Background: Storage roots are an ecologically and agriculturally important plant trait that have evolved numerous times in angiosperms. Storage roots primarily function to store carbohydrates underground as reserves for perennial species. In morning glories, storage roots are well characterized in the crop species sweetpotato, where starch accumulates in storage roots. This starch-storage tissue proliferates, and roots thicken to accommodate the additional tissue. In morning glories, storage roots have evolved numerous times. The primary goal of this study is to understand whether this was through parallel evolution, where species use a common genetic mechanism to achieve storage root formation, or through convergent evolution, where storage roots in distantly related species are formed using a different set of genes. Pairs of species where one forms storage roots and the other does not were sampled from two tribes in the morning glory family, the Ipomoeeae and Merremieae. Root anatomy in storage roots and fine roots was examined. Furthermore, we sequenced total mRNA from storage roots and fine roots in these species and analyzed differential gene expression. Results: Anatomical results reveal that storage roots of species in the Ipomoeeae tribe, such as sweetpotato, accumulate starch similar to species in the Merremieae tribe but differ in vascular tissue organization. In both storage root forming species, more genes were found to be upregulated in storage roots compared to fine roots. Further, we find that fifty-seven orthologous genes were differentially expressed between storage roots and fine roots in both storage root forming species. These genes are primarily involved in starch biosynthesis, regulation of starch biosynthesis, and transcription factor activity. Conclusions: Taken together, these results demonstrate that storage roots of species from both morning glory tribes are anatomically different but utilize a common core set of genes in storage root formation. This is consistent with a pattern of parallel evolution, thus highlighting the importance of examining anatomy together with gene expression to understand the evolutionary origins of ecologically and economically important plant traits. Keywords: Comparative transcriptomics, Gene expression, Ipomoea, Ipomoea batatas (sweetpotato), Parallel evolution, Root anatomy, Storage roots Background phenotypically and functionally similar but utilizing differ- Parallel and convergent evolution of complex morpho- ent genetic mechanisms e.g. [1–3]. Alternatively, traits logical traits has long been of interest to evolutionary evolving in parallel have the same genetic basis [4–6]. biologists, who have noted that functionally and mor- Often, differentiating between these alternative evolution- phologically similar phenotypes have evolved independ- ary scenarios is difficult. Studies comparing morphology, ently in unrelated lineages. Studies characterizing the anatomy, gene expression and other aspects of a trait can genetic basis of independent phenotypic evolution have provide insights into whether a trait evolved convergently concluded that many traits evolve convergently, appearing or in parallel. Morning glories offer an ideal system in which to address hypotheses regarding convergent versus parallel * Correspondence: email@example.com Plant Biology Department, University of Georgia, Athens, GA 30602, USA evolution. In morning glories, storage root formation Present address: Conservation and Research Department, Atlanta Botanical has been either lost or gained at least ten times inde- Garden, Atlanta, GA 30309, USA pendently, and storage roots are found in many diverse Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Eserman et al. BMC Plant Biology (2018) 18:95 Page 2 of 11 morning glory lineages such as those containing I. bata- and share few to no differentially expressed orthologous tas, I. lindheimeri,and Distimake dissectus; however, it is genes, this would support the hypothesis that storage unclear whether the ancestor of all morning glories was roots evolved independently in storage root forming able to form storage roots . Studies characterizing stor- lineages (convergent evolution). Using this comparative age root development in sweetpotato have demonstrated approach, we can better understand the genetic mecha- that a storage root is simply a modification of the tap- nisms and evolutionary origins of storage root formation. root, an adventitious root, and/or one or more lateral Through this work we ultimately seek to understand the roots such that the root cambium expands and the genetic basis of storage root formation and whether inde- starch-storage tissue proliferates [8–12]. The prolifera- pendent lineages utilize the same or different genetic tion of starch-storage tissue expands the root so that mechanisms during storage root development across the storage roots are much greater in diameter than roots morning glory phylogeny. which do not function in long-term starch storage. Studies analyzing gene expression differences between Methods fine and storage roots in sweetpotato (Ipomoea batatas) Plant material have found that genes in the starch biosynthesis path- Three pairs of closely-related species were selected from way are highly expressed and lignin biosynthesis genes across the morning glory phylogeny, where one member have reduced expression in storage roots compared to of the species pair produces storage roots and fine roots fine roots . Studies have also implicated three genes and the other produces only fine roots. The three stor- in the development of storage roots, two of which are age root forming species are Ipomoea batatas (L.) Lam. MADS-box transcription factors [13, 14] and the other (sweetpotato), I. lindheimeri A. Gray and Distimake is an alpha-expansin gene . However, these studies dissectus (Jacq.) Simoes & Staples (formerly Merremia were strictly limited to sweetpotato. Comparative stud- dissecta), and the species that produce only fine roots ies may reveal genes involved in storage root formation are I. trifida G. Don, I. nil (L.) Roth, and D. quinquefo- across distantly related species. lius (L.) Simoes & Staples (formerly Merremia quinque- In addition to the evolutionary importance, storage folia). All three pairs of species were utilized for roots have economic and ecological significance as well. anatomical observations. Four species, I. batatas, I. tri- Sweetpotato [Ipomoea batatas (L.) Lam.] ranks among fida, D. dissectus and D. quinquefolius, were used for the ten most important crop species for human nutri- transcriptome sequencing so that we could directly tion. In 2014, over 100 million tonnes of sweetpotato contrast gene expression of different observed root ar- were produced worldwide . The large storage roots chitectures. Plant material was obtained from outside are an important source of carbohydrates and vitamin A sources, including USDA GRIN, seed companies, and the in developing countries . More generally, storage seed collections of R. Miller and J. Ekrut (Additional file 1: roots play a key role in the life history and ecological Table S1). Three biological replicates were chosen for each strategies of plants, as perennial species tend to mobilize species except I. trifida, where RNA-seq libraries for one starch to roots year-round and thus form storage roots sample consistently failed. Sweetpotato (I. batatas)is but annual species cease starch mobilization after only a hexaploid, and I. trifida is diploid . Ploidy of D. dissec- few months . Additionally, root carbohydrate reserves tus and D. quinquefolius have not been previously deter- are necessary for resprouting after cutting or large-scale mined; therefore, we attempted root tip squashes but were events such as fire [19–21]. not able to get separation among chromosomes for count- Given what is known about the developmental biology ing. Genome size estimates are often used to infer ploidy and anatomy of storage roots, lineages that form storage in some species; however, because chromosome number roots may represent instances of either convergent or has never been determined in these species and genome parallel evolution. In this study, we aim to: 1) to under- size varies widely among morning glories of the same ploi- stand the anatomy of storage roots in morning glories dal level [22–24], we are not able to use this method to and 2) to characterize gene expression during an early determine ploidy. stage of storage root formation. If we observe that stor- Sweetpotato is vegetatively propagated, so cuttings age roots from distantly related morning glory lineages were planted of the three cultivars (Beauregard, Jewel, are anatomically similar and share an overlapping set of and Tinian) with three true leaves. Seeds of the other differentially expressed orthologous genes, this would five species were scarified before planting. Seeds and provide evidence supporting the hypothesis that storage cuttings were planted in Fafard 3B mix in 4″ square roots evolved prior to the diversification of morning pots. Seeds were allowed to germinate for 1 week in the glories and were subsequently lost in lineages that do UGA Greenhouses. Plants were then moved to a growth not form storage roots (parallel evolution). However, if chamber under an 8 h photoperiod and 30°/25 °C day/ we observe that storage roots are anatomically dissimilar night temperatures . Previous studies have found that Eserman et al. BMC Plant Biology (2018) 18:95 Page 3 of 11 storage root formation occurs within four to six weeks slight modifications. Libraries were amplified with 15 PCR after planting in sweetpotato [8, 25]; therefore, plants in cycles. An initial test set of libraries showed adapter dimer this study were grown for six weeks prior to sampling. peaks; therefore, the adapter was diluted 1.25 μMrather Roots were sampled using the following procedure: roots than the standard 1.5 μM, which eliminated adapter dimer were removed from medium, washed in tap water, and peaks in future libraries. The library preparation protocol rinsed a final time in nuclease-free molecular biology used in this experiment implements the dUTP method grade water. The primary root was dissected from the  to generate stranded libraries. whole plant, and fine lateral roots were then dissected Libraries were quantified using quantitative real-time from the primary root. Fresh root tissue was flash frozen PCR prior to sequencing. Libraries were diluted to 10 nM in liquid nitrogen and was subsequently stored at − 80 °C for sequencing. Barcoded and diluted libraries were until RNA isolation. Alternatively, fresh root tissue was pooled before sequencing. All libraries were sequenced at used immediately for anatomical observations. the Georgia Genomics Facility on the Illumina NextSeq platform with paired-end 150 bp reads. Illumina sequence Anatomical observations data used to assemble transcriptomes has been deposited Fresh root tissue was sectioned by hand with a razor to the GenBank Sequence Read Archive database under blade. A main goal of this was to observe the spatial de- Bioproject PRJNA448837. position of starch in cross sections of the root; therefore, fresh sections were necessary because starch is removed during standard tissue clearing . Serial sections were Transcriptome analysis taken from fine roots and from two places on the taproot Reads for each species were assembled separately into or storage root: 1) after the 4th lateral root, and 2) after transcripts with the Trinity software suite version the 10th lateral root. Sections were stained with Lugol’s r20140717 . Within-species transcriptome assembly iodine, a solution of iodine and potassium iodide which and analysis followed the developed Trinity pipeline indicates the presence of starch, or phloroglucinol-HCl, [33, 34]. Read quality was assessed with FastQC. Prior which stains lignin , immediately following sectioning. to assembly, reads were quality trimmed with Trimmo- Stained sections were mounted in a filtered 20% CaCl matic as implemented in the Trinity package. Bases at solution . Mounted sections were viewed with a Zeiss the beginning and end of a read with a phred score less Axio microscope with attached camera under either a than 5 were removed. In addition, reads less than 50 bp 2.5× or 10× objective lens. Sections too large to be viewed long were removed. Reads for each library were digitally in a single field of vision using the 2.5× objective lens were normalized to a maximum of 50× coverage within Trinity captured in multiple images which were then stitched (−-normalize_reads) to accelerate the assembly process. together using the image stitching plugin for the Fiji Reads were considered paired-end in the assembly, where distribution of ImageJ [29–31]. Field of vision length was the first read of the pair was considered the reverse read determined using a standard microscope scale, and scale and the second was the forward read (−-SS_lib_type RF). bars were added to images in ImageJ. We then filtered assemblies to remove poorly supported isoforms and contaminants. We used RSEM version RNA isolations and library construction 1.2.20  to estimate gene and transcript abundances as Total RNA was isolated from frozen root tissue using the implemented in the Trinity package (align_and_estima- standard Trizol protocol (Life Technologies). RNA was te_abundance.pl script). Non-normalized reads were eluted in molecular biology grade H 0 following isolation. mapped to each transcriptome assembly with Bowtie 2 DNA was removed using the TURBO DNA-free kit . Isoforms which were supported by less than 30% of (Thermo Fisher Scientific). Prior to library construction, the total reads for a gene from two or more biological rep- RNA quality was assessed with the Agilent Bioanalyzer licates or had an FPKM less than 2 were removed, as these 2100 using the RNA 6000 Nano kit (Agilent Technologies, represent possible assembly artifacts. Filtering was per- Santa Clara, CA). mRNA was isolated from total RNA formed using the perl script filter_fasta_by_rsem_values.pl using the NEBNext Poly(A) mRNA Magnetic Isolation in the Trinity software package . To remove contami- Module (New England Biolabs, Inc.). The first mRNA nants, we annotated the assembled transcriptomes in Tri- isolations performed using the recommended total RNA notate  using a blastx of the filtered assembly against input yielded low mRNA concentrations. Therefore, the the Uniprot database. Transcripts with annotations from amount of total RNA added to the mRNA isolation proto- any taxon other than Viridiplantae with an e-value greater col was increased to 5 μg, the maximum recommended than 1e-5 and 40% identity were removed as potential RNA input. Libraries were constructed with the NEBNext contaminants. Finally, the program DeconSeq version Ultra Directional RNA Library Prep Kit for Illumina (New 0.4.2  was used to further filter any remaining bacter- England Biolabs, Inc.) using the standard protocol with ial, viral, and human contaminant sequences. Eserman et al. BMC Plant Biology (2018) 18:95 Page 4 of 11 RSEM  and Bowtie 2  were again used to map similar, accumulate very little starch, and do not show reads from individual libraries back to the filtered tran- evidence of proliferation of starch-accumulating cells. scriptome assemblies and calculate transcript abundances. Third, storage roots of the three storage root forming EdgeR  was then used to assess differentially expressed species showed similar starch accumulation, specifically, genes between storage roots and fine roots of sweetpotato proliferation of the starch-accumulating cells that oc- and Distimake dissectus using perl scripts from the Trinity curred within the bounds of the endodermis. Finally, analysis pipeline . EdgeR was run separately for each the vascular tissue in storage roots of sweetpotato and species and incorporated biological replicates for each tis- I. lindheimeri appeared visually similar, where the sue type. FPKM values for each library were normalized starch-accumulating cells disrupted the organization of by library size. This normalization process is referred to as the vascular tissue. In contrast, the vascular tissue of “Trimmed Mean of M-values”, or TMM, normalization storage roots of Distimake dissectus appeared markedly . Only TMM-normalized FPKM values were used for different such that vascular tissue was tightly organized differential expression analysis (Additional file 2: Table S2, in the center of the cross section. Additional file 3: Table S3, Additional file 4: Table S4, and Additional file 5: Table S5). Transcripts were considered Transcriptome assembly statistics significantly differentially expressed at a false discovery The final dataset included seventeen RNA-seq libraries rate (FDR) less than 0.05 and a log fold change of 2 (Add- from two pairs of morning glory species. Transcriptome itional file 6: Table S6 and Additional file 7: Table S7). We assembly statistics are shown in Table 1. Before filtering, then generated Euclidean distances among transcripts and the Distimake quinquefolius transcriptome had the lar- libraries and used a complete linkage clustering approach gest number of transcripts, and the I. trifida transcrip- on the Euclidean distance matrices to cluster transcripts tome had the fewest assembled transcripts. Transcript and libraries in edgeR. N50 ranged from 952 to 1277 nt. We then filtered the Protein coding regions were identified from the final raw assemblies by isoform percentage and FPKM, which filtered assemblies using the program Transdecoder . resulted in a 42–70% reduction in the number of tran- Protein sequences shorter than 50 amino acid residues scripts in the assembly (Table 2). This step removed po- long were not kept in the final set of peptide sequences. tentially erroneous transcripts that were not supported Functional annotation utilized the standard Trinotate by re-mapped reads. Further filtering of bacterial, fungal, pipeline , which incorporated a blastx search of the algal, and viral transcripts using Swiss-prot annotations assembled transcripts against the Uniprot database and and DeconSeq resulted in an additional ca. 3900–5700 a blastp search of the peptide sequences inferred from transcripts removed from each assembly. Only the tran- Transdecoder against the Uniprot database. These re- scriptomes filtered by isoform percentage and FPKM sults as well as gene ontology (GO) term annotations of and which had contaminants removed were used for the best gene match in Uniprot were incorporated into a downstream analyses. SQLite database using Trinotate  (Additional file 8: Table S8, Additional file 9: Table S9, Additional file 10: Within species differential gene expression Table S10, and Additional file 11: Table S11). We assessed differential gene expression between stor- Peptide sequences from the final filtered assemblies age roots and fine roots in sweetpotato and Distimake from all four species were sorted into gene families dissectus separately. After accounting for multiple com- with OrthoFinder  to determine orthology among parisons, there were 2643 genes differentially expressed transcripts from the four species. Coding sequences of (DE) between storage roots and fine roots in sweetpotato the gene families estimated from OrthoFinder were and 219 DE genes in D. dissectus at a FDR < 0.05 (Fig. 2a, aligned in SATé-II . Gene trees were estimated in b). In both species, there were more transcripts highly RAxML , and node support was determined using expressed in storage roots than in fine roots. As a con- 500 bootstrap replicates. vention, upregulated transcripts refers to those more highly expressed in storage roots vs. fine roots and Results downregulated refers to transcripts with reduced expres- Root anatomy sion in storage roots compared to fine roots. In sweetpo- Results of the root anatomical observations are shown tato, 1642 transcripts were upregulated and 1001 in Fig. 1. There were three main results from this. transcripts were downregulated. In Distimake dissectus, First, fine roots of all six species are anatomically there were 178 upregulated transcripts and 41 downreg- similar and exhibit the typical eudicot root anatomy ulated transcripts. with highly organized vascular tissue in the center and The top ten most abundant gene ontology annotations a larger cortex. Second, we found that the taproot of for the differentially expressed genes in sweetpotato and the species that do not form storage roots appear D. dissectus are found in Table 3. When we compare the Eserman et al. BMC Plant Biology (2018) 18:95 Page 5 of 11 Fig. 1 Root cross sections from three pairs of species, where one member of the species pair forms storage roots and the other does not. To the left is a phylogeny depicting the evolutionary relationships among the six species with arrows denoting the two tribes, Ipomoeeae and Merremieae. The left-most three columns are root sections stained with Lugol’s iodine, which indicates starch a dark blue to black color. The right-most three columns are root sections stained with phloroglucinol-HCl, which stains lignin orange to pink. Scale bars are included with each section. Black bars are 1 mm, and blue bars are 0.5 mm in length. White arrows indicate starch-storage tissue ten most abundant GO annotations from genes DE in with OrthoFinder . We then queried the orthologous sweetpotato and D. dissectus, we find that eight of these groups for known sets of transcripts differentially GO terms overlap. Additionally, many of the most expressed (DE) between storage and fine roots in sweetpo- enriched GO terms were involved in transcription or tato and Distimake dissectus. We found there were 57 are annotated as having transcription factor activity orthologous genes DE between storage roots and fine (Table 3). roots of both species (Fig. 2c). We then examined GO term annotations for the set of orthologous DE transcripts Between species differential gene expression (Table 4). Transcripts annotated with amyloplast or starch To compare gene expression between orthologs of differ- biosynthetic activity were found to represent a larger per- ent species, we sorted transcripts into orthologous groups cent of the total GO annotations in the set of shared DE Table 1 Transcriptome assembly statistics I. batatas I. trifida D. dissectus D. quinquefolius Total reads (PE 150 bp) 39,632,572 15,657,942 44,877,650 64,267,290 No. of transcripts 245,140 119,153 254,174 363,820 %GC 40.97 42.1 39.67 39.37 Transcript N50 952 1125 1277 952 Median transcript length 416 455 446 417 Mean transcript length 663.57 732.7 777.18 663.29 Eserman et al. BMC Plant Biology (2018) 18:95 Page 6 of 11 Table 2 Assembly statistics after successive filtering by IsoPct and FPKM, Swiss-prot annotations, and Decon-Seq I. batatas I. trifida D. dissectus D. quinquefolius Transcripts in original assembly 245,140 119,153 254,174 363,820 Transcripts filtered by IsoPct, FPKM 158,267 (64.6%) 51,181 (43.0%) 176,584 (69.5%) 209,593 (57.6%) Transcripts filtered by Swiss-prot annotations 5097 (2.1%) 5262 (4.4%) 3529 (1.4%) 3665 (1.0%) Transcripts filtered by Decon-Seq 619 (0.3%) 491 (0.4%) 441 (0.2%) 595 (0.2%) Total removed by filtering 163,983 (66.9%) 56,934 (47.8%) 180,554 (71.0%) 213,853 (58.8%) transcripts than in the DE transcripts from sweetpotato in storage roots and had reduced expression in fine roots, and D. dissectus analyzed separately (Tables 3, 4). Similarly, except for GLGL1 in D. quinquefolius (Fig. 3). Furthermore, we examined the functional annotation of these transcripts we examined expression of transcripts annotated as having and found that some of these DE genes share close hom- transcription factor activity, where orthologs were differen- ology with transcription factors, alpha-expansin genes, tially expressed in both sweetpotato and D. dissectus (Fig. 4). genes that function in the starch biosynthetic pathway, In all cases, orthologs of the shared differentially expressed and one that functions in the starch degradation pathway. transcription factors were more highly expressed in storage roots than fine roots (Fig. 4). Among species differential gene expression We then wanted to examine the expression of genes in the Discussion starch biosynthetic pathway (Fig. 3). Most genes in the Root anatomy starch biosynthesis pathway were found to have reduced Results of the root anatomical work clearly show that the expression. However, orthologs of GLGL1 and SSG1 were storage roots of species in the tribe Ipomoeeae (sweetpotato significantly differentially expressed in sweetpotato and Dis- and I. lindheimeri) are anatomically quite different from timake dissectus (Fig. 3). These genes were highly expressed storage roots of Distimake dissectus, a member of the sister Fig. 2 Heat map of genes differentially expressed between storage roots and fine roots of sweet potato, Ipomoea batatas (a) and Distimake dissectus (b). Each row in the heatmap is depicting the expression patterns of each transcript, and each column represents each library. A dendrogram illustrating clustering of libraries is shown above each heatmap, and a dendrogram showing clustering of transcript expression patterns is to the left of each heatmap. c Number of transcripts differentially expressed between storage and fine roots and the number that were orthologous between I. batatas and D. dissectus Eserman et al. BMC Plant Biology (2018) 18:95 Page 7 of 11 Table 3 Top ten most abundant gene ontology (GO) categories represented in genes differentially expressed between storage roots and fine roots of Ipomoea batatas and Distimake dissectus considering each species separately Species % of Total GO annotation Type Specific I. batatas 4.50 GO:0016021 cellular_component integral component of membrane I. batatas 3.12 GO:0005634 cellular_component nucleus I. batatas 2.96 GO:0005886 cellular_component plasma membrane I. batatas 2.45 GO:0005524 molecular_function ATP binding I. batatas 1.98 GO:0046872 molecular_function metal ion binding I. batatas 1.89 GO:0006351 biological_process transcription, DNA-templated I. batatas 1.69 GO:0005576 cellular_component extracellular region I. batatas 1.68 GO:0009507 cellular_component chloroplast I. batatas 1.63 GO:0003700 molecular_function sequence-specific DNA binding transcription factor activity I. batatas 1.57 GO:0003677 molecular_function DNA binding D. dissectus 4.37 GO:0016021 cellular_component integral component of membrane D. dissectus 3.06 GO:0005634 cellular_component nucleus D. dissectus 2.51 GO:0005886 cellular_component plasma membrane D. dissectus 2.51 GO:0003700 molecular_function sequence-specific DNA binding transcription factor activity D. dissectus 2.18 GO:0006351 biological_process transcription, DNA-templated D. dissectus 2.07 GO:0005524 molecular_function ATP binding D. dissectus 1.86 GO:0009507 cellular_component chloroplast D. dissectus 1.53 GO:0006355 biological_process regulation of transcription, DNA-templated D. dissectus 1.53 GO:0003677 molecular_function DNA binding D. dissectus 1.42 GO:0009501 cellular_component amyloplast tribe Merremieae. Starch-accumulating cells proliferated in which found an approximately equal number of genes all three storage root forming species; however, xylem up- and downregulated in storage roots compared to organization differed greatly in D. dissectus compared to fine roots . We sampled roots six weeks after planting storage roots of the other two species (Fig. 1). Our findings in contrast to Firon et al. (2013), which sampled roots at are consistent with other studies examining root anatomical four weeks. Given that we are sampling at a slightly later structure of sweetpotato [8–12]. However, we had no a growth stage, perhaps we are capturing a more active priori expectations with regard to root anatomy of all other stage of storage root bulking in this study. In the future, speciesincludedinthisstudy, as thisisthe firsttodocu- closer examination of the anatomical and gene expres- ment root anatomy of I. lindheimeri, I. nil, I. trifida, D. sion changes during the very early stages of storage root dissectus,and D. quinquefolius. formation would provide further insights into the devel- opment of this trait. Comparison of gene expression in all species Based on the anatomical results, we can generate expec- tations with respect to the transcriptome experiment. Starch biosynthetic pathway Starch accumulation occurred similarly in storage roots Starch biosynthesis occurs as part of a complex and dy- of all three species; however, xylem organization was namic pathway and the enzymesand transportproteins quite different in storage roots of D. dissectus. Therefore, involved depend heavily upon the tissue in which starch it is likely that genes involved in starch biosynthesis and is being synthesized. The process differs in photosyn- cell proliferation will be differentially expressed between thetic and heterotrophic tissues . Therefore, we storage and fine roots in both species, but genes in- focused on the starch pathway that has been character- volved in xylem organization may not show the same ized in potato tubers from Bahaji et al. because it gene expression patterns between species. is the most well-characterized starch biosynthetic path- At a broad level, more genes were found to be upregu- way in heterotrophic tissue in a species closely related lated in storage roots compared to fine roots in both to sweetpotato. sweetpotato and D. dissectus. Interestingly, this result is Whereas in photosynthetic tissue sucrose is broken in contrast to a previous RNA-seq study in sweetpotato into fructose and glucose prior to starch synthesis, in Eserman et al. BMC Plant Biology (2018) 18:95 Page 8 of 11 Table 4 Top ten most abundant gene ontology (GO) categories represented in the set of orthologous genes differentially expressed between storage roots and fine roots in both Ipomoea batatas and Distimake dissectus Species % of Total GO annotation Type Specific I. batatas 3.89 GO:0009507 cellular_component chloroplast I. batatas 3.53 GO:0016021 cellular_component integral component of membrane I. batatas 3.18 GO:0009501 cellular_component amyloplast I. batatas 3.18 GO:0005634 cellular_component Nucleus I. batatas 2.83 GO:0005524 molecular_function ATP binding I. batatas 2.12 GO:0019252 biological_process starch biosynthetic process I. batatas 2.12 GO:0003700 molecular_function sequence-specific DNA binding transcription factor activity I. batatas 1.77 GO:0006351 biological_process transcription, DNA-templated I. batatas 1.77 GO:0005886 cellular_component plasma membrane I. batatas 1.77 GO:0003677 molecular_function DNA binding D. dissectus 4.00 GO:0009507 cellular_component chloroplast D. dissectus 4.00 GO:0016021 cellular_component integral component of membrane D. dissectus 3.20 GO:0009501 cellular_component amyloplast D. dissectus 3.20 GO:0005634 cellular_component Nucleus D. dissectus 2.40 GO:0005576 cellular_component extracellular region D. dissectus 2.40 GO:0005524 molecular_function ATP binding D. dissectus 2.40 GO:0003677 molecular_function DNA binding D. dissectus 2.00 GO:0019252 biological_process starch biosynthetic process D. dissectus 2.00 GO:0006351 biological_process transcription, DNA-templated D. dissectus 2.00 GO:0003700 molecular_function sequence-specific DNA binding transcription factor activity heterotrophic tissues, sucrose is directly converted to sweetpotato and D. dissectus (Fig. 3). GLGL acts down- UDP-glucose before starch biosynthesis . In stream in the pathway, directly upstream of SSG, which is addition, the downstream conversion of UDP-glucose to involved in the synthesis of amylose . Generally, amyl- starch intermediates differs between eudicot and ose content in sweetpotato cultivarsishigh, ranging from monocot heterotrophic tissues. UDP-glucose is con- 20 to 33% of total starch content [44, 45], much higher than verted to glucose-1-phosphate by the enzyme UDP-glucose in other starch-rich root and tuber crops such as cassava pyrophosphorylase (UGPA). Glucose-1-phosphate is then . either transported from the cytosol into the amyloplast or This examination must be taken with the caveat that converted in the cytosol to glucose-6-phosphate by the en- starch accumulation and bulking may occur through dif- zyme phosphoglucomutase (PGMP). Glucose-6-phosphate ferent mechanisms in sweetpotato and potato. First, is then transported into the amyloplast by the transport sweetpotato storage roots and potato tubers arise from protein glucose-6-phosphate translocator (GPT) where it is different tissue types; storage roots from root tissue and converted back to glucose-1-phosphate by PGMP. tubers from stem tissue . Furthermore, tuber forma- Glucose-1-phosphate is converted to ADP-glucose by the tion in potato is controlled by a homologue of flowering action of ADP-glucose pyrophosphorylase (GLGL), which locus T (SP6A), and the process of tuber initiation is requires an input of ATP. ADP-glucose is then converted dependent on photoperiod . However, experimental to the main components of starch by granule-bound starch evidence has demonstrated that sweetpotato storage root synthase (SSG) to generate amylose or starch synthase initiation occurs under both long and short day regimes (SSY) and starch branching enzymes (GLGB) to generate . Future functional genomic research involving sweet- amylopectin. potato and its close relatives is necessary to elucidate the In the context of this study, we found that orthologs of exact mechanisms of starch biosynthesis and storage. two genes involved in starch biosynthesis had significantly higher expression in storage roots compared to fine roots Transcription factors in both sweetpotato and Distimake dissectus (Fig. 3). In this Of the fifty-seven orthologous genes differentially expressed study, GLGL1 and SSG1 were significantly differentially between storage roots and fine roots, seven were annotated expressed between storage roots and fine roots of as having transcription factor activity (Fig. 4). When we Eserman et al. BMC Plant Biology (2018) 18:95 Page 9 of 11 Fig. 4 Mean TMM-normalized FPKM values for the seven transcription factors found to be significantly differentially expressed between storage and fine roots in both Ipomoea batatas and Distimake dissectus at a FDR < 0.05. The heatmap depicts mean TMM-normalized FPKM values for orthologs of the transcription factors in each tissue type for all four species (bata = Ipomoea Fig. 3 Starch biosynthetic pathway adapted from Bahaji et al. 2014. batatas, trif = I. trifida, diss = Distimake dissectus, and quin = D. Metabolites are shown in black, and enzymes are shown in green. quinquefolius). The heatmap was colored by percentile, where genes Shown are TMM-normalized FPKM values for homologs in all four in the 10th percentile were colored yellow and those in the 90th species (bata = Ipomoea batatas, trif = I. trifida, diss = Distimake percentile were colored dark blue. No ortholog of KN1 could be dissectus, and quin = D. quinquefolius). Grey boxes indicate genes identified in the transcriptome assembly of I. trifida, and no ortholog where orthology could not be determined. Stacked boxes indicate of HAT22 could be identified in D. quinquefolius homologs of a particular gene. Gene names with an asterisk were found to be significantly differentially expressed at a FDR < 0.05 in both I. batatas and D. dissectus. The heatmap is colored by Arabidopsis thaliana, and functions specifically within percentile, where genes in the 10th percentile were colored the cambium of stems and roots [53, 54]. Perhaps this yellow and those in the 90th percentile were colored dark blue gene plays a role in the proliferation of starch-storage tissue that we observe in storage roots of sweetpotato further examine the annotated functions of these genes, and D. dissectus. two stand out as potential candidate regulators of storage root formation. Conclusions IDD5, also called RAVEN, has been shown to posi- The anatomical results suggested that storage roots tively regulate starch synthase in Arabidopsis thaliana differ from fine roots in starch content, deposition and . Additionally, IDD5 is part of a larger regulatory vasculature patterning. As expected, we found signifi- network that, among other functions, regulates spatial cantly higher expression of genes involved directly in patterning of root tissue through asymmetric cell div- starch biosynthesis in both storage root forming species ision [49–51]. Many members of the larger regulatory and increased expression of IDD5, a transcription fac- network to which IDD5 belongs were found to be differ- tor known to regulate starch biosynthesis in Arabidop- entially expressed between SRs and FRs in sweetpotato sis . Similarly, we found significant upregulation of cv. Georgia Jet and Xushu [8, 52], suggesting a possible WOX4, a gene known to be involved in vasculature role of IDD5 and members of this regulatory network in proliferation in Arabidopsis [53, 54]. Given the large storage root formation. number of orthologous genes DE between storage roots Similarly, WOX4 orthologs were DE between storage and fine roots, we hypothesize that there was a single roots and fine roots of both sweetpotato and D. dissec- origin of storage roots before the divergence of the tus. This gene has been shown to play a critical role in morning glory tribes Ipomoeeae and Merremieae given vasculature proliferation and secondary growth in that storage roots in the species examined are superficially Eserman et al. BMC Plant Biology (2018) 18:95 Page 10 of 11 anatomically different but store starch in a similar man- Acknowledgements We are grateful for funding from a Rosemary Grant Award from the Society ner. To further support this hypothesis, we find that many for the Study of Evolution, a Palfrey award from the UGA Plant Biology of the same genes were differentially expressed between Department, a Grant in Aid of Research from the Society for Integrative storage roots and fine roots in sweetpotato and Distimake and Comparative Biology and the USDA. We would like to thank Craig Yencho, Chung-Jui Tsai, Shumei Chang, Russell Malmberg, Rick Miller, dissectus. However, an alternative hypothesis, that storage Saravanaraj Ayyampalayam, and members of the Leebens-Mack lab for roots evolved multiple times independently using the helpful discussions on experimental design and data analysis. We would same genetic mechanisms, cannot be directly rejected by like to thank Magdy Alabady for assistance with RNA isolation and library preparation and Beth Richardson, Brigitte Bruns, and Zheng-Hua Ye for these results. Therefore, much more work must be done assistance with microscopy. Finally, we are grateful for assistance to test these hypotheses in a rigorous framework. The maintaining plants from the UGA greenhouse staff. findings presented here present a first step in understand- Funding ing the evolution and development of a plant trait that has Funding for this project was made available through a Rosemary Grant received little attention to date but is economically and Award from the Society for the Study of Evolution, a Palfrey award from the ecologically important. These results further demonstrate UGA Plant Biology Department, a Grant in Aid of Research from the Society for Integrative and Comparative Biology and the USDA. None of these the power of comparative studies to understand the devel- agencies were involved in the design or implementation of the study. opment of a trait and its evolution in a deeper way than to examine a single species. Availability of data and materials Data generated and analyzed in this study are available from the GenBank Sequence Read Archive database under Bioproject PRJNA448837. Supporting data have been provided as supplementary files. Additional files Authors’ contributions Additional file 1: Table S1. Accession information for the plant material LE, RJ, and JL-M designed the study. LE performed all data collection and used in this experiment. (XLSX 10 kb) analysis. LE and JL-M interpreted major results and prepared the manuscript for publication. All authors read and approved this manuscript. Additional file 2: Table S2. TMM-normalized FPKM values for fine root samples in Ipomoea batatas. Statistics were output from edgeR analysis. Ethics approval and consent to participate (XLSX 4456 kb) Not applicable. Additional file 3: Table S3. TMM-normalized FPKM values for fine root samples in Ipomoea trifida. Statistics were output from edgeR analysis. Competing interests (XLSX 1833 kb) JL-M is a member of the editorial board of BMC Plant Biology. Additional file 4: Table S4. TMM-normalized FPKM values for fine root samples in Distimake dissectus. Statistics were output from edgeR analysis. (XLSX 4129 kb) Publisher’sNote Springer Nature remains neutral with regard to jurisdictional claims in Additional file 5: Table S5. TMM-normalized FPKM values for fine root published maps and institutional affiliations. samples in Distimake quinquefolius. Statistics were output from edgeR analysis. (XLSX 5130 kb) Author details Additional file 6: Table S6. Genes differentially expressed between Plant Biology Department, University of Georgia, Athens, GA 30602, USA. storage roots and fine roots in Ipomoea batatas. Statistics from edgeR on U.S. Department of Agriculture, Plant Genetic Resources Conservation Unit, transcripts found to be significantly differentially expressed between Griffin, GA 30223, USA. Present address: Conservation and Research storage roots and fine roots in Ipomoea batatas. (XLSX 142 kb) Department, Atlanta Botanical Garden, Atlanta, GA 30309, USA. Additional file 7: Table S7. Genes differentially expressed between storage roots and fine roots of Distimake dissectus. Statistics from edgeR Received: 23 October 2017 Accepted: 8 May 2018 on transcripts found to be significantly differentially expressed between storage roots and fine roots in Distimake dissectus. (XLSX 23 kb) Additional file 8: Table S8. Functional annotation of genes and References transcripts in Ipomoea batatas from Trinotate. (XLSX 12149 kb) 1. Ng J, Smith SD. Widespread flower color convergence in Solanaceae via alternate biochemical pathways. New Phytol. 2015;209:407–17. Additional file 9: Table S9. Functional annotation of genes and 2. Yoon H-S, Baum DA. Transgenic study of parallelism in plant morphological transcripts in Ipomoea trifida from Trinotate. (XLSX 10975 kb) evolution. Proc Natl Acad Sci. 2004;101:6524–9. Additional file 10: Table S10. Functional annotation of genes and 3. Wittkopp PJ, Williams BL, Selegue JE, Carroll SB. Drosophila pigmentation transcripts in Distimake dissectus from Trinotate. (XLSX 13051 kb) evolution: divergent genotypes underlying convergent phenotypes. Proc Additional file 11: Table S11. Functional annotation of genes and Natl Acad Sci. 2003;100:1808–13. transcripts in Distimake quinquefolius from Trinotate. (XLSX 23712 kb) 4. Haas O, Simpson GG. Analysis of some phylogenetic terms, with Attempts at Redefinition. Proc Am Philos Soc. 1946;90:319–49. 5. Scotland RW. What is parallelism? Evol Dev. 2011;13:214–27. 6. Des Marais DL, Rausher MD. Parallel evolution at multiple levels in the Abbreviations origin of hummingbird pollinated flowers in Ipomoea. Evolution. DE: Differentially expressed; FDR: False discovery rate; FPKM: Fragments Per 2010;64:2044–54. Kilobase of transcript per Million mapped reads; GLGB: Starch branching 7. Eserman LA, Tiley GP, Jarret RL, Leebens-Mack JH, Miller RE. Phylogenetics enzyme; GLGL: ADP-glucose pyrophosphorylase; GO: Gene ontology; and diversification of morning glories (tribe Ipomoeeae, Convolvulaceae) GPT: Glucose-6-phosphate translocator; IDD5: Protein indeterminate-domain based on whole plastome sequences. Am J Bot. 2014;101:92–103. 5; PGMP: Phosphoglucomutase; RNA-seq: RNA sequencing; SSG: Granule- 8. Firon N, LaBonte D, Villordon A, Kfir Y, Solis J, Lapis E, et al. Transcriptional bound starch synthase; SSY: Starch synthase; TMM: Trimmed Mean of M- profiling of sweetpotato (Ipomoea batatas) roots indicates down-regulation values; UGPA: UDP-glucose pyrophosphorylase; WOX4: WUSCHEL-related of lignin biosynthesis and up-regulation of starch biosynthesis at an early homeobox 4 stage of storage root formation. BMC Genomics. 2013;14:460. Eserman et al. BMC Plant Biology (2018) 18:95 Page 11 of 11 9. Artschwager E. On the anatomy of the sweet potato with notes on internal 37. Schmieder R, Edwards R. Fast identification and removal of sequence breakdown. J Agric Res. 1924;27:157–66. contamination from genomic and metagenomic datasets. PLoS One. 10. Lowe SB, Wilson LA. Comparative analysis of tuber development in six 2011;6(3):e17288. sweet potato (Ipomoea batatas (L.) lam.) cultivars. 1. Tuber initiation, tuber 38. Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package for growth and partition of assimilate. Ann Bot. 1974;38:307–17. differential expression analysis of digital gene expression data. Bioinformatics. 2010;26:139–40. 11. Lowe SB, Wilson LA. Comparative analysis of tuber development in six 39. Robinson M, Oshlack A. A scaling normalization method for differential sweet potato (Ipomoea batatas (L.) lam.) cultivars. 2. Interrelationships expression analysis of RNA-seq data. Genome Biol. 2010;11:R25. between tuber shape and yield. Ann Bot. 1974;38:319–26. 40. Emms DM, Kelly S. OrthoFinder: solving fundamental biases in whole 12. Wilson LA, Lowe SB. The anatomy of the root system in west Indian sweet genome comparisons dramatically improves orthogroup inference accuracy. potato (Ipomoea batatas (L.) lam.) cultivars. Ann Bot. 1973;37:633–43. Genome Biol. 2015;16:157. 13. Noh SA, Lee H-S, Huh EJ, Huh GH, Paek K-H, Shin JS, et al. SRD1 is involved 41. Liu K, Warnow TJ, Holder MT, Nelesen SM, Yu J, Stamatakis AP, et al. SATe-II: in the auxin-mediated initial thickening growth of storage root by very fast and accurate simultaneous estimation of multiple sequence enhancing proliferation of metaxylem and cambium cells in sweetpotato alignments and phylogenetic trees. Syst Biol. 2012;61:90–106. (Ipomoea batatas). J Exp Bot. 2010;61:1337–49. 42. Stamatakis A. RAxML version 8: a tool for phylogenetic analysis and post- 14. Ku AT, Huang Y-S, Wang Y-S, Ma D, Yeh K-W. IbMADS1 (Ipomoea batatas analysis of large phylogenies. Bioinformatics. 2014;30:1312–3. MADS-box 1 gene) is involved in tuberous root initiation in sweet potato 43. Bahaji A, Li J, Sánchez-López ÁM, Baroja-Fernández E, Muñoz FJ, Ovecka M, (Ipomoea batatas). Ann Bot. 2008;102:57–67. et al. Starch biosynthesis, its regulation and biotechnological approaches to 15. Noh SA, Lee H-S, Kim Y-S, Paek K-H, Shin JS, Bae JM. Down-regulation of the improve crop yields. Biotechnol Adv. 2014;32:87–106. IbEXP1 gene enhanced storage root development in sweetpotato. J Exp 44. Waramboi JG, Dennien S, Gidley MJ, Sopade PA. Characterisation of Bot. 2013;64:129–42. sweetpotato from Papua New Guinea and Australia: physicochemical, 16. FAO. Food and Agriculture Organization of the United Nations. 2016. www. pasting and gelatinisation properties. Food Chem. 2011;126:1759–70. fao.org. 45. Walter WM, Truong VD, Wiesenborn DP, Carvajal P. Rheological and 17. Hotz C, Loechl C, de Brauw A, Eozenou P, Gilligan D, Moursi M, et al. A physicochemical properties of starches from moist- and dry-type large-scale intervention to introduce orange sweet potato in rural sweetpotatoes. J Agric Food Chem. 2000;48:2937–42. Mozambique increases vitamin a intakes among children and women. Br J 46. Mejia-Aguero LE, Galeno F, Hernandez-Hernandez O, Matehus J, Tovar J. Nutr. 2012;108:163–76. Starch determination, amylose content and susceptibility to in vitro 18. De Souza JG, Viera Da Silva J. Partitioning of carbohydrates in annual and amylolysis in flours from the roots of 25 cassava varieties. J Sci Food Agric. perennial cotton (Gossypium hirsutum L.). J Exp Bot. 1987;38:1211–8. 2012;92:673–8. 19. Vriet C, Smith AM, Wang TL. Root starch reserves are necessary for vigorous 47. Xu X, Pan S, Cheng S, Zhang B, Mu D, Ni P, et al. Genome sequence and re-growth following cutting back in Lotus japonicus. PLoS One. 2014;9:1–7. analysis of the tuber crop potato. Nature. 2011;475:189–95. 20. Bowen BJ, Pate JS. The significance of root starch in post-fire shoot recovery 48. Loretan PA, Bonsi CK, Mortley DG, Wheeler RM, Mackowiak CL, Hill WA, et al. of the resprouter Stirlingia latifolia R. Br. (Proteaceae). Ann Bot. 1993;72:7–16. Effects of several environmental factors on sweetpotato growth. Adv Sp 21. Bell TL, Pate JS, Dixon KW. Relationships between fire response, Res. 1994;14:277–80. morphology, root anatomy and starch distribution in south-west Australian 49. Ingkasuwan P, Netrphan S, Prasitwattanaseree S, Tanticharoen M, Epacridaceae. Ann Bot. 1996;77:357–64. Bhumiratana S, Meechai A, et al. Inferring transcriptional gene regulation 22. Ozias-Akins P, Jarret RL. Nuclear DNA content and ploidy levels in the genus network of starch metabolism in Arabidopsis thaliana leaves using graphical Ipomoea. J Soc Hortic Sci. 1994;119:110–5. Gaussian model. BMC Syst Biol. 2012;6:100. 23. Jones A. Chromosome numbers in Ipomoea and related genera. J Hered. 50. Welch D, Hassan H, Blilou I, Immink R, Heidstra R, Scheres B. Arabidopsis 1968;59:99–102. JACKDAW and MAGPIE zinc finger proteins delimit asymmetric cell division 24. Jones A. Chromosome numbers in the genus Ipomoea. J. Hered. 1964;55:216–9. and stabilize tissue boundaries by restricting SHORT-ROOT action. Genes 25. Nakatani M, Tanaka M, Yoshinaga M. Physiological and Anatomical Dev. 2007;21:2196–204. characterization of a late-storage root-forming mutant of Sweetpotato. J 51. Hassan H, Scheres B, Blilou I. JACKDAW controls epidermal patterning in the Am Soc Hortic Sci. 2002;127:178–83. Arabidopsis root meristem through a non-cell-autonomous mechanism. 26. Jensen WA. Botanical histochemistry: principles and Practice. San Francisco: Development. 2010;137:1523–9. W. H. Freeman; 1962. 52. Tao X, Gu Y-H, Wang H-Y, Zheng W, li X, Zhao C-W, et al. digital gene 27. Turrell FM, Fisher PL. The proximate chemical constituents of Citrus woods, expression analysis based on integrated de novo transcriptome assembly of with special reference to lignin. Plant Physiol. 1942;17:558–81. sweet potato [Ipomoea batatas (L.) lam]. PLoS One. 2012;7:e36234. 28. Herr JM. New uses for calcium chloride solution as a mounting medium. 53. Suer S, Agusti J, Sanchez P, Schwarz M, Greb T. WOX4 imparts auxin Biotech Histochem. 1992;67:9–13. responsiveness to cambium cells in Arabidopsis. Plant Cell. 2011;23:3247–59. 29. Preibisch S, Saalfeld S, Tomancak P. Globally optimal stitching of tiled 3D 54. Etchells JP, Provost CM, Mishra L, Turner SR. WOX4 and WOX14 act microscopic image acquisitions. Bioinformatics. 2009;25:1463–5. downstream of the PXY receptor kinase to regulate plant vascular 30. Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, proliferation independently of any role in vascular organisation. et al. Fiji: an open-source platform for biological-image analysis. Nat Development. 2013;140:2224–34. Methods. 2012;9:676–82. 31. Schindelin J, Rueden CT, Hiner MC, Eliceiri KW. The ImageJ ecosystem: an open platform for biomedical image analysis. Mol Reprod Dev. 2015;82:518–29. 32. Parkhomchuk D, Borodina T, Amstislavskiy V, Banaru M, Hallen L, Krobitsch S, et al. Transcriptome analysis by strand-specific sequencing of complementary DNA. Nucleic Acids Res. 2009;37:e123. 33. Grabherr MG, Haas BJ, Yassour M, Levin JZ, Thompson DA, Amit I, et al. Full- length transcriptome assembly from RNA-Seq data without a reference genome. Nat Biotechnol. 2011;29:644–52. 34. Haas BJ, Papanicolaou A, Yassour M, Grabherr M, Philip D, Bowden J, et al. De novo transcript sequence reconstruction from RNA-Seq: reference generation and analysis with trinity. Nat Protoc. 2013;8:1–43. 35. Li B, Dewey CN. RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinformatics. 2011;12:323. 36. Langmead B, Salzberg SL. Fast gapped-read alignment with bowtie 2. Nat Methods. 2012;9:357–9.
BMC Plant Biology – Springer Journals
Published: May 29, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
All the latest content is available, no embargo periods.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud