Preliminary investigation of glyphosate resistance mechanism in giant ragweed using transcriptome analysis

Giant ragweed ( ) is a highly competitive annual weed prevalent Ambrosia trifida mainly in the United States across the eastern Corn Belt. Glyphosate has been a key herbicide to help tackle the spread of giant ragweed in the past few decades. Recently, there have been reports of widespread resistance to glyphosate in giant ragweed, with the mechanism of resistance yet to be determined. We designed a single-replicate RNA sequencing experiment to study the genes differentially expressed between glyphosate-resistant and glyphosate-sensitive biotypes of giant ragweed. We used a assembly de novo of the giant ragweed transcriptome to determine key marker genes that could help explain the mechanism of resistance. 1 2 3 2


Introduction
Giant ragweed (Ambrosia trifida) is a problematic annual weed in the United States and Canada, particularly in fields where corn and soybean are grown 1,2 .Due to the extensive use of weed control measures such as the growth of genetically-modified glyphosate-resistant crops, giant ragweed populations have been kept in check 3 .But due to the overuse of glyphosate and strong selective pressure, resistant weeds have been reported across the world (http://www.weedscience.org) 4,5.The mechanism of resistance to glyphosate in other common weeds such as Malaysian goosegrass, Italian ryegrass, and rigid ryegrass have been identified [6][7][8] .However, the glyphosate resistance mechanism in giant ragweed is still unknown.In this study, we compare gene expression differences between the glyphosate resistant and sensitive giant ragweed plants using a time course experiment and identify genes that could be involved in glyphosate resistance.

Methods
Glyphosate-resistant and glyphosate-sensitive seeds of giant ragweed were collected from Noble County, Indiana (N41° 28.470 W85° 29.371) and Darke County, Ohio (N40° 15.9 W84° 42.7) respectively.After seed germination, plants at the five-node growth stage were selected for herbicide treatment.mRNA was extracted from leaf disks following a protocol adapted from Eggermont et al. 2 cm diameter leaf disks from the first fully developed leaf were punched out, frozen in liquid nitrogen, and total RNA was extracted in a 2 ml test tube.mRNA was sequenced using Illumina TruSeq for four time points -pre-treatment (0 hour), 3 hours, 8 hours and 12 hours after treatment with glyphosate at a rate of 0.7kg ae ha -1 (recommended field rate) sprayed using a compressed-air bench top track sprayer with a nozzle pressure of 249 kPa delivering a volume of 187 L of spray solution ha -19 .RNA sequences were assembled using the Trinity package (version r2012-10-05) from paired-end reads 10 .
RNA-seq reads were mapped to the transcriptome assembly, and the counts per million transcripts (CPM) value was determined using RSEM (version 1.2.8) 11 .Since we observed clear systemic changes in gene expression even at the first time point, we used a set of genes previously published in rice analyses as controls to normalize the expression values 12 .The consistent expression of these genes with respect to each other across the time course was verified (Table 1).Gene level counts that were less than 1 CPM in all time points were excluded from further analysis.Expression ratios were then calculated for each assembly, comparing the expression levels in the glyphosate resistant and sensitive strains at each time point.Numbers larger than 1 therefore reflect genes (transcript assemblies) with higher expression in the resistant variety.Sampling variation was controlled by the addition of a pseudo count of 0.5 CPM before calculating expression ratios.Assemblies with expression ratios greater than 4, or less than -4 were considered to be differentially expressed and were further examined.Annotations of the transcriptome were done using Trinotate (version r2013-02-25) 13 .

Results
There is a clear difference in gene expression patterns between the resistant and sensitive plants even before plants were treated with herbicide (Table 2).The top differentially expressed transcripts in resistant and sensitive plants before treatment are shown in Table 3 and Table 4.The response to glyphosate is rapid, and a large number of genes are significantly differentially expressed within the first three hours after treatment compared to pre-treatment expression levels (Table 5).
The genes with at least a four-fold change in expression level were identified in resistant and sensitive plants, and pathways associated with significantly over-represented genes identified using agriGO (cutoff P < 1e-7) 14,15 .Pathways with terms such as "response to other organisms" and "lipid biosynthetic process", both of which are known to be related to pathogen response, were the most significantly over-represented 16 .Contrastingly, pathways that are typically over-represented in the sensitive biotype are annotated with terms like "response to stress", "response to oxidative stimulus" and "lignin biosynthesis", which are known stress response indicators 17 .This leads us to speculate that, not only do resistant giant ragweed plants react to glyphosate treatment in a manner resembling pathogen defense reactions, but they are already primed by alterations in stress response processes to hyper-react.This is consistent with the rapid necrosis reaction observed in resistant giant ragweed biotypes used in this study.

Conclusion
The complete transcriptome assembly of giant ragweed has been deposited in the NCBI BioProject database (http://www.ncbi.nlm.nih.gov/bioproject/) and is publicly available under accession PRJNA267208.The preliminary time-course experiment presented here identified groups of genes that may explain glyphosate resistance in giant ragweed.A more extensive transcriptome analysis study, with multiple replicates of sensitive and resistant giant ragweed biotypes, from a broader range of geographic sources, and with shorter time intervals will be useful to overcome the limitations of this preliminary study.This dataset contains two data files (csv) and a text file providing description of the data.
Table 1.Normalization of expression values using control genes from rice.
12 genes from the list of 25 genes identified by Jain (2009) showed relatively stable expression across all time points, and thus were used for determining the scaling factor for the normalization 12 .

Open Peer Review
Current Referee Status: Glyphosate is a widespread herbicide since 1994, when transgenic crops were commercially released into the environment.Several cases of glyphosate resistance have been reported, after repeated use of this herbicide, since then.Previous cases have included species of Poaceae or Amaranthaceae.
The manuscript by Padmanabhan and colleagues presents an interesting comparison between the transcriptomic profiles of giant ragweed plants collected from populations that are either resistant or sensitive to the use of glyphosate.This paper is relevant, because the authors investigate an agricultural weed that is not a Poaceae or Amaranthaceae.This contribution is useful to addressing if glyphosate resistance follows common metabolic pathways in distantly related plants.Moreover, they do so following an experimental set up, that can potentially be replicated and expanded; this is an additional contribution.
We summarize our main concerns on the present status of the work under review, and provide comments or suggestions that may help improve this work.
A more comprehensive transcriptomic comparison would greatly contribute to the objetives of the work under review.
We recommend: Transcriptomic analyses made between resistant and sensitive ragweed individuals sampled at different time points during the experiment -which the authors summarize in Tables 3 and 4 and compare on table 5-with an explicit discussion of their results in light of the transcriptomic profiles (and highly expressed genes) described for other glyphosate-resistant tweeds, in order to discuss if common metabolic pathways that are not lineage specific, are being selected upon through pervasive use of glyphosate in the corn belt in the USA, or not.
The paper would also benefit from a more thorough discussion concerning the type and function of genes that were found to be highly expressed in resistant and/or sensitive giant ragweed biotypes.We recommend using other databases, such as KEGG.It would be particularly important to focus on genes involved in secondary growth, lateral meristem activity, or any aspect related to perennial habit, in contrast to non-perennial habit of close relatives, if there are any.Also, comparison to functional information available in other perennial and non perennial herbs that have evolved 4.

5.
functional information available in other perennial and non perennial herbs that have evolved resistance, would be very useful.
The fact that the genes that are highly expressed in sensitive giant ragweed are those involved with chloroplast function, which is the function that is primarily and directly affected by glyphosate, through the inhibition of aromatic aminoacid synthesis.This is not addressed by the authors aside from Table 4.Such discussion would also improve the significance of the paper.
In the last paragraph of the discussion, the authors propose a hypothesis concerning the highly expressed genes in resistant ragweed.The following statement made by the authors is confusing and requires clarification: "This is consistent with the rapid necrosis reaction observed in resistant giant ragweed biotypes used in this study".Necrosis is a common response to biotic and abiotic stress but this phrase could be further clarified if the authors maybe by explaining to the reader what is the main physiological response to glyphosate observed in resistant biotypes: necrosis of all leaves?Of some?Necrosis and then new leaf generation?Are any of these observed in non resistant biotypes, ever?
In the conclusions the authors suggest how a more thorough analysis could be carried on in the future for giant ragweed resistance.This made it clear that some details were missing in the methods section of the present paper: how many plants did they sample (two for each biotype for biological replicates; same individual plant, but two samplings for technical replicates?).These details should be clarified, please.A useful review article that may be cited and help tackle questions pertaining the possible metabolic pathways affected by glyphosate: F1000Research validated, genetically or/and functionally.No isogenic populations of sensitive and resistant glyphosate plants were used.Whereas the paper is well written and the data clearly presented, this work is a good start, but it is in a too early phase to be indexed.The article might be, from my point of view, re-focused, on the transcriptome assembly and annotation.This would be acceptable.
I have read this submission.I believe that I have an appropriate level of expertise to state that I do not consider it to be of an acceptable scientific standard, for reasons outlined above.
No competing interests were disclosed.Competing Interests:

Table 3 . Genes with greater than four-fold higher expression in resistant plants compared to sensitive plants.
Genes expressed higher in resistant plants tend to play important roles in pathogen response regulation.

Table 4 . Genes with greater than four-fold higher expression levels in sensitive plants compared to resistant plants.
Genes expressed at a higher level in sensitive plants seem to impact control of stress response.

Table 5 . Gene expression differences between resistant and sensitive biotypes of giant ragweed after treatment with glyphosate.
After treatment with glyphosate, the number of differentially expressed genes increases rapidly within the first three hours, and continues to increase at later time points.