Introduction

Mikania micrantha Kunth (“mile-a-minute” weed) is an extremely fast-growing, sprawling, perennial vine belonging to the family Asteraceae and native to tropical America1. The vine is listed as one of the top 100 worst invasive species by the International Union for Conservation of Nature (IUCN)2. It is problematic in tropical Asia, in parts of Papua New Guinea, on Indian Ocean islands and Pacific Ocean islands and in Florida in the United States3,4, causing serious damage to the natural ecosystems1,5. The invasion of M. micrantha has caused economic losses related to forests and crop production of up to US$4000 ha−1 6,7 by climbing plants to the canopy, and blocking the sunlight and has led to a loss of genetic and species diversity, a decline in soil and food web stability, and altered nutrient cycling4,8,9,10,11.

M. micrantha possesses a variety of biological characteristics associated with successful invasive plant species. It germinates early in the growing season; grows extremely fast (with a maximum mean growth rate of 20 cm day−1)12; roots from each vine node; has a smothering habit; produces very large numbers of widely dispersed seeds (170,000 m−2)13, which are very small (8.92 × 10−5 g 1000 grains−1)14,15; and has an ability to survive harsh conditions15. M. micrantha also exhibits a high degree of morphological and physiological plasticity in response to different light environments16. In addition, M. micrantha has strong allelopathic effects on other plants and soil microbes10,17,18. Allelopathy is the influence of a plant over a target species, through the release into the environment of compounds that influence the growth and development of biological systems19. Various allelochemicals, that are produced by a plant must escape into the environment and subsequently influence the growth and development of other plants, such as Mikania sesquiterpene lactones (STLs)20,21,22 and phenolic compounds23 have been isolated from different M. micrantha plant tissues. However, the molecular mechanism underlying the fast growth of M. micrantha and the biosynthetic pathways of its characteristic allelochemicals are not yet clear.

Asteraceae is not only one of the largest, most valuable plant families, but also contains the largest number of invasive alien species worldwide. Genomes of some important species of Asteraceae, including lettuce24, sunflower25, artichoke26, and horseweed27, have been sequenced, deepening our understanding of Asteraceae evolution and promoting biological studies of plants in Asteraceae. Although some studies have examined the biological characteristics of M. micrantha, genomic and molecular data are scarce, which limits in-depth studies of this invasive species. Here, we generate a reference genome of M. micrantha and perform analyses of transcriptomic, metabolomics, and metagenomic data to explore possible reasons causing its rapid growth.

Results

Chromosome-level genome assembly and recent LTR-RT expansion

We generated 228 Gb (122 × coverage) of single molecule real-time (SMRT) sequences on the PacBio RS II and Sequel platform (Supplementary Table 1); these sequences were used to assemble a haploid reference genome by Canu-1.6, followed by filtering of the extra heterozygous fragments (Supplementary Fig. 1 and Supplementary Note 1). The assembled genome includes 4414 contigs with a total length of 1.79 Gb, a contig N50 of 1352 kb, and a contig N90 of 180 Kb (Table 1 and Supplementary Table 2). Based on the distribution of k-mer frequencies, the estimated genome size of M. micrantha is 1.86 Gb (Supplementary Fig. 2); thus, 96% of the M. micrantha genome has been successfully assembled. With Hi-C technology, 1.61 Gb (90.3%) of contigs was anchored and oriented into the 19 linkage groups (Fig. 1a and Supplementary Table 2), with the longest being 124 Mb and the shortest being 47 Mb (Table 1).

Table 1 Statistics of the Mikaina micrantha genome and gene prediction.
Fig. 1: Landscape of the Mikania micrantha genome.
figure 1

a Integration of genomic and expression data. (a) Distribution of GC content; (b) distribution of the Gypsy family of long terminal repeats retrotransposons (LTR-RTs); (c) distribution of the Copia family of LTR-RTs; (d) distribution of coding genes; (e–h) expression of organ-specific genes (from outside to inside tracks: root, stem, leaf and flower). b Comparison of repetitive sequences in asterids, including Mikania micrantha, sunflower (Helianthus annuus), artichoke thistle (Cynara cardunculus), Artemisia annua, chrysanthemum (Dendranthema morifolium), lettuce (Lactuca sativa), and coffee (Coffea canephora). c The insertion time distribution of intact LTRs in the M. micrantha genome. Mya indicates million years ago.

The protein-coding genes in the reference genome were predicted by the EVidenceModeler pipeline (Supplementary Table 3). In total, 46,351 gene models were predicted in the assembled genome as the reference gene set, with coding regions spanning ~57.6 Mb (3.2%) of the genome (Table 1 and Supplementary Table 4). In the evaluation of completeness, 91% of eukaryote core genes from OrthoDB (http://www.orthodb.org) were identified as complete in the reference gene set by BUSCO (Table 1), comparable to the number for other published genomes (Supplementary Fig. 3). A total of 42,804 (92%) coding proteins were annotated by functional databases (Supplementary Table 5), including eggNOG, KEGG, Interpro, and uniprot. In addition, we identified 3633 long noncoding RNA (lncRNAs) classified into three subclasses: 2489 intergenic lncRNAs (lincRNAs), 551 intronic lncRNAs (ilncRNAs), and 593 antisense lncRNAs (NAT-lncRNAs) (Supplementary Figs. 46).

The high coverage and continuity of the M. micrantha genome enabled a comprehensive analysis of transposable elements (TEs). In total, more than three quarters of the M. micrantha genome consists of TEs (Fig. 1b and Supplementary Table 6), 60% of which are long terminal repeat retrotransposons (LTR-RTs). Notably, the most abundant LTR-RTs family present in M. micrantha was Gypsy, accounting for 70.1% of all LTR elements, followed by Copia (26.1%) (Supplementary Table 7). LTR-RTs are the most abundant DNA components in all investigated plant species and are largely responsible for plant genome size variation. Interestingly, over 80% of the LTR-RT lineages in M. micrantha are very young, exhibiting minimal sequence divergence compared to those of other plant species, owing to significant expansion in the past 1 million years (Fig. 1c), which is consistent with previous reports in sunflower25.

Genome-wide duplication and subsequent fragment duplication

To assess the paleohistory of the asterids, we performed a comparative genomic investigation of M. micrantha with sunflower (Helianthus annuus), lettuce (Lactuca sativa), artichoke thistle (Cynara cardunculus), chrysanthemum (Dendranthema morifolium), Artemisia annua and coffee (Coffea canephora) with grape (Vitis vinifera) as an outgroup, showing that the M. micrantha diverged from grape ~109 million years ago (Mya), from artichoke ~51 Mya, and from lettuce ~43 Mya. Notably, M. micrantha and sunflower were diverged ~31 Mya (Fig. 2a).

Fig. 2: Genome evolution of Mikania micrantha.
figure 2

a The paleohistory of the asterids, including M. micrantha, sunflower, lettuce, artichoke, chrysanthemum, Artemisia annua, and coffee, with grape as an outgroup. A whole-genome duplication event (WGD-2) occurring before M. micrantha-sunflower divergence (approximately 32–36 Mya), and the shared ancestral whole genome triplications WGT-γ (approximately 122–164 Mya) and WGT-1 (approximately 53–62 Mya). Circles indicate the ages of WGT-1 (blue), WGD-2 (gray) and SD-3 (red). b Ks distribution of syntenic orthologues from M. micrantha, sunflower and artichoke. The y-axis shows the ratio of gene pairs in the syntenic block. The polyploidization (WGT-1, WGD-2, and SD-3) and speciation (M. micrantha-sunflower) events are referenced on the peaks. c Dot plots of syntenic orthologues in two chromosomes between M. micrantha and sunflower. d The functional enrichment of WGT-1, WGD-2, and SD-3 duplicated genes in M. micrantha by Gene Ontology (GO) classification. The significantly enriched GO terms for the SD-3 and WGT-1/WGT2 duplicated genes are marked by red and blue, respectively.

To study the conservation of genomic structure, we identified the interspecies syntenic genes between M. micrantha and sunflower, artichoke, and grape, and intraspecies syntenic genes within the M. micrantha, sunflower, lettuce, artichoke, and coffee genomes, then calculated the Ks values of the syntenic fragments for orthologous pairs (Fig. 2b, c and Supplementary Figs. 7 and 8). These results, together with the results of previous reports25, showed that the M. micrantha genome has a complex paleo-polyploidy history, including the ancient whole-genome triplication WGT-γ (approximately 122–164 Mya) in dicotyledons, the whole-genome triplication WGT-1 (approximately 53–62 Mya) shared with asterids II, and the whole-genome duplication WGD-2 (approximately 32–36 Mya) shared with sunflower. In contrast to the genomes of other species in Asteraceae, the M. micrantha genome exhibits a large number of recent segmental duplications. With Ks values lower than 0.2, we identified 127 syntenic blocks ranging from 46 to 8955 kb in length and containing 23% of all syntenic genes, indicating that the M. micrantha genome experienced a recent explosion of lineage-specific segmental duplications (SD-3, less than 1 Mya; Fig. 2b).

Large-scale genome duplications produce abundant duplicated genes, which are important for the diversity of gene functions and adaptation to changing environments. The WGT-1, WGD-2, and SD-3 duplicated genes were significantly enriched in the functions of catalytic and hydrolase activity, nitrogen compound metabolism, transcription factor, response to stimulus and chemical (Fig. 2d and Supplementary Figs. 9 and 10). It demonstrated that extensive gene fractionation has occurred during the evolutionary history of the M. micrantha genome, which retained the genes that are essential for survival and lost other redundant genes. Next, gene families related to fast growth were analyzed. A total of 189 photosynthesis related genes were identified in the M. micrantha genome, and the number of genes related to photoreaction was significantly higher than that in other Asteraceae species (p < 0.01, Z-test; Supplementary Fig. 11). The large-scale duplication prompted photosynthesis gene expansion, with 5% of these genes produced after the WGT-1 and WGD-2 events and 21% of these genes produced after recent SD-3 events. In addition, auxin signaling and jasmonic acid (JA) biosynthesis genes were significantly expanded in the M. micrantha genome, compared with the genomes of other Asteraceae species (p < 0.01, Z-test; Supplementary Fig. 12a, b). The duplicated genes, which were enriched in categories of genes encoding interacting products, photosynthesis, hormone, and response to stress, play an important role in the environmental adaptability of plants.

High photosynthesis of the leaves benefits rapid growth

In this study, the diurnal carbon isotope (δ13C) changes in M. micrantha leaves were not significantly different between day and night based on analysis of the carbon isotope ratio (Supplementary Fig. 13), indicating that M. micrantha has a typical C3 carbon assimilation type, consistent with the findings of previous reports1. However, the net photosynthetic rate of leaves was higher than that in other C3 plants and similar to that in C4 plants1. Previous studies revealed that all plants contain the genes necessary for Crassulacean acid metabolism (CAM) photosynthesis, but the genes regulating the CAM pathway are often silent or inactivated in non-CAM plants28. However, the expression of genes related to CAM photosynthesis is activated in response to changes in the external environment or in certain growth stages29,30. To adapt to environmental changes, a few plants have evolved to use both the C3 and CAM photosynthesis pathways, with interconversion between growth stages31.

Here, we identified genes related to the CAM pathway based on homology to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. The M. micrantha genome contains 51 candidate genes involved in the carbon fixation module of CAM, including carbonic anhydrase (CA), phosphoenolpyruvate carboxylase (PEPC), phosphoenolpyruvate carboxykinase (PEPCK), malic enzyme (ME), malate dehydrogenase (MDH), and pyruvate orthophosphate dikinase (PPDK) (Fig. 3a). The gene copy number of the CAM pathway in M. micrantha was 1.7 and 1.4 times greater than that in pineapple (a typical CAM plant) and sunflower (a C3 plant), respectively. In addition, the activity of PEPC in the leaves of M. micrantha (2.8 μmol mg−1 protein) was higher than that in pineapple leaves (2.0 μmol mg−1 protein).

Fig. 3: The special photosynthetic system in Mikania micrantha.
figure 3

a The number of putative carbon fixation genes in M. micrantha, pineapple and sunflower. The gene number of pineapple was obtained from the literature28; carbonic anhydrase (CA), phosphoenolpyruvate carboxylase (PEPC), phosphoenolpyruvate carboxykinase (PEPCK), malic enzyme (ME), malate dehydrogenase (MDH), and pyruvate orthophosphate dikinase (PPDK). b Expression pattern of CAM pathway key genes in M. micrantha across the diurnal expression data. The R-package (pheatmap)-transformed transcripts per kilobase million (TMP) expression profiles are shown. The gene names of M. micrantha are shown on the right. The gene expression of each replicate sample in two time points (9 A.M. and 9 P.M.) presented in the heatmap. c The diurnal variation pattern of key enzyme activities and malic acid content in M. micrantha leaves at 9:00 A.M. (day) and 9:00 P.M. (night). The y-axis on the left shows the activity of PEPC and NADP-linked ME. Statistical significance (n = 3) determined using the two-sided Student’s t test. On each box plot, the central mark indicates the median, the bottom and top edges of the box indicate the interquartile range (IQR) and the whiskers represent the maximum and minimum data points. d Scanning electron microscopy (SEM) of stomata in M. micrantha leaves during the day and at night. SEM images of the samples were taken using a scanning electron microscope (Q25, FEI, USA). e Schematic of the CAM and C3 photosynthesis proposed pathways of M. micrantha. The enzymes in the CAM pathway of M. micrantha are shown in red. During the day, carbon dioxide is absorbed directly by the leaves through the C3 pathway. At night, carbon dioxide is fixed nocturnally by PEPC and stored transiently as malic acid in large vacuoles. Furthermore, carbon dioxide is released from organic acids through NADP-linked ME and used directly in the Calvin cycle in the daytime. The source data underlying Fig. 3b–d are provided as a Source Data file.

To investigate the enzyme activity patterns of the CAM pathway in day and night, we collected RNA-seq samples at 9:00 A.M. (day) and 9:00 P.M. (night) from leaf tissues of potted M. micrantha (see Supplementary Note 2). On the basis of contrasting expression patterns in the leaf tissues between day and night, we identified the diurnal variation of M. micrantha genes involved in carbon fixation (Fig. 3b). The eight key genes of CAM pathway (encoding three copies of CA, one copy of PEPC and two copies of NADP-linked ME) had a diurnal expression pattern in the leaf tissue (Fig. 3b). In addition, the enzyme activity of M. micrantha PEPC in the leaf tissue was higher in the night than that during the day (Fig. 3c). Carbon dioxide was fixed nocturnally by PEPC and stored transiently as malic acid in large vacuoles; consequently the malic acid and citric acid content in the evening was nearly 10 times and two times as high as that in the daytime, respectively (Fig. 3c and Supplementary Fig. 14a). The stored malic acid was decarboxylated, and the carbon dioxide was released by NADP-linked ME during the daytime; thus, the activity of NADP-linked ME showed a pattern in which the activity in the daytime was significantly higher than that at night (p = 0.027, two-sided Student’s t test, Fig. 3c).

In addition, the stomata of C3 and C4 plants open to absorb carbon dioxide during the day, while they close at night, except for in CAM plants, whose stomata exhibit a unique pattern of diurnal variation. Surprisingly, the stomata of M. micrantha leaves were not only always open during the day (Fig. 3d and Supplementary Fig. 14b, c), but also partly open at night (Fig. 3d and Supplementary Fig. 14b, c), implying that the CAM pathway was involved in the carbon dioxide fixation. Together, these results indicate that one of the reasons for the higher photosynthetic capacity of M. micrantha leaves is that CO2 may be absorbed at night by photosynthetic pathway similar to CAM and stored in vacuoles as organic acids to supplement the carbon fixation capacity during the day (Fig. 3e).

High photosynthesis of the stem enhances rapid growth

In most plants, the leaves are the major photosynthetic organs; however, some studies have reported that many nonleaf green organs or tissues also have photosynthetic activity32. To investigate the photosynthetic ability of stems in M. micrantha, we measured their net photosynthetic rate (Pn), respiration and gross photosynthetic rate and compared these values with those of five other species. The Pn of the M. micrantha stem was approximately 1.5 μmol CO2 m−2 s−1, and was significantly higher than that in Paederia scandens, Pueraria lobata, Stephania longa, Merremia hederacea, and Ipomoea cairica (p < 0.05, Duncan-test; Fig. 4a), which was close to or less than 0 μmol CO2 m−2 s−1. The gross photosynthetic rate of the M. micrantha stem showed a similar pattern (p < 0.05, Duncan-test; Supplementary Fig. 15).

Fig. 4: Identification of stem photosynthesis in Mikania micrantha.
figure 4

a Comparison of the net photosynthetic rate of M. micrantha to that of five neighboring species, namely, Paederia scandens, Pueraria lobata, Stephania longa, Merremia hederacea, and Ipomoea cairica. Statistical significance (n = 3) determined using one-way ANOVA with Duncan’s multiple comparison test. b The change in chlorophyll content in the stem under defoliation conditions compared to that in the control. Statistical significance (n = 3) determined using the two-sided Student’s t test. On each box plot, the central mark indicates the median, the bottom and top edges of the box indicate the interquartile range (IQR) and the whiskers represent the maximum and minimum data points. c KEGG pathway enrichment of upregulated genes. The significant enrichment of KEGG pathways related to photosynthesis, carbon fixation and hormone signal transduction are marked in red. d Comparison of key genes related to the Calvin cycle in the M. micrantha stem between defoliation conditions and the control. The significantly upregulated genes are marked in red. e Expression pattern of differentially expressed genes involved in auxin, cytokinin, gibberellin, and chlorophyll biosynthesis and plant hormone signal transduction. The gene names of M. micrantha are shown on the right. Error bars indicate average value ± s.e.m. of indicated replicates. The source data underlying Fig. 4a, b, e are provided as a Source Data file.

To confirm that the photosynthesis of the M. micrantha stem could maintain plant growth, we conducted a defoliation experiment with both M. micrantha and its neighboring species P. scandens (Supplementary Note 3). All the M. micrantha samples grew normally without leaves for 30 days, but only 10% of the P. scandens samples survived (Supplementary Fig. 16). In addition, the outer epidermis of the stem of M. micrantha gradually changed from red to green in the defoliation treatment (Supplementary Fig. 17a). The major reason for this change was a significant reduction in the anthocyanin content (p = 0.001, two-sided Student’s t test), which was more than ten times lower than that in the stems of the undefoliated samples (Supplementary Fig. 17b), indicating that the stem of M. micrantha can compensate for its photosynthetic capacity by reducing anthocyanin accumulation and absorbing more external light energy. In addition, the stem of M. micrantha had the same photosynthetic components as the leaves, such as stomata and chlorophyll. The chlorophyll content in the stem of M. micrantha increased significantly in the defoliation treatment, but no significant changes occurred in P. scandens (Fig. 4b). Moreover, the number of grana lamellae increased, and the synthesized starch grains were also significantly larger than those in the control stems, which were similar to the patterns in the leaves (Supplementary Fig. 18). In addition, we found that the plant hormone (auxin, gibberellin, and cytokinin) contents in the M. micrantha stem after defoliation were significantly higher than that in the control group, and similar to that in the leaves (Supplementary Fig. 19a–c).

To investigate the gene expression patterns of the photosynthesis pathway genes in the stem of M. micrantha, we collected RNA-seq samples from defoliated and control stems. The upregulated genes were significantly enriched in the photosynthesis (p = 1.6 × 10−4, Deseq2), carbon fixation (p = 2.7 × 10−9, Deseq2), and plant hormone signal transduction pathways (p = 5.4 × 10−5, Deseq2; Fig. 4c). The anthocyanin contents were markedly reduced in the defoliation treatment because the expression of anthocyanin synthesis genes in this treatment were significantly downregulated compared with that in the control (Supplementary Fig. 17c, d). Six key genes in the Calvin cycle of the photosynthesis pathway showed significant upregulation in the defoliation treatment (Fig. 4d and Supplementary Fig. 20). Ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco) is an enzyme involved in the first major step of carbon fixation and plays an important role in the process of atmospheric carbon dioxide conversion by plants. The gene expression of five Rubisco copies was four times higher than that in the control group (Supplementary Fig. 20), suggesting that the stem maintains the plant growth by regulating the expression of Rubisco in the absence of leaves. Furthermore, the chlorophyll and plant hormone signal transduction and synthesis pathways displayed similar gene expression patterns, in which the genes were significantly upregulated in the treatment group (Fig. 4e). This evidence indicated that under the defoliation stress, the M. micrantha gained enough energy for rapid stem growth and development by improving its photosynthetic rate and increasing its chlorophyll and hormone contents.

Genetic pathway of Mikania STLs

STLs, which are mainly derived from the precursor germacrene A acid (GAA), are known for their allelopathic effects and are characteristic of the family Asteraceae19,33,34. Several enzymes acting downstream of GAA have been elucidated recently, such as HaG8H, LsCOS, and Ih8H, that lead to the formation of different STLs with distinctive stereochemistry35,36,37. However, the biosynthetic pathways of STLs in Mikania remain unknown. Lactone ring formation in STLs is dependent upon the (C6 or C8) position and stereoselective hydroxylations of GAA, which can result in STLs with four unique configurations (12, 6α-, 12, 6β-, 12, 8α-, and 12, 8β-olides derivatives of GAA)37. Ih8H homolog is important enzyme convert to 7, 8-trans lactone from GAA. As all the detected STLs in this study had a 7, 8-trans lactone, we were investigating for a homolog of Ih8H, an enzyme capable of synthesizing 7, 8-trans inunolide, the putative precursor of these M. micrantha STLs.

We reconstructed a genome-scale co-expression network for M. micrantha and focused on the metabolic pathways involved in inunolide synthesis, yielding a total of 38 genes in this pathway (Fig. 5a). Five genes previously reported to be involved in GAA synthesis, namely, germacrene A oxidases (GAOs), germacrene A synthase (GAS), and GAA 8-hydroxylase (Ih8H), which have high identity with the genes in H. annuus, A. annua, L. sativa, and Inula hupehensis (Supplementary Table 8 and Supplementary Fig. 21), were identified in the inunolide synthesis pathway and showed the high expression in the leaf and flower (Supplementary Fig. 22). Inunolide is a key intermediate of STL products, which can be identified and detected in fresh leaves by ultra-performance liquid chromatography (UPLC)-quadrupole time-of-flight-mass spectrometry (MS) (Supplementary Fig. 23). In addition, we have developed a liquid chromatograph MS (LC–MS) method to determine the content of inunolide (Fig. 5b), which displayed the same pattern as the expression of the five genes in different tissues (Supplementary Fig. 22). This suggests that inunolide is likely as intermediate to the STLs metabolites in Mikania. Thus, the proposed pathway of STLs in M. micrantha was displayed in green arrows (Fig. 5c). According to the co-expression network, the candidate genes of STLs biosynthesis were detected, which may regulate the pathway producing STLs in M. micrantha (Fig. 5a and Supplementary Table 9). Based on the previous research38, the candidate genes of STLs biosynthesis were assigned to CYP71 subfamily of P450 enzyme. Quantitative analysis of five available STLs (anhydroscandenolide, deoxymikanolide, dihydromikanolide, scandenolide, and 3-epi-dihydroscandenolide) showed a significant higher STLs content in the leaf and flower (p < 0.05, Duncan-test), but almost no STLs were detected in stem and root (Fig. 5d and Supplementary Fig. 27a). Microscopic studies revealed the high density of glandular trichomes occurred in flowers (petal) followed by leaves, less glandular trichomes were observed in stem surface (Supplementary Fig. 24). Most of STLs were extracted from glandular trichomes with the chloroform dipping extraction method (Supplementary Fig. 25). The content of five STLs in flowers and leaves trichomes present a higher abundance, but less in the stem trichomes (Supplementary Fig. 26), which further proved the STLs were derived from Mikania trichomes. These results are consistent with those of gene expression, which supported the proposed pathway of STLs in M. micrantha.

Fig. 5: Investigation of sesquiterpene lactones and proposed biosynthetic pathway.
figure 5

a Coexpression network of the sesquiterpene lactones (STLs) biosynthesis pathway elements with a correlation coefficient (>0.7). The genes (GAO germacrene A oxidase, GAS germacrene A synthase, Ih8H germacrene A acid 8-hydroxylae), that were reported in previous studies, are colored in orange. The candidate genes involved in STLs biosynthesis from inunolide are colored in red. b The relative inunolide content in different fresh plant tissues (n = 3). L, F, S, and R represent the leaf, flower, stem, and root, respectively. c The hypothesized biosynthesis pathway of STLs from germacrene A acid. Green arrows indicate the proposed metabolic pathways of Mikania micrantha STLs. The candidate genes of the biosynthesis pathway of M. micrantha STLs are indicated by blue boxes. The black arrows indicate the biosynthetic pathway of STLs in sunflower from a previous report36. FPP farnesyl pyrophosphate, GAA germacrene A acid. d The amount of five STLs in fresh tissues (n = 3). Anh anhydroscandenolide, Deo deoxymikanolide, Dih dihydromikanolide, Sca scandenolide, 3-epi 3-epi-dihydroscandenolide. e Pot experiments to simulate the release of M. micrantha STLs into soil. Five treatments: Add leaves, dried M. micrantha leaves were added to the soil surface in each pot to measure STLs in the soil, mainly coming from the fallen leaves. Add flowers, dried flowers were added to the soil surface in each pot to measure STLs mainly from fallen flowers. Nothing added, nothing was added to determine STLs in the soil from root exudates. Leaching, the plants were watered to wet the growing leaves and thus determine the STLs that were mainly derived from leaching. Natural growth, this was used as the control experiment. n = 4 biologically independent samples. d, e Duncan-test (p < 0.05) was used to determine statistical significance. Different letters above the error bars indicate significant differences among means for the different parts of the plant or treatments. Error bars indicate average value ± s.d. of indicated replicates. The source data underlying Fig. 5b, d, e are provided as a Source Data file.

Five STLs, including the dominant constituent dihydromikanolide and scandenolide, were identified in invaded soils collected from different regions (Supplementary Fig. 27b). The STLs content in soil differed among the collection regions, which may have been related to community age, i.e., invasion history, or other ecological factors in the field. To better understand how the STLs are released into surrounding soil, a pot experiment was conducted to determine where the STLs in soil come from (litter fall from the aerial part, root exudates or leaching) (Fig. 5e and Supplementary Note 4). Five STLs were also detected in the soil after 15 days of treatment, and the STL levels in the treatments with added M. micrantha leaves and added flowers were higher than those in the treatments with root exudates (nothing added) and leaching (Fig. 5e). Five STLs that were almost undetectable in the root by LC–MS analysis accumulated in the leaves and flowers, which suggested that these five STLs were released into the environment by litter fall and rain, not by root exudates. In addition, the five STLs were detected in soil invaded by M. micrantha, providing reliable evidence that STLs are typical allelochemicals of the “mile-a-minute” weed. The proposed STLs pathway based on metabolic and genetic data will promote further investigation of the function of allelochemicals in M. micrantha.

Reinforced microbial function promotes nutrient utilization

To investigate the potential effects of M. micrantha on soil nutrient cycling, we conducted a pot experiment with M. micrantha and its two neighboring native species, namely, Polygonum chinense and Paederia scandens (see Supplementary Note 5). The total nitrogen in M. micrantha soil significantly decreased, compared to that in the soil of its two neighboring species and the control which no plants were planted (Fig. 6a). Conversely, the available nitrogen, particularly ammonium nitrogen, in M. micrantha soil increased significantly more than that in the soil of its two neighboring native species (Fig. 6a and Supplementary Table 10). In addition, the variation in phosphorus and potassium content in soil showed a similar trend (Supplementary Table 10). Furthermore, we determined the contents of nitrogen, phosphorus and potassium contents in different plant tissues. The accumulation of the three elements in the different tissues of M. micrantha was significantly higher than that in its two neighbors (P. chinense and P. scandens) (Supplementary Fig. 28).

Fig. 6: Effect of Mikania micrantha metabolites on soil nitrogen cycling.
figure 6

STLs: sesquiterpene lactones. a Soil nitrogen content in the pot experiment of M. micrantha and its two neighboring native species. n = 6 biologically independent soil samples. We used Duncan’s multiple range test (p < 0.05) method. Data with different letters in the same column are significantly different. Control indicates that no plants were planted. Error bars indicate average value ± s.d. of indicated replicates. b Box plots of the relative abundance of genes involved in nitrogen metabolism pathway. Adjusted p value was calculated using Dunn’s test. n = 5 biologically independent soil samples. c The relative abundance of key genes of soil microorganisms involved in assimilatory nitrogen reduction (nasA), dissimilatory nitrogen reduction (napA), ammonification (ureC), and denitrification (norB) between M. micrantha and its two neighboring species (Polygonum chinense and Paederia scandens) as well as in the blank control. Adjusted p value was calculated using Dunn’s test. n = 5 biologically independent soil samples. On each box plot, the central mark indicates the median, the bottom and top edges of the box indicate the interquartile range (IQR) and the whiskers represent the maximum and minimum data points. d A schematic of nitrogen cycling in the M. micrantha soil. Red and green indicate that the abundance of the gene in M. micrantha soil was upregulated and downregulated compared to that in the soil of the two neighboring species (P. chinense and P. scandens), respectively. Black, orange, red, blue, gray, and purple arrows indicate the denitrification, nitrification, nitrogen fixation, dissimilatory nitrogen reduction, assimilatory nitrogen reduction, and ammonification, respectively. The source data underlying Fig. 6a–c are provided as a Source Data file.

Soil metagenomics provides a cultivation-independent assessment of the largely untapped genetic reservoir of soil microbial communities, which has already led to the identification of functional biomolecules39. To investigate the effects of whole soil microbial communities on the growth of M. micrantha and its neighboring species, we collected soil DNA samples for metagenomic research. With all the soil samples, a total of 25,597,567 nonredundant genes were recovered, with an average open reading frame length of 537 bp. The rarefaction analysis of all soil samples showed a curve approaching saturation (Supplementary Fig. 29), suggesting that the vast majority of microbial genes in the soil where the three species grew were present in our gene catalog. Genes related to carbohydrate metabolism, energy and nitrogen metabolism, membrane transport, replication, and repair and cell motility in M. micrantha root microbiome was significantly enriched compared to that in its two neighbors and the control (p < 0.05, Duncan-test; Fig. 6b and Supplementary Figs. 30 and 31), indicating that the root microbiome of M. micrantha was more active than that of the two neighboring species.

In natural soils, the vast majority of N, P, and K atoms are minimally bioavailable for plants growth, but can be metabolized by various soil microbes, which means that the mineralization processes mediated by soil microbes promoted nutrient cycling in natural ecosystems40. In this study, the relative abundance of key genes involved in nitrogen fixation (nif), assimilatory nitrogen reduction (nirB and nasA), dissimilatory nitrogen reduction (napA and nirB/D), and ammonification (ureC) in M. micrantha soil was higher than that in the soil of its two neighboring native species (P. chinense and P. scandens) soil and the control (Fig. 6c, d and Supplementary Fig. 32), indicating that as a result of the soil microbe functioning, more available forms of nitrogen were reduced to ammonia nitrogen and ammonium was produced and accumulated so as to speed up the nitrogen cycle in M. micrantha soil. In contrast, the relative abundance of key genes involved in denitrification (norB) in M. micrantha soil was significantly lower than that in P. chinense and P. scandens soil (Fig. 6c, d), suggesting that there were fewer denitrifying bacteria which use nitrates in M. micrantha soil to carry out respiration so as to decrease the loss of available nitrate nitrogen. In addition, the population density of cultured Azotobacter, ammoniated bacteria, and phosphorus- and potassium-solubilizing bacteria (Supplementary Fig. 33) in M. micrantha soil were significantly higher than that in the soil of its two neighboring species soil.

Plant metabolites play an important role in plant survival in a given environment and establish ecological relationships between plants and other organisms. To investigate the metabolites effects of M. micrantha on soil nutrient cycling, we added five STLs (anhydroscandenolide, deoxymikanolide, dihydromikanolide, scandenolide, and 3-epi-dihydroscandenolide) separately to the bare soil near an M. micrantha monoculture to investigate the metabolite-microbe interactions. The STLs significantly increased not only the concentration of carbon dioxide (Supplementary Fig. 34), but also the relative abundance of nitrogen fixation bacteria (Supplementary Fig. 35), indicating that the activity of microorganisms was improved. Furthermore, the available nutrients, e.g., ammonium nitrogen, were mineralized efficiently in the soil in five STLs treatments, which was significantly higher than that in the control (Supplementary Table 11). These evidences indicated that the STLs of soil may play an important role in enriching microorganisms, which in turn accelerating the nutrient cycling (Fig. 6d). On the other hand, root zone is the primary site for interactions between microbes and plants. They provide a very attractive, nutrient-rich niche that harbors large numbers of active microbes that are important for plant health and nutrient uptake41,42,43.

Discussion

Given the common name “mile-a-minute”, one of the most distinctive features of M. micrantha is its fast growth, which strongly relies on its photosynthesis ability1. In this study, we found that the variation pattern of malic acid content, citric acid content, stomatal conductance and key genes expression, such as PEPC, CA, NADP-ME, are consistent with classic CAM plants. This result indicates that CO2 may be absorbed at night by photosynthesis pathways which was similar to CAM and stored in vacuoles as organic acids to supplement the carbon fixation capacity during the day, helping to absorb more carbon dioxide at night in M. micrantha. The CAM pathway is mostly found in dry-desert plants and is rare in normal plants28,44,45, which was contrast with tropical plant M. micrantha. In addition, the M. micrantha leaves are not succulent, which was a morphological prerequisite for vacuolar storage capacity of organic acids46. These evidences suggested that M. micrantha may develop a special mechanism allowing it to achieve a higher photosynthetic rate that enables fast growth, and the details of the photosynthesis mechanism needs further research. Moreover, the stem of M. micrantha also has a high photosynthetic capacity, further enhancing the growth ability of stems. Thus, the special photosynthesis pathways, together with the greater photosynthetic ability of the stem, is thought to be one of the most important factors contributing to the fast growth of this species. Previous studies reported that photosynthetic efficiency could increase by modification of the photosynthetic pathway, such as introducing an intact maize gene for one of the key enzymes in C4 photosynthesis into rice47 and the alternative photorespiratory pathways48, implying a photosynthetic pathway by similar CAM at night in C3 crop could be used to improve their carbon fixation capacity through a modified photosynthetic pathway in the future.

The well-known allelochemicals STLs21,35,36 have a certain allelopathic effect on associated plants and soil microbes49,50,51. We found that STLs mainly accumulated in the leaves and flowers and were then released into the surrounding soil by litter fall. The STLs in the soil increased the ammonium nitrogen content by regulating microbial activity. Our results showed that the M. micrantha STLs could not only promote nitrogen fixation, assimilatory nitrogen reduction, dissimilatory nitrate reduction to ammonium, and ammonification, but also inhibited denitrification in the soil, indicating that the soil microbes of M. micrantha promoted nitrogen mineralization and decreased the loss of available nitrogen, thereby accelerating nitrogen cycling, which allowed M. micrantha to utilize more nutrients as well as grow rapidly. In addition, we also proposed a putative biosynthesis pathway of STLs, which may be valuable for producing biological herbicides and pharmacological chemicals derived from STLs, benefiting environmental biosecurity and human health.

Methods

Plant materials and sequencing

Genomic DNA was extracted from a M. micrantha sample, which was collected form Neilingding Island, Shenzhen, China, for constructing four Illumina PCR-free libraries with 350-bp insert size and sequenced on Illumina HiSeq x ten platforms. Totally, 339 and 14 SMRT cells with 20-kb insert library were constructed and sequenced on RSII and Sequel, respectively. Total RNA was extracted from four tissues (root, stem, leaf, and flower, three biological repeats) using the RNA extraction kit (Huayueyang, ZH120, China). The mRNA from all samples was purified using the Truseq RNA Sample Prep Kit (illumina), followed to be sequenced on an Illumina HiSeq 2500 sequencer.

Genome assembly and annotation

The Illumina raw reads were filtered by trimming the adapter sequence and low-quality part by in-house software clean_adapter and clean_lowqual (https://github.com/fanagislab/assembly_2ndGeneration/tree/master/clean_illumina), resulting in a clean and high-quality reads data with average error rate <0.001. The raw PacBio reads were assembled by the software canu-1.652, and contaminating heterozygous fragments were filtered by using the result of falcon/falcon-unzip53. We used the alternative heterozygous haplotype sequence (hapotig) assembled by falcon/falcon-unzip to align to the contigs assembled by canu and d filtered out the heterozygous contigs from canu assembled result (Supplementary Fig. 1). Then Illumina reads were aligned to the contigs by BWA-MEM54, and single base errors in the contigs were corrected by Pilon-v1.2255. For the Hi-C-based proximity-guided assembly, we removed duplications and kept reads that uniquely mapped to the reference genome. The assembly package, Lachesis was applied to do clustering, ordering and orienting56.

The gene models were predicted by EVidence Modeler v1.1.157, embedded in a pipeline that integrating evidences from ab initio predictions, homology-based searches, and full-length transcriptome and RNA-seq alignments. Then, the protein-coding sequences were mapped by RNA-seq data and functionally annotated using UniProt58 and KEGG databases59. Finally, the gene models were retained if they had at least one supporting evidence. Gene functional annotation was performed by aligning the protein sequences to UniProt, eggNOG and KEGG databases using BLASTP v2.3.0+ (http://blast.ncbi.nlm.nih.gov/Blast.cgi) with E-value cutoff of 10−5. The pathway analysis and functional classification were conducted based on KEGG database. InterProScan60 was used to assign preliminary GO terms, Pfam domains and IPR domains to the gene models.

A de novo repeat library for M. micrantha was constructed by RepeatModeler (v1.0.4; http://www.repeatmasker.org/RepeatModeler/). TEs were identified by RepeatMasker (v4.0.6; http://www.repeatmasker.org/) using both Repbase library and the de novo library. Tandem repeats were predicted using Tandem Repeats Finder v4.07b61. The LTR-RTs were annotated with an in-house pipeline that uses LTR-finder62 and LTR-retriever63. The LTR insert time is estimated by LTR-retriever.

Evolution analysis

Duplicated genomic fragments were identified by MCscanX64, requiring at least ten paralogous gene pairs per collinear block, and the duplicate_gene_classifier in MCscanX was implemented to classify the origins of the duplicate genes into different types. Orthologous and paralogous gene families were assigned by OrthoFinder65 with default parameters.

Gene families that contain only one gene for each species were selected to construct the phylogenetic tree. The protein sequences of each gene family were independently aligned by muscle v3.8.3166 and then concatenated into one super-sequence. The phylogenetic tree was constructed by maximum likelihood (ML) using PhyML v3.067 with best-fit model (MtMAN) that was estimated by ProtTest368. The Bayesian Relaxed Molecular Clock (BRMC) approach was adopted to estimate the neutral evolutionary rate and species divergence time using the program MCMCTree, implemented in PAML v4.9 package69. The calibration time interval (110–124 Mya) of grape was adopted from TimeTree (http://www.timetree.org). We used the the inter- and intra-syntentic gene pairs to calculate the synonymous mutation rate (Ks) values by KaKs_Calculator 2.0 with default parameter70.

Transcriptome data analysis

Transcriptome reads were filtered by in-house software clean_adapter and clean_lowqual (https://github.com/fanagislab/assembly_2ndGeneration/tree/master/clean_illumina) and then mapped to the reference genome of M. micrantha using TopHat v. 2.1.071 with default settings. Reads count were calculated with HTSeq v. 0.9.172 using BAM results from TopHat v. 2.1.0, and TPM values were then calculated for every gene in the samples. We continue to identify the differentially expressed genes by DEseq273. Co-expression network analysis was performed by WGCNA package74 using all transcriptomes. WGCNA network construction and module detection was identified using an unsigned type of topological overlap matrix with a soft-thresholding power β of 28 (correlation coefficient R2 > 0.9).

The STLs measure

Five STLs were isolated from the aerial part of M. micrantha by repeated chromatography. The structures of STLs were elucidated by nuclear magnetic resonance and high resolution electrospray ionization MS. UPLC coupled to MS was used as a method for quantitative analysis of five STLs (Supplementary Table 12) and the chemical constituents of different plant tissues (for details see the Supplementary Information).

Statistical analysis

All statistical tests were performed using SPSS 11.5 software (SPSS Inc., USA). Soil chemical characteristics, nitrogen, phosphorus, potassium content of plant tissue and the population density of soil cultured bacteria were, respectively, compared using one-way ANOVA, followed by least significant difference tests at p < 0.05 or Duncan’s multiple range test, p < 0.05. All observations are independent of one another and scores in groups are normally distributed. A univariate F-test for each variable was used to interpret the respective effects. The equality of error variances was tested by using Levene’s test and the error variance of the dependent variable was considered to be equal across groups when p > 0.05. In the soil metagenomic analysis, the significant functional differences between M. micrantha and two neighboring species, control samples were determined by the Duncan-test.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.