subtracted) that no eukaryotic viruses are identified from the sequencing data of the samples from the chemostat from 50 hours and onwards.
However, for future experiments we will consider adding a known quantity of a marker virus to the inoculum and monitoring its concentration (e.g., by qPCR) throughout the culture process. Importantly, if the resulting virome is meant for in vivo testing, this marker virus should be inert to the receiving organism.
(2) Questions regarding Fig 3F,
(a) How can the medium have 'the baseline viral content' ?
As we have previously seen persistent eukaryotic viral signals in metagenomics sequencing data from chemostat experiments, we sampled and sequenced the culture medium. As is seen from Figure 3F, this only concerns Dicistroviridae, as the patterns of the remaining eukaryotic viral signals before and after medium subtraction are virtually similar. For some reason, a component of the culture medium contains a genetic signal from this entity. Since all culture components are sterilized, it is most likely genomic traces that are then continuously supplied with the medium and appears in all culture samples. As it is unlikely to derive from intact viruses, the in vivo implications are deemed minimal.
(b) What is the statistical significance of relative abundance of specific eukaryotic viruses?
The same as any statistical comparison on single OTU level in a nucleotide sequencing dataset. As commented above, it does not prove a quantitative depletion of eukaryotic virus throughout the chemostat process but given the context a reduction in relative abundance supports the notion that eukaryotic viruses are indeed depleted when the culture medium is exchanged. The relevant question to us is: What is the magnitude of depletion? Which is particularly relevant since the clinical data indicates a delay and not a prevention of side effects after transplantation. Hence, as proposed above, the use of a marker virus would provide us with that answer.
(c) The hosts for some of the listed eukaryotic viruses are neither pigs or human, as such the significance of a decrease in these viruses to humans is unclear.
Dicistroviridae is not present in the inoculum and shows up only when medium is added. Picobirnavirus and Astrovirus are relevant mammalian intestinal viruses, whereas Smacoviridae is less well described (dois: 10.3389/fvets.2020.615293 and 10.3390/v8020042). Genomoviridae as a fungal virus indeed appears to be less relevant in the case of the mammalian intestine. Indeed, at any given time point in any given individual, be it a pig or a human, it would carry with it several viral species that are incapable of infecting it, most likely transiting after being ingested with food, or in the case of pigs through rummaging. It is no secret that we have been searching for a causative agent responsible for the clinical side effect patterns related with FVT, but there seems to be no consistent viral agent that is overabundant in diarrheal piglets. Hence, in this study, we are mostly interested in the proof-of-concept for overall eukaryotic virus reduction through chemostat propagation, and we believe we have presented data in support of this.
(3) In this study, pH 6.5 was selected as the pH value for chemostat cultivation, but considering the different adaptability of different bacteria to pH, it is recommended to further explore the effect of pH on bacteria and virus groups. In particular, it was optimized to maintain the growth of beneficial bacteria such as Lactobacillaceae and Bacteroides in order to improve the effect of chemostat cultivation.
We agree that pH is a key parameter in shaping microbial communities during chemostat cultivation. As noted, we selected pH 6.5 to balance physiological relevance and bacterial viability, but we acknowledge that this pH may not be optimal for supporting the growth of certain potentially beneficial taxa such as Lactobacillaceae. We explicitly address this in the discussion (lines 736–741), where we state that the selected pH may have limited engraftment and that future studies should investigate pH optimization to better support bacterial groups and improve the overall effectiveness of the cultivation system.
(4) Please improve the quality of the images, charts, error bars and statistical significance markers throughout and mark the n's. used in each experiment.
We have carefully reviewed all figures and could not identify any general image quality issues. If some specific images or panels appear unclear or problematic, we would appreciate it if the reviewer could point them out so we can address them directly.
Regarding sample sizes, the number of animals (n) is indicated in Fig. 5A and its legend, as well as in Fig. 8A. We have now also added this information to the legend of Fig. 8 for clarity.
To improve the clarity of statistical findings, we have added asterisks to denote significance in panels 6A, 6F, and 7A, as requested.
To improve the clarity of Fig. 3B, we have added a dashed line to separate LAC and LAC-MO.
Reviewer #3 (Public review):
Major revisions
This study investigated the in vitro amplification of donor fecal virus using chemostat culturing technology, aiming to reduce eukaryotic virus load while preserving bacteriophage community diversity, thereby optimizing the safety and efficacy of FVT. The research employed a preterm pig model to evaluate the effects of chemostat-propagated viromes (CVT) in preventing necrotizing enterocolitis (NEC) and mitigating adverse effects such as diarrhea.
Strengths:
Enhanced Safety Profile: Chemostat cultivation effectively reduced eukaryotic virus load, thereby minimizing the potential infection risks associated with virome transplantation and offering a safer virome preparation method for clinical applications.
Process Reproducibility: The chemostat system achieved stable amplification of bacteriophage communities (Bray-Curtis similarity >70%), mitigating the impact of donor fecal variability on therapeutic efficacy.
Weaknesses:
Loss of Phage Functionality: The chemostat cultivation resulted in a reduction in phage diversity (e.g., the loss of Lactobacillaceae phages), which may compromise their protective effects against NEC (potentially linked to the immunomodulatory functions of Lactobacilli). The authors should explicitly address this limitation in the discussion section, particularly if additional experiments cannot be conducted to resolve it within the current study.
We appreciate the reviewer’s concern and agree that the loss of phage diversity during chemostat cultivation, especially phages targeting Lactobacillaceae, is an important limitation with potential implications for NEC protection.
We already described the depletion of Lactobacillaceae in the chemostat and its implications in the discussion (lines 742-751 + 787-793), along with our plans to address this in future work by adjusting culture pH. However, we acknowledge that the significance of losing phage diversity deserves more explicit attention. Accordingly, we have expanded the discussion to highlight the possible consequences of this loss and its impact on phage functionality (see lines 758–762), as suggested by the reviewer.
Limitations in Experimental Design: The low incidence of NEC lesions in the control group reduced the statistical power of the study. This limitation undermines the ability to conclusively evaluate the efficacy and safety of the chemostat-propagated virome as a novel intervention for NEC. Future studies should optimize experimental conditions (e.g., using a more NEC-susceptible model or diet) to ensure adequate disease incidence for robust statistical comparisons.
We agree that the low NEC incidence in Experiment 1 limited the statistical power to evaluate efficacy. To address this, we designed Experiment 2 using a more NEC-inducing diet (formula 2), which resulted in a higher level of baseline lesions. This allowed for a more conclusive assessment, demonstrating that the MO-propagated chemostat virome did not provide NEC protection when using the donor feces and culture conditions applied in this experiment.
We acknowledge that this was too unclear in the original manuscript. Please see the response to the first comment by Reviewer 2, where we have highlighted several revisions to improve clarity.
However, we do believe the data are robust enough to conclude that the level of diarrhea — and thereby safety — was improved in the piglet model, which is why we chose to focus on this aspect in the paper’s title.
Recommendations for the authors:
Reviewer #3 (Recommendations for the authors):
The manuscript presents a well-structured study investigating the feasibility of using chemostat-based culturing of the fecal virome to reduce the transfer of eukaryotic viruses during fecal virome transfer (FVT). Utilizing both in vitro fermentation systems and a preterm piglet model, the authors explore whether this method could be a safer and equally effective alternative to raw FVT for treating neonatal intestinal diseases, such as necrotizing enterocolitis (NEC). This study introduces a novel mitigation strategy for FVT through chemostat fermentation. However, a significant revision is recommended before the manuscript can be considered for publication.
Major Changes:
- A central aim of the study was to assess whether chemostat-cultured viromes maintain protective effects against NEC. However, this key outcome remains "unresolved" due to the low incidence of NEC in the control group. The discussion should address this limitation.
We fully acknowledge this limitation and agree that our study cannot conclude whether the NEC effect of FVT was maintained without demonstrating an effect of this native virome. Please see our response to a similar concern raised by Reviewer 1, where we describe the revisions made to the discussion (lines 793-797) and conclusion (lines 817-818).
- The section on viral particle enrichment should be expanded and discussed in more detail. It would be beneficial to examine its efficiency in separating bacteria from viral-like particles (VLPs) compared to findings from previously reported studies. The authors should clarify the rationale behind the selected dose of VLPs used in the experiments and their role in virus engraftment results.
We selected the virome isolation method based on previous experiments within our lab, demonstrating efficient separation of bacteria and virus particles, using a 0.45 um filter syringe. Filtrates were quality assessed by fluorescence microscopy, showing absence of intact bacteria. Using a diverse mock virus community, we also showed a high degree of preservation of infective viruses in the FVT following the isolation procedures. We have now expanded the description of the separation method in the results section with a reference to this work (lines 188-190). We did however choose to increase the molecular weight cut off (MWCO) to enhance the exclusion of non-viral components.
We acknowledge that the rationale and importance of the VLP dose was lacking in the discussion. This has now been added (line 758-762).
- The viral richness of chemostat viromes was significantly lower than that of native feces. The authors should discuss how this may impact microbiome and virome outcomes.
We have included this point in the new section about VLP dose in the discussion. Please see lines 758-762.
- The immune response was assessed through THP-1 cells and a limited piglet cytokine panel. These may not fully represent the intestinal epithelial or mucosal immune responses. Thus, authors should acknowledge these limitations in the discussion section.
Thank you for the comment. The limitation of using THP-1 cells as an in vitro model is already acknowledged in the results section (line 545): “Since fecal-derived eukaryotic viruses mainly infect intestinal cells, an
in vivo stimulation may reveal a different response pattern. ”
The limited panel of porcine cytokines was not intended as a comprehensive assessment of the mucosal immune response, but rather as supportive data for NEC-associated inflammation, as we have previously demonstrated (reference 37: https://doi.org/10.1016/j.yexmp.2024.104936). To obtain a comprehensive view of the immune response, a few days after diarrhoea onset, we additionally performed RNA-Seq analyses of the intestinal lymph node.
- While the manuscript is comprehensive, it is also lengthy and text-heavy. Some sections could be condensed for clarity.
The manuscript has been through multiple revisions by authors. While it is indeed lengthy, we have removed non-essential information and redundancies and now feel that the balance between data, text, figures, and supplementary information is acceptable.
- Several figures (e.g., Figs. 1-5) contain significant data but need clearer summaries in their captions.
We appreciate the suggestion and have revised the captions for Figs. 1-8 to provide clearer, more informative summaries of the data they present.
