Fructans Are Differentially Distributed in Root Tissues of Asparagus

Round 1

Reviewer 1 Report

The authors present a really nice use of modern proteomic approaches to tackle the tissue localization of fructans in Asparagus. The authors use of multiple advanced mass spectrometry approaches is quite well executed and has created some unique results that will be of interest to the potential biotechnological engineering of fructans.

Minor Issues:

1) As is the case with many studies predominantly using omics technologies, I would suggest re-working the manuscript to be a combined results/discussion. This just makes sense for these papers in terms of flow and readability as well as integration of known literature. 

Major issues:

1) It is imperative that the methods (ideally accompanied by a supplemental figure) clearly describe how the separate root sections (inner, middle, outer) were deduced and harvested for proteomic LFQ analysis. Alternatively, the authors could have tried to validate there offline sectioning by using their maldi laser setup to do some imaging MS. That said, I think it is sufficient to add some clarity on how this was performed.

2) Why use a t-test? Why not use an ANOVA with a post-hoc to take into account study variability as well as deduce where specific changes were occuring. 

3) Did the authors have a protein ID FDR analysis? If so, please provide details as to how was this achieved? 

4) I have not commonly seen the selection criteria of "10% protein coverage". Can the authors please elaborate on to how this determination was made. Typically for protein abundance LFQ analyses a min of 2 peptides is required.

4a) What percent of the proteins assess for differential expression fit this criteria?

5) Why only KEGG analysis? Why not any other bioinformatics analyses of this data. For example, heat maps that cluster based on co-abundance change profiles to assess what pathways are doing what where? The current analysis seems very cursory and could stand to have more context other than listing the KEGG pathways changing.

Author Response

Rebuttal Letter

Reviewer 1

The authors thank the reviewer for the thorough evaluation of our manuscript and the valuable comments and suggestions. Please find below the point-by-point explanations of the revisions in the manuscript and our responses to the reviewers' comments.

Reviewer 1:

The authors present a really nice use of modern proteomic approaches to tackle the tissue localization of fructans in Asparagus. The authors use of multiple advanced mass spectrometry approaches is quite well executed and has created some unique results that will be of interest to the potential biotechnological engineering of fructans.

Minor Issues:

1) As is the case with many studies predominantly using omics technologies, I would suggest re-working the manuscript to be a combined results/discussion. This just makes sense for these papers in terms of flow and readability as well as integration of known literature.

Answer:

We clearly agree with the reviewer that, in case of purely descriptive studies, combining the results and discussion sections is a valuable option. In our study, we have used MALDI-MS imaging technology for the visualization of tissue particularities for carbohydrates of different molecular weight. The detailed structure of the underlying fructooligosaccharides has then be elucidated by HPAEC with PAD detection and combined to peak annotations by classical mass spectrometric analysis. To identify isoforms of metabolic enzymes related to the observed tissue particularities we additionally investigated the proteome of the isolated root tissues. In the discussion we placed our results in the scientific context at three levels: a) a technical level related to the state of the art of polysaccharide analysis and its limitations, b) a physiological level discussing possible functions of the different fructooligosaccharides in the different root tissues, and c) a biochemical level discussing molecular specificities of the different root tissues with specific emphasis on fructan biosynthetic enzymes. Combining these discussed aspects with the respective results sections would lead to a higher degree of redundancy. Also, we feel that the manuscript would lose structure, such as making it harder for the reader to follow the rational of our selection of approaches in the results description. Therefore, we prefer to keep the current structure of the manuscript unless advised differently by the editor.

Major issues:

1) It is imperative that the methods (ideally accompanied by a supplemental figure) clearly describe how the separate root sections (inner, middle, outer) were deduced and harvested for proteomic LFQ analysis. Alternatively, the authors could have tried to validate there offline sectioning by using their maldi laser setup to do some imaging MS. That said, I think it is sufficient to add some clarity on how this was performed.

Answer:

Sample processing for sugar profiling and proteomic analysis is described in “2. Materials and Methods” sub-section “2.1. Plant material and tissue preparation” lines 128-132, including how core pieces have been taken to result in inner, middle and outer root regions. In addition, the Supplementary Figure S1 visualizes how core pieces have been taken. We have now also included an explaining sentence at the beginning of section 2.3. (line 159) and 2.4 (line 180). Hence, we believe that this method has been described in detail.

We agree with the reviewer that validating the sampled root sections by MSI would have given additional confidence. However, root cross sections made for MSI analysis need to be as thin as possible, in our approach 16 µm in thickness. For punching out the core pieces, cross sections needed to be considerably thicker, in our approach 150 µm, to keep the tissue intact as well as possible. Initial attempts to punch out core pieces from 16 µm thin sections were not successful and resulted in the destruction of the sample.

2) Why use a t-test? Why not use an ANOVA with a post-hoc to take into account study variability as well as deduce where specific changes were occuring.

Answer:

Indeed, the description of the applied statistical methods fell short in the manuscript. We have added information on both, peptide identification and protein abundance quantification, to that chapter. In detail, the statistical significance in the data set was determined running the t- test (post hoc) after an analysis of variance (ANOVA) test. Missing values were filled by low abundance resampling (missing values are replaced with random values sampled from the lower five percent of detected values). An FDR-corrected p-value < 0.05 was also determined, using the Benjamini-Hochberg correction (lines 194-206).

3) Did the authors have a protein ID FDR analysis? If so, please provide details as to how was this achieved?

Answer:

Database searches implemented in Proteome Discoverer 2.4 and Sequest HT search engine (Thermo Scientific) included the generation of a decoy database and calculation of FDR values from searches against this artificial database. The thresholds for highly confident identifications of peptides and proteins were set to an FDR of 0.01. We have included the FDR search parameters in the Materials and Methods section (lines 195-197). With our dataset the following results were obtained for high confidence peptide identifications: 0.010 (16003 targets, 159 decoys) for peptides, and 0.010 (9073 targets, 89 decoys) for proteins.

4) I have not commonly seen the selection criteria of "10% protein coverage". Can the authors please elaborate on to how this determination was made. Typically for protein abundance LFQ analyses a min of 2 peptides is required.

Answer:

We agree with the reviewer that the conventional reporting for protein identification is based on two independent peptides, which we also applied in our study. However, doing this applies a bias against small proteins that, upon enzymatic digestion, generate only a few peptides. Supplementary table S2 provides the number of peptides used for identification (worksheet “identified”). Proteins identified by one peptide, representing a protein coverage > 10%, had an average size of 14.1 kDa (± 5.7 kDa, standard deviation). The 10% level was chosen since it allowed the inclusion of proteins that are a bit larger (20-30 kDa), but that might be difficult to cleave by trypsin due to the lack of respective amino acids in the protein sequence, e.g. such as often the case for hydrophobic proteins. We have included an explanation in this section (lines 202-204).

4a) What percent of the proteins assess for differential expression fit this criteria?

Answer:

The number of proteins with statistical significantly different abundance between the analyzed root regions, identified by this criterion, was 16, leading to an amount of 7 % from all differential expressed proteins.

5) Why only KEGG analysis? Why not any other bioinformatics analyses of this data. For example, heat maps that cluster based on co-abundance change profiles to assess what pathways are doing what where? The current analysis seems very cursory and could stand to have more context other than listing the KEGG pathways changing.

Answer:

Thank you for this remark. The proteome dataset was first evaluated by principle component analysis and hierarchical clustering of expression patterns of all identified proteins based on Euclidean distance and complete linkage (Supplementary Figure S5). These analyses revealed a close grouping of technical replicate runs and a clear separation between protein extracts derived from the three different root tissues. The data set revealed a higher similarity in protein expression pattern between the middle and the outer region, as compared to the inner region of the root, which was distinctly separated from the other samples. To gain insights in the functionality of the underlying proteins we further evaluated only highly significantly identified proteins which were significantly differentially expressed. The asparagus genome sequence is poorly annotated and multiple tools for bioinformatics analyses cannot be applied. We believe that gene/protein annotation is the background of data interpretation. Building co-expression networks without the background of gene annotation will probably not lead to new information. In our study, blasting the protein sequences against KEGG resulted in the annotation of 77% of all differentially abundant proteins and provided a robust method to get a valuable impression on the data set (Figure 4).

However, we agree that the proteomics data set has not been elucidated to its finest details. Primarily, it was the intention of the proteome approach to get insight into the spatial distribution of fructan/carbohydrate metabolism (as shown in Figure 5). The heat map shown in Figure 4 served to describe the complete proteome data set in an overarching structure, to demonstrate that these three root regions express a number of different proteins (10% of all identified proteins) and that these proteins might be functionally related. We agree with the reviewer that the proteome data set might reveal other interesting insights into the spatial distribution of other major proceedings in the root (e.g. water uptake: seven identified aquaporins are generally higher abundant in the inner region; e.g. root hair development: six identified linoleate 9S-lipoxygenases are higher abundant in the outer region). However, this was not the focus of our research. Upon publication, we will make the raw files and processed data set public available for other researchers to be explored.

Reviewer 2 Report

In this manuscript (cells-875991), authors present a work centered on fructans distribution in the root tissues of asparagus.

In particular, they investigate on spatial distribution of fructans in asparagus storage roots by the application of MALDI MSI and fructan profiling by HPLC-PAD. Moreover, they performed the proteome profiling in order to increase the knowledge about fructan metabolic enzymes.

I think the experimental design is linear and coherent with the aim of the work. However, it lacks some valuable insights to make it acceptable for publication in Cells. Detailed comments below.

Keywords do not very much help positioning the paper.

The introduction section should be reformulated and enriched with more information regarding about fructan accumulation in root and fructan distribution in root.

The author performed a proteome analysis aimed at better understand the specific proteins involved in carbohydrate metabolism and specifically in fructan biosynthesis.

However, the results they obtain did not lead to indicated conclusions (lane 437-438).

At this purpose, a Real Time analysis, investigating the specific gene expression of the enzymes involved in fructan biosynthesis, should be performed in the different part of the root.

Moreover, how the author can explain about sucrose localization that mainly concenrs the outer and middle root tissues while Invertase (β-fructofuranosidase, EC 3.2.1.26) were mainly found in the inner tissue?

Minor changes:

Lane 111: “Seeds were cultivated” change with “Seeds were germinated”

Lane 153: Oligosaccharide extraction and profiling: indicate the amount of starting material that you use in your experiments (g) and the ratio for the extraction.

Increase figures resolution.

Author Response

Rebuttal Letter

Reviewer 2

The authors thank the reviewer for the critical evaluation of our manuscript and the valuable comments and suggestions. Please find below the point-by-point explanations of the revisions in the manuscript and our responses to the reviewers' comments.

Reviewer 2

In this manuscript (cells-875991), authors present a work centered on fructans distribution in the root tissues of asparagus.

In particular, they investigate on spatial distribution of fructans in asparagus storage roots by the application of MALDI MSI and fructan profiling by HPLC-PAD. Moreover, they performed the proteome profiling in order to increase the knowledge about fructan metabolic enzymes.

I think the experimental design is linear and coherent with the aim of the work. However, it lacks some valuable insights to make it acceptable for publication in Cells. Detailed comments below.

Keywords do not very much help positioning the paper.

Answer:

Generally, we tried to avoid repeating keywords from the manuscript title as this is already included in search engines indexing and ranking processes. However, we tried to follow this hint and amended the list of key words (line 31-32).

The introduction section should be reformulated and enriched with more information regarding about fructan accumulation in root and fructan distribution in root.

Answer:

Many thanks for this suggestion. Fructans accumulate in all parts of plants and are enriched in storage roots of a wide range of plant species. However, to our knowledge except for Taraxacum officinale (Van den Ende et al., 2000), for Campuloclinium chlorolepis (Vilhalva et al, 2011), for Chrysolaena obovata (Portes et al., 2006; and Rigui et al., 2015), and for Gomphrena marginata Seub. (Joaquim et al., 2018), particularities of spatial distribution of fructans and fructan metabolizing enzymes in rhizophores/roots have not been reported in the literature yet. Therefore, we targeted this topic in our study for asparagus. We have extended the information on spatial distribution of fructans in roots both, in the introduction (lines 76-80) and the discussion (lines 488-491).

The author performed a proteome analysis aimed at better understand the specific proteins involved in carbohydrate metabolism and specifically in fructan biosynthesis.

However, the results they obtain did not lead to indicated conclusions (lane 437-438).

At this purpose, a Real Time analysis, investigating the specific gene expression of the enzymes involved in fructan biosynthesis, should be performed in the different part of the root.

Answer:

In lanes 437-438 of the previously submitted version we wrote: “We also demonstrate that specific proteins involved in carbohydrate metabolism and specifically fructan biosynthesis, co-locate with the fructans.” We still feel that this assumption is supported by our data. Our results showed that short chain FOS locate to outer tissues, while long chain FOS and fructopolysaccharides accumulate in the middle and inner tissues of the root cortex (Figures 1, 2, and 3). The proteome profiling then revealed, among the six proteins involved in ‘fructan metabolism’, two proteins with highest expression in the inner root tissue (Figure 5). One was a 6G-FFT (Swissprot ID Q5FC15), involved in the formation of neoseries-type fructans and described in asparagus previously (Ueno et al., 2005). The other one was assigned as ß-fructofuranosidase (NCBI ID gi:1150742863), derived and annotated from a genomic sequence from asparagus, which has not yet been functionally characterized and revealed a highly significant differential expression. The enzyme product of this gene likely exhibits fructan:fructan 1-fructosyltransferase (1-FFT) activity, mediating the transfer of fructose units from one fructan to another and thus leading to the enhanced formation of higher DP inulin neoseries-type fructans in the inner root tissues. We have stated this in the discussion lanes 549-557.

The reason for choosing a proteome approach over a transcriptome (or qPCR) approach is that peptide-based analyses are more applicable for investigating the expression of closely related isoforms, such as sucrose synthases, phosphoenolpyruvate carboxylases and many more proteins involved in carbohydrate / fructan metabolism (please see Supplementary Table S3). Label-free LC-MS/MS analysis allows the simultaneous and specific measurement of highly homologous isoforms by measuring the surrogate peptides that are specific to each isoform. In addition, the detection of a particular gene product in a transcript-based experiment does not confirm the presence or absence of the resulting protein product. It is also commonly understood that quantitative differences in the transcript of a particular gene may not necessarily correlate with the corresponding protein abundance, which has also been observed in our group (Witzel, K., Abu Risha, M., Albers, P., Börnke, F., & Hanschen, F. S. (2019). Identification and characterization of three epithiospecifier protein isoforms in Brassica oleracea. Frontiers in Plant Science, 10(1552), doi 10.3389/fpls.2019.01552). Therefore, we believe that a qPCR profiling of fructan biosynthetic genes would not lead to novel insights into the spatial partitioning. We rather believe that future experiments will benefit from enzymatic activity measurements of extracts from the three root regions to better understand fructan spatial accumulation.

Moreover, how the author can explain about sucrose localization that mainly concenrs the outer and middle root tissues while Invertase (β-fructofuranosidase, EC 3.2.1.26) were mainly found in the inner tissue?

Answer:

Generally, sucrose levels accounted for the highest peak in all three root tissues, while highest values were observed in middle and outer root tissues. In plants, sucrose is transported from source to sink organs over long distances in solution in the phloem sap via the network of sieve element cells (Lemoine 2000). As sucrose must pass several membrane barriers on its way several sucrose transporters were shown to be involved in this process (Durand et al. 2018). The high concentration of sucrose in the sieve elements increases turgor pressure allowing sucrose transport in the sieve tubes to the sink organ according to the Mass-Flow model (Münch 1930). Unloading of sucrose from the phloem can occur through symplastic or apoplastic pathways depending on tissue types or development stages, but always where the turgor pressure drops. Apoplastic sucrose unloading involves sucrose transporters in sink organs or its conversion to hexoses by a cell-wall invertase (please refer to Durand et al., 2018 and references therein). Therefore, in asparagus roots central vessel cells, including sieve elements, will force sucrose to be released from the inner root tissues towards the outer root tissues. It is very likely that for this purpose the sucrose concentration, and thus turgor pressure, must be kept low in the tissues adjacent to the central vessel cells. This fits well to our observation that for all ‘Sucrose metabolism’ related enzymes, such as alpha,alpha-trehalose-phosphate synthase (α,α-THP synthase), sucrose-phosphate synthase (SP synthase), sucrose synthase (Susy) and sucrose phosphatase (Suc-phosphatase), isoforms with highest abundance in inner root tissues and lowest abundance in outer root tissues were detected. Accordingly, for almost all other ‘carbohydrate metabolism’ related enzymes highest abundance levels in inner tissues were detected in our study. The striking abundance of one ß-fructofuranosidase (NCBI ID gi:1150742863) in the inner tissues is possible related to both, the diminishment of sucrose and the synthesis of higher DP inulin neoseries-type fructans in this tissue. We have extended the discussion regarding this information (lanes 532-549).

Minor changes:

Lane 111: “Seeds were cultivated” change with “Seeds were germinated”

Answer.

We have changed the wording accordingly.

Lane 153: Oligosaccharide extraction and profiling: indicate the amount of starting material that you use in your experiments (g) and the ratio for the extraction.

Answer:

Thanks for this remark. Generally, very low sample amounts were needed for our analyses, as the extraction of soluble sugars can be performed in “excess of solvent” conditions. We have extended the method description in this regard (lanes 159-162). The sample amounts for sugar extraction were as follows:

Increase figures resolution.

Answer:

Many thanks for this hint. As required by the journal’s instructions figures were implemented in the manuscript file directly. However, this process usually results in a decrease of image quality and resolution. We are very happy to provide the original files for final publication processing.

Round 2

Reviewer 1 Report

The authors have made a nice effort in revision.