In Silico Design and Subsequent Expression of Human Papillomavirus-16 and -18 L1 Vaccine Antigens in Broccoli

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This is an interesting manuscript about experiments aiming to find an effective and affordable HPV-vaccine, which is a very relevant issue and interesting for this journal. Fusion constructs were created (presumably?, see below) between heatlabile enterotoxin B subunit (as adjuvant) and modified L1 capsid proteins from the high-risk types HPV16 and HPV18, and successfully expressed in broccoli after nuclear transformation. Extracts from these broccoli lines were shown to generate a significant immune response in mice immunized either orally or subcutaneously.

In addition, the authors present the results of several bioinformatics analyses on the L1 proteins of HPV16 and HPV18, but here there are some very major issues (see below) and these experiments seem very poorly integrated in the design of the antigen constructs. In addition it is unclear what they contribute to the available experimental data.

Overall, the strategy of producing HPC vaccines from broccoli lines is interesting and shows some nice initial results, but there are several very serious issues in the presentation of the results and some of the bioinformatics experiments seem poorly designed.

Major changes are necessary.

 

Major comments:

Naming confusion:

A major issue with how this manuscript is written, is the inadequate descriptions of the different L1-containing sequences, which severely compromise the presentation of the results. In several cases, contradictory information is provided. For example, HPV16-L1 is reported to contain 503 residues in the Methods (L140), a His6-tagged version of HPV16-L1 in Figure 2 has 503 residues, and the HPV16-L1 vaccine construct (which supposedly also contains LTB and a linker sequence, in addition to the His6-tag) has also 503 residues according to Table 2 and L375. This cannot all be correct. The name LVC is equally loosely defined and used. This confusion has to be cleared up throughout the manuscript. In addition it would be helpful if different L1 versions would be given a separate name, and if this was used consistently.

And e.g. L140: …503 amino acids…:             These numbers do not add up either. HPV16-L1 in Uniprot has 505 residues, not 503. The sequence in Figure 2 has 503 residues, but this sequence is not just L1 alone, because it has a His6-tag. Subtracting the 1^st 10 residues from a 505-residue sequence (according to the Methods), and adding 6 C-terminal His residues gives a sequence length of 501…, not 503.  Please provide a better description for the sequences, and make it easier for the reader to distinguish between the different variants of both L1 sequences. It is almost impossible to know for certain which sequence is used where, the wild-type, the modified L1 sequences, the sequences used in bionformatics, or the sequences used in the constructs?

 

What species is detected in Figure 9?

The VLP positive controls have the same apparent molecular weight (56 kDa) as the plant-expressed L1 constructs. These constructs are reported to contain also LTB, a linker sequence and a His6-tag, which theoretically would make them significantly larger at  ~65-70 kDa in size. This discrepancy needs an explanation.

Taken together with the naming confusion mentioned above, this even begs the question: Is it certain that these L1-containing proteins detected in Western blot and ELISA are indeed LTB-linker-L1-His6 fusions? If yes, these results warrant a closer investigation, and further experiments to establish what protein species these plant-derived samples contain are strongly recommended. Western blots using antibodies against LTB and polyhistidine would be enlightening. Is LTB present anywhere to fulfil its role as adjuvant? If not as part of the fusion protein, then separately?

 

Tertiary structure prediction of vaccine constructs (methods 2.4, 2.5, results 3.4, 3.5, figures 2-5):

Here there are some issues in the experimental approach. The authors appear to choose to create a homology model of one subunit of the pentameric L1 capsomer using a server, even though experimentally determined models of at least the HPV-16 L1 capsomer exist (10.1016/s1097-2765(00)80449-9). Arguably, modeling a single subunit would lead to inferior results of the model itself, especially in the subunit-subunit interfaces, which could adopt a different, biologically non-relevant, conformation.

In addition, it is not explained why the vaccine constructs used in the actual wet-lab experiments (including LTB) were not modelled.

Using these L1-only models subsequently in B-cell epitope mapping and molecular docking could lead to distorted results, e.g. identifying inaccessible epitopes, or improbable docking interfaces which would be buried inside the capsomer or sterically blocked by LTB.

Modelling the L1-capsomers, fused to the LTB pentamers, should produce more relevant results for the vaccine candidate itself, and for the subsequent experiments using these models.

 

Molecular docking of LVC with TLR4 (methods 2.5, results 3.5, Figure 5):

This is a very curious experiment on many levels, which cast severe doubts on the results.

The most fundamental concern here is that no reason is given for modeling an interaction between TLR-4 and HPV L1 subunits. Is there a rationale for including it in a manuscript about vaccine antigens? TLR-4 mainly binds LPS and LPS-like molecules. As far as I am aware, no direct interaction between TLR-4 (or TLR-4 and MD-2) and HPV-L1 has been reported in literature. If there are studies showing there is an interaction between HPV L1 and TLr-4/MD-2, then these should be detailed in the text. If there are no indications for an interaction, then these docking studies should be removed from the manuscript completely.

Also, the experimental approach to docking elicits some concerns. To begin with, there is the choice to choose to model an interaction using the single subunit of L1, which is likely to result in interactions which are biologically not relevant. Indeed, from the very limited infromation about the interactinos which can be seen from Figure 5, it looks like that is indeed the case. The interaction interface from L1 looks like it is a subunit interface inside the pentameric capsomer. Not including LTB also diminishes the relevance for the present vaccine study.

Next, the choice for the model for the other partner in the docking experiment, TLR-4/MD-2. Here, a model of the activated dimer, containing a TLR-4 dimer, 2 bound MD-2 molecules and 2 bound LPS molecules buried in the interfaces, was taken as a basis. LPS was then removed, which in real life would destabilize the dimer. TLR-4 and MD-2 are both glycoproteins and it is unclear if the glycans were present in the models used for docking. Omitting them could potentially lead to docking results involving sites which in real life would be covered by glycans. It is not clear from the limited information if that is the case.

Considering the many serious issues, removing the docking results altogether is recommended.

 

Formula 1 and quantification:

It is not made clear why Formula 1 should apply here and can be used for quantification. Data should be provided supporting the unwritten assumption that performance in this ELISA of L1 in the VLPs (the standard) and of LTB-L1 in the capsomers (from broccoli) is the same.

In addition, the formula should be modified to address the differences in weight between the L1 in the VLPs and the LTB-L1, at least if the concentrations are expressed in weight/volume, and also the difference (if any) in MW between the LTB-L1-constructs from HPV16 and HPV18.

 

Oral vaccine:
The interest in this research lies primarily in its  possible use as an oral vaccine. It feels like a missed opportunity not to have included (mashed) broccoli leaves in the immunogenicity assay in addition to TSP.

 

Methods versus Results versus Discussion:

There is considerable overlap between Methods and Results sections, with many small experimental details mentioned both in methods and results, where a single mention in the methods would suffice. The reverse also applies. The same can also be said for Results and Discussion sections.

 

Section 3.3 and table 2:

Why are the physicochemical properties calculated for L1 constructs other than the ones used for expression in broccoli? It would make more sense to provide information on these constructs.

 

Discussion:

The first two paragraphs could be shortened drastically because they repeat the introduction. 

In general, much of the bioinformatics results are not really discussed here. There is quite a lot of literature about these properties and they should be compared with the results from this study.

In addition, both HPV VLPs and capsomers have previously been expressed, even in plants, and by some of the same authors. A discussion and comparison of the results would be expected here in the discussion.

 

L583-585: “The evaluation of the vaccine physiochemical features proved that both the vaccine constructs were quite stable but insoluble upon over-expression.”   -- This is evidently incorrect! The constructs can be detected in TSP extracts, thereby proving that they are soluble. This illustrates that the bioinformatics results should be treated with more caution. This particular statement should be removed, and instead the discrepancies between predicted and observed properties (not only solubility, but also molecular weight) should be discussed.

 

Figure 11:

What do the error bars represent?

 

 

 

Minor comments:

L94-98: “Prophylactic vaccine designing is the ultimate ray of hope followed by in silico analysis of the HPV target that can not only serve as protection from laborious trial-and-error experimental failures, but also minimize the expenditures on vaccine synthesis by providing a solid direction for designing the vaccine”.   – This sentence needs clarification.

L112 and l119: the same argument is made twice

L121: Additionally,…    : This is not a entirely separate argument, because the use of plants in raw or partially processed form reduces cold-chain maintenance and transportation costs.

L128: the 3D structure … was generated:  a structural model was generated

L140, L143: please provide a database number and/or reference for the L1 sequences

L142: why was this aspartate substituted?

L168: define LVC better, see above

L169: provide a database nr or reference for the LTB sequence used here.

L184-187: Which criteria were used in searching for homologous sequences?

L218-219: “This webserver also determines whether a target gene is being expressed in the host system by analyzing its GC content”:    This is incorrect. Codon optimization will probably change the GC-content of the sequence towards the GC-content of broccoli, but GC-content in itself does not determine whether a gene is expressed or not. This needs to be changed in methods, results and discussion.

L237: please provide information about the cultivar of the broccoli used, or its provenance.

L295: please provide more information about the generation of the baculovirus-derived VLPs, or a reference. Which L1 was used in the VLPs? Did they contain the complete native sequence?

L299: please provide origin of MD2H11 (and MAB 1.3.5.15 and P4543). Do all these antibodies bind equally well to both types of L1?

Formula 1: (TSP -> % TSP

L346: what were the doses actually used in immunization?

L465: please provide observations on the appearance of the plants. Are there any differences with regular broccoli plants or are they similar?

L492: which L1 protein construct has a MW of 56.5 kDa. What is the MW of the other L1 construct? What construct is this exactly?

L501: is the Ritti01 the same antibody as the one mentioned in the Methods? Please add a reference.

L586: “significantly non-homologous” What does this mean?

L589-590: “good quality”: see above for a different opinion.

L638: “yield of 0.33 and 0.35 % of total soluble protein”. These yields should be discussed with respect to the minumum yields required for an oral vaccine.

Figure 1: is the LTB present in the construct? It would be nice to include it in the figure.

Figure 2: Figure 2A shows both the primary sequence and secondary structure. How was the 2ndary structure generated? Predictions, from the refined 3D model, or other?

Figure 9: The VLP positive controls used here are unspecified: Which L1 was used here?

Figure 10: how many replicates were used?

Author Response

Reviewer’s Reports

Reviewer 1

 

Comments and Suggestions for Authors

This is an interesting manuscript about experiments aiming to find an effective and affordable HPV-vaccine, which is a very relevant issue and interesting for this journal. Fusion constructs were created (presumably?, see below) between heatlabile enterotoxin B subunit (as adjuvant) and modified L1 capsid proteins from the high-risk types HPV16 and HPV18, and successfully expressed in broccoli after nuclear transformation. Extracts from these broccoli lines were shown to generate a significant immune response in mice immunized either orally or subcutaneously.

In addition, the authors present the results of several bioinformatics analyses on the L1 proteins of HPV16 and HPV18, but here there are some very major issues (see below) and these experiments seem very poorly integrated in the design of the antigen constructs. In addition it is unclear what they contribute to the available experimental data.

Overall, the strategy of producing HPC vaccines from broccoli lines is interesting and shows some nice initial results, but there are several very serious issues in the presentation of the results and some of the bioinformatics experiments seem poorly designed.

Major changes are necessary.

Response: 

Thank you very much for your comments. We appreciate your time for reviewing the manuscript thoroughly, and providing valuable suggestions that improved our manuscript. We have revised the manuscript and the changes are highlighted for quick view.  Below point-by-point answers to the comments are provided.

 

Major comments:

Comment 1: Naming confusion:

A major issue with how this manuscript is written, is the inadequate descriptions of the different L1-containing sequences, which severely compromise the presentation of the results. In several cases, contradictory information is provided. For example, HPV16-L1 is reported to contain 503 residues in the Methods (L140), a His6-tagged version of HPV16-L1 in Figure 2 has 503 residues, and the HPV16-L1 vaccine construct (which supposedly also contains LTB and a linker sequence, in addition to the His6-tag) has also 503 residues according to Table 2 and L375. This cannot all be correct. The name LVC is equally loosely defined and used. This confusion has to be cleared up throughout the manuscript. In addition it would be helpful if different L1 versions would be given a separate name, and if this was used consistently.

And e.g. L140: …503 amino acids…:             These numbers do not add up either. HPV16-L1 in Uniprot has 505 residues, not 503. The sequence in Figure 2 has 503 residues, but this sequence is not just L1 alone, because it has a His6-tag. Subtracting the 1^st 10 residues from a 505-residue sequence (according to the Methods), and adding 6 C-terminal His residues gives a sequence length of 501…, not 503.  Please provide a better description for the sequences, and make it easier for the reader to distinguish between the different variants of both L1 sequences. It is almost impossible to know for certain which sequence is used where, the wild-type, the modified L1 sequences, the sequences used in bioinformatics, or the sequences used in the constructs?

Response

We fully agree with the reviewer’s observation. In response, the manuscript has been carefully revised to eliminate any ambiguity related to sequence naming and description. All modifications are clearly indicated and can be tracked in the revised version in sections 2.1 and 2.3. Specifically, all L1-related sequences are now consistently named throughout the manuscript to clearly distinguish between the wild-type L1 sequences (L1-WT16 and L1-WT18), the modified capsomeric L1 variants (L1-Cap16 and L1-Cap18), and the final vaccine constructs (LTB–L1-Cap16-His and LTB–L1-Cap18-His). In addition, all sequence lengths have been recalculated and corrected, and previous inconsistencies in amino acid numbering have been resolved. To further improve clarity, database accession numbers for the original sequences have now been provided.

 

Major Comment 2: What species is detected in Figure 9?

The VLP positive controls have the same apparent molecular weight (56 kDa) as the plant-expressed L1 constructs. These constructs are reported to contain also LTB, a linker sequence and a His6-tag, which theoretically would make them significantly larger at  ~65-70 kDa in size. This discrepancy needs an explanation.

Taken together with the naming confusion mentioned above, this even begs the question: Is it certain that these L1-containing proteins detected in Western blot and ELISA are indeed LTB-linker-L1-His6 fusions? If yes, these results warrant a closer investigation, and further experiments to establish what protein species these plant-derived samples contain are strongly recommended. Western blots using antibodies against LTB and polyhistidine would be enlightening. Is LTB present anywhere to fulfil its role as adjuvant? If not as part of the fusion protein, then separately?

Response

The full-length LTB–L1 fusion protein with an expected molecular weight of approximately 73 kDa was originally anticipated for both HPV-16 and 18, Western blot analysis consistently revealed a predominant band at ~56.5 kDa, corresponding to the molecular weight of the L1 monomer. This finding is now interpreted as post-translational cleavage of the LTB adjuvant in plants, most likely mediated by endogenous plant proteases. Similar cleavage events are well documented in plant-based expression systems and do not necessarily compromise antigen integrity or immunogenicity. Importantly, we now clearly state in both the Results section 3.9 and Discussion that the immunogenic species detected correspond predominantly to L1 rather than the intact fusion protein. This interpretation is fully consistent with the antigen-capture ELISA data and the mouse immunization results, which together confirm that conformational L1 epitopes remain intact, functional, and biologically relevant. While we agree that probing with anti-LTB antibodies would provide additional insight, this has now been acknowledged as a future optimization strategy.

 

Major Comment 3: Tertiary structure prediction of vaccine constructs (methods 2.4, 2.5, results 3.4, 3.5, figures 2-5):

Here there are some issues in the experimental approach. The authors appear to choose to create a homology model of one subunit of the pentameric L1 capsomer using a server, even though experimentally determined models of at least the HPV-16 L1 capsomer exist (10.1016/s1097-2765(00)80449-9). Arguably, modeling a single subunit would lead to inferior results of the model itself, especially in the subunit-subunit interfaces, which could adopt a different, biologically non-relevant, conformation.

In addition, it is not explained why the vaccine constructs used in the actual wet-lab experiments (including LTB) were not modelled.

Using these L1-only models subsequently in B-cell epitope mapping and molecular docking could lead to distorted results, e.g. identifying inaccessible epitopes, or improbable docking interfaces which would be buried inside the capsomer or sterically blocked by LTB.

Modelling the L1-capsomers, fused to the LTB pentamers, should produce more relevant results for the vaccine candidate itself, and for the subsequent experiments using these models.

Response

We have revised and expanded the in silico analyses to ensure that they are fully aligned with the experimentally expressed vaccine antigen. Specifically, all computational analyses, including tertiary structure prediction, structural refinement, epitope mapping, and molecular docking were repeated using the complete LTB–L1 fusion construct rather than the isolated L1 subunit.

Although high-resolution structures of HPV L1 capsomeres and virus-like particles are available, these structures do not incorporate the LTB fusion partner or the linker region used in our vaccine design. Importantly, this revised modeling strategy enables evaluation of the influence of the LTB moiety on overall protein folding, surface exposure of B-cell epitopes, and immunologically relevant factors that cannot be explained from native capsomer structures alone. The Methods, Results, and Discussion sections have been updated accordingly to reflect this revised computational approach.

 

Major Comment 4: Molecular docking of LVC with TLR4 (methods 2.5, results 3.5, Figure 5):

This is a very curious experiment on many levels, which cast severe doubts on the results.

The most fundamental concern here is that no reason is given for modeling an interaction between TLR-4 and HPV L1 subunits. Is there a rationale for including it in a manuscript about vaccine antigens? TLR-4 mainly binds LPS and LPS-like molecules. As far as I am aware, no direct interaction between TLR-4 (or TLR-4 and MD-2) and HPV-L1 has been reported in literature. If there are studies showing there is an interaction between HPV L1 and TLr-4/MD-2, then these should be detailed in the text. If there are no indications for an interaction, then these docking studies should be removed from the manuscript completely.

Also, the experimental approach to docking elicits some concerns. To begin with, there is the choice to choose to model an interaction using the single subunit of L1, which is likely to result in interactions which are biologically not relevant. Indeed, from the very limited infromation about the interactinos which can be seen from Figure 5, it looks like that is indeed the case. The interaction interface from L1 looks like it is a subunit interface inside the pentameric capsomer. Not including LTB also diminishes the relevance for the present vaccine study.

Next, the choice for the model for the other partner in the docking experiment, TLR-4/MD-2. Here, a model of the activated dimer, containing a TLR-4 dimer, 2 bound MD-2 molecules and 2 bound LPS molecules buried in the interfaces, was taken as a basis. LPS was then removed, which in real life would destabilize the dimer. TLR-4 and MD-2 are both glycoproteins and it is unclear if the glycans were present in the models used for docking. Omitting them could potentially lead to docking results involving sites which in real life would be covered by glycans. It is not clear from the limited information if that is the case.

Considering the many serious issues, removing the docking results altogether is recommended.

Response

The revised manuscript now clearly gives the interpretation of the molecular docking analysis. We explicitly state that HPV L1 is not a canonical ligand for TLR4 and that no claim of direct biological binding is being made. Instead, docking was employed as a supportive computational approach, commonly used in immunoinformatics studies, to explore structural compatibility between vaccine antigens and innate immune receptors. This approach shows that HPV capsid proteins and virus-like particles can activate dendritic cells through TLR-associated signaling pathways, thereby justifying the relevance of such exploratory analyses.

Docking was performed using the intact TLR4–MD2 complex to preserve native receptor conformation, and the absence of glycosylation modeling has been explicitly acknowledged as a limitation of the docking platform. Overall, the interpretation of the docking results has been carefully refined in section 3.5 and in discussion section under heading “Contribution of In Silico Analysis to Vaccine Design.”

 

Major Comment 5: Formula 1 and quantification:

It is not made clear why Formula 1 should apply here and can be used for quantification. Data should be provided supporting the unwritten assumption that performance in this ELISA of L1 in the VLPs (the standard) and of LTB-L1 in the capsomers (from broccoli) is the same.

In addition, the formula should be modified to address the differences in weight between the L1 in the VLPs and the LTB-L1, at least if the concentrations are expressed in weight/volume, and also the difference (if any) in MW between the LTB-L1-constructs from HPV16 and HPV18.

 

Response

We thank the reviewer for this important clarification query. The use of Formula 1 and the overall quantification strategy are directly based on the methodology described by Verma et al., 2008 [31], which is widely cited for quantifying recombinant proteins expressed in plant systems.

In the study by Verma et al. (2008), recombinant proteins expressed in plant tissues were quantified by:

1. Measuring total soluble protein (TSP) from plant extracts.
2. Estimating the amount of target antigen (TP) using ELISA with a well-characterized antigen standard.
3. Expressing antigen accumulation as percentage of total soluble protein (%TSP) using the same formula applied in our study:

 

This approach is explicitly used by Verma et al. to provide relative expression levels of plant-produced antigens and is not intended to represent absolute mass quantification. Accordingly, in our study:

• The ELISA signal reflects relative antigen abundance, normalized to total soluble protein, which is standard practice in plant molecular farming.
• The use of L1 VLPs as a reference standard follows the same conceptual framework, as the ELISA is conformation-dependent and detects properly folded L1 rather than molecular mass.
• Importantly, Western blot analysis demonstrated that the predominant antigen accumulating in planta corresponds to the L1 monomer, not the intact LTB–L1 fusion, minimizing molecular weight discrepancies between the plant-derived antigen and the VLP standard.

To address the reviewer’s concern, the revised manuscript now explicitly states that:

• The reported %TSP values represent approximate, relative expression levels, consistent with Verma et al. (2008).

These clarifications have been added to the Methods and Discussion sections to avoid overinterpretation and to align the manuscript more closely with established methodology.

 

Major Comment 6: Oral vaccine

The interest in this research lies primarily in its possible use as an oral vaccine. It feels like a missed opportunity not to have included (mashed) broccoli leaves in the immunogenicity assay in addition to TSP.

Response

The oral vaccine potential was indeed explored in the present study using TSP. However, due to limitation of transgenic plant tissues, edible vaccine potential could not be investigated, which can be explored in future study.

 

Major Comment 7: Methods versus Results versus Discussion:

There is considerable overlap between Methods and Results sections, with many small experimental details mentioned both in methods and results, where a single mention in the methods would suffice. The reverse also applies. The same can also be said for Results and Discussion sections.

Response

The manuscript has been structurally reorganized to improve clarity and readability. Redundant methodological details that were previously repeated in the Results and Discussion sections have been removed, and the Discussion has been streamlined to focus on interpretation of the findings and comparison with existing literature.

 

Major Comment 8: Section 3.3 and table 2:

Why are the physicochemical properties calculated for L1 constructs other than the ones used for expression in broccoli? It would make more sense to provide information on these constructs.

Response

All bioinformatics analyses have been repeated, and the physicochemical property assessment has been recalculated for the full-length LTB–L1 vaccine constructs (LTB–L1-Cap16-His and LTB–L1-Cap18-His) that were used for Agrobacterium-mediated transformation and expression in broccoli. Accordingly, all parameters reported in Section 3.3 and Table 2—including molecular weight, instability index, aliphatic index, GRAVY score, theoretical isoelectric point, and predicted solubility—now correspond exclusively to the expressed LTB–L1 fusion constructs rather than the isolated capsomeric L1 sequences. To avoid any potential ambiguity, previous references to physicochemical properties of isolated L1 constructs have been removed from the manuscript.

 

Major Comment 9: Discussion

The first two paragraphs could be shortened drastically because they repeat the introduction. 

In general, much of the bioinformatics results are not really discussed here. There is quite a lot of literature about these properties and they should be compared with the results from this study.

In addition, both HPV VLPs and capsomers have previously been expressed, even in plants, and by some of the same authors. A discussion and comparison of the results would be expected here in the discussion.

Response

In response to this comment, the opening paragraphs of the Discussion have been revised and condensed to eliminate redundancy with the Introduction and to focus more directly on interpreting the key findings. The revised Discussion now provides a clearer evaluation of the bioinformatics analyses, structural modeling, physicochemical characterization, and docking results. We have expanded the Discussion to compare our findings with previous reports on HPV L1 VLPs and capsomer-based vaccines, including plant-based expression systems, highlighting similarities and differences in antigen design, expression levels, and immunogenic outcomes.

 

Major Comment 10:

L583-585: “The evaluation of the vaccine physiochemical features proved that both the vaccine constructs were quite stable but insoluble upon over-expression.”   -- This is evidently incorrect! The constructs can be detected in TSP extracts, thereby proving that they are soluble. This illustrates that the bioinformatics results should be treated with more caution. This particular statement should be removed, and instead the discrepancies between predicted and observed properties (not only solubility, but also molecular weight) should be discussed.

Response

The original statement suggesting that the vaccine constructs were “insoluble upon over-expression” has been corrected to reflect probability-based computational predictions rather than definitive conclusions. In the revised Results section 3.3, we now specify that the in silico solubility analysis predicted a reduced likelihood of solubility under overexpression conditions, with probability scores of 0.752058 for the HPV-16 construct and 0.672943 for the HPV-18 construct, indicating a theoretical tendency toward insolubility at high expression levels. Importantly, the Discussion has been updated to clearly emphasize that these values represent computational estimates only and do not contradict experimental observations. Consistent with this clarification, the recombinant L1 protein was successfully detected in the total soluble protein fraction, demonstrating that the antigen remains at least partially soluble in plant.

 

Major Comment 11: What do the error bars represent?

Response

This has been clarified in the legend. Error bars explicitly represent ± standard deviation (SD) (n = 5 mice per group).

 

 

Minor Comments

L94-98: “Prophylactic vaccine designing is the ultimate ray of hope followed by in silico analysis of the HPV target that can not only serve as protection from laborious trial-and-error experimental failures, but also minimize the expenditures on vaccine synthesis by providing a solid direction for designing the vaccine”.   – This sentence needs clarification.

Response: The sentence has been revised and rephrased for clarity.

 

L112 and 119: the same argument is made twice

Response: The repetition is omitted in the revised version.

 

L121: Additionally,…    : This is not a entirely separate argument, because the use of plants in raw or partially processed form reduces cold-chain maintenance and transportation costs.

Response: The repetitive argument is removed.

 

L128: the 3D structure … was generated:  a structural model was generated

Response: Modified as suggested

 

L140, L143: please provide a database number and/or reference for the L1 sequences

Response: Provided

 

L142: why was this aspartate substituted?

Response: All this section has been revised in the light of Reviewer’s comments related to queries raised regarding sequences.

 

L168: define LVC better, see above

Response: Use of LVC is removed. Instead, different nomenclature is used for clarity of expressed constructs. 

 

L169: provide a database nr or reference for the LTB sequence used here.

Response: Provided

 

L184-187: Which criteria were used in searching for homologous sequences?

Homology analysis was conducted using BLASTP (NCBI) to evaluate potential sequence similarity between the designed HPV vaccine constructs and human proteins, with the aim of minimizing the risk of unintended cross-reactivity or adverse immunogenic effects. Sequence similarity was assessed using standard BLASTP statistical parameters, including E-value, percentage identity, alignment length, and query coverage. To ensure a conservative and biologically meaningful assessment, alignments with an E-value ≤ 1 × 10⁻⁵, amino acid identity of at least 30%, and substantial query coverage (≥70%) were considered potentially significant.

L218-219: “This webserver also determines whether a target gene is being expressed in the host system by analyzing its GC content”:    This is incorrect. Codon optimization will probably change the GC-content of the sequence towards the GC-content of broccoli, but GC-content in itself does not determine whether a gene is expressed or not. This needs to be changed in methods, results and discussion.

Response: sentence is removed.

 

L237: please provide information about the cultivar of the broccoli used, or its provenance.

Response: Brassica oleracea var. italica cultivar ‘Marathon’ was used throughout the study. The cultivar name has been added to the Materials and Methods section.

 

L295: please provide more information about the generation of the baculovirus-derived VLPs, or a reference. Which L1 was used in the VLPs? Did they contain the complete native sequence?

Response: Baculovirus-derived VLPs, were obtained from German Cancer Research Centre (DKFZ), Heidelberg, Germany. The information is added in the revised version. These VLPs contained the native sequence.

 

L299: please provide origin of MD2H11 (and MAB 1.3.5.15 and P4543). Do all these antibodies bind equally well to both types of L1?

Response: Antibodies were also obtained from German Cancer Research Centre (DKFZ), Heidelberg, Germany. The information is added in the revised version. MAB 1.3.5.15 is conformation-specific antibody suitable for use in ELISA for binding to properly folded L1, while MDH11 is suitable for use in Western blotting binding to linear form.

 

Formula 1: (TSP -> % TSP

Response: Corrected

 

L346: what were the doses actually used in immunization?

Response: Each mouse received an equivalent dose of 10 µg of transgenic L1 protein per immunization. This data is now provided in the section 2.12 of Materials and Methods.

L465: please provide observations on the appearance of the plants. Are there any differences with regular broccoli plants or are they similar?

Response: Plants were similar to the regular broccoli plants. The statement has been added in the revised version, section 3.7.

 

L492: which L1 protein construct has a MW of 56.5 kDa. What is the MW of the other L1 construct? What construct is this exactly?

Response: Both of the L1 HPV-16 and HPV-18 constructs have approximately MW of 56.5 kDa. It has been specified in the section 3.9 of the revised version.

 

L501: is the Ritti01 the same antibody as the one mentioned in the Methods? Please add a reference.

Response: Yes, it has now been specified in the section 2.10 (Methods).

 

L586: “significantly non-homologous” What does this mean?

Response: this means they showed no significant sequence homology to human proteins

 

L589-590: “good quality”: see above for a different opinion.

Response: The discussion is rewritten in the revised version.

 

L638: “yield of 0.33 and 0.35 % of total soluble protein”. These yields should be discussed with respect to the minimum yields required for an oral vaccine.

Response: The accumulation levels of 0.33% and 0.35% of total soluble protein (TSP) for HPV-16 and HPV-18 L1, respectively, are within the range reported for several plant-produced oral vaccine antigens. Previous studies have demonstrated that oral immunization using plant tissues expressing antigens at levels as low as 0.01–1% of TSP can successfully elicit systemic and mucosal immune responses, particularly when antigens are bio-encapsulated within plant cells and protected from gastric degradation (Streatfield 2007; Dus Santos et al., 2005). In this context, the expression levels achieved in broccoli are considered sufficient for proof-of-concept oral immunization, especially given the robust humoral immune responses observed in mice following oral administration of transgenic plant extracts.

1. Streatfield SJ. Plant-derived antigens as mucosal vaccines. Vaccine. 2007;25(47):8079–8085.
2. Dus Santos MJ, Carrillo C, Ardila F, Ríos RD, Franzone PM, Piccone ME, Wigdorovitz A, Borca MV. Development of transgenic alfalfa plants containing the foot and mouth disease virus structural polyprotein gene P1 and its utilization as an experimental immunogen. Vaccine. 2005;23(15):1838–1843. doi:10.1016/j.vaccine.2004.11.014.

 

Figure 1: is the LTB present in the construct? It would be nice to include it in the figure.

Response: Yes. In the revised version, we have labeled LTB separately.

 

Figure 2: Figure 2A shows both the primary sequence and secondary structure. How was the 2ndary structure generated? Predictions, from the refined 3D model, or other?

Response: Yes, from refined model.

 

Figure 9: The VLP positive controls used here are unspecified: Which L1 was used here?

Response: VLPs were Baculovirus-derived, which have been specified now in the revised version.

 

Figure 10: how many replicates were used?

Response: 3 replicates were use. It has been specified now in the figure description in the revised version.  

 

Reviewer 2 Report

Comments and Suggestions for Authors

In their paper "In Silico Designing and Subsequent Expression of Human Papillomavirus 16 and 18 L1 Vaccine Antigens in Broccoli," the authors focused, as the title and text of the paper suggest, on the design and production of antigens for certain HPV strains in plant producers. This problem is highly relevant (and the authors address this issue in the introduction). Furthermore, the problem of HPV spread in the human population can be solved with available means, which is of particular interest.

The submitted manuscript, however, is not without its shortcomings. In particular:

1. Authors should expand abbreviations upon their first occurrence to avoid ambiguity and confusion.

2. From the text of Subsection 3.1. The Gene Sequences of HPV-16 and 18 L1 are completely unclear as to how exactly the sequences were modified, and the subsection itself contains redundant information that should be in the "Materials and Methods" section (although the tool used for in silico translation is irrelevant—the process itself is entirely unambiguous).

3. The following subsections are not results per se and should be combined with the corresponding sections of the "Materials and Methods" section:

   1. 3.6. Codon optimization and final transformation vector (437)

   2. 3.7. Regeneration of transformed broccoli plants (457)

   3. 3.8. Confirmation of the L1 gene within the broccoli nuclear genome (468)

4. -Paragraph (613-622) contains redundant, previously cited information and can be safely removed from the text.

5. Overall, the Discussion section should be thoroughly rewritten (with an emphasis on comparing the obtained results with existing data) and significantly shortened.

6. Given such a large number of experimental steps, it would be useful to present the experimental workflow graphically for ease of understanding (and possible use by other researchers).

7. Line 497 is an extra period.

The manuscript itself could be significantly shortened without losing useful information, which would only improve its readability and overall impression. Nevertheless, the presented work is of practical interest and may be of potential interest to readers. This implies the possibility of its publication after technical revision by the authors, taking into account the comments.

Author Response

Reviewer 2

 

Comments and Suggestions for Authors

In their paper "In Silico Designing and Subsequent Expression of Human Papillomavirus 16 and 18 L1 Vaccine Antigens in Broccoli," the authors focused, as the title and text of the paper suggest, on the design and production of antigens for certain HPV strains in plant producers. This problem is highly relevant (and the authors address this issue in the introduction). Furthermore, the problem of HPV spread in the human population can be solved with available means, which is of particular interest.

Response: Thank you very much for your comments and suggestions. We appreciate your time and effort for reviewing our manuscript. Below, we are giving answers to the comments.

 

The submitted manuscript, however, is not without its shortcomings. In particular:

1. Authors should expand abbreviations upon their first occurrence to avoid ambiguity and confusion.

Response: All abbreviations are now introduced at their first appearances.

 

1. From the text of Subsection 3.1. The Gene Sequences of HPV-16 and 18 L1 are completely unclear as to how exactly the sequences were modified, and the subsection itself contains redundant information that should be in the "Materials and Methods" section (although the tool used for in silico translation is irrelevant—the process itself is entirely unambiguous).

Response: We have rewritten the section describing the gene sequences and modifications more elaborately. Redundant information has also been removed in the revised version.

 

1. The following subsections are not results per se and should be combined with the corresponding sections of the "Materials and Methods" section:
1. 3.6. Codon optimization and final transformation vector (437)
2. 3.7. Regeneration of transformed broccoli plants (457)
3. 3.8. Confirmation of the L1 gene within the broccoli nuclear genome (468)

Response: Thank you for your suggestion. In revised version, we have restructured the sections to reduce redundancy, and descriptive procedural content has been minimized in Results.

1. -Paragraph (613-622) contains redundant, previously cited information and can be safely removed from the text.

Response: Overall, the discussion has been restructured and rewritten to reduce such redundancy.

 

1. Overall, the Discussion section should be thoroughly rewritten (with an emphasis on comparing the obtained results with existing data) and significantly shortened.

Response: Done as suggested. The Discussion has been substantially rewritten, shortened, and refocused, emphasizing comparison with existing literature and interpretation of findings.

 

1. Given such a large number of experimental steps, it would be useful to present the experimental workflow graphically for ease of understanding (and possible use by other researchers).

Response:  We appreciate your suggestion. However, since manuscript describes the sections in adequate details, we prefer to retain the textual description of the experimental procedures.

 

1. Line 497 is an extra period.

Response: Corrected.

 

The manuscript itself could be significantly shortened without losing useful information, which would only improve its readability and overall impression. Nevertheless, the presented work is of practical interest and may be of potential interest to readers. This implies the possibility of its publication after technical revision by the authors, taking into account the comments.

Response: Overall, we have rewritten and restructured many sections of the manuscript. We believe that this have improved the readability and flow of the manuscript.

 

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript presents an interesting and timely study integrating immunoinformatics, plant-based expression, and in vivo immunogenicity assessment to develop HPV-16 and HPV-18 L1 capsomere-based vaccine candidates in broccoli. The work is well-structured and addresses an important global health challenge, particularly regarding vaccine accessibility in low-resource settings. The study has clear potential, and the following comments are offered to strengthen the manuscript and enhance its scientific impact.

1. The authors intentionally designed L1 variants to form capsomeres rather than full VLPs. This is a central conceptual choice and deserves deeper discussion. Please expand on:

• Comparative immunogenicity between capsomeres and VLPs
• Stability and production advantages of capsomeres
• Supporting literature demonstrating capsomeres as effective vaccine antigens

This will help readers understand why capsomeres were selected and how they compare to existing VLP-based HPV vaccines.

1. While broccoli is identified as a potential platform, the authors are encouraged to elaborate on the specific advantages of using broccoli over other established plant systems (e.g., tobacco or lettuce). A more detailed discussion on its biomass yield, tissue-specific expression, and its unique potential as an "edible vaccine" candidate would enhance the significance of the platform choice.
2. The Western blot results indicate that the L1 protein was detected at approximately 56.5 kDa, suggesting a cleavage of the LTB-L1 fusion protein. It would be helpful to include a brief discussion on whether this cleavage was an intended design (e.g., via specific protease sites) or an observation of in vivo processing. Adding references regarding the stability of such fusion proteins in plant cells would clarify this point for the readers.
3. The in silico analysis provides a strong foundation for the study. To further enhance the manuscript, the authors might consider deepening the correlation between the computational predictions (such as B-cell and T-cell epitopes) and the observed experimental results (e.g., ELISA data). This would highlight the predictive power of immunoinformatics in plant-based vaccine design.
4. The study reports L1 accumulation levels of approximately 0.33% and 0.35% of Total Soluble Protein (TSP). It would be highly informative to provide a comparative discussion with other plant-based HPV L1 expression studies. This contextualization would help underscore the efficiency of the current system and its competitiveness within the field of molecular farming.
5. In the Introduction, the authors refer to 2022 epidemiological data. To ensure the highest level of contemporary relevance, it is suggested to include brief mentions or citations of recent challenges in global vaccine distribution (e.g., Gardasil 9 rollout) in developing countries. This would further reinforce the "low-cost" value proposition of the research.
6. Since broccoli is an edible plant, a short discussion on future prospects for oral delivery would be valuable. This may include: Antigen stability during digestion / Processing methods (e.g., lyophilization) / Dosage considerations for edible vaccines. Even a brief forward-looking paragraph would strengthen the translational relevance of the platform.
7. The BLASTP analysis to avoid homology with human proteins is important. If possible, consider briefly summarizing:

• Thresholds used
• Whether any partial similarities were detected
• A short table or supplementary figure listing top hits

This will reassure readers regarding the safety profile of the designed constructs.

9. In the Results section, subsections 3.9. and 3.10. appear to contain an extra dot (“.”) in the numbering. This seems to be a simple typing or formatting error and can be corrected easily during revision.

This manuscript presents a meaningful contribution to plant-based vaccine development and demonstrates a well-integrated workflow from in silico design to in vivo validation. Addressing the points above—particularly the comparative discussions, clarification of fusion protein processing, and contextualization of expression levels—will significantly strengthen the manuscript and enhance its impact.

Comments for author File: Comments.pdf

Author Response

Comments and Suggestions for Authors

 

The manuscript presents an interesting and timely study integrating immunoinformatics, plant-based expression, and in vivo immunogenicity assessment to develop HPV-16 and HPV-18 L1 capsomere-based vaccine candidates in broccoli. The work is well-structured and addresses an important global health challenge, particularly regarding vaccine accessibility in low-resource settings. The study has clear potential, and the following comments are offered to strengthen the manuscript and enhance its scientific impact.

• The authors intentionally designed L1 variants to form capsomeres rather than full VLPs. This is a central conceptual choice and deserves deeper discussion. Please expand on:
• Comparative immunogenicity between capsomeres and VLPs
• Stability and production advantages of capsomeres
• Supporting literature demonstrating capsomeres as effective vaccine antigens

This will help readers understand why capsomeres were selected and how they compare to existing VLP-based HPV vaccines.

Response: Thank you very much for your valuable suggestions. These points have been added in the discussion of the revised version under the heading “Capsomer-based HPV Vaccines as Alternative to VLPs”. All the changes are highlighted for quick review.

 

1. While broccoli is identified as a potential platform, the authors are encouraged to elaborate on the specific advantages of using broccoli over other established plant systems (e.g., tobacco or lettuce). A more detailed discussion on its biomass yield, tissue-specific expression, and its unique potential as an "edible vaccine" candidate would enhance the significance of the platform choice.

Response: We have provided this information in the introduction paragraph 7.

 

2. The Western blot results indicate that the L1 protein was detected at approximately 56.5 kDa, suggesting a cleavage of the LTB-L1 fusion protein. It would be helpful to include a brief discussion on whether this cleavage was an intended design (e.g., via specific protease sites) or an observation of in vivo processing. Adding references regarding the stability of such fusion proteins in plant cells would clarify this point for the readers.

Response: In discussion of the revised manuscript under heading “LTB–L1 Fusion and Evidence of LTB Cleavage in Plants” we have mentioned that the cleavage was observation of in vivo processing and not intended. Appropriate citations are also added.

 

4. The in silico analysis provides a strong foundation for the study. To further enhance the manuscript, the authors might consider deepening the correlation between the computational predictions (such as B-cell and T-cell epitopes) and the observed experimental results (e.g., ELISA data). This would highlight the predictive power of immunoinformatics in plant-based vaccine design.

Response: Thank you for your suggestion. We have highlighted this in the revised manuscript section 3.10. Antigen-capture ELISA.

 

5. The study reports L1 accumulation levels of approximately 0.33% and 0.35% of Total Soluble Protein (TSP). It would be highly informative to provide a comparative discussion with other plant-based HPV L1 expression studies. This contextualization would help underscore the efficiency of the current system and its competitiveness within the field of molecular farming.

Response: We have revised the last paragraph of Discussion section, in which we compared our nuclear expression levels to the reported ranges of plant expression platforms and with other studies previously published of HPV L1 capsomere expression. These supplements explain the fact that the levels of accumulation obtained in broccoli coincide with previously published ranges of nuclear plant transformation system.

 

6. In the Introduction, the authors refer to 2022 epidemiological data. To ensure the highest level of contemporary relevance, it is suggested to include brief mentions or citations of recent challenges in global vaccine distribution (e.g., Gardasil 9 rollout) in developing countries. This would further reinforce the "low-cost" value proposition of the research.

Response: We appreciate the reviewer’s insightful comment. Accordingly, some points describing challenges in global HPV vaccine distribution has now been incorporated into the Introduction (5^th paragraph) with appropriate citations.

 

7. Since broccoli is an edible plant, a short discussion on future prospects for oral delivery would be valuable. This may include: Antigen stability during digestion / Processing methods (e.g., lyophilization) / Dosage considerations for edible vaccines. Even a brief forward-looking paragraph would strengthen the translational relevance of the platform.

Response: We have included in the revised version in “Conclusions and Future Perspectives’

 

8. The BLASTP analysis to avoid homology with human proteins is important. If possible, consider briefly summarizing:

➢ Thresholds used

➢ Whether any partial similarities were detected

➢ A short table or supplementary figure listing top hits

This will reassure readers regarding the safety profile of the designed constructs.

Response: We thank the reviewer for this valuable suggestion. In the revised manuscript, we have clarified the BLASTP screening strategy used to evaluate potential similarity with human proteins. The analysis was performed using the NCBI BLASTP server against the human protein database with an E-value threshold of 0.05. No biologically significant homology with human proteins was detected under these criteria.

However, at the time of this revision, the BLAST server was intermittently unavailable.  Therefore, the analysis could not be re-run and thus a supplementary figure or table could not be provided.

 

1. In the Results section, subsections 3.9. and 3.10. appear to contain an extra dot (“.”) in the numbering. This seems to be a simple typing or formatting error and can be corrected easily during revision.

Response: Corrected

 

This manuscript presents a meaningful contribution to plant-based vaccine development and demonstrates a well-integrated workflow from in silico design to in vivo validation. Addressing the points above—particularly the comparative discussions, clarification of fusion protein processing, and contextualization of expression levels—will significantly strengthen the manuscript and enhance its impact.

Response: Thank you very much for your positive comments. Your suggestions have greatly improved the worth of the manuscript.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have addressed several of the comments in the revised manuscript. The overall clarity of the manuscript writing has improved, in particular the description of the wet-lab experiments.

However, problems with the bio-informatics experiments persist. Here, some issues have not been addressed, and new, and even larger, issues have been introduced, as detailed below.

My overall recommendation is therefore to reject the manuscript.

 

Major new issues:

The structural models of L1 presented in Fig 2B and 2C cannot be correct. HPV L1 has in reality a β jelly roll (as it had in the previous version of the manuscript!) and does not have the helical structure such as displayed in Figures 2B and 2C. The secondary structure as indicated in figure 2A (with many β strands) does appear to be plausible.

These helical models also appear to have been the basis of the bioinformatics results presented in Figures 3,4 and 5.

All these results therefore can not be correct.

I would suggest a thorough evaluation to identify the cause of what went wrong here.

 

 

Response to major comment 4:

Thank you for providing some further information on this. However, the response, and the section concerning this in the discussion fail to address the real problems with this experiment.

Docking is attempted between a protein species which does not exist (a hypothetic monomeric L1) with a very hypothetical possible interaction partner (activated TLR4-MD2). In the previous manuscript, this almost inevitably resulted in an interaction interface involving residues stably buried within the intra-capsomeric L1-L1 interfaces, which is completely implausible. The chances of such docking experiments resulting in a relevant interaction are vanishingly small and I would therefore recommend a complete re-evaluation of the docking approach.

The references provided in the discussion are also not supporting an interaction between L1 (monomers or capsomers) and TLR4-MD2. Experiments in reference [54] may suggest an interaction between L1 VLPs and dendritic cells. However, any effect on dendritic cells is abolished with a L1 mutation which mostly results in capsomers. Therefore, reference [54] is at best an argument against attempting a docking.

Worse, supporting reference [55] is unfindable.

 

Response to Major comments 2 and 5:

Post-translational cleavage of the LTB portion is indeed a plausible explanation. Confirmation of this with a Western blot with anti-LTB antibodies would still be recommended. The hypothesized cleavage would also remove most of the issues with applying the quantification formula, as there is no MW-correction necessary and the reference-protein assumption, that VLPs and capsomers of essentially the same nature behave similarly in the ELISA experiment, is also more plausible.

 

 

Author Response

Reviewer 1:

The authors have addressed several of the comments in the revised manuscript. The overall clarity of the manuscript writing has improved, in particular the description of the wet-lab experiments.

However, problems with the bio-informatics experiments persist. Here, some issues have not been addressed, and new, and even larger, issues have been introduced, as detailed below.

My overall recommendation is therefore to reject the manuscript.

 

Major new issues:

The structural models of L1 presented in Fig 2B and 2C cannot be correct. HPV L1 has in reality a β jelly roll (as it had in the previous version of the manuscript!) and does not have the helical structure such as displayed in Figures 2B and 2C. The secondary structure as indicated in figure 2A (with many β strands) does appear to be plausible.

These helical models also appear to have been the basis of the bioinformatics results presented in Figures 3,4 and 5.

All these results therefore can not be correct.

I would suggest a thorough evaluation to identify the cause of what went wrong here.

 

Response

We thank the reviewer for this important and accurate observation regarding the structure of HPV L1. We fully agree that HPV L1 adopts a characteristic β-jelly roll fold, and that the predominantly helical conformations shown in the earlier versions of Figures 2B and 2C were not biologically plausible.

In the revised version, we undertook a thorough re-evaluation of the structural modeling workflow and identified the source of this discrepancy. In earlier versions of the manuscript (initially submitted version), I-TASSER was used to model the L1 protein alone (without LTB), yielding structures consistent with the expected β-jelly roll topology. Subsequent refinement using ModRefiner and GalaxyRefine preserved the β-strand–rich architecture of L1, confirming that this approach is valid for the L1 protein.

However, in the first revised version, when the full-length LTB–L1 fusion sequence was used as input in I-TASSER, refinement with ModRefiner and GalaxyRefine resulted in the loss of substantial regions of the sequence, including the linker and adjacent segments. This is because ModRefiner is specifically designed to refine well-folded, globular protein domains by improving atomic geometry and optimizing backbone and side-chain packing. During this refinement process, regions that are intrinsically disordered, poorly modeled, highly flexible, or associated with low structural confidence, such as linker regions may be truncated or removed and this led to truncated models of approximately 150 residues and the emergence of artificially helix-rich conformations. This reflects a limitation of refinement algorithms when applied to long, multi-domain fusion proteins that lack a single continuous structural template. As a consequence, the refined fusion models failed to preserve the native β-jelly roll fold of L1, despite secondary structure predictions that remained plausible at the sequence level.

To confirm that this issue was methodological rather than biological, LTB and L1 were modeled independently using appropriate experimental templates in I-TASSER. In these domain-specific models, the β-jelly roll architecture of L1 was fully retained. Visualization of the independently modeled domains together in PyMOL demonstrated that the L1 fold remained intact (as β jelly roll), confirming that the previously observed structural distortion arose from fusion-protein refinement artifacts rather than from the L1 sequence itself.

Due to these scenarios, we revised our structural modeling strategy. The LTB–L1 fusion protein was re-modeled using the SWISS-MODEL server. SWISS-MODEL successfully generated the 3D structure, preserving the experimentally validated β-jelly roll fold of L1 while modeling the full-length fusion construct without truncation or helical distortion. As a result of this re-analysis, in the revised version, Figures 2B and 2C have been replaced with corrected models that maintain the characteristic β-jelly roll architecture of L1. In addition, all downstream bioinformatics analyses (Figures 3, and 4) have been repeated using the corrected structural models.

The relevant sections have been revised accordingly to clearly describe the updated modeling strategy. All changes are highlighted in the revised version.

 

Response to major comment 4:

Thank you for providing some further information on this. However, the response, and the section concerning this in the discussion fail to address the real problems with this experiment.

Docking is attempted between a protein species which does not exist (a hypothetic monomeric L1) with a very hypothetical possible interaction partner (activated TLR4-MD2). In the previous manuscript, this almost inevitably resulted in an interaction interface involving residues stably buried within the intra-capsomeric L1-L1 interfaces, which is completely implausible. The chances of such docking experiments resulting in a relevant interaction are vanishingly small and I would therefore recommend a complete re-evaluation of the docking approach.

The references provided in the discussion are also not supporting an interaction between L1 (monomers or capsomers) and TLR4-MD2. Experiments in reference [54] may suggest an interaction between L1 VLPs and dendritic cells. However, any effect on dendritic cells is abolished with a L1 mutation which mostly results in capsomers. Therefore, reference [54] is at best an argument against attempting a docking.

Worse, supporting reference [55] is unfindable.

 

Response

We thank reviewer for pointing out. In the revised version, we have completely omitted all TLR4 docking-related data from the manuscript (from all sections of introduction, Methods, Results and Discussion).

 

 

 

Response to Major comments 2 and 5:

Post-translational cleavage of the LTB portion is indeed a plausible explanation. Confirmation of this with a Western blot with anti-LTB antibodies would still be recommended. The hypothesized cleavage would also remove most of the issues with applying the quantification formula, as there is no MW-correction necessary and the reference-protein assumption, that VLPs and capsomers of essentially the same nature behave similarly in the ELISA experiment, is also more plausible.

 

Response

We really appreciate for accepting our explanation. We are also thankful for the valuable comments that improved our manuscript substantially. For LTB detection in Western blot with anti-LTB antibodies, we would like to point out that since L1 is successfully expressed and validated till immunological analysis (both in vitro and in vivo), it seems unnecessary to do this experimentation. Doing this experiment will significantly delay the manuscript publication.

 

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have significantly improved the manuscript. The work may be accepted for publication.

Author Response

Thank you very much for your positive and constructive input. 

Round 3

Reviewer 1 Report

Comments and Suggestions for Authors

This is a revised manuscript where the authors have addressed the comments from before. The manuscript is greatly improved by replacing the previous modelling approaches with a model for the designed pentameric assemblies. These new experiments lack however detail about in particular the LTB part of the models, which is important in interpreting the results.

 

Paragraphs 2.4 and 3.4

The description for generating the structural models of vaccine constructs is very limited and misses some important information.

The structural models are for fusions between L1s and LTB. L1s are engineered to form pentamers, while LTB also tends to assemble into pentamers. Although this is not stated explicitly, presumably both L1 and LTB portions were designed to contribute to the pentameric assembly of the vaccine construct (LTB wouldn’t bind to its receptors otherwise). It is not described how this was treated in the modeling approach and in SWIS-MODEL. Presumably the L1 pentamer was selected as the base of the biological assemblies (Fig 2B/C seem to suggest this). However, information about the LTB part is missing. What approach did the authors take to  modeling of the LTB part. Does it indeed form pentamers in the assemblies? Is the linker between long enough to allow this?

Providing the quality scores from SWISS-MODEL as well would be useful. They are a bit more up-to-date than the PROCHECK and ERRAT scores.

 

 

Figure 2

Fig 2A. The legend lacks information about the secondary structures indicated in Fig2A. Are they predictions from a program, derived from a database, or based on the new 3D models?

Fig2B/C: It is very difficult to interpret these Figures. Are the LTB portions in the figures? Highlighting one subunit in the pentamers and indicating the L1 and LTB portions would make the viewing much easier.

 

Figure 3 and 4: The rows of residue names in the panels are very crowded and in 3C are so small that they become unreadable. The information could be put in the legend instead.

Author Response

Comments and Suggestions for Authors

This is a revised manuscript where the authors have addressed the comments from before. The manuscript is greatly improved by replacing the previous modelling approaches with a model for the designed pentameric assemblies. These new experiments lack however detail about in particular the LTB part of the models, which is important in interpreting the results.

Response: Thank you very much for taking time to review our manuscript and providing critical suggestions for improvement. Below, we are providing responses to individual comments. The changes made in the manuscript are highlighted for quick.

 

Comment

Paragraphs 2.4 and 3.4

The description for generating the structural models of vaccine constructs is very limited and misses some important information.

The structural models are for fusions between L1s and LTB. L1s are engineered to form pentamers, while LTB also tends to assemble into pentamers. Although this is not stated explicitly, presumably both L1 and LTB portions were designed to contribute to the pentameric assembly of the vaccine construct (LTB wouldn’t bind to its receptors otherwise). It is not described how this was treated in the modeling approach and in SWIS-MODEL. Presumably the L1 pentamer was selected as the base of the biological assemblies (Fig 2B/C seem to suggest this). However, information about the LTB part is missing. What approach did the authors take to  modeling of the LTB part. Does it indeed form pentamers in the assemblies? Is the linker between long enough to allow this?

Response:

We have revised the manuscript as per the raised queries. In the revised version, the whole LTB–L1 amino acid sequence was submitted into SWISS-MODEL server. Instead of finding a single continuous template that covered the whole fusion sequence, SWISS-MODEL found discrete high-confidence templates that corresponded to the L1 and LTB domains independently, as would be expected for a lengthy multi-domain fusion protein, due to the limitation of the modeling software. As it happened in case of ITASSER software in which Full-length homology modeling of the LTB–L1 fusion protein using automated refinement tools resulted in truncation and loss of known secondary structure features, particularly the β-jelly-roll fold of L1, a domain-wise modeling strategy was adopted in new approach.

Four structural models were produced in SWISS-MODEL after giving them the whole LTB-L1 sequence, two of which matched the L1 domain and two to the LTB domain. The model with the best sequence coverage and the highest quality metrics was chosen for each domain. The distinctive β-jelly-roll fold and pentameric capsomeric architecture were preserved in the L1 domain models, and classic pentameric assembly feature was also adopted by the LTB domain models. The modeled pentameric L1 and pentameric LTB structures were visualized in PyMOL that showed the complete structure of the vaccine construct (revised version: Figure 2C & 2D).

Both domains were able to maintain their original pentameric configurations. The 14-amino acid linker that connects the two domains is lengthy and flexible enough to allow for the autonomous folding and oligomerization of each domain without steric interference, as well as the spatial separation of the LTB and L1 moieties.

 

Comment:

Providing the quality scores from SWISS-MODEL as well would be useful. They are a bit more up-to-date than the PROCHECK and ERRAT scores.

Response: Given as suggested.

 

Comment: 

Figure 2

Fig 2A. The legend lacks information about the secondary structures indicated in Fig2A. Are they predictions from a program, derived from a database, or based on the new 3D models?

Response: Provided in the figure legend of the revised version

 

Comment:

Fig2B/C: It is very difficult to interpret these Figures. Are the LTB portions in the figures? Highlighting one subunit in the pentamers and indicating the L1 and LTB portions would make the viewing much easier.

Response:

We have modified all figures and the labels inside the figures have been updated. LTB and L1 domains are marked on the figures.

Comment:

Figure 3 and 4: The rows of residue names in the panels are very crowded and in 3C are so small that they become unreadable. The information could be put in the legend instead.

Response: Done as suggested.

 

Round 4

Reviewer 1 Report

Comments and Suggestions for Authors

This is a revised manuscript where the authors have addressed the comments from before.

Comments to the previous version highlighted the lack of details about the generation of the structural model. In response, the authors have changed the model. There are three issues with this.

1. Naming confusion. The authors are still using the term vaccine construct very loosely, sometimes it seems to mean the whole construct, and sometimes only the L1 part. In comments on the previous version, it was assumed that the structural model was for the whole LTB-L1 fusion, i.e., that the naming was consistent between Fig2A and 2B/C. From the response, it is now apparent that this was not the case, and the model was only for the L1 pentamer, while the sequence was for the whole construct. That may have been improved in this version, but that is not clear. Consistency in naming is important.
2. Even though the authors now explain their (flawed) approach in a bit more detail, they do this only in the response. These details have still not been included in the manuscript. Yet, the journal explicitly states that full experimental details must be provided so that the results can be reproduced.
3. Unfortunately, the eventual model in the current version of the manuscript is also not fit for purpose. The model for the LTB-L1 pentamers seems to have been fused together from individual pentamers for L1 and LTB (manually?), where LTB and L1 parts are at best only connected for 1 subunit, instead of all 5. This is not a correct modelling approach and in addition still does not clarify which L1 epitopes would have been accessible and which would have been blocked by LTB. Also, ligands seem to have been included in the LTB portion (in Fig2C/D, at least), which were not mentioned at all.

The discontinuous epitopes in Figures 3 and 4 on this inaccurate model were now also generated for the LTB part. As these LTB epitopes are not part of HPV, they have a different relevance for vaccine development and should probably be omitted. The epitopes can also more economically be presented by their sequence range (they are not very discontinuous), instead of listing each residue individually (e.g. Fig3A residues 491-509).

Author Response

Comments and Suggestions for Authors

This is a revised manuscript where the authors have addressed the comments from before.

Comments to the previous version highlighted the lack of details about the generation of the structural model. In response, the authors have changed the model. There are three issues with this.

1. Naming confusion. The authors are still using the term vaccine construct very loosely, sometimes it seems to mean the whole construct, and sometimes only the L1 part. In  comments on the previous version, it was assumed that the structural model was for the whole LTB-L1 fusion, i.e., that the naming was consistent between Fig2A and 2B/C. From the response, it is now apparent that this was not the case, and the model was only for the L1 pentamer, while the sequence was for the whole construct. That may have been improved in this version, but that is not clear. Consistency in naming is important.

Response: Thank you very much for reviewing the manuscript critically. We appreciate your time and effort invested in the improvement of the manuscript quality.

We have modified several sections in the revised manuscript to ensure consistency of using the term vaccine construct. These sections include last paragraph of introduction, sections 2.1, 2.2, 2.3 and 2.4. We tried our best to remain consistent for using the term vaccine construct, pertaining to LTB-L1, throughout the manuscript. Further, in the relevant sections, we have also clarified that the modeling of L1 and LTB was done while using the sequence of whole LTB-L1 construct.

 

 

2. Even though the authors now explain their (flawed) approach in a bit more detail, they do this only in the response. These details have still not been included in the manuscript. Yet, the journal explicitly states that full experimental details must be provided so that the results can be reproduced.

Response: In the revised version, we have tried our best to include all the details to facilitate reproducibility.

 

1. Unfortunately, the eventual model in the current version of the manuscript is also not fit for purpose. The model for the LTB-L1 pentamers seems to have been fused together from individual pentamers for L1 and LTB (manually?), where LTB and L1 parts are at best only connected for 1 subunit, instead of all 5. This is not a correct modelling approach and in addition still does not clarify which L1 epitopes would have been accessible and which would have been blocked by LTB. Also, ligands seem to have been included in the LTB portion (in Fig2C/D, at least), which were not mentioned at all.

Response: We sincerely thank the reviewer for this important observation. We agree that the previously presented manually fused pentamer model was not an appropriate representation of a biologically assembled LTB–L1 fusion complex. In the revised version, the fused pentamer model has been removed, and independent models of LTB and L1 structures based on accessible experimental templates are now added to give structural analysis at the domain level.
We clarify that due to lack of an experimentally defined template for an LTB–L1 fusion protein restricts the ability to accurately model the entire pentameric fusion structure of both LTB and L1 together. For modeling, we selected complete LTB-L1 vaccine construct as the base of the biological assemblies, but server gave us independent models for LTB and L1. Both of them form pentamers. Complete LTB-L1 sequence was used for the analysis. We used the complete sequence of LTB-L1 in SWISS MODEL server; it gave us independent LTB and L1 models rather than the continuous LTB-L1 fusion model. Hence, we assembled them in Pymol to show a hypothetical illustration of LTB-L1 fusion model, which connected the LTB domain to L1 as 1 subunit rather than all 5. Now, in the revised version, we have modeled L1 and LTB separately, obtained by providing LTB-L1 sequence to the software.

 

 

4. The discontinuous epitopes in Figures 3 and 4 on this inaccurate model were now also generated for the LTB part. As these LTB epitopes are not part of HPV, they have a different relevance for vaccine development and should probably be omitted. The epitopes can also more economically be presented by their sequence range (they are not very discontinuous), instead of listing each residue individually (e.g. Fig3A residues 491-509).

Response: We thank the reviewer for this important observation. In the revised manuscript, discontinuous B-cell epitopes corresponding to the LTB portion have been removed. Only epitopes located within the HPV L1 region are now presented. Furthermore, epitopes are now reported using sequence ranges instead of listing individual residues.