Protein secretion routes in fungi are mostly determined by the length of the hydrophobic helix in the signal peptide

  1. Tristan Sones-Dykes
  2. Edward Wallace

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

Secreted proteins are translocated across membranes through multiple routes. In eukaryotes, secreted proteins with N-terminal signal sequences can use either the signal recognition particle (SRP) or the alternative Sec63 translocon to cross the endoplasmic reticulum membrane. Large-scale experiments on the substrates of these pathways are primarily from the model yeast Saccharomyces cerevisiae , but less is known about conservation of translocation pathways. Here we take a computational approach to analyse secretion signals across the fungi. Computational predictions by the Phobius model robustly separate known SRP-dependent from Sec63-dependent proteins in S. cerevisiae . Prior work suggested that this separation is driven by the compound hydropathy of the signal peptide’s hydrophobic helix, i.e., its length multiplied by maximum hydropathy. Instead, we find that the length of the hydrophobic helix is the major discriminator in native proteins: 8-13 amino acids for Sec63-dependent proteins and 16-27 amino acids for SRP-dependent proteins. Secreted proteins in diverse fungal species also separate into distinct populations by Phobius predictions and by the length of the hydrophobic helix. Our analysis across fungi shows that distinct functional groups of proteins, including fungal cell wall proteins and extracellular proteins, have cleaved signal peptides with short hydrophobic helices, similarly to Sec63-dependent proteins in S. cerevisiae . Our results suggest that the Sec63 translocon is critical for cell wall biogenesis and protein secretion in fungi, including secretion of major virulence factors in fungal pathogens of plants and animals.

Author Summary

Fungal cells have their “stomachs on the outside” – they secrete proteins to the environment that act as enzymes that digest macromolecules and importers of nutrients back across the cell membrane. For fungi that cause infections, the environment is the infected host animal, plant, or other fungus. Fungal cells are also protected by a cell wall, built by proteins that are secreted across the cell membrane and other proteins that use the cell’s secretory system and are then retained in the membrane. These secreted proteins are translocated across membranes by cellular machinery called translocons, and different secreted proteins use different translocons. Detailed studies of translocons in brewers’ yeast ( Saccharomyces cerevisiae ) identified the features in secreted proteins that make the protein use one translocon compared to another. Specifically, there is a helical structure near the start of secreted proteins, and current understanding is that the hydrophobicity of this helix directs the protein through the matched transposon. Here, we show that the length of the hydrophobic helix, not necessarily its maximum hydrophobicity, matters most in fungal cells: shorter helices using a translocon including the Sec63 component and longer helices using a translocon involving the “Signal Recognition Particle”. Because these translocons differ between yeast and animals, and even differ across the wide diversity of fungi, we next asked if the hydrophobic helices that direct secretion have similar properties in diverse fungi including pathogens of humans, frogs, and plants. Indeed, our computational predictions separate these these helices into short and long helix groups across fungi. The conservation of short helices in cleaved signal peptides across fungi is consistent with their using the Sec63 translocon and not the Signal Recognition Particle, which merits further investigation. Secreted proteins with cleaved signal peptides, that likely use the Sec63 translocon, include cell wall proteins, digestive enzymes, and “effectors” that manipulate the infected host to promote fungal infections.

Article activity feed

  1. Review Commons

    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

    General Statements

    Our study identifies characteristics of secretory signal peptides in fungi, and how their sequence determines which alternative pathways that proteins take to the endoplasmic reticulum. All 3 reviewers grasp this, and agree that the study is publishable. Reviewer 3 puts it well, that we "convincingly show that the length of the hydrophobic helix in a signal peptide is the main factor distinguishing [...] pathways. This simplifies a previous model [...] provides a modest but important advancement to the field of protein secretion. ... The study extends its computational analysis beyond the model yeast Saccharomyces cerevisiae to a diverse range of fungal species."

    Thank you to all the reviewers: we found the reviews fair and constructive. and have addressed them in full.

    In the process of responding to reviews, we softened the claim in the title to "Protein secretion routes in fungi are predicted by the length of the hydrophobic helix in the signal sequence". We also reorganised the manuscript to put the cross-fungal analysis first, followed by the more detailed mechanistic analysis. We feel that this leads a broader audience through the story more effectively. This reorganisation also moved some material from introduction to discussion. Also on larger-scale changes, we reformatted the materials and methods section as requested.

    Reviewer #1 (Evidence, reproducibility and clarity (Required)):

    Summary:

    In this manuscript the authors analyze characteristics of secretory signal peptides in fungi. They identify length of the hydrophobic core rather than overall hydrophobicity as the parameter that determines whether proteins use SRP-dependent cotranslational import through the Sec61 channel, or SRP-independent posttranslational translocation through the hetero-heptameric Sec complex to enter the ER.

    Major comments

    The authors need to adequately use the existing nomenclature in the field:

    There is no 'Sec63 translocon'. Proteins with more hydrophobic signal sequences are targeted to the ER by SRP and its receptor, and these proteins are translocated cotranslationally by the Sec61 channel (aka the translocon). Proteins with less hydrophobic signal sequences are imported into the ER postranslationally by the Sec complex consisting of the Sec61 channel and hetero-tetrameric Sec63 complex (Sec62, Sec63, Sec71, Sec72).

    Sec63 on its own also contributes to co-translational import (Brodsky et al, PNAS, 1995), so the term 'Sec63 translocon' is really confusing and should be replaced by the standard nomenclature as above throughout the paper.

    We sincerely appreciate the advice in correctly navigating terminology in the secretion and translocation field. We now say "Sec complex", and not the incorrect "Sec63 translocon". In the same spirit, we have replaced the terminology "Sec63-dependent" with "Sec-dependent", which is a more accurate description of the overall role of the Sec complex. For example, Ast et al. primarily assayed dependence on the Sec complex using sec72∆ strains.

    The paper should contain a proper methods section.

    We have reformatted the manuscript with a separate materials and methods section in the main manuscript, per Genetics/G3 journal family guidelines.

    The authors should explain more explicitly the differences of the Phobius and DeepTMHMM algorithms. Why was that particular algorithm chosen for comparison to Phobius?

    We initially focused on algorithms that distinguish SPs and TM sequences in a single tool, which both Phobius and DeepTMHMM do. This differs from other algorithms such as the SignalP family, that do not also predict TM sequences - SignalP version 4.0 onwards was indeed trained to exclude TM sequences from their predictions (PMID: 21959131).

    In response to this and the similar comment from reviewer 2, we expanded our analysis to compare with the SignalP6.0 algorithm as well as DeepTMHMM.

    Minor comments

    • p2, para 2: ER protein import has been studied for 50 years, and its complexity been obvious for well over a decade

    We corrected this to "However, detailed functional investigations of secretion mechanisms in eukaryotes have focused on a handful of model yeasts and mammalian cells, revealing unexpected complexity"

    • p2, para 3: ref for the signal sequence should be one of the original Blobel papers instead of [8]

    We added the citation to Blobel and Sabatini, 1971, and kept the 1979 citation as we find the additional context is helpful to readers.

    • p3, para 1: ref for SRP should be Walter, Ibrahimi, & Blobel, JCB 1981, instead of [11]

    We added the original citation, and again kept the more modern citation that summarizes the field in decades following initial discovery.

    • p3, para 1: NB: SRP and its receptor do NOT translocate anything, they TARGET proteins to the ER

    We have corrected this, thank you.

    Reviewer #1 (Significance (Required)):

    The authors report an interesting observation which is of interest to the field and sufficiently well documented in this manuscript to be convincing. The paper does extend our understanding of the critical characteristics of secretory signal peptides.

    A limitation of all signal peptide prediction by current algorithms is that they are trained on 'standard' signal peptides and tend to miss ones that do not sufficiently conform to the standard parameters.

    Thank you for this point, the "standard/non-standard" conceptualization is helpful and we now mention this in our expanded discussion. We agree that testing the limits of these models would involve experimental screening of non-standard or non-natural sequences.

    Reviewer's expertise: SRP and Sec61 channel structure/function analysis, cell-free assays for ER protein import, yeast genetics

    Reviewer #2 (Evidence, reproducibility and clarity (Required)):

    Review of manuscript of Sones-Dykes et al. entitled: 'Protein secretion routes in fungi are mostly determined by the length of the hydrophobic helix in the signal peptide'

    This manuscript deals with the important question of how different fungi exhibit variety in protein targeting to the secretory pathway mostly using bioinformatic sequence analysis. This is important for understanding the evolution of the diverse targeting routes within the early secretory pathway, but also for biotechnology since diverse fungi are used as "biofactories" in biotechnological production of secreted proteins. While the results of the current study mostly confirm the analyses already carried out in S.cerevisiae, the work is important and warrants publication in a suitable journal.

    We appreciate this positive and balanced appraisal.

    Major points:

    Could the authors elaborate what was the motivation to use Phobius and not some other signal peptide predictor? I am wondering because of the cited Ast et al. paper is already several years old and new improved prediction tools such as the latest SignalP iteration have been developed since that study.

    The main motivation to use Phobius, and check with DeepTMHMM, was that these tools simultaneously predict cleaved signal peptides and transmembrane helices, unlike other tools that predict only cleaved signal peptides and can give false positives with N-terminal transmembrane helices.

    To clarify this point, we also emailed Prof. Henrik Nielsen, the lead developer of SignalP. I asked: "Although we mostly used Phobius prediction and also compared to DeepTMHMM, reviewers have asked us to also compare to SignalP. A critical part of our argument is about predictions of the h-region length, so we would like to compare h-region lengths to SignalP4.1 HMM mode in addition to SignalP6.0."

    Prof. Nielsen replied:

    As for your question, I must tell you that SignalP 4.1 does not have an HMM mode at all. The last SignalP version to have an HMM mode was 3.0. Therefore, 4.0, 4.1, and 5.0 do not output signal peptide regions; this was first reintroduced with version 6.0. See also the FAQ tab at the website.

    *You could try to install version 3.0, but for your purpose, I would not recommend it. The old HMM module had a strong preference for certain h-region lengths because of a specific kind of overtraining. This was, at least partially, solved in Phobius through regularization of the length distribution. Since h-region length is a crucial parameter in your analysis, I would not trust the region assignments by SignalP 3.0. You are welcome to cite me for that to the reviewers, if needed. *

    But comparing the region assignments between Phobius and SignalP 6.0 will be interesting.**

    Regarding SignalP3.0, we now cite Liaci et al., who analysed all experimentally verified eukaryotic signal peptides using SignalP 3.0, and Xue et al., who analysed S. cerevisiae signal peptides, and both arrived at similar conclusions that cleaved signal peptides have hydrophobic regions of length 8-14 amino acids.

    Also, we have expanded our analysis to also compare Phobius and SignalP6.0 predictions of entire signal peptides and of h-regions. The comparisons are now in Figures 4, S3, and S4.

    I am slightly puzzled by the analysis of the annotation of the Sec63- and SRP-dependent targeting sequences presented in Fig. 1. Could the "SRP-dependent" sequences with long hydrophobic sequences simply be called transmembrane helices? Based on structure of the SPC, it has been proposed that cleavable signal peptides with h-regions beyond 18 residues are extremely rare so I would imagine that majority of these sequences are longer transmembrane segments.

    The point of this figure is to compare lists of proteins that are experimentally verified to be Sec-dependent or SRP-dependent in their targeting, so that's the correct way to refer to them for the purpose of this analysis. Yes, the conclusion of this paper and other work (e.g. Ast et al.) is that these SRP-dependent sequences with long hydrophobic sequences are mostly transmembrane (TM) helices.

    I appreciate the analysis of protein targeting features in evolutionarily distinct fungal species, but since the authors highlight importance of fungi in heterologous industrial protein production, it would have been satisfying to see some of these fungi included in this analysis. In particular, Pichia pastoris and Trichoderma reesei are commonly used fungi with apparently a highly specialized secretory machinery capable of very high production levels of different secretory proteins. I would urge the authors to consider the aspect of selecting optimal secretion signals for these industrial fungi and perhaps include some discussion of it in this manuscript.

    We added Pichia pastoris (Komagataella phaffii) and Trichoderma reesei to the analysis. We appreciate the suggestion to discuss optimal secretion signals, however, our analysis doesn't directly address that so we chose to leave that point out.

    Minor points:

    The authors state that both Sec63 and SRP pathways converge at the Sec61 translocon. However, we now know that targeting of proteins to Sec61 is even more complicated and for example the EMC is a complex that delivers some proteins to Sec61. It might be appropriate to cite some recent reviews on complexity of early protein targeting to Sec61 in the Introduction.

    As a review of complexity of early protein targeting, we cite a Aviram and Schuldiner 2017 (Targeting and translocation of proteins to the endoplasmic reticulum at a glance). We could add other citations if the reviewer considers this to be necessary.

    Page 5. The authors repeat the compound hydropathy analysis of Ast et al. and used the earlier reported 9-amino acid window for this. Is this analysis result robust with other window sizes?

    Ast et al., checked that this result is robust to window sizes of 9, 11, or 19 aa, in their Figure S1A, which we now specifically mention. In our manuscript, we instead check robustness to different hydropathy scales and prediction algorithms.

    Page 12. Authors state that "cleaved signal peptides do not need to span a membrane". A recent structure of the signal peptidase complex (PMID: 34388369) directly suggests that the signal peptide does span the membrane immediately before its final cleavage. Importantly, the SPC thins the membrane in this region to accommodate the shorter signal peptide h-region and this is proposed as a basis for SPC discriminating between signal peptides and longer transmembrane segments. It would be appropriate to cite this paper in the Discussion.

    Thank you for bringing this important paper to our attention. We have clarified our wording here and cited Liaci et al (PMID: 34388369) in the updated manuscript. Both for the detailed structural discussion, and for similarly concluding that in mammals "Signal peptides possess short h-regions".

    Reviewer #2 (Significance (Required)):

    Protein targeting into the early secretory pathway is an important general concept, and recent years have revealed many new aspects into the diverse mechanisms that cells employ for targeting of proteins with diverse folding needs by use of protein-specific targeting sequences. Also, how proteins are targeted is an important biotechnological question as choice of e.g. the signal peptide can have a dramatic impact on quantity and quality of the produced protein.

    This work is generally interesting to cell biologists studying mechanisms of protein targeting, but the results are mostly confirmatory. Still, no-one has carried out such analysis and fungi are remarkably diverse with potential for new innovations in protein targeting and therefore, the work should be published in my opinion. The suitable audience in my view is quite specialized and could be cell biologists with high interest in fungal protein secretion or biotechnologists using fungi for heterologous expression. For the latter, I would request the authors to extend the data analysis to a few more most biotechnologically relevant fungi and add some discussion on choice of signal peptide in biotechnological protein production in fungi.

    We appreciate this fair perspective. Indeed, we have added analyses of the biotechnologically relevant fungi Komagataella phaffii (Pichia pastoris), and Trichoderma reesei.

    Reviewer #3 (Evidence, reproducibility and clarity (Required)):

    Summary:

    This manuscript revisits the analysis of hydrophobic forces driving endoplasmic reticulum translocation in fungi. Sones-Dykes and Wallace convincingly show that the length of the hydrophobic helix in a signal peptide is the main factor distinguishing SRP-dependent and Sec63-dependent pathways. This simplifies a previous model that relied on a compound hydropathy score, which incorporated both length and hydrophobicity. The analysis, confirmed by Phobius and DeepTMHMM, indicates that length alone is an equally effective and simpler metric for predicting the translocation route in fungi. The study extends its computational analysis beyond the model yeast Saccharomyces cerevisiae to a diverse range of fungal species. It finds that the bimodal distribution of hydrophobic helix lengths-short for predicted Sec63-dependent and long for SRP-dependent proteins-is highly conserved. By broadly identifying proteins with short hydrophobic helixes, the research suggests that the Sec63 translocation route is crucial for cell wall biogenesis and secretion (likely encompassing and the secretion of virulence factors). This provides a functional and pathological context for the translocation pathway choice.

    The manuscript was well written, and its central messages were clear.

    We appreciate this, and are glad that the messages came across clearly.

    Major points:

    Extension of analysis to human secretome: In Fig 4, the helix length analysis is extended to additional organisms, among them Homo sapiens. It is observed that 'h-region lengths in humans had a similar distribution'. However, as the authors themselves note in the introduction, the functional thresholds of signal peptides are dramatically different in mammalian cells. Without overlaying 'ground truth' data of Sec63-dependence in humans, it is difficult to draw any conclusions about the meaning of h region length on human translocation preferences. I would suggest either: (1) Performing an analysis similar to that done in Fig 1 for the human secretome (2) Removing the human outgroup from the analysis in Fig 4.

    We appreciate the reviewer's point, but decided to keep the human analysis as an outgroup in Fig 4. only. This manuscript focuses on fungi by extrapolating and testing results from S. cerevisiae on other fungi. A mechanistic interpretation of signal peptides in human cells is out of scope due to the mentioned differences in functional thresholds of signal peptides in human cells. However, including humans gives a context that we feel readers would ask for if we did not include it.

    If we wanted to analyse the human signal peptides thoroughly then it would be interesting to extend to a more diverse range of eukaryotes, and extend beyond signal peptide prediction algorithms to structural modeling of signal peptides into cognate translocon structures. That's a whole different project.

    Incorporate additional cross-validation: Since the key findings from this paper stem from hydrophobic segment predictions, it would be beneficial to augment the conclusions with another independent analysis. The Hessa scale (PMID: 15674282) has the advantage of being a 'biological' hydrophobicity scale defined by transmembrane helix insertion. It would be important to show that the findings obtained with Phobius (e.g. no improvement in categorization with compound score) also hold with this scale.

    Thank you for this helpful and important point. We also performed the analysis with the Hessa scale, included in the updated manuscript as Figure S2. The Hessa scale looks like a better predictor than the Kyte-Doolittle or Rose scales in that the distributions are clearly different for SRP-dependent and Sec63-dependent proteins. However, there is no improvement in classification, both because the Hessa maximum hydrophobicity distributions for SP and TM groups overlap, and also because the 97.5% accuracy of the length-based prediction is already so good that there's no room to improve in classifying this set of S. cerevisiae sequences.

    Minor points:

    Incorporate GO analysis in Fig 4: Visualization of the GO analysis referenced in the text (Fig 4) may be useful to drive home the point of .

    We have indicated the top enriched GO terms in the paper, and also provided the full GO results in the supplementary data at https://github.com/TristanSones-Dykes/TMSP_Pub. There's not really more information in these GO analyses that makes it worth plotting. For example, for predicted signal peptides in all annotated fungi, "extracellular region" and "cell wall" come up as very highly enriched with extremely low p-values.

    Cite origin of 'ground truth' protein list: The authors cite 83 and 107 bona-fide Sec63-dependent and SRP-dependent proteins which were used to define the 'ground truth' lists. It would be informative to define how these lists were collected; for example, the Ast et al. paper referenced appears to validate ~40-50 proteins as Sec63-dependent.

    The 'ground truth' protein list was collected and curated in the paper by Ast et al., and thoroughly explained there. In our expanded methods section, we now explain their classification based on localisation/mislocalisation of GFP-tagged proteins in sec72∆ (Sec63 complex deficient) strains. After careful checking, we didn't find any flaws in their analysis or any better yeast datasets more recent than 2013. So, we think the approach of giving a brief description here and referring to Ast et al. for a thorough description is most helpful for readers.

    Reviewer #3 (Significance (Required)):

    This manuscript by Sones-Dykes and Wallace provides a modest but important advancement to the field of protein secretion. While previous work has already identified that Sec63-dependent proteins in baker's yeast have moderately hydrophobic signal peptides, this paper refines this concept and extends it for additional fungal species. It will be of interest to researchers studying protein translocation/secretion pathways and fungal biology.

    Thank you for supporting the main point of our paper. We agree with the assessment, and that this analysis needed to be done to discover if and how results from S. cerevisiae extend to other fungi. We hope that this paper will encourage new work on mechanisms of protein secretion in other fungi, especially of the role of the Sec63 complex.

  2. 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

    Summary:

    This manuscript revisits the analysis of hydrophobic forces driving endoplasmic reticulum translocation in fungi. Sones-Dykes and Wallace convincingly show that the length of the hydrophobic helix in a signal peptide is the main factor distinguishing SRP-dependent and Sec63-dependent pathways. This simplifies a previous model that relied on a compound hydropathy score, which incorporated both length and hydrophobicity. The analysis, confirmed by Phobius and DeepTMHMM, indicates that length alone is an equally effective and simpler metric for predicting the translocation route in fungi. The study extends its computational analysis beyond the model yeast Saccharomyces cerevisiae to a diverse range of fungal species. It finds that the bimodal distribution of hydrophobic helix lengths-short for predicted Sec63-dependent and long for SRP-dependent proteins-is highly conserved. By broadly identifying proteins with short hydrophobic helixes, the research suggests that the Sec63 translocation route is crucial for cell wall biogenesis and secretion (likely encompassing and the secretion of virulence factors). This provides a functional and pathological context for the translocation pathway choice. The manuscript was well written, and its central messages were clear.

    Major points:

    • Extension of analysis to human secretome: In Fig 4, the helix length analysis is extended to additional organisms, among them Homo sapiens. It is observed that 'h-region lengths in humans had a similar distribution'. However, as the authors themselves note in the introduction, the functional thresholds of signal peptides are dramatically different in mammalian cells. Without overlaying 'ground truth' data of Sec63-dependence in humans, it is difficult to draw any conclusions about the meaning of h region length on human translocation preferences. I would suggest either: (1) Performing an analysis similar to that done in Fig 1 for the human secretome (2) Removing the human outgroup from the analysis in Fig 4.
    • Incorporate additional cross-validation: Since the key findings from this paper stem from hydrophobic segment predictions, it would be beneficial to augment the conclusions with another independent analysis. The Hessa scale (PMID: 15674282) has the advantage of being a 'biological' hydrophobicity scale defined by transmembrane helix insertion. It would be important to show that the findings obtained with Phobius (e.g. no improvement in categorization with compound score) also hold with this scale.

    Minor points:

    • Incorporate GO analysis in Fig 4: Visualization of the GO analysis referenced in the text (Fig 4) may be useful to drive home the point of .
    • Cite origin of 'ground truth' protein list: The authors cite 83 and 107 bona-fide Sec63-dependent and SRP-dependent proteins which were used to define the 'ground truth' lists. It would be informative to define how these lists were collected; for example, the Ast et al. paper referenced appears to validate ~40-50 proteins as Sec63-dependent.

    Significance

    This manuscript by Sones-Dykes and Wallace provides a modest but important advancement to the field of protein secretion. While previous work has already identified that Sec63-dependent proteins in baker's yeast have moderately hydrophobic signal peptides, this paper refines this concept and extends it for additional fungal species. It will be of interest to researchers studying protein translocation/secretion pathways and fungal biology.

  3. 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

    Review of manuscript of Sones-Dykes et al. entitled: 'Protein secretion routes in fungi are mostly determined by the length of the hydrophobic helix in the signal peptide'

    This manuscript deals with the important question of how different fungi exhibit variety in protein targeting to the secretory pathway mostly using bioinformatic sequence analysis. This is important for understanding the evolution of the diverse targeting routes within the early secretory pathway, but also for biotechnology since diverse fungi are used as "biofactories" in biotechnological production of secreted proteins. While the results of the current study mostly confirm the analyses already carried out in S.cerevisiae, the work is important and warrants publication in a suitable journal.

    Major points:

    1. Could the authors elaborate what was the motivation to use Phobius and not some other signal peptide predictor? I am wondering because of the cited Ast et al. paper is already several years old and new improved prediction tools such as the latest SignalP iteration have been developed since that study.
    2. I am slightly puzzled by the analysis of the annotation of the Sec63- and SRP-dependent targeting sequences presented in Fig. 1. Could the "SRP-dependent" sequences with long hydrophobic sequences simply be called transmembrane helices? Based on structure of the SPC, it has been proposed that cleavable signal peptides with h-regions beyond 18 residues are extremely rare so I would imagine that majority of these sequences are longer transmembrane segments.
    3. I appreciate the analysis of protein targeting features in evolutionarily distinct fungal species, but since the authors highlight importance of fungi in heterologous industrial protein production, it would have been satisfying to see some of these fungi included in this analysis. In particular, Pichia pastoris and Trichoderma reesei are commonly used fungi with apparently a highly specialized secretory machinery capable of very high production levels of different secretory proteins. I would urge the authors to consider the aspect of selecting optimal secretion signals for these industrial fungi and perhaps include some discussion of it in this manuscript.

    Minor points:

    1. The authors state that both Sec63 and SRP pathways converge at the Sec61 translocon. However, we now know that targeting of proteins to Sec61 is even more complicated and for example the EMC is a complex that delivers some proteins to Sec61. It might be appropriate to cite some recent reviews on complexity of early protein targeting to Sec61 in the Introduction.
    2. Page 5. The authors repeat the compound hydropathy analysis of Ast et al. and used the earlier reported 9-amino acid window for this. Is this analysis result robust with other window sizes?
    3. Page 12. Authors state that "cleaved signal peptides do not need to span a membrane". A recent structure of the signal peptidase complex (PMID: 34388369) directly suggests that the signal peptide does span the membrane immediately before its final cleavage. Importantly, the SPC thins the membrane in this region to accommodate the shorter signal peptide h-region and this is proposed as a basis for SPC discriminating between signal peptides and longer transmembrane segments. It would be appropriate to cite this paper in the Discussion.

    Significance

    Protein targeting into the early secretory pathway is an important general concept, and recent years have revealed many new aspects into the diverse mechanisms that cells employ for targeting of proteins with diverse folding needs by use of protein-specific targeting sequences. Also, how proteins are targeted is an important biotechnological question as choice of e.g. the signal peptide can have a dramatic impact on quantity and quality of the produced protein.

    This work is generally interesting to cell biologists studying mechanisms of protein targeting, but the results are mostly confirmatory. Still, no-one has carried out such analysis and fungi are remarkably diverse with potential for new innovations in protein targeting and therefore, the work should be published in my opinion. The suitable audience in my view is quite specialized and could be cell biologists with high interest in fungal protein secretion or biotechnologists using fungi for heterologous expression. For the latter, I would request the authors to extend the data analysis to a few more most biotechnologically relevant fungi and add some discussion on choice of signal peptide in biotechnological protein production in fungi.

  4. 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 manuscript the authors analyze characteristics of secretory signal peptides in fungi. They identify length of the hydrophobic core rather than overall hydrophobicity as the parameter that determines whether proteins use SRP-dependent cotranslational import through the Sec61 channel, or SRP-independent posttranslational translocation through the hetero-heptameric Sec complex to enter the ER.

    Major comments

    1. The authors need to adequately use the existing nomenclature in the field: There is no 'Sec63 translocon'. Proteins with more hydrophobic signal sequences are targeted to the ER by SRP and its receptor, and these proteins are translocated cotranslationally by the Sec61 channel (aka the translocon). Proteins with less hydrophobic signal sequences are imported into the ER postranslationally by the Sec complex consisting of the Sec61 channel and hetero-tetrameric Sec63 complex (Sec62, Sec63, Sec71, Sec72).

    Sec63 on its own also contributes to co-translational import (Brodsky et al, PNAS, 1995), so the term 'Sec63 translocon' is really confusing and should be replaced by the standard nomenclature as above throughout the paper.

    1. The paper should contain a proper methods section.
    2. The authors should explain more explicitly the differences of the Phobius and DeepTMHMM algorithms. Why was that particular algorithm chosen for comparison to Phobius?

    Minor comments

    • p2, para 2: ER protein import has been studied for 50 years, and its complexity been obvious for well over a decade
    • p2, para 3: ref for the signal sequence should be one of the original Blobel papers instead of [8]
    • p3, para 1: ref for SRP should be Walter, Ibrahimi, & Blobel, JCB 1981, instead of [11]
    • p3, para 1: NB: SRP and its receptor do NOT translocate anything, they TARGET proteins to the ER

    Significance

    The authors report an interesting observation which is of interest to the field and sufficiently well documented in this manuscript to be convincing. The paper does extend our understanding of the critical characteristics of secretory signal peptides.

    A limitation of all signal peptide prediction by current algorithms is that they are trained on 'standard' signal peptides and tend to miss ones that do not sufficiently conform to the standard parameters.

    Reviewer's expertise: SRP and Sec61 channel structure/function analysis, cell-free assays for ER protein import, yeast genetics

  5. Version published to 10.1101/2025.07.30.667231 on bioRxiv