Gloriosa superba and Colchicum autumnale multi-tissue transcriptome analysis for colchicine pathway and rhizome development candidate gene identification CURRENT STATUS: POSTED

Background The continued emergence of side-effects caused by synthetic drugs underscores the need for plant-based drugs in human medicine. Medicinal rhizomatous crops are a “goldmine for modern drugs”, and include such species as Gloriosa superba L. and Colchicum autumnale L., the producers of colchicine, a plant-based medicine. The natural isomer of bioactive colchicine is used to effectively treat major diseases such as cancer, cardiovascular disease, and gout. The medicinal properties of colchicine are well characterized, however, almost nothing is known about its biosynthesis. The paucity of information on the colchicine biosynthetic pathway is a significant barrier to biomanufacturing of this biomedicine. A comparative transcriptome study of G. superba and C. autumnale serves as a sequence resource to aid with identification of this biomedicine pathway and rhizome development genes for synthetic biotechnology toolbox, which will enable improved colchicine biomanufacturing. Result Transcriptomes of two colchicine synthesizing monocots G. superba and C. autumnale were interrogated to identify putative cDNAs encoding enzymes and transcription factors involved in the colchicine biosynthetic pathway and rhizome development. Mining of the transcriptomes using Blast2GO led to the identification from G. superba and C. autumnale, respectively, of 20 and 29 candidate colchicine biosynthetic genes N-methyltransferases, 3-O-methyltransferases, cytochrome P450s, a class that could catalyze several steps in the pathway, and N-acetyltransferases. Similarly, 19 and 15 candidate rhizome developmental genes, which belongs to several classes including GIGANTEA, CONSTANS, Phytochrome B, Sucrose Synthase), Flowering Locus T, and REVOLUTA. Likewise, about 16 and 12 transcription factors involved in regulating rhizome development and secondary metabolic pathways in rhizomes such as MADS-box, AP2-EREBP, bHLH, MYB, NAC, and WRKY were also found in G. superba and C. autumnale, respectively. of different groups of secondary metabolites in biorhizome and rhizome developmental genes, which could serve as a comprehensive resource for molecular mechanism research of colchicine biosynthesis in G. superba . This provides a molecular platform and resource for future genetic and functional rhizomatous medicinal crops genomic research.


Abstract Background
The continued emergence of side-effects caused by synthetic drugs underscores the need for plantbased drugs in human medicine. Medicinal rhizomatous crops are a "goldmine for modern drugs", and include such species as Gloriosa superba L. and Colchicum autumnale L., the producers of colchicine, a plant-based medicine. The natural isomer of bioactive colchicine is used to effectively treat major diseases such as cancer, cardiovascular disease, and gout. The medicinal properties of colchicine are well characterized, however, almost nothing is known about its biosynthesis. The paucity of information on the colchicine biosynthetic pathway is a significant barrier to biomanufacturing of this biomedicine. A comparative transcriptome study of G. superba and C. autumnale serves as a sequence resource to aid with identification of this biomedicine pathway and rhizome development genes for synthetic biotechnology toolbox, which will enable improved colchicine biomanufacturing.

Result
Transcriptomes of two colchicine synthesizing monocots G. superba and C. autumnale were interrogated to identify putative cDNAs encoding enzymes and transcription factors involved in the colchicine biosynthetic pathway and rhizome development. Mining of the transcriptomes using Blast2GO led to the identification from G. superba and C. autumnale, respectively, of 20 and 29 candidate colchicine biosynthetic genes N-methyltransferases, 3-O-methyltransferases, cytochrome P450s, a class that could catalyze several steps in the pathway, and N-acetyltransferases. Similarly, 19 and 15 candidate rhizome developmental genes, which belongs to several classes including GIGANTEA, CONSTANS, Phytochrome B, Sucrose Synthase), Flowering Locus T, and REVOLUTA.
Likewise, about 16 and 12 transcription factors involved in regulating rhizome development and secondary metabolic pathways in rhizomes such as MADS-box, AP2-EREBP, bHLH, MYB, NAC, and WRKY were also found in G. superba and C. autumnale, respectively.

Conclusion
The predicted genes in G. superba and C. autumnale encode colchicine pathway enzymes that provide fundamental information for plant-based biomedicine engineering in biorhizomes and microorganisms, a potentially important area of synthetic biotechnology. Additionally, increasing our understanding of rhizome functional genomics will lead to improved colchicine biomanufacturing, and generate important knowledge that can be applied to many other medicinal plant species, allowing for the engineered production of additional biomedicines in medicinal rhizomes.

Background
The growing need for ultrapure plant-based medicines to treat human disease often cannot be met due to a lack of feasible upstream biomanufacturing processes. For example, therapeutic colchicine alkaloid, a drug used to treat cancer, cardiovascular disease, and gout , is uniquely biosynthesized by the Colchicaceae family and extracted commercially from Gloriosa superba and Colchicum autumnale, important rhizome crops and the prime pharmaceutical source of colchicine.
Despite the availability of sequence data for the G. superba and C. autumnale transcriptomes (from the Medicinal Plant Transcriptome https://medplantrnaseq.org/) and their chloroplast genomes, little is known about the colchicine biosynthetic pathway and its regulation in the plant [46]. The efficiency of colchicine biosynthesis likely depends on the interaction of gene circuit elements with other components within the biosynthetic network and how those gene circuits are regulated. This situation with colchicine production is not unusual-as more plant-based medicinally important compounds are discovered, especially where high purity and large amounts are required. To overcome the plantbased therapeutic colchicine production limitation, an advanced non-dormant in vitro biorhizome technology from G. superba has been established [47]. Biorhizomes are non-transgenic cultures that produce important biomedicine, which also serves as asexual reproductive organs, and are an advanced biotechnological platform compared to root and cell cultures due to their continuous and rapid colchicine production [48]. Nevertheless, the biochemical pathways and regulatory networks in the biorhizomes that control colchicine biosynthesis are yet to be characterized, leaving a significant barrier to improving colchicine biomanufacturing. Therefore, the first steps in building a synthetic biology toolbox for colchicine production include analysis of genes from the different Colchicaceae species in order to identify regulatory steps and factors that can then be adjusted to enhance colchicine biomanufacturing in the biorhizomes.
The current understanding of colchicine biosynthesis in planta is based on radiolabeling studies [49-51] and the transformation of O-methylandrocymbine to demecolcine by microsomes prepared from immature C. autumnale seeds [52]. The phenylalanine precursor in the phenylpropanoid pathway and trihydroxylated phenethylisoquinoline in the colchicine pathway have been studied [53-56]. However, research has not been performed at the molecular level to uncover enzymes and regulatory proteins in the colchicine biosynthetic pathway in G. superba or C. autumnale. To fully understand how colchicine is biosynthesized in Gloriosa biorhizomes, the key genes and enzymes that control the colchicine pathway from the alkaloidal precursor trihydroxylated phenethylisoquinoline to colchicine must be identified. To this end, we constructed a full-length cDNA library using mRNA isolated from G. superba leaves that consists of 2,790 processed sequences, of which 1,379 were assembled into 292 contigs and 1,411 singletons [47]. The cDNA library contains gene families expected to be involved in the colchicine pathway, including NMT, 3-OMT, cytochrome P450s: CYP96T1, CYP82E10, and NAT.Furthermore, we manually curated these gene families and identified specific genes whose corresponding enzymes are excellent candidates for involvement in the colchicine pathway, from the alkaloid formation steps to later steps in the pathway (Figure 1), including: 1) an NMT enzyme that catalyzes the conversion of the trihydroxylated phenethylisoquinoline intermediate to (S)autumnaline; 2) a P450: CYP96T1 that catalyzes (S)-autumnaline to isoandrocymbine; 3) an OMT enzyme that catalyzes the transformation of isoandrocymbine to 3-O-methylandrocymbine; 4) a P450 that catalyzes O-methylandrocymbine to demecolcine; 5) an additional P450CYP82E10 that catalyzes the conversion of demecolcine to deacetylcolchicine; and 6) a NAT enzyme that catalyzes the transfer of an acetyl group from acetyl-CoA to a deacetylcolchicine nitrogen group to yield colchicine [47].
Although significant progress has been made in understanding rhizome-specific functions in plants, the mechanisms underlying the regulation of Gloriosa biorhizome growth, and the dormancy of fieldgrown G. superba and C. autumnale are not yet known [57][58][59]. Similar to many other rhizomatous species, natural G. superba and C. autumnale rhizomes undergo a dormancy period in their normal growth cycle, but the biorhizomes do not go dormant. Extensive studies examined the phenotypic variation between plant species, but why dormancy-free biorhizomes from different Gloriosa species produce different levels of colchicine in the controlled bioreactor environment remains unclear. The function of dormancy-associated genes (such as specific transcription factors) in biorhizomes and the core regulatory machinery that controls differential colchicine biosynthesis between species are not known. We hypothesize that G. superba and C. autumnale genes and gene networks are comparable, but that subtle differences in their regulation lead to changes in colchicine accumulation between the species. Comparison of the transcriptomes of these species will aid in filling the identified knowledge gaps.

Results
Benchmarking universal single-copy orthologue (BUSCO) analysis: Three in vivo tissues (leaf, fruit, and rhizome) were previously used to generate a combined RNA-seq dataset of G. superba and C. autumnale (https://medplantrnaseq.org/). This investigation of the dataset indicated that the N50 for the assembly was fairly long at 2,134, given that contigs of ≥ 100 (instead of ≥ 200) were included in the assembly and that the data were generated from 50 bp single-end reads. The average contig length was somewhat short, however, as the statistic was skewed due to the inclusion of several contigs that were shorter than 200. BUSCO analysis, which in this case examined the core eukaryotic genes in plants, indicated that the dataset was ~89% complete (64% of the genes detected were found as a single sequence in the assembly, and 25% had duplicated sequences). In comparison, a total transcriptome analysis that we conducted in a different sepcies and that focused on only a single sample type (containing combined RNA samples from the 1 st and 2 nd Asian Citrus Psyllid instars, including six biological replicates and utilizing 150 bp reads), led to BUSCO analysis results suggesting that the dataset was 96% complete (13% single sequence and 83% duplicate sequences), with only 1.4% fragments of core genes and 2.6% missing core genes. Thus, although the sequence data used in this analysis were of high quality, it is likely that some important genes may nevertheless still be missing from the dataset. suggesting that many of the transcripts belong to non-canonical pathways (e.g., are not involved in known signaling pathways or biosynthetic pathways). Many of these reads were characterized as "unknown", as is still typical in transcriptomic analysis of non-model plant species [60][61]. G. superba cDNA library was also annotated by BlastKOALA, which annotated 66 of 848 (7.8%) sequences sets of proteins based on their role in specific pathways ( Figure 4).
The mining of G. superba and C. autumnale transcriptomes revealed a total of 1299 genes that were possible initial candidates for involvement in the colchicine pathway, which included 647 sequences from G. superba and 652 from C. autumnale. From 647 candidate sequences in G. superba, 186 coded for NMTs, 105 for OMTs, 19 for NATs, and 337 for P450s; while from 652 candidate sequences in C. autumnale, 16 were putative NMTs, 106 OMTs, 20 NATs, and 510 P450s ( Figure 5). Further, these transcripts were narrowed down to candidates for the genes that catalyze the reactions in the colchicine biosynthetic pathway. Moreover, a total of 339 putative rhizome developmental genes identified based on comparison to known rhizome developmental genes (119 sequences) from Nelumbo nucifera and tuber developmental genes (142 sequences) from Solanum tuberosum such as GI, CO, PHYB, SuSy, FT, and REV were also identified in G. superba and C. autumnale transcriptomes ( Figure 6). Notably, 113 sequences from G. superba and226 from C. autumnale. In addition, a total of 1146 transcription factor sequences associated with rhizome development namely MADS-box, AP2-EREBP, bHLH, MYB, NAC, and WRKY were analyzed ( Figure 7). This included 481 sequences from G. superba and 665 from C. autumnale. Phylogenetic trees of colchicine pathway, rhizome development, and transcription factors (with a total of 65, 58, and 59, respectively) of full-length amino-acid candidate and reference genes were constructed using the PHYLIP-3.697 software with the objective to examine the genetic divergence between two colchicine producing plants, G. superba and C.
autumnale, and their reference sequences.

Discussion
What genes are expressed in rhizomes? Several recent investigations have identified rhizome-specific genes in different rhizomatous plants by directly comparing leaves, other tissues, and rhizomes, leading to the detection of genetic mechanisms responsible for controlling rhizome development, growth, and metabolism. Many genes exhibit significantly altered expression during rhizome development but genes associated with auxin hormone signaling appear to trigger rhizome induction [57-59]. The rhizome developmental gene REV is highly expressed in bamboo rhizome buds and plays an important role in meristem initiation [62]. In potato plants, calmodulin-binding protein plays a regulatory role in signal transduction for tuber formation [63]. In addition, the FT, CO, and GI genes are involved in the transduction of photoperiodic signals, which may promote rhizome budding in potatoes [64][65]. Tubers are not rhizomes, but there are some similarities to their growth that may include involvement of analogous genes. There are 14 other important rhizome formation-related genes, including a MADS-box that could be involved in rhizome enlargement [66]. Genes encoding PHYB, CO, GI, FT and SuSy were identified in Lotus rhizomes, but their expression and regulation differed in the shoot and rhizome [60]. Transcription factor families such as AP2-EREBP, bHLH, MYB, NAC, and WRKY were reported to be important in regulating specialized metabolic pathways in rhizomes [61]. Bioactive small molecules such as curcuminoid and candidate genes for gingerol synthesis are also highly expressed in rhizomes [67][68][69][70][71]. Notably, biosynthetic genes involved in benzylisoquinoline alkaloid formation were highly upregulated during bulb development in Corydalis yanhusuo [72]. In wild rice, microRNAs were differentially expressed in aerial shoots and rhizomes [73]. However, the exact roles that the corresponding genes might play in biorhizomes are not known.
This information provides a framework for our analysis of biorhizome-expressed genes and the identification of specific genes involved in biorhizome growth, resistance to dormancy and production of colchicine in this interesting biotechnology platform.
Why are colchicine pathway candidate NMT, P450s (CYP96T1 and CYP82E10), 3-OMT, and NAT genes from the transcriptomes the best targets? First, a cDNA contig consisting of three sequences showed significant sequence similarity to Ricinus communis NAT and partial similarity to an analogous gene from Oryza sativa. This contig, which contains a full-length transcript (cDNA Gloriosa 148), is a potential candidate for encoding the enzyme that catalyzes the acetyl transfer from acetyl-CoA to deacetylcolchicine to form colchicine, the final step in the proposed colchicine pathway [52]. The putative NAT gene also showed 100% homology to a full-length transcript in the G. superba transcriptome (Gloriosa-20120814|26859). Similarly, two NAT catalyzed reactions were identified in melatonin biosynthesis in O. sativa [74][75], carried out by SNAT1 and SNAT2 enzymes. The SNAT1 transcript Q5KQI6.1 is a possible alternative reference sequence, with an identity of 69%, and 77%, respectively, to transcripts in the G. superba and C. autumnale transcriptomes (Gloriosa-20120814|70209_1 and Colchicum_20101112|6813; see Table 1 and Figure 8).
Second, the cDNA clone (Gloriosa 14D06) identified as encoding a putative NMT is a partial clone consisting mainly of the 3´ end of the cDNA (Table 1 and Figure 8). The putative NMT clone had high sequence homology to a Coptis japonica NMT, which catalyzes a similar N-methylation reaction involved in (S)-N-methylcoclaurine formation in the benzylisoquinoline pathway and has a near identical full-length transcript in the G. superba transcriptome (Gloriosa-20120814_33123). The putative NMT is a one candidate for the N-methylation step that converts the trihydroxylated phenethylisoquinoline intermediate to (S) autumnaline, the first alkaloidal precursor formation step in the proposed colchicine pathway [52,76]. Other candidates include, the G. superba putative NMT transcript Gloriosa-20120814|64082_1, which was found to be 58% identical to C. japonica coclaurine NMT (BAB74802.1), and Colchicum_20101112|85, a transcript with 48.7% identity to pavine NMT (PNMT, sp|C3SBW0.1) from Thalictrum flavum (PNMT) converts (S)-tetrahydropapaverine to (S)laudanosine in the benzylisoquinoline alkaloid pathway that is common in many plants [77]. PNMT is also considered a possible alternative reference sequence due to its capability of adding a methyl group to (S)-tetrahydropapaverine where the nitrogen is present, which is similar to the colchicine pathway mechanism. These putative transcripts also share homology with reticuline NMT involved in biosynthesis of the aporphine alkaloid magnoflorine in opium poppy roots [78].
Third, the cDNA clone (Gloriosa 8E03) identified as a putative P450 is a partial clone with 87% sequence homology to Narcissus pseudonarcissus CYP96T1, which catalyzes a C-C phenol coupling reaction in noroxomaritidine biosynthesis in the haemanthamine pathway and has an identical fulllength transcript in the C. autumnale transcriptome (Colchicum-20101112_3005). This putative enzyme might be a candidate for the para-para phenol-tropolone oxidative coupling bridge-forming P450 CYP96T1, that is NADPH and O 2 -dependent and converts (S)-autumnaline to isoandrocymbine [79][80]. In Papaver somniferum, salutaridine synthase enzyme (CYP719B1)is responsible for the C-C phenol coupling converting (R)-reticuline to salutaridine by connecting the 12 and 13 carbon [81][82].
P. somniferum CYP719B1 enzyme could also be considered as an alternative reference gene because of its C-C phenol-coupling mechanisms in the morphine pathway. Furthermore, in C. japonica, CYP80G2 has been shown to convert (S)-reticuline to (S)-corytuberine through C-C phenol-coupling in an isoquinoline alkaloid pathway [83]. The G. superba and C. autumnale transcriptomes contained transcripts (Gloriosa-20120814|30999_1 and Colchicum_20101112|76071) that shared of 52.7% and 51.7% identity, respectively, with CYP80G2 (sp|A8CDR5.1; see Table 1 and Figure 8). Fifth, the cDNA clone (Gloriosa 1F02) identified as another putative P450 is a partial clone with 79% sequence homology to CYP82E10 from Fragaria x ananassa and Nicotiana tabacum, which catalyzes an N-demethylation reaction from nicotine to nornicotine and has a full-length homolog in the C.
CYP82E4 is the major nicotine demethylase enzyme responsible for converting nicotine to nornicotine [89]. The demethylase P450 enzymes CYP82E5v2and CYP82E10 were expressed in nonsenescent green leaves and/or root tissue in N. tabacum [87,[89][90][91][92][93]. The transcripts, Gloriosa-20120814|5091 and Colchicum_20101112|74349 were found to possess 53.4%, and 54% identity, respectively, to NP_001312976.1 (CYP82E4; see Table 1 and Figure 9). Thus, enzymes belonging to the predicated NMT, 3-OMT, CYP96T1, CYP82E10, and NAT families are likely to be early targets for cloning and characterization in the colchicine pathway. However, the possibility exists that one or more of the candidate genes will not function as predicted. In this case, other genes from the G. superba and C.
autumnale transcriptomes that show homology to colchicine pathway enzymes would also be considered as candidate genes.
What genes are candidate rhizome developmental genes from G. superba and C. autumnale?
Underground rhizome (storage organ) development and the molecular mechanisms controlling rhizome enlargement and dynamics remain unclear, even in well-studied medicinal and invasive plants [60, [68][69]. Biorhizomes can be a suitable system to study rhizome growth and biomedicine biomanufacturing because they are much more accessible than plant tissues growing in the soil. In addition, it is useful to identify genes related to rhizome development as they likely control key resources involved in biomass yield. Several genes have been identified from other species that affect rhizome formation in different rhizomatous crops, but comparable genes have not been characterized in the Colchicaceae. Therefore, characterizing rhizome developmental genes from G. superba and C.
autumnale could allow clarification of molecular intricacies involved in biorhizome biomass production and lead to enhanced biomanufacturing. To screen for rhizome developmental genes in colchicine producing species, the transcriptomes from G. superba and C. autumnale were analyzed as described above for metabolism related genes. Specific results related to major development-related genes are outlined below.
The photoreceptor PHYB is involved tuber induction and microRNA, miR172 highly expressed in potato tuber [100]. It has been shown in N. nucifera that PHYB and/or other phytochromes might measure the length of the light period to affect rhizome girth enlargement [61]. The PHYB candidate gene Gloriosa-20120814|36050_1 and Colchicum_20101112|1221 shared an identity of 81% with the reference sequence from N. nucifera (gi|720038316|ref|XP_010267948.1|). The PHYB reference sequence from S. tuberosum (PGSC0003DMT400061712) shared an identity of 81% with candidate sequence Gloriosa-20120814|85746_1) and 80% with Colchicum_20101112|1221 (Table 2 and Figure   9).
Phylogenetic analysis of colchicine pathway, rhizome developmental, and transcription factor protein in G. superba and C. autumnale: A total of 20 and 29 candidate colchicine pathway enzymes in G. superba and C. autumnale transcriptomes, respectively, were split into four primary clusters and mapped against 4 cDNA and 12 reference enzymes (Table 1 and Figure 10A). Among these, step 1 NMT contained 7 genes in the clade such as 3 from G. superba transcriptome, 1 from G. superba cDNA, and 1from C. autumnale transcriptome. Both step 2 and step 4 P450s genes were closely related and aligned in the same clade but step 2 P450 genes was nested within step 4. Within the Additionally, rhizome developmental gene identification of potential orthologous between G. superba and C. autumnale is one of the most important bottlenecks in transcriptomics [120]. Notably, rhizome developmental genes such as GI, CO, FT, PHYB, SuSy, and REV have been mapped in G. superba and C. autumnale transcriptomes using reference genes from S. tuberosum, N. nucifera, A. thaliana, and P. praecox. A total of 19 G. superba and 15 C. autumnale candidate genes were mapped against 24 reference genes to identify possible rhizome developmental unigenes ( Figure 10B). Furthermore, rhizome developmental phylogenetic analysis shows 6 distinct clusters, among them CO represents the largest clade. In addition, the transcription factor plays a crucial role in rhizome development [121]. A total of 16 and 12 transcription factors were mapped in the G. superba and C. autumnale transcriptomes, respectively ( Figure 10C). To better reveal the genes associated with rhizome development, 6 transcription factor families were analyzed such as MADS-box, AP2-EREBP, bHLH, MYB, NAC, and WRKY that showed good correlation between the clusters of protein families. The mapped rhizome developmental gene families explains the conserved sequence homology between reference and candidate genes, which indicates a network of genetic mechanisms in G. superba and C. autumnale are crucial for rhizome development.
Interestingly, the phylogenetic trees of colchicine pathway, rhizome developmental, and transcription factors indicate that the candidate genes of G. superba and C. autumnale were common to Colchicaceae due to high homology and evolution similarity of the candidate genes. This suggests that the predicted candidate genes not only are involved in colchicine biosynthesis in rhizomes but are also involved in the rhizome metabolism. In comparison with tuber producing S. tuberosum and N. nucifera, rhizome developmental and transcription factor genes from G. superba and C. autumnale showed nearly the same transcriptional regulation map that are possible downstream targets of rhizome initiation or biomass production. The annotated genes from G. superba and C. autumnale can contributes to identifying candidate genes in the biosynthesis of different groups of secondary metabolites in biorhizome and rhizome developmental genes, which could serve as a comprehensive resource for molecular mechanism research of colchicine biosynthesis in G. superba. This provides a molecular platform and resource for future genetic and functional rhizomatous medicinal crops genomic research.

Conclusions
In this study, we have interrogated two evolutionarily diverse transcriptomes of G. superba and C.
autumnale to predict genes that encode candidate proteins of colchicine biosynthesis and rhizome metabolism. G. superba and C. autumnale are rhizomatous medicinal plants that lack reference genomes. Collectively, transcriptomes and cDNA approaches were applied to select candidate genes for each predicted colchicine pathway step, which should help to elucidate the colchicine biosynthetic pathway. Additionally, our work will be useful to identifying rhizome developmental genes and transcription factors in G. superba and C. autumnale, with the ultimate goal of improving biorhizome biomass. The predicted genes in this work now need to be functionally validated, and are an important resource for metabolic engineering or synthetic biotechnology that could improve colchicine biomanufacturing.

Methods cDNA library
A full-length cDNA library was constructed from a month-old G. superba biorhizome derived leaves.
Total RNA isolation and cDNA library construction were performed according to previous protocols [122]. The assembled cDNA sequences were BLASTed against the National Center for Biotechnology Information (NCBI) database (http://www.ncbi.nlm.nih.gov/), and G. superba and C. autumnale transcriptomes.

Prediction of colchicine pathway and rhizome developmental proteins
The G. superba and C. autumnale transcriptomes were obtained from our Medicinal Plant Transcriptome database (https://medplantrnaseq.org/). BUSCO analysis of transcriptomes were processed according to Waterhouse et al. [123]. Colchicine pathway and rhizome developmental along with its transcription factors encoding proteins were predicted from the NCBI database.
Functional annotation of transcriptomes, colchicine pathway and rhizome developmental proteins The G. superba and C. autumnale transcriptomes GO classification of the identified know plant proteins was performed using the web-accessible Blast2GO v5 annotation system (https://www.blast2go.com/) [124]. Blast2GO is an all-in-one bioinformatics software for protein functional prediction and the genome-wide analysis of annotation data. KEGG pathway enrichment analysis was used to analyze the functional significance of biochemical pathways using BlastKOALA and GhostKOALA. The first step in Blast2GO is to align cDNA nucleotide sequences against the NCBI non-redundant database by Basic Local Alignment Search Tool protein (BLASTp/BLASTn) with an expectation value of 1e − 5 . Next, transcriptomes FASTA protein sequences were uploaded to Blast2GO for BLAST analysis to identify homologous sequences, then mapping and annotation were performed.