Horizontal transfer of an antimicrobial peptide across insects

Article activity feed

1. 

   Review Commons

   Oct 16, 2026

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

   Learn more at Review Commons

   ---

   Reply to the reviewers

   Response to reviews is provided as an attached file.

   Read the original source
2. 

   Review Commons

   Oct 16, 2026

   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

   Summary

   In this manuscript, the authors report how they have investigated the presence of the Drosomycin gene in insects, starting from the unexpected discovery of five Drosomycin-like genes in Drosophila busckii of the subgenus dorsilopha whereas it had been found only in the Melanogaster group of the Sophophora subgenus and has not been identified in closer relatives such as other groups of the Sophophora subgenus or the Drosophila subgenus. They document that the presence of Drosomycin-like clusters in both D. melanogaster and D. busckii may result from horizontal gene transfer based on synteny arguments. Importantly, two D. …

   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

   Summary

   In this manuscript, the authors report how they have investigated the presence of the Drosomycin gene in insects, starting from the unexpected discovery of five Drosomycin-like genes in Drosophila busckii of the subgenus dorsilopha whereas it had been found only in the Melanogaster group of the Sophophora subgenus and has not been identified in closer relatives such as other groups of the Sophophora subgenus or the Drosophila subgenus. They document that the presence of Drosomycin-like clusters in both D. melanogaster and D. busckii may result from horizontal gene transfer based on synteny arguments. Importantly, two D. busckii Drosomycin-like genes appear to be preferentially induced by a Gram-positive bacterial challenge than one by Gram-negative bacteria, in keeping with a possible regulation of their expression by the Toll and not Immune deficiency pathway. They then extend their bioinformatics strategy to discover further Drosomycin-like genes in sequence or expression data from 24 winged insect orders. This analysis reveals that only a limited number of orders and a limited number of species within orders have acquired Drosomycin-like genes in their genomes. They conclude that the acquisition of the Drosomycin-like genes has proceeded through Horizontal Gene Transfer (HGT) followed by some cases of secondary losses.

   Major comments

   The authors make a compelling argument for the hypothesis of seven events of HGT vs. more than 50 secondary loss events required to account for their data if a common ancestor were assumed (Fig. S5). This reviewer is convinced by this study and has mostly minor comments to improve the manuscript, except for two major comments detailed below.

   Fig. 5 should include the Drosomycin gene discovered in Musca domestica. By the same token, the text in the summary is inaccurate: " We show that in Diptera...". It should read: "We show in Drosophilidae..". Actually, the authors of ref. 44 have not relied solely on a protein database entry as wrongly claimed by the authors. Ref. 44 states: "In addition, our BLASTP search retrieved a hit from Musca domestica that has identical sequence to Drosomycin. We also confirmed its presence in the M. domestica genome by PCR amplification and DNA sequencing (Fig. 6b)." Thus, the authors should be more cautious in their statements: this apparent mistake weakens their overall argument and questions the sensitivity of their approach. At the very least, they should check by PCR amplification and DNA sequencing on several isolates of M. domestica whether they can amplify the Drosomycin gene as reported by Zhu & Gao.

   Anoither shortcoming is the poor description of the RTqPCR data that seem to depend on only three experiments. It is not described how many times the experiments have been independently performed, whether the only three data points correspond to biological replicates of a single experiment, and which type of statistical analysis has been used (of note, in theory, a parametric test can be applied only if the data have been shown to follow a Normal law of distribution...). If the data correspond to three biological replicates of five flies each, then the experiment cannot be validated, especially given the variability of the signal for Drsl4 after a mixed challenge (Fig. 1D). The advice is to have more independent experiments to increase the confidence in the presented data, which is not expected to take much time and should very easily be completed within two months (a week if flies are available...).

   Minor comments

   1. "sanger" sequencing: while becoming a common name is a testimony to how popular a device or method has become, the authors may want to remember that Prof. Fred Sanger is the only researcher awarded two Nobel Prices in the same discipline...
   2. p7, left column: "to this end, we screened up to 15 insect genomes". In the right-hand column, one reads: "However, within Orthoptera (N=23)"...
   3. Ref. 75, included in the Discussion on the antifungal activity of Drosomycin and Drosomycin-like Defensin peptides: this reference deals with Metchnikowin, which is a proline-rich antimicrobial peptide and does not belong to the Defensin family.
   4. Argument on the short promoters that may be transferred (interesting idea): while indeed NF-kappaB binding sites are often found in the proximal region of the promoter, they can also be found in other positions, e.g., the third binding site of the Diptericin promoter that is more than a kb away from the transcription start site. If this reviewer recalls correctly, there is also a kappaB site that lies downstream of the D. melanogaster Drosomycin gene. Thus, the authors might want to validate their concept more carefully in future work by performing a more detailed analysis of potential kappaB binding sites in the Drosomycin-like genes regions.
   5. SRA: the definition is given only in the Supplementary Text and not in the main text.
   6. Several figures are really difficult to read as they are too small: Fig. 3, Fig. 4, Fig. S2 and also Fig. S4 (names of the species difficult to read). Even at high magnification on the computer, it is very difficult to read the sequences displayed in 4A, due to poor resolution.
   7. Fig. S2: the figure legend does not adequately describe the figure. What are the turquoise and yellow arrows? The exact sequences of the kappaB sites should be provided in a second panel of this figure.
   8. Legends Fig. S3 and Fig. S7: please, give the reference numbers, respectively 61-62-3 and 112.
   9. Legend Fig.S4. It may help the reader to indicate that Liposcelis sp. sequences are on top of the panel. It would also be desirable to highlight the D. mel sequence when present in multiple alignments.
   10. Ref. 76 is not correct; the article has recently been published in PLoS Pathogens and not in Immunology.

   Significance

   This study provides a careful analysis of the distribution of Drosomycin-like genes in insects and provides a compelling argument for HGT being the mechanism behind the remarkably patchy distribution of this antifungal gene within insect order and species. This however is not a surprise in that the antifungal function of Drosomycin was discovered upon realizing that it presented sequence similarity to that of a plant antifungal peptide (rs-AFP2; Terras et al., FEBS Letters 1993, 316, 233-240.). An even more striking observation is that the Drosomycin gene has been found to be transferred to nematodes through HGT (ref. 44) and has also been found in humans (ref. 74). It would have been worth mentioning specifically these observations that generalize even more the work made by the authors on insects. Indeed, these disparate observations may fit well with the authors' proposition of an initial development of this type of molecules in bacteria in their warfare against fungi that might have been repeatedly transferred to eukaryotes. How related at the sequence level are the Drosomycins found in plants, ecdysozoans, and humans? A limitation of the study acknowledged by the authors is that no "wet" experiments have been performed to determine whether the different peptides identified in this study are bona fide antifungal effectors of the innate immune response. This approach would be needed to fully understand the evolution of function of these sequences in the different recipient hosts.¬¬

   This work will be of interest to evolutionary immunologists and more generally to anyone interested in antimicrobial peptides and innate immunity.

   This reviewer is an expert of the study of Drosophila innate immunity and is not a specialist of evolutionary processes.

   Read the original source
3. 

   Review Commons

   Oct 16, 2026

   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

   Summary

   This study investigates the evolutionary origin of the drosomycin (Drs) across insects using comparative genomics, transcriptomics, phylogenetics, and gene expression analyses. The authors screen >670 genomes/transcriptomes across 24 insect orders and report a highly discontinuous distribution of Drs genes. They identify Drs homologs in Drosophila (restricted clades), cockroaches, mantises, select beetles, a katydid, and pseudogenes in booklice, but not in most other insect lineages. To explain this pattern, the authors argue that multiple horizontal gene transfer (HGT) events (~7) are more parsimonious than extensive gene …

   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

   Summary

   This study investigates the evolutionary origin of the drosomycin (Drs) across insects using comparative genomics, transcriptomics, phylogenetics, and gene expression analyses. The authors screen >670 genomes/transcriptomes across 24 insect orders and report a highly discontinuous distribution of Drs genes. They identify Drs homologs in Drosophila (restricted clades), cockroaches, mantises, select beetles, a katydid, and pseudogenes in booklice, but not in most other insect lineages. To explain this pattern, the authors argue that multiple horizontal gene transfer (HGT) events (~7) are more parsimonious than extensive gene loss (>50 independent losses). Supporting evidence includes: Genomic synteny analyses suggesting independent insertions, Codon conservation (serine usage bias) inconsistent with convergence, Paraphyletic phylogenetic clustering of Drs sequences. Evidence of immune inducibility in at least one non-Drosophila lineage. The study concludes that HGT is an underappreciated mechanism for immune gene innovation in insects, potentially enabling rapid acquisition of pathogen defense traits.

   General comment

   This is an ambitious and conceptually important study addressing a long-standing question in immune gene evolution: how similar antimicrobial peptides appear across distantly related taxa. The dataset is large and the integrative approach (genomics + expression + phylogenetics) is a major strength. However, the central claim-multiple HGT events of an AMP gene-is not yet fully supported at the level required for such a strong evolutionary conclusion. While the authors present suggestive evidence, key alternative explanations (e.g., extreme gene loss, convergence under constraint, contamination, or assembly artifacts) are not fully excluded. The manuscript would benefit from stronger phylogenetic and genomic evidence of HGT, more rigorous exclusion of contamination and assembly issues, and clearer framing of conclusions as probabilistic rather than definitive Overall, the study is promising and potentially impactful but requires substantial revision to strengthen evidence and temper conclusions.

   Major comments

   Strength of evidence for HGT vs alternative explanations The core claim-that Drs distribution is best explained by HGT-is plausible but not conclusively demonstrated. The argument relies heavily on parsimony (7 gains vs >50 losses) , but parsimony alone is insufficient in deep evolutionary contexts where gene loss is common. Alternative explanations are not fully ruled out: Extensive lineage-specific gene loss, Convergent evolution under structural constraints, and hidden homologs missed due to rapid divergence Specific revisions: Reframe conclusions as "consistent with HGT" rather than definitive evidence. Provide statistical model comparison (e.g., likelihood-based gain/loss models vs HGT scenarios) Phylogenetic analysis is insufficiently resolved The phylogenetic trees show low posterior support and limited resolution due to short sequence length (~80-130 nt). Paraphyly is used as evidence of HGT, but low support weakens this claim. No clear donor-recipient relationships are identified

   3.3 Functional validation is lacking The manuscript assumes Drs homologs are functional AMPs, but no antimicrobial assays are performed. Expression data are limited (only one lineage tested experimentally) This weakens claims that HGT contributes to functional immune innovation

   3.4 Contamination and assembly artifacts not fully excluded The study relies heavily on public genomic datasets. Some assemblies are acknowledged as previously contaminated (e.g., cockroach genome issues). Limited validation (PCR in some taxa) is performed

   3.5 Limited mechanistic evidence for HGT The manuscript speculates on mechanisms (e.g., transposons), but no clear vector or mechanism identified. No consistent association with mobile elements. Please clearly separate speculation from evidence

   3.6 Sampling bias and absence inference Absence of Drs is used heavily in arguments, but benome completeness varies across taxa. Absence detection is uncertain, Authors should provide quantitative completeness metrics (e.g., BUSCO) for all datasets

   3.7 Overinterpretation of codon usage analysis The serine codon argument is interesting but not definitive as codon conservation could reflect functional constraint, not shared ancestry. Limited number of informative sites

   Minor comments - Clarify replication and statistical tests for qPCR experiments. Provide exact sample sizes and variance measures. Define thresholds for HMM/BLAST detection more explicitly

   Significance

   General assessment

   The study addresses a highly significant and unresolved question in evolutionary immunology: the origin of antimicrobial peptides across taxa. Its strength lies in large-scale comparative analysis and integration across disciplines. There are some minor issues as Evidence for HGT remains circumstantial rather than conclusive and functional relevance is largely inferred Audience

   Reproducibility and rigor

   Strengths:

   • Large dataset (673 genomes/transcriptomes)
   • Multiple complementary methods (HMM, BLAST, phylogeny, expression)

   Concerns:

   • Limited methodological detail for:
     • Filtering criteria
     • Statistical analyses
   • Reproducibility would benefit from:
     • Public pipelines/code
     • Full dataset availability

   1. Suggested major revisions (summary)

   Required

   1. Temper conclusions regarding HGT; present as best-supported hypothesis
   2. Strengthen phylogenetic analyses and report support values clearly
   3. Improve contamination controls and validation
   4. Clarify reproducibility (methods, thresholds, statistics)
   5. Expand discussion of alternative explanations

   Optional

   1. Functional validation of Drs peptides
   2. Advanced phylogenetic reconciliation analyses
   3. Mechanistic exploration of HGT vectors

   Final recommendation

   Major revision

   The manuscript has strong potential and addresses an important question, but key claims require more rigorous support and more cautious interpretation before publication.

   Reviewer expertise - insect biology and genomics.

   Read the original source
4. 

   Review Commons

   Oct 16, 2026

   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

   In this manuscript, Aumont and colleagues conduct a thorough search of numerous insect genomes for the presence of genes encoding Drosomycin family antimicrobial peptides. They take advantage of a large set of existing genome sequences and use an appropriate search algorithm to find Drosomycin paralogs. They then map the presence or absence of paralogs onto a known phylogeny of the insects. Based on the patterns of presence and absence across the tree, they infer that an excessively large number of independent evolutionary losses would be required to explain the disparate distribution of Drosomycins across insects, and that repeated …

   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

   In this manuscript, Aumont and colleagues conduct a thorough search of numerous insect genomes for the presence of genes encoding Drosomycin family antimicrobial peptides. They take advantage of a large set of existing genome sequences and use an appropriate search algorithm to find Drosomycin paralogs. They then map the presence or absence of paralogs onto a known phylogeny of the insects. Based on the patterns of presence and absence across the tree, they infer that an excessively large number of independent evolutionary losses would be required to explain the disparate distribution of Drosomycins across insects, and that repeated horizontal gene transfer is a more likely explanation.

   The detection of Drosomycin paralogs is performed well, and the degree of similarity in primary amino acid across distantly related species is striking. The inference that horizontal transfer is more likely than evolutionary convergence or repeated independent loss is logical and probably true, but is not formally tested against any evolutionary model other than parsimony. There are opportunities for more complete evolutionary analyses, including examination of gene turnover rate within the Drosomycin family after each putative horizontal transfer. The extreme amino acid similarity in the mature peptides contrasts sharply with high divergence in the signal peptides, but this contrast is not explored. Are the Drosomycin mature peptides more similar across species than other housekeeping genes are, as would be expected under HGT? By inferring ancestral sequences of the Drosomycins, is it possible to guess which taxonomic groups were the origins of each transfer? If the signal peptides are assumed to evolve approximately neutrally, it might be possible to use them to date the age or order of gene duplication events after each horizontal transfer, or to date the transfers themselves after inferring an ancestral sequence. Not all such analyses may be fruitful and the authors could select the most informative ones for inclusion in the manuscript, but the total absence of any of them leaves the paper feeling more a like an interesting anecdote than a thorough work of evolutionary biology.

   Gene family turnover dynamics are alluded to throughout the manuscript but are never formally described. The first paragraph of Results section could be a good place to do that. The paragraph describes restriction of Drosomycin genes to the Melanogaster subgroup, but the Drosomycins are not described as a gene family in that paragraph. Thus, the occurrence of 5 Drosomycin-like paralogs from D. busckii in the next paragraph appears to be exceptional in number as well as evolutionary placement. But I believe there are 12 Drosomycin-like genes in melanogaster, so busckii actually has fewer. Establishing Drosomycins as encoded by dynamic multigene families will give context to the remainder of the manuscript. At the end of the last paragraph of the right column in page 3, the authors could indicate explicitly whether they believe the busckii family results from a single transfer and subsequent expansion, and based on what evidence.

   The D. busckii Drsl genes 1-3 were not induced by infection, left column of Results page column 5. In D. melanogaster, Drosomycin-like peptides are expressed in the gut. The authors could dissect D. busckii guts and look for expression of the DbDrsl genes. The logic presented in the right column of page 12, that the genes are small enough that their regulatory sequences could be transferred with them, would also apply to enhancers that drive expression in the gut.

   The passages on detection of Drosomycins from cockroaches, mantids, and katydid are very interesting and I agree with the authors that they probably occur by ancient horizontal transfer. However, the authors have not formally rejected a hypothesis of evolutionary convergence in these passages. I think that alternative hypothesis should be easy to reject based on the similarity of the mature peptides as shown in Figure 3, but it should be formally tested.

   The analysis of the Serine codons is clever and forms a convincing argument against convergent evolution. The authors could quantify the probability of each of the 4 serines using the same codon across all taxa under convergent evolution and it is probably vanishingly small. On the other hand, identical codon usage is exactly what you would predict if the entire Drosomycin family is descended from a paralog that was present in the ancestor of insects. The HGT model as applied to this data does assume a common ancestor at the base of Insecta, and that could be explained in this passage.

   The last paragraph of the left column and first paragraph of the right column on page 8 describe unexpected similarity of signal peptides between cockroaches and beetles, but the sequence divergence between these groups is neither quantified nor benchmarked to any expectation. That results in these paragraphs coming across as vague and unrigorous.

   In Figure 4, I would suggest adding phylogeny of the insect taxa. Some readers of this paper will not be entomologist and will not know how the species groups relate to each other. Seeing the insect tree will allow comparison of the Drosomycin tree for concordance. A larger insect tree is given in Figure 5, so I would give just the focal taxa in Figure 4.

   Minor comments

   It would be easier to refer to specific passages in the manuscript if line numbering was employed.

   In the middle of the right column on page 2, the authors mention convergent evolution of a similar structure in Defensins and cite two recent papers. This same observation was made more than 30 years ago by Broekaert et al (1995) Plant Physiol. 108:1353-8. The authors could decide at their discretion whether they would like to include a reference to the earlier paper.

   If the authors wish for a contextual example of repeated horizontal transfer in insect defenses, they could look to the aip and cdt genes of aphids and Drosophila.

   The second paragraph of the right column on page 3 states that the annotated Drs in housefly is probably a contamination but does not provide any evidence to support that statement. The paragraph should more clearly state that the authors examined the Musca genome carefully and could not find evidence of the reported paralog. It might also be worth noting whether the alleged Musca paralog matches the sequence of any other known Drosomycin, thereby suggesting a possible source of contamination (eg, is it a D. melanogaster sequence?).

   In the left column of page 5, I'm not sure why it is surprising that the DptC of D. busckii is induced by infection by both G+ and G- bacteria. The Toll=G+, IMD=G- dichotomy of D. melanogaster is probably restricted to melanogaster and closely related species. One of the authors of this paper has previously shown that D. busckii has 8 DptC genes (https://www.biorxiv.org/content/10.1101/2025.06.06.658223v1) so it is not clear which paralog is being referenced in this paragraph.

   Delete the period after "pomonae" in the left column of page 8.

   The right column of page 9 refers to "a remarkably specific pattern of evolution" but I did not catch what specific pattern is being referenced. The repeated loss of Drosomycins but not other genes? There is no formal analysis here, only verbal argument and intuition so the verbal argument should be precise.

   The discussion might benefit from an explanation of how transposon-mediated horizontal transfer between distantly related species would work. Especially because the authors do not find sequence signatures of TEs, this suggestion appears weak. However, there does not appear to be any other plausible argument either.

   In the top of the left paragraph on page 14, it is unclear whether the suggested dietary shift is inferred to cause pressure on gut-expressed Drosomycins, risk of systemic infection, or both.

   The passage at the top of the right column of page 14 describes shifting understanding of the evolutionary history of Drosocin and Attacins. As written, this passage will probably be difficult to understand for readers who are not already familiar with those genes.

   I do not think this paper needs a summary conclusion section. I would delete that section.

   Significance

   In this manuscript, Aumont and colleagues conduct a thorough search of numerous insect genomes for the presence of genes encoding Drosomycin family antimicrobial peptides. They take advantage of a large set of existing genome sequences and use an appropriate search algorithm to find Drosomycin paralogs. They then map the presence or absence of paralogs onto a known phylogeny of the insects. Based on the patterns of presence and absence across the tree, they infer that an excessively large number of independent evolutionary losses would be required to explain the disparate distribution of Drosomycins across insects, and that repeated horizontal gene transfer is a more likely explanation.

   Read the original source
5. 

   Version published to 10.64898/2026.03.03.709459 on bioRxiv

   Mar 5, 2026