Multi-omic analysis of bat versus human fibroblasts reveals altered central metabolism

  1. N. Suhas Jagannathan
  2. Javier Yu Peng Koh
  3. Younghwan Lee
  4. Radoslaw M Sobota
  5. Aaron Irving
  6. Lin-Fa Wang
  7. Yoko Itahana
  8. Koji Itahana
  9. Lisa Tucker-Kellogg

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Bats have unique characteristics compared to other mammals, including increased longevity and higher resistance to cancer and infectious disease. While previous studies have analyzed the metabolic requirements for flight, it is still unclear how bat metabolism supports these unique features, and no study has integrated metabolomics, transcriptomics, and proteomics to characterize bat metabolism. In this work, we performed a multi-omics data analysis using a computational model of metabolic fluxes to identify fundamental differences in central metabolism between primary lung fibroblast cell lines from the black flying fox fruit bat ( Pteropus alecto ) and human. Bat cells showed higher expression levels of Complex I components of electron transport chain (ETC), but, remarkably, a lower rate of oxygen consumption (OCR). Computational modeling interpreted these results as indicating that Complex II activity may be low or reversed, similar to an ischemic state. An ischemic-like state of bats was also supported by decreased levels of central metabolites and increased ratios of succinate to fumarate in bat cells. Ischemic states tend to produce reactive oxygen species (ROS), which would be incompatible with the longevity of bats. However, bat cells had higher antioxidant reservoirs (higher total glutathione and higher ratio of NADPH to NADP) despite higher mitochondrial ROS levels. In addition, bat cells were more resistant to glucose deprivation and had increased resistance to ferroptosis, one of the characteristics of which is oxidative stress. Thus, our studies revealed distinct differences in the ETC regulation and metabolic stress responses between human and bat cells.

Article activity feed

  5. Review Commons

    Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    The authors do not wish to provide a response at this time.

  6. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    In this work, the authors generate a multi-omic dataset (RNA, proteomic and metabolomic) from fibroblast cell-lines of human and bat origins, in study of the specific differences in bat that allows them to have a good cancer resistance and longevity. They specifically focus on metabolic differences between humans and bats. They perform differential analysis followed by GO enrichment analysis to highlight differences related to the electron transport chain both at the level of RNA and protein abundance. They then use FBA sampling and specific constraints to propose an hypothesis of reverse direction of the second complex of the ETC, as well as better resistance to ROS, which they support with several subsequent experiments.

    Overall, the paper is very well written, the findings are presented clearly and efficiently. For the most part, the assumption and limits of the study are clearly stated by the author (notably with respect to the limits of using only cell lines). In my opinion, the goal of the paper, which is presented as a stepping stone into further characterisation of the metabolic differences between human and bat for potential oncological research benefits, is clearly stated and appropriate.

    There are however several points that I think are important to address inorder to improve the quality of the scientific work and its interest for the rest of the scientific community.

    Major

    The authors state:
    "We then set the lower bound of the PaLung Complex I reaction flux to a value equal to 70% of its theoretical maximum. Similarly, we set the upper bound of the WI-38 Complex I reaction at a value equal to 30% of its theoretical maximum value. This ensured that the PaLung model would have higher flux through the Complex I reaction, in comparison to the WI-38 model."

    How do the results hold with different thresholds ? Are these findings robust with e.g. in ranges between 10 to 50% (90-50%) (instead of only 30% and 70%). Furthermore, the histogram figures doesnt seem to reflect a 70% of maximum lower bound for complex I (threshold at a value of 30 seems like extremity of tail).

    Number of differentially expressed genes is extremely high because such cutoffs are not really meaningful given the comparison between two organisms. No need to refer to the 6247 above cutoff as differentially regulated genes (see: https://elevanth.org/blog/2023/07/17/none-of-the-above/ and https://daniel-saunders-phil.github.io/imagination_machine/posts/if-none-of-the-above-then-what/ for pointers toward current best practice in biological statistics). Enough to simply note that 6247 are above the cutoffs, which suggest a drastic (and expected) difference in expression profiles between the two organisms.

    Please highlight the RNA and proteomic analysis assumption and present results within those boundaries (e.g. how are the transcript matched between human and bat, the use of human gene ontologies, etc...). Are the human GO set definitions relevant in bat (it is a common practice with mice and rats, are bats close ?)?

    Are oxphos and hypoxia responses the most extreme pathway scores in the GSEA ? Instead of barcode plots that are generally not a very useful use of figure space, use fig 1C to show the top e.g.20 (positive and negative) pathway scores so that we can see how much those two actually stand out. Same for the proteomic analysis. Also, need to show an unbiased side by side comparison of the pathway enrichments for RNA and proteomic, the reported results in main text and figures are too cherry picked to be of interest as they stand.

    Finally, and very importantly, please upload ALL the code used for the analysis, with instructions to run it and all the required inputs and source files. The computational analysis is only as credible as it is easy to reproduce.

    Minor

    Introduce GeTMM, what are its key specificities ?

    Fig 1C code bar plot useless, simply report ES and NES and pathway absolute rank in text.

    Report Foldchange/p-value/rank of complex-I members and other genes of interest for the narrative of the paper.

    Referees cross-commenting

    I also think the comments from the other reviewers are appropriate.

    Significance

    In my opinion, the goal of the paper, which is presented as a stepping stone into further characterisation of the metabolic differences between human and bat for potential oncological research benefits, is clearly stated and appropriate.

    Broadly interesting for oncological research.

    My espertise is multi-omic data analysis and integration with prior knowledge in the context of complexe diseases.

  7. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    Jagannathan et al. performed a multi-omics comparative analysis between fibroblast cells from bats of the species P. alecto and humans. Using a combination of transcriptomics, proteomics and metabolomics, the authors showed differences in central metabolism between the cells of the two species. Specifically, the authors noted higher expression of Complex I components of the electron transport chain, as well as low activity of the Complex II. The computational modeling suggested that the latter is indicative of a state resembling the state of ischemia. Furthermore, the expression of antioxidant components was interpreted as higher in the bat compared to human cells, which is in accordance with previous reports.

    Overall, it is a comprehensive multi-omics approach performed with a very interesting biological object, such as a bat. However, some aspects, especially the part of bioinformatic analysis, needs to be enhanced. The existence of differences in mitochondrial metabolism is also not surprising, given the evolutionary distance and very different ecology and lifestyle between the two species, but a more mechanistic follow-up would be of greater interest. Nevertheless, it is the first study using integrated omics approach of that sort done on bats.

    Major:

    1. The authors compared a fibroblast cell line derived from adult bats with a human embryonic cell line. Please discuss whether mitochondrial metabolism in embryonic cells might be different and how it could have affected the obtained results. Please describe in more detail how the cells were established, what population doubling they were used at (both bat and human cells). Were the cells cultured in atmospheric oxygen or low-oxygen conditions. The exposure of cells to atmospheric oxygen might affect the many mitochondrial parameters measured in this study and could influence the main finding about ischemic-like state. Additionally, please mention in the limitations of the study that only biological n=1 was compared (since cells only from 1 individual per species was used in experimental groups), despite n=3 technical replicates.
    2. Reference genomes for bats are not as well annotated as for human. Downregulation of a pathway may result from some genes being excluded from the analysis because of poor annotation of the P. Alecto genome compared to human. The authors state: "Genes with counts per million (CPM) < 1 in more than 3 out of 6 samples were discarded from downstream analysis". So, if the gene was not annotated, was it assigned a zero value and discarded? Was it discarded if it was zero in one species (e.g. bat) or set to 0? If such genes were excluded, while in reality not being mapped, they could have skewed the pathway analysis.
    3. All conclusions are based on high-throughput data, however it is accepted that some validation should be provided. Please provide qPCR or WB (if good antibodies are available) validation for several most significantly differentially expressed genes supporting the pathways identified in Figure 2 (preferably supporting the conclusions about Complexes I/II).
    4. The major findings of this paper were based on the omic data, followed by some experimental validations. However, the quality of these omic data or the results are not solid enough to motivate the authors to validate these findings. For example, both of the GO terms enriched by the DEGs in Fig.1 are not the top terms as claimed by the authors (not even significant after multiple test correction). Also, even though the 2 GO terms in Fig.2 are quite significant, the expression pattern seems not very consistent among the replicates, which make the enrichments not so solid. This highlights an inconsistency among different omic datasets, which may generate some conflicting results. For example, the low level of metabolites from TCA cycle (Fig.4c) seems not consistent with the high level of TCA-related protein, as described in Fig.2c & d. For the purpose of improving the manuscript quality, the authors may have to evaluate the consistency among the multiple omic datasets or to optimize their bioinformatic pipeline to enhance the results.
    5. The dominant up-regulation of complex I in ETC is interesting and is the main finding of this paper. However, no experimental evidence was provided to prove the greater activity of Complex I, for example, metabolites changes. In addition, the genes encoding proteins belong to ETC complex I, II, III and IV vary a lot, with much more genes encoding complex I. Therefore, the author should consider the background gene number when they compare the up-regulated gene number differences in each complex. For example, a fisher-exact test could be done to see if complex I has significantly more genes been up-regulated than a random expectation.
    6. If the main findings of this paper can be further confirmed by additional experiments or data, it will be a very nice paper. This could be a potential mechanism that bats used to switch metabolism modes between two metabolic extremes: flight and hibernation, which require high and low energy. However, the usage of only the lung fibroblasts of human and bat may limit the ability of generalizing this 'ischemic-like state' of ETC in most of the bats tissue/organs. While I agree what the authors mentioned in the discussion section, that to extend to primary cells of other species can help generalize this finding, studying the metabolism state of different cell type of bats (e.g., muscle cells responsible for flight; myocytes and neurons for hibernation) probably can provide more insights into the evolution of various interesting phenotypes of bats.

    Minor:

    • the author may have to add the p value or FDR for each GSEA plot, even though some of the FDR are not significant. Also, it will be better to show the normalized enrichment score (NES) instead of the ES.
    • the gene set name in several supplementary tables contains many '%' characters and those needs to be removed.
    • in Line 302, "...combined with the earlier findings of downregulated OxPhos expression and low OCR, we conclude...". If my understanding is right, the authors only mentioned the up-regulation of Oxphos expression, instead of down-regulation. This sentence may need to be clarified.
    • How did mitochondrial DNA content per cell compared between the two species? Could the results be affected by the number and size of the mitochondria per cell in each species? An indirect measurement of mitochondrial DNA yield in the fractionation experiment would be the total DNA amount that was obtained in mitochondrial fractions per cell lysed.

    Significance

    The work is significant considering the limitations stated above. This may be considered a pilot study of brand significance.

  8. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    Summary

    In this study, the authors conducted a multi-omics analysis comparing cells from the long-lived bat, Pteropus alecto, and human cells. Their findings revealed that bat cells express higher levels of mitochondrial complex I components and exhibit a lower rate of oxygen consumption. Moreover, computational modeling suggested that the activity of complex II in bat cells might be low or even reversed, similar to the conditions observed during ischemia. The decrease in central metabolites and the increased ratio of succinate to fumarate in bat cells might indicate an ischemia-like metabolic state. Despite having high mitochondrial ROS levels, bat cells exhibit higher levels of total glutathione and a higher ratio of NADPH to NADP. Additionally, bat cells showed resistance to glucose deprivation and induction of ferroptosis.

    Major comments

    1. Regarding Figure 1A, the authors mention 'n = 3' for a single cell line. Does this refer to three different passages or three independent experiments? Please provide a more detailed description to clarify.
    2. In relation to Figures 1C and 1D, the authors state in the figure legend that the 'GSEA analysis identifies Respiratory electron transport and Cellular response to hypoxia as the top metabolic pathways that are differentially regulated between PaLung and WI-38 cells.' (Lines 140-144). However, the criteria for selecting these terms as the top metabolic pathways is not clear. In the lists in Supplementary Tables 2 and 3, the authors' proposed term, 'Respiratory electron transport,' is ranked 126th, and 'Cellular response to hypoxia' is ranked 79th. Conversely, terms related to the TCA cycle are ranked 66th and 82nd, and another term that seems to be related to hypoxia, 'OXYGEN-DEPENDENT PROLINE HYDROXYLATION OF HYPOXIA-INDUCIBLE FACTOR ALPHA,' is ranked 62nd. Could the authors please provide a clarification for their choice of 'Respiratory electron transport' and 'Cellular response to hypoxia' as the top metabolic pathways?
    3. In the Materials and Methods section (lines 419-421), the authors mention, 'GSEA was run against the complete Gene ontology biological process (GO BP) gene set list (containing 18356 gene sets).' However, they narrow down the gene dataset for analysis (lines 136-138, 'we filtered our gene dataset to contain only genes listed under the Gene ontology category Cellular Metabolic Process (GO ID:0044237), resulting in a truncated list of 4794 genes.'). I'm concerned that this selective approach might introduce bias into the resultant pathways. Is this selective approach commonly employed in this type of analysis? And isn't there a need for adjustments to avoid potential bias?
    4. The authors noted that the number of differentially expressed genes (DEGs) is quite high (6,247 out of 14,986) as per lines 134-135, stating that "The number of differentially expressed genes (6,247) was extremely high, suggesting that multiple pathways are differentially regulated between the two species." However, this large number of DEGs could indicate either an improper correction procedure or a need for a more stringent threshold. The authors should address this issue to avoid potential misinterpretation of the results.
    5. In Figure 2B, the samples labeled as W1 and P1 appear to be outliers. This raises questions about the integrity of the sampling or analysis process. Please describe about this.
    6. Regarding the GSEA analysis of Fig. 2, they are using the full set of GSEA. However, this reviewer is wondering if this is appropriate when analyzing mitochondrial fractions, as I believe using the entire GSEA set could introduce a bias. Is this a common approach? Shouldn't the authors be focusing on mitochondrial-related sets within the GSEA, and then determining the upregulated and downregulated pathways from there?
    7. The authors describe in lines 195-197, "GSEA-flagged upregulation in OxPhos was driven mostly by the upregulation of Complex I subunits, for both the proteomic and transcriptomic data (Figure 2G, Supplementary Figure S1D)." However, within this analysis, the number of genes composing each subgroup of the mitochondrial Complexes are 44 for Complex I, 4 for Complex II, 10 for Complex III, and 19 for Complex IV (https://www.genenames.org/data/genegroup/#!/group/639). The authors mention that the genes of Complex I were dominant in the ETC, but, might this just be reflecting the original difference in the number of genes? As this reviewer believes this could have a significant impact on the authors' current claims, this reviewer suggest the authors to carefully reconsider this point, comparing the actual results with the proportion expected from the difference in gene numbers. (Even in Fig. S1D, it appears to correlate with the number of genes: C1 39.3%, C3 10.7%, C4 10.7%, C2 3.5%)
    8. As pointed out in Major Point 7, if the authors' claim of enrichment in Complex I is indeed due to the large number of genes included in the Complex I subgroup (https://www.genenames.org/data/genegroup/#!/group/639), can the assumption of High Complex I flux truly be considered valid? In that case, this constraints model would become inappropriate, and the validity of the inferred low or reverse activity of Complex II would be diminished. Therefore, a careful re-examination is desirable.
    9. (option, takes about 1-2 months). This reviewer believes that the authors' most important claim, concerning the high activity of Complex I and the low activity of Complex II, lacks strong evidence as no biochemical data of the activities of each mitochondrial complex are presented to substantiate this. Unless additional biochemical experimental data is provided, the assertions should be toned down. While the abstract mentions "complex II activity may be low or reversed," it is stated with certainty in line 108 of the introduction, "associated with the low or reverse activity of Complex II." Based on the present data, this reviewer believes that the claim remains speculative. Therefore, I suggest moderating the overall argument or adding the biochemical data. While the results from metabolomics are supportive, they do not serve as direct evidence.
    10. Regarding Figure 5, the title of the figure states "lower antioxidant response", but it doesn't seem that the data in the figure actually shows a lower antioxidant response.
    11. In lines 109-110 of the Introduction, the authors state, "we confirmed our prediction of ischemic-like basal metabolism in PaLung cells by characterizing the response of bat cells to cellular stresses such as oxidative stress, nutrient deprivation, and a type of cell death related to ischemia, viz. ferroptosis." However, can the assertion that the cells are in an ischemic-like state be confirmed simply because they are resistant to several types of cellular stress?

    Minor points:

    1. The authors mention the use of cufflinks/Tophat for mapping/quantification. However, support for these software programs has ended and the creators of these programs themselves recommend using the successor programs. I recommend re-analysis using a more current pipeline (such as HISAT2/StringTie, STAR/RSEM, etc.). Furthermore, the transcriptomics section of the methods should also include the program used for cleaning and trimming.
    2. As for the Oxygen Consumption Rate (OCR) data presented in Figure 2F, it makes sense that it's low at the basal level. However, it's perplexing that it is also low even under uncoupled conditions, especially considering the high energy demand associated with flight in this species. Could the authors provide their interpretation on this apparent contradiction?
    3. In line 156, the authors mention that 'Profiling detected a total of 1,469 proteins.' Please provide more details in the explanation. Specifically, does this total of 1,469 proteins represent a combined count from both humans and bats, or is this the number of proteins for which orthologs could be identified in both species, just like the authors did with the transcript results.
    4. In Supplementary Table 4, only 127 mitochondrial proteins are listed out of the 405 proteins mentioned in "Of these 405 proteins, we identified 127 to be core mitochondrial proteins (lines 161-163)". As there is no explanation for this within Supplementary Table 4, it would be better to include one.
    5. In line 472, the phrase "GO BB gene set list" is used. Could this potentially be a typographical error, and should it instead be "GO BP gene set list"?
    6. In the volcano plot of Fig. S3B, it appears that the side with lower P/W values generally corresponds with lower p-values. I wonder if there might have been any oversight or mistake in the data analysis process that could explain this observation?
    7. In lines 249-252, it is stated, "The low or negative flux values for Complex II in our PaLung simulations indicate that the electrons obtained from Complex I may accumulate at Complex II or potentially even get consumed by Complex II operating in reverse (bypassing the rest of the ETC) in PaLung cells." However, isn't the basic process of electron transfer done through Complex I-III-IV, independent of Complex II?
    8. Regarding Figure 4F, the authors state, 'PaLung cells displayed higher viability than WI-38 cells after glucose deprivation (Figure 4F).' However, in addition to the cell images, it would be beneficial to perform experimental quantification of cell death to provide more rigorous data. Additionally, the cells appear to be over-confluent, which might influence the results. Also, scale bars should be included in all photos, including Fig. 6.
    9. Regarding Figure 5B, it is stated that 'the expression levels of differentially expressed antioxidant genes' are shown, but it includes those that are not significant. It would be helpful if the authors could clarify how this gene set was selected.
    10. Regarding Figure 6C, the values for total glutathione seem to significantly differ from those in Figure 5C. An explanation for this discrepancy would be appreciated to ensure the consistency and reliability of the data.

    Referees cross-commenting

    I think the comments from the other reviewers are appropriate.

    Significance

    Collectively, these intriguing results from the interspecies comparison provide novel insights into the differences in metabolism and cellular characteristics between bat and human cells. However, the study has some limitations, notably certain weaknesses in the data and potential overstating of certain interpretations. Addressing these issues would enhance the overall quality and robustness of the manuscript. Furthermore, if feasible, conducting a biochemical analysis of each mitochondrial complex activity would solidify the authors' main conclusions.

  9. Version published to 10.1101/2023.05.08.537600v1 on bioRxiv