Integrated multi-omics investigation revealed the importance of phenylpropanoid metabolism in the defense response of Lilium regale Wilson to fusarium wilt

Abstract Lilies (genus Lilium) play a significant role in the global cut-flower industry, but they are highly susceptible to fusarium wilt caused by Fusarium oxysporum. However, Lilium regale, a wild lily species, exhibits remarkable resistance to F. oxysporum. To investigate the quantitative resistance of L. regale to fusarium wilt, a comprehensive multi-omics analysis was conducted. Upon inoculation with F. oxysporum, L. regale roots showed a significant accumulation of phenylpropane metabolites, including lignin precursors, flavonoids, and hydroxycinnamic acids. These findings were consistent with the upregulated expression of phenylpropanoid biosynthesis-related genes encoding various enzymes, as revealed by transcriptomics and proteomics analyses. Furthermore, metabolomics and proteomics data demonstrated differential activation of monoterpenoid and isoquinoline alkaloid biosynthesis. Colorimetry and high-performance liquid chromatography analyses revealed significantly higher levels of total flavonoids, lignin, ferulic acid, phlorizin, and quercetin contents in L. regale scales compared with susceptible lily ‘Siberia’ scales during F. oxysporum infection. These phenylpropanes exhibited inhibitory effects on F. oxysporum growth and suppressed the expression of pathogenicity-related genes. Transcriptional regulatory network analysis suggested that ethylene-responsive transcription factors (ERFs) may positively regulate phenylpropanoid biosynthesis. Therefore, LrERF4 was cloned and transiently overexpressed in the fusarium wilt-susceptible Oriental hybrid lily ‘Siberia’. The overexpression of LrERF4 resulted in increased levels of total flavonoids, lignin, ferulic acid, phlorizin, and quercetin, while the silencing of LrERF4 in L. regale through RNAi had the opposite effect. In conclusion, phenylpropanoid metabolism plays a crucial role in the defense response of L. regale against fusarium wilt, with LrERF4 acting as a positive regulator of phenylpropane biosynthesis.


Introduction
Plants face numerous stresses throughout their life cycle, including fungal diseases that have a detrimental impact on their growth and productivity.Over time, plants have developed unique immune mechanisms to combat pathogens [1].Plant pattern recognition receptors detect pathogen-associated molecular patterns, triggering pattern-triggered immunity, which is closely linked to quantitative resistance (QTR) [2].Pathogens release effector proteins to infect plants, but plants produce resistance proteins encoded by resistance genes to counteract these effectors, resulting in effector-triggered immunity, also known as single-gene resistance or qualitative resistance [3].QTR, controlled by multiple genes and involving the biosynthesis of resistance-related metabolites, is considered the superior disease resistance mechanism in plants [4,5].
Recent multi-omics analyses have provided comprehensive insights into the mechanisms underlying QTR.Metabolome and transcriptome analyses of Zea mays infected with Puccinia sorghi demonstrated higher accumulation of phenylpropanes and terpenes in the resistant line (H95:Rp1-D) compared with the susceptible line (H95) [6].Additionally, transcription factors (TFs) such as WRKY, MYB, and bHLH (basic helix-loop-helix) were identified as regulators of the H95:Rp1-D response to P. sorghi infection.Metabolome and proteome analyses of Triticum aestivum indicated higher accumulation of hydroxycinnamic acid amides and f lavonoids in the Fusarium graminearum-resistant wheat line HC374, leading to a thickened cell wall and enhanced resistance to F. graminearum [7].Furthermore, integrated transcriptomic and metabolomic analyses revealed that the Citrus sinensis mutant MT exhibited increased resistance to Penicillium digitatum compared with the wild-type C. sinensis due to enhanced jasmonate biosynthesis [8].
Metabolites play a crucial role in plant defense responses, with several phenylpropanes derived from phenylalanine, terpenoids derived from methyl-glutaric acid, and alkaloids serving as secondary metabolites contributing to stress resistance [9].For example, cinnamaldehyde, a phenylpropane, inhibits the growth of Fusarium sambucinum [10].The β-costic acid in Z. mays restricts the growth of Fusarium verticillioides and F. graminearum both in vivo and in vitro [11].Harmane, a β-carboline alkaloid extracted from Peganum harmala, damages the cell membrane of Fusarium oxysporum and induces its death [12].Previous research has demonstrated that identifying and characterizing resistancerelated genes using forward genetics-based methods, along with analyses of resistance-related metabolites, can elucidate the molecular basis of QTR [13].
Lilies (Lilium spp.) are highly valued cut-f lower species worldwide, but many lilies are susceptible to fusarium wilt, primarily caused by F. oxysporum, which negatively impacts lily production [20].Fusarium oxysporum infection leads to root and scale decay, vascular bundle blockage, and hindered nutrient transport, ultimately resulting in wilting and death [21].Fortunately, some wild lilies, including Lilium regale Wilson, exhibit resistance to fusarium wilt.Previous studies have shown that WRKY TFs in L. regale modulate the expression of resistance genes, such as pathogenesisrelated protein 10 (PR10), chitinase, and defensin, positively regulating resistance to F. oxysporum [22][23][24].However, the mechanism underlying L. regale's resistance to F. oxysporum remains unknown.This study aims to explore the metabolites associated with fusarium wilt resistance in L. regale using multi-omics techniques and investigate the related TFs as well as the antifungal effects of these metabolites.

Transcriptome profile of L. regale responding to an F. oxysporum infection
Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis revealed that 7661 unigenes in the L. regale transcriptomic data were associated with metabolism, with a significant number of unigenes involved in secondary metabolite biosynthesis (Fig. 1A).Consequently, a substantial number of metabolismrelated unigenes were further identified.
In comparison with the control group (mock inoculation with sterile water), the expression of 752, 1150, and 4297 unigenes was upregulated at 2, 4, and 6 days post-inoculation (dpi), respectively, in the F. oxysporum-infected samples.Conversely, 2060, 2195, and 1224 unigenes exhibited downregulated expression levels at 2, 4, and 6 dpi, respectively, in the F. oxysporum-infected samples.KEGG pathway enrichment analysis of the metabolism-related differentially expressed genes (DEGs) indicated that the DEGs at 2 dpi were primarily associated with phenylpropanoid, f lavonoid, and isoquinoline alkaloid biosynthesis, glutathione metabolism, and oxidative phosphorylation (Fig. 1B and Dynamic Fig. 1).At 4 dpi, many DEGs were related to mitogen-activated protein kinase (MAPK) signaling, oxidative phosphorylation, and phenylpropanoid biosynthesis (Fig. 1C).At 6 dpi, numerous DEGs were relevant to MAPK signaling, ABC transporters, glutathione metabolism, and stilbene/phenylpropanoid/diarylheptanoid/gingerol biosynthesis (Fig. 1D).The phenylpropanoid biosynthesis pathway serves as an upstream pathway for the synthesis of lignin/lignan, f lavonoids, and coumarin.Moreover, the MAPK signaling pathway plays a role in regulating biotic and abiotic stress responses through interactions with plant hormone signaling pathways and TFs [25].Additionally, downregulated oxidative phosphorylation, indicative of inhibited energy metabolism, may contribute to maintaining a balance between growth and immune responses.Therefore, the transcriptome analysis uncovered the induction of biosynthesis pathways for resistance-related metabolites and signaling pathways in F. oxysporum-infected L. regale.

Characteristics of the proteome of L. regale inoculated with F. oxysporum
The L. regale proteins affected by F. oxysporum infection were identified using ultra-high-performance liquid chromatographytandem mass spectrometry (UPLC-MS/MS).Out of the 473 595 secondary spectra obtained, 51 691 matched the spectra of known proteins.Additionally, 27 558 peptides were identified, including 26 053 unique peptides.A total of 6501 proteins were assembled and quantified based on the proteome data analysis.The differentially expressed proteins (DEPs) identified underwent KEGG pathway enrichment analysis.At 2, 4, and 6 dpi, 595, 652, and 657 upregulated DEPs were observed, respectively, while 398, 353, and 450 DEPs were downregulated, respectively.

Overview of the metabolome of L. regale infected with F. oxysporum
The metabolome of F. oxysporum-inoculated L. regale was analyzed using UPLC-MS/MS, which detected 22 387 negative ions and 20 050 positive ions.A total of 15 341 ions were annotated as different metabolites in the KEGG database, with 12 172 ions representing phytochemical compounds, including 2460 ions corresponding to secondary metabolites.Among these, 1909 secondary metabolites showed significant differences in accumulation levels and were designated as differentially accumulated metabolites (DAMs).At 2, 4, and 6 dpi, 1340, 1115, and 1071 DAMs were upregulated, while 569, 794, and 838 DAMs were downregulated, respectively.The DAMs were classified into eight major categories according to the KEGG database (Fig. 3A).Terpenoids, phenylpropanes, and alkaloids represented the three major secondary metabolite groups that exhibited the most significant differential accumulation in L. regale, with 1007, 572, and 552 compounds classified as such, respectively.Notably, 80% of the identified phenylpropanes showed differential accumulation between the inoculated and control samples at 2 dpi, highlighting their importance during the early stages of F. oxysporum infection.Furthermore, 83 and 82% of the identified terpenoids showed differential accumulation between the inoculated and control samples at 4 and 6 dpi, respectively.

Integrated analysis of multi-omics data revealed the importance of phenylpropanes for L. regale resistance to F. oxysporum
The integrated analysis of transcriptome, proteome, and metabolome data revealed that the phenylpropanoid biosynthesis pathway was highly upregulated in F. oxysporum-infected L. regale roots (Fig. 4).The accumulation of hydroxycinnamic acids, such as ferulic acid and p-coumaric acid, increased at 2 dpi.Notably, the ferulic acid content increased 2.5-fold at 2 dpi.Additionally, the biosynthesis of lignin/lignans was also activated, as evidenced by the 3.14-, 2.44-, and 1.38-fold increases in the accumulation of (+)-lariciresinol, (+)-pinoresinol, and secoisolariciresinol at 2, 4, and 2 dpi, respectively.The lignin biosynthesis precursors were also upregulated.The biosynthesis of coumarins, including esculetin, scopoletin, and fraxetin, was induced, too.Similarly, the f lavonoid contents (e.g.luteolin, kaempferol, quercetin, and pinostrobin) also increased.We further analyzed the effects of the F. oxysporum infection on the expression levels of genes and the accumulation of proteins related to phenylpropanoid biosynthesis.Key enzymes involved in phenylpropanoid biosynthesis, such as phenylalanine ammonia lyase (PAL), showed significant increases at the transcriptome and proteome levels.The hydroxycinnamic acid synthesisrelated enzyme caffeic acid 3-O-methyltransferase (COMT) was upregulated at 2 dpi , with 2.84-and 1.72-fold increase  S1.
(relative to the corresponding control levels) at transcription and protein level, respectively.Moreover, the unigene and protein of ferulic acid 5-hydroxylase (F5H), another hydroxycinnamic acid synthesis-related enzyme, was upregulated at 4 dpi with 4.6-and 1.3-fold increase (relative to the corresponding control levels), respectively.Additionally, the production of cinnamoyl-CoA reductase (CCR), POD, dirigent (DIR), and laccase (LAC) was also upregulated at the transcriptome and proteome levels.Furthermore, the first enzyme of the f lavonoid biosynthesis pathway, chalcone synthase (CHS), was significantly upregulated according to the transcriptome and proteome.Moreover, the expression levels of chalcone isomerase (CHI), f lavonoid 3 -monooxygenase (F3M) and phlorizin synthase (PGT1), which affect f lavonoid biosynthesis, increased at the transcript and protein levels during the infection of L. regale by F. oxysporum.
Overall, the integrated analysis of multi-omics data highlighted the significant accumulation of phenylpropanes in F. oxysporuminfected L. regale, indicating their importance in the defense response against the pathogen.Furthermore, the biosynthesis pathways of alkaloids and terpenoids were significantly affected by the infection.These findings provide valuable insights into the molecular mechanisms underlying the defense response.

Phenylpropanes accumulated more in the resistant L. regale than in the susceptible lily 'Siberia' during F. oxysporum infection
Multi-omics analysis revealed that the accumulation of phenylpropanes is crucial for the defense response of L. regale against F. oxysporum.To compare the accumulation of certain phenylpropanes during F. oxysporum infection, L. regale (resistant) and Oriental hybrid lily 'Siberia' (susceptible) were examined.At 6 dpi, 'Siberia' leaves exhibited severe chlorosis (yellowing), and the inoculation sites showed clear decay, consistent with the susceptibility of 'Siberia' to F. oxysporum (Fig. 5A).Subsequently, the lignin, total f lavonoid, ferulic acid, phlorizin, and quercetin contents were determined in the scales of both lilies following F. oxysporum infection.HPLC chromatograms of ferulic acid, phlorizin, and quercetin (Fig. 5B) revealed significant differences in the accumulation of these phenylpropanes between L. regale and 'Siberia' scales.After inoculation with F. oxysporum, ferulic acid showed higher accumulation in L. regale compared with 'Siberia' (Fig. 5C).At 6 dpi, the ferulic acid content increased 2.53-fold in L. regale, while it only increased 1.31-fold in 'Siberia'.Additionally, phlorizin content increased in L. regale, peaking at 6 dpi (128.53 mg/g dry weight), in contrast to the decrease observed in 'Siberia'.Similarly, quercetin showed a significant increase (1.53-fold) in L. regale at 6 dpi, but decreased in 'Siberia'.Regarding f lavonoids, at 2 dpi the content was higher in L. regale (242.18mg/g) compared with 'Siberia' (146.35mg/g) (Fig. 5D), indicating the rapid accumulation of f lavonoids in L. regale in response to F. oxysporum infection.At 6 dpi, lignin content increased 1.78-fold in L. regale (290.01 mg/g) (Fig. 5D), exceeding the lignin content in 'Siberia' (191.48 mg/g).Consequently, compared with 'Siberia', F. oxysporum infection led to a higher accumulation of phenylpropanes in L. regale scales.

Phenylpropanes inhibited F. oxysporum growth and pathogenicity-related gene expression
The antifungal effects of the ferulic acid, phlorizin, quercetin, and total f lavonoids extracted from L. regale were evaluated.The antifungal assay demonstrated that these phenylpropanes had significant inhibitory effects on F. oxysporum mycelial growth, with the inhibitory effects increasing as the concentrations of the compounds increased (Fig. 6A).Notably, 0.6 mg ferulic acid exhibited the most potent inhibition, resulting in a growth inhibition area of 220.11 mm 2 (Fig. 6B).Even at smaller content (0.2 mg), f lavonoids, ferulic acid, phlorizin, and quercetin inhibited F. oxysporum mycelial growth, with growth inhibition areas of 25.89, 78.49, 52.62, and 32.57mm 2 , respectively.Furthermore, the addition of f lavonoids, ferulic acid, phlorizin, and quercetin (2 mg/mL) to the liquid culture medium adversely affected the germination of F. oxysporum conidia (Fig. 6C).
The expression of F. oxysporum pathogenicity-related genes was analyzed after supplementing the culture medium with total f lavonoids, ferulic acid, phlorizin, and quercetin (0.1, 0.2, and 0.3 mg/mL for all treatments).The treatments with these compounds led to the suppression of pathogenicity-related gene expression in F. oxysporum (Fig. 6D).For instance, treatment with 0.2 mg/L total f lavonoids significantly downregulated the expression of xylanase transcriptional activator (FoXlnR), which inf luences the growth of saprophytic and pathogenic F. oxysporum.The expression levels of two F. oxysporum virulence genes, mitogenactivated protein kinase (FoFmk1) and guanine nucleotide-binding protein (FoFgb1), were only 0.2 and 6.7% of the corresponding control levels after treatment with 0.3 mg/L total f lavonoids and ferulic acid, respectively.The expression of two plant cell wall degradation-related genes, beta-1,3-glucanosyltransferase (FoGas1) and pectate lyase (FoPl1), was downregulated by treatment with 0.2 and 0.3 mg/L total f lavonoids, with expression levels reaching only 0.3 and 1.2% of the corresponding control levels, respectively.Moreover, 0.1 mg/L quercetin treatment significantly inhibited the expression of F. oxysporum GTPase-binding protein (FoGBP).

Transcriptional network regulating phenylpropanoid biosynthesis during the L. regale defense response to F. oxysporum
TFs play a crucial role in regulating defense responses in plants.
To investigate the transcriptional regulation of phenylpropanoid biosynthesis, we analyzed the correlation between phenylpropane contents determined by multi-omics analysis and the expression patterns of differentially expressed transcription factors (DETFs) identified from the transcriptomics data.A total of 1526 unigenes were identified as TFs, with 104, 114, and 67 DETFs found to be differentially expressed at 2, 4, and 6 dpi, respectively.These DETFs belonged to 38 TF families, including Trihelix, MYB, bZIP (basic region leucine zipper), and GRF (growth-regulating factor) families (Fig. 7A).The majority of DETFs belonged to the Trihelix, WRKY, ERF, and C2H2 families.
Additionally, 34 TFs were found to regulate 24 phenylpropane synthesis pathways (Fig. 7B).Among them, ERF, bZIP, and WRKY family members were found to regulate the synthesis of 16, 15, and 13 phenylpropanes, respectively, with a Pearson correlation coefficient (ρ) of >0.7 or <−0.7.TFs with a ρ value >0.8 or <−0.8 are listed in Table 1.The ERF family was found to positively regulate the synthesis of phenylpropanes such as phlorizin and ferulic acid, based on the high ρ value (0.85) for the correlation between ferulic acid/5-hydroxyconiferaldehyde and ERF.Additionally, one glabrous-enhancer-binding protein (GeBP) was predicted to modulate sinapic acid biosynthesis.On the other hand, the WRKY family was predicted to negatively regulate the biosynthesis of phenylpropanes such as 4-coumarate, 4-hydroxycinnamyl aldehyde, 5-hydroxyferulic acid, caffeic aldehyde, caffeyl alcohol, and trans-cinnamate.Furthermore, the C2H2 family was predicted to negatively regulate the biosynthesis of 4-coumaryl alcohol/5hydroxyconiferyl alcohol.

Positive regulatory effects of LrERF4 on L. regale resistance to F. oxysporum
The above-mentioned analysis of the transcriptional regulatory network suggests that ERF TFs likely have a positive regulatory role in the biosynthesis of phenylpropanes, such as f lavonoids and hydroxycinnamic acid.Therefore, we cloned and characterized an ERF gene from L. regale (LrERF4) to investigate its function.We also examined the expression profiles of ERF4 in F. oxysporuminfected L. regale and lily 'Siberia' plants.Compared with the control, the expression of LrERF4 increased 2.5-fold at 2 dpi, while the expression level of ERF4 decreased in F. oxysporum-infected 'Siberia' (Fig. 8A).This suggests that a rapid increase in LrERF4 expression may contribute to the resistance of L. regale to F. oxysporum.
Furthermore, we transiently overexpressed LrERF4 in lily 'Siberia' scales and inoculated them with F. oxysporum.The average lesion area on the LrERF4-overexpressing (OE) scales (1.88 mm 2 ) was only 9% of that on the control scales containing the empty vector (Fig. 8B).This indicated that the transient overexpression of LrERF4 reduced the susceptibility of 'Siberia' to F. oxysporum.Quantitative real-time PCR (qPCR) analysis confirmed the expression of LrERF4 in the 'Siberia' scales  biosynthesis-related genes naringenin 3-dioxygenase (F3H) and CHS was upregulated by 1.76-and 1.05-fold in the LrERF4-OE scales, respectively.Furthermore, the expression of F5H, which contributes to hydroxycinnamic acid biosynthesis, increased by 1.93-fold in the LrERF4-OE scales.The lignin, total f lavonoid, ferulic acid, phlorizin, and quercetin contents also increased in the LrERF4-OE scales (Fig. 8D).Specifically, the lignin and total f lavonoid contents increased by 1.33-and 1.29-fold, respectively.The accumulation levels of ferulic acid, phlorizin, and quercetin in 'Siberia' scales overexpressing LrERF4 increased by 1.30-, 1.21-, and 1.07-fold, respectively.Therefore, it appears that LrERF4 positively regulates the biosynthesis of phenylpropanes, including lignin, f lavonoids, ferulic acid, phlorizin, and quercetin.
To suppress the expression of LrERF4 in L. regale, the RNAi vector pHellsgate2-LrERF4 was introduced into L. regale scales using Agrobacterium tumefaciens.After 72 h, the L. regale scales were infected with F. oxysporum, and the disease symptoms were observed at 6 dpi.Compared with the control L. regale scales expressing the empty RNAi vector, the resistance of L. regale scales expressing pHellsgate2-LrERF4 was reduced (Fig. 9A), with an average lesion area 12.65-fold larger than that of the control scales.QPCR analysis confirmed that the expression of LrERF4 decreased in the scales transformed with pHellsgate2-LrERF4 (Fig. 9B).This decreased expression resulted in the downregulation of phenylpropanoid biosynthesis-related genes.Notably, the relative expression levels of genes involved in lignin/lignan biosynthesis, such as PAL, 4CL, F5H, CCoAOMT, CCR, and coniferyl-alcohol dehydrogenase (CAD), decreased by 72, 77, 82, 51, 32, and 45%, respectively, compared with the corresponding control levels.The relative expression of f lavonoid biosynthesisrelated genes, including CHI, F3M, and F3H, decreased to 27, 81, and 31% of the corresponding control levels, respectively, following the RNAi-mediated silencing of LrERF4.Moreover, the lignin, total f lavonoid, ferulic acid, phlorizin, and quercetin contents decreased to 80, 66, 94, 88, and 87% of the corresponding control levels, respectively, after the silencing of LrERF4 by RNAi (Fig. 9C).

Discussion
In plants, broad-spectrum and long-lasting resistance to stress, known as QTR, is mediated by genes related to defense responses, metabolite production, and phytohormone production [26,27].In Z. mays, ZmCCoAOMT2 and ZmCAD, which are involved in lignin biosynthesis, were confirmed to encode components of the QTR mechanism protecting against leaf diseases [28,29].As a highly resistant wild lily, L. regale is valuable for breeding disease-resistant lily varieties [30].Although previous research has identified several WRKY TFs that positively regulate L. regale resistance to fusarium wilt, the QTR mechanism has not been thoroughly investigated.This study was conducted to characterize the QTR mechanism that protects L. regale from F. oxysporum, thus providing insights into the molecular basis of its resistance.
Multi-omics techniques are useful for comprehensively exploring plant defense mechanisms activated by pathogens.Integrated transcriptome and proteome analysis has confirmed the importance of phenylpropanoid biosynthesis and linoleic acid metabolism in the resistance of Z. mays to F. verticillioides [31].Additionally, protein-lysine 6-oxidase 2 was found to positively regulate maize resistance to F. verticillioides through the salicylic acid-mediated pathway [31].In this study, multi-omics analysis revealed that pathways involved in phenylpropanoid, f lavonoid, and isoquinoline alkaloid biosynthesis were rapidly differentially expressed during early F. oxysporum infection (2 dpi) (Figs 1B, 2D  and 3B).The f lavone and f lavonol biosynthesis pathways were also differentially expressed during the interaction between L.

Figure 8.
LrERF4 enhanced the resistance of 'Siberia' to F. oxysporum by upregulating the biosynthesis of lignin, total f lavonoids, ferulic acid, phlorizin, and quercetin.A ERF4 expression patterns were significantly different in L. regale and 'Siberia' during F. oxysporum infection.B Susceptibility of lily 'Siberia' scales to F. oxysporum reduced after LrERF4 overexpression, which was evidenced by the decreased rot area compared with the control.C Expression of genes related to biosynthesis of lignin, total f lavonoids, ferulic acid, phlorizin, and quercetin was upregulated in 'Siberia' scales undergoing LrERF4 overexpression.D Contents of lignin, total f lavonoids, ferulic acid, phlorizin, and quercetin in 'Siberia' scales increased after LrERF4 overexpression.The t-test was used to reveal statistical differences between different samples ( * * P < 0.01).Different lowercase letters indicated statistically significant differences between two samples (P < 0.05).regale and F. oxysporum (4 dpi), as were the phenylpropanoid, f lavonoid, and isoquinoline alkaloid biosynthesis pathways (Figs 1C, 2E, and 3C).In the late F. oxysporum infection stage (6 dpi), the phenylpropanoid and f lavonoid biosynthesis pathways were still differentially expressed according to the multi-omics data (Figs 1D, 2F, and 3D).These pathways, along with glutathione metabolism, oxidative phosphorylation, plant hormone signal transduction, and MAPK signaling, were significantly altered by F. oxysporum infection (Figs 1B-D and 2D-F).These results suggested that phenylpropane metabolism, hormone signaling, and MAPK signaling contribute to the resistance of L. regale to fusarium wilt.
In addition to phenylpropanes, other secondary metabolites, such as monoterpenoids and isoquinoline alkaloids, play important roles in plant defense responses [32,33].The biosynthesis of monoterpenoids was found to increase during Ophiostoma bicolor infection in Picea koraiensis [34].A monoterpenoid called citral enhanced the resistance of N. benthamiana to tobacco mosaic virus [35].Metabolomics analysis has shown that certain metabolites, like l-dopa and tyramine, accumulate more in resistant T. aestivum compared with a susceptible strain during Blumeria graminis infection [36].Thalicfoetine isolated from Thalictrum foetidum was found to inhibit Bacillus subtilis growth [37].The multi-omics analysis conducted in the current study revealed that, in addition to phenylpropane production, the biosynthesis of monoterpenoids (geranylgeranyl diphosphate) and isoquinoline alkaloids (dopamine) might play an important role during the interaction between L. regale and F. oxysporum (Figs 2-4 and Supplementary Data Figs S1 and S2).These results ref lected that the regulation by secondary metabolism of resistance of L. regale to fusarium wilt is complex, and indicate that the contributions of these regulatory networks to the defense response against F. oxysporum need future studies.
Pathogenic fungal infections in plants lead to the accumulation of resistance-related metabolites such as phenylpropanes.Phenylpropanes derived from phenylalanine are mainly divided into the precursors of lignin and lignans, f lavonoids and Figure 9. Resistance of L. regale was downregulated after LrERF4 RNAi, accompanied by decreased lignin, total f lavonoid, ferulic acid, phlorizin, and quercetin contents.A Resistance of L. regale scales to F. oxysporum was damaged after expression of LrERF4 RNAi vector, which was supported by the increased rot area of F. oxysporum infection.B Expression of genes related to lignin, total f lavonoids, ferulic acid, phlorizin, and quercetin biosynthesis was suppressed in L. regale scales by LrERF4 RNAi.C Accumulation of lignin, total f lavonoids, ferulic acid, phlorizin, and quercetin in L. regale scales decreased after LrERF4 RNAi.The t-test was used to reveal statistical differences between samples ( * P < 0.05, * * P < 0.01).Different lowercase letters indicated statistically significant differences between two samples (P < 0.05).hydroxycinnamic acid, which play essential roles associated with plant development and disease resistance [38].The metabolome analysis in this study revealed an increase in phenylpropanoid content in F. oxysporum-resistant L. regale at different time points (Fig. 4).Further analysis confirmed the accumulation of phenylpropanes, such as lignin, f lavonoids, ferulic acid, phlorizin, and quercetin, during F. oxysporum infection of L. regale, with 'Siberia' serving as a control (Fig. 5).During F. oxysporum infection, the abundance of these phenylpropanes changed considerably less in 'Siberia' than in L. regale.Total f lavonoids, ferulic acid, phlorizin, and quercetin showed inhibitory effects on F. oxysporum growth and the expression of pathogenicity-related genes (Fig. 6).Therefore, it appears that the activation of phenylpropanoid metabolism is crucial for the QTR mechanism that protects L. regale from fusarium wilt.
The phenylpropanoid metabolism-based defense response to pathogens has been partially characterized in plants.In Oryza sativa, MYB30 activates lignin biosynthesis genes, resulting in increased lignin accumulation and resistance to Magnaporthe oryzae [39].The application of certain phenylpropanes, such as kaempferide and apigenin, enhances the resistance of wheat to F. graminearum [40].Chlorogenic acid, a hydroxycinnamic acid derivative, restricts Fusarium fujikuroi growth by inducing a burst of reactive oxygen species [41].Flavonoids can disrupt the plasma membrane of pathogens, disrupt mitochondrial functions, and inhibit cell wall formation, all of which have detrimental effects on pathogen growth [42].In this study, the total f lavonoids, ferulic acid, phlorizin, and quercetin extracted from L. regale exhibited inhibitory effects on F. oxysporum growth and the expression of pathogenicity-related genes (Fig. 6).These results indicated that the phenylpropanes that accumulate in L. regale contribute to the physical and chemical defense barriers during F. oxysporum infection.
TFs play a crucial role in the transcriptional reprogramming of plants during pathogen infections, inf luencing the accumulation of resistance-related metabolites.MdMYB1r1 positively regulates the expression of MdBGLU40, promoting the synthesis of coumarin, which inhibits Cytospora mali infection [43].In Gossypium hirsutum infected with F. oxysporum, VIIIc WRKY TFs activate GhMKK1 expression, leading to the biosynthesis of f lavonoids through an MAPK cascade involving GhNTF6 and GhMYC2 [44].Chrysanthemum morifolium WRKY6-1 negatively regulated chrysanthemum resistance to F. oxysporum by reducing salicylic acid biosynthesis, inducing reactive oxygen species production and inhibiting CmWRKY15-like expression [45].In Vitis davidii, MYB1 upregulates VdSTS2 expression, increasing resveratrol content and resistance to Erysiphe necator [46].In this study, our analysis of the transcriptional regulatory network suggested that ERF, bZIP, and WRKY TFs might regulate phenylpropanoid biosynthesis (Fig. 7).Through reverse genetics, we found that LrERF4 regulates the accumulation of phenylpropanes during F. oxysporum infection (Figs 8 and 9).Transient overexpression of LrERF4 protected 'Siberia' from F. oxysporum by increasing the contents of lignin, total f lavonoids, ferulic acid, phlorizin, and quercetin.Conversely, RNAi-mediated suppression of LrERF4 expression in L. regale prevented the accumulation of these phenylpropanes and decreased resistance to F. oxysporum.Therefore, LrERF4 positively regulates the resistance of L. regale to fusarium wilt by modulating the biosynthesis of defense-related phenylpropanes.

Conclusion
This study analyzed the transcriptome, proteome, and metabolome profiles of L. regale in response to F. oxysporum.The integrated multi-omics analysis revealed the importance of phenylpropanoid metabolism in protecting L. regale from F. oxysporum.Moreover, some f lavonoids and hydroxycinnamic acids were more abundant in resistant L. regale compared with susceptible 'Siberia'.These phenylpropanes showed inhibitory effects on F. oxysporum growth, conidial germination, and pathogenicityrelated gene expression, suggesting their contribution to the chemical and physical defenses of L. regale.Furthermore, LrERF4 was identified as a key TF regulating the biosynthesis of defenserelated phenylpropanes in L. regale.The findings highlight the importance of phenylpropanoid metabolism in the high-level resistance of L. regale to Fusarium wilt.

Lily materials and F. oxysporum inoculation
Lilium regale and Oriental hybrid lily 'Siberia' plants were cultivated in a greenhouse.'Siberia' plants showing symptoms of fusarium wilt were used to isolate F. oxysporum, which was then preserved in our laboratory [47].Prior to use, F. oxysporum was cultured on potato dextrose agar (PDA).
Fusarium oxysporum conidial suspension (5 × 10 6 conidia/mL) was used to infect healthy L. regale plants by dipping the roots.Plants subjected to mock inoculation with sterile water were used as controls.At 6 h post-inoculation, the plants were transferred to sterile soil.The roots were collected at 2, 4, and 6 dpi for omics analysis, with three biological replicates.

RNA isolation, sequencing, and transcriptome analysis
Total RNA was extracted from L. regale roots (sample collection is described in section 'Lily materials and F. oxysporum inoculation') to isolate mRNA.The mRNA was fragmented, and a cDNA library was synthesized and sequenced using the Illumina NovaSeq™ 6000 system (LC Sciences, USA).
High-quality clean sequencing reads were selected using Cutadapt (v.1.9.1), and de novo assembly of the high-quality reads was performed using Trinity (v.2.4.0) [48].The longest sequence in each cluster was selected as the unigene.The unigenes were aligned with sequences in the KEGG and eggNOG databases using DIAMOND [49].Salmon (v.0.8.2) [50] was used to analyze unigene expression levels.Significant DEGs (fold change >1.5 or <0.5 relative to control levels and P < 0.05) were obtained using the R package edgeR [51].KOBAS was used for KEGG enrichment analysis of DEGs.

Protein extraction, annotation, and proteome analysis
The concentration of protein extracted from L. regale roots (sample collection is described in section 'Lily materials and F. oxysporum inoculation') was determined using a BCA kit (Beyotime, China).The protein samples were then treated with dithiothreitol and iodoacetamide.After precipitation and washing with acetone, peptides were obtained by trypsin digestion.The peptides were labeled using TMTpro™ 16plex Label Reagent (Thermo, USA) and pre-separated using the LC-20AD system (Shimadzu, Japan).The labeled peptides were analyzed as described in Liu et al. [52].
Proteome Discoverer software was used for protein identification, which was subsequently annotated in the KEGG database.
Significant DEPs (fold change >1.5 or <0.5 relative to control levels and P < 0.05) were mapped to KEGG pathways using KEGG Mapper.Enriched pathways were identified using a two-tailed Fisher's exact test.

Metabolite extraction, determination, and metabolome analysis
Metabolites were extracted from L. regale roots (sample collection is described in section 'Lily materials and F. oxysporum inoculation') for UPLC-MS/MS analysis.Separation and detection of metabolites were performed according to the methods described in Liu et al. [52].The acquired MS data were processed to obtain retention time and exact mass for identification of ions.Metabolites were annotated by matching the m/z values of samples with those in the KEGG database.
Metabolome data were statistically analyzed as described in Liu et al. [53].The DAMs (fold change >1.5 or <0.5 relative to control levels) were classified based on phytochemical compounds in the KEGG database.KOBAS was used to identify enriched pathways.

Inoculation and collection of L. regale and 'Siberia' scales for HPLC analysis of metabolite contents
Differences in phenylpropane accumulation between 'Siberia' and L. regale were examined.Fusarium oxysporum conidial suspension (1 × 10 6 conidia/mL) was used to infect leaves of 'Siberia' and L. regale, and fungal infection damage was recorded at 6 dpi.Additionally, healthy L. regale and 'Siberia' plants were infected with F. oxysporum conidial suspension (5 × 10 6 conidia/mL), and scales were harvested at 2, 4, and 6 dpi for analysis of lignin, total f lavonoid, ferulic acid, phlorizin, and quercetin contents as described in section 'Analyses of lignin, total f lavonoid, ferulic acid, phlorizin, and quercetin contents'.

Antifungal effects of total flavonoids, luteolin, and quercetin extracted from L. regale
To obtain f lavonoids, dried L. regale scale samples (1 g) were added to a 25-mL NaOH solution (pH 10) for a 45-min ultrasonic extraction at 60 • C.After cooling the supernatant the extract was harvested through centrifugation, and the pH of the solution was adjusted to 3.5 before incubation for 12-16 h.After centrifugation the supernatant was removed, and the sediment was washed with ether and distilled water.Total f lavonoids were dissolved in methanol after the sediment had been dried to a constant weight.
After pre-culturing F. oxysporum on PDA to a fungal colony diameter of 2 cm, four sterile paper discs were evenly placed on the periphery of F. oxysporum colonies, and the angle between two neighbouring discs was 90 • .The inhibitory effects of the total f lavonoid, ferulic acid, phlorizin, and quercetin solutions on F. oxysporum mycelial growth were examined in separate plates.More specifically, the following reagents were added to the discs: 40 μL f lavonoid solution (5, 10, and 15 mg/mL), 40 μL ferulic acid solution (5, 10, and 15 mg/mL), 40 μL phlorizin solution (5, 10, and 15 mg/mL), or 40 μL quercetin solution (5, 10, and 15 mg/mL).We added 40 μL methanol to the control disc in each plate.Fungal growth was recorded after incubation at 28 • C for 4 days using a digital camera (Nikon, Japan), and then Photoshop 7.0 was used to analyze the effects of various solutions on fungal growth.The antifungal assay was repeated three times.
After pre-culturing F. oxysporum on PDA for 5 days, an F. oxysporum conidial suspension was prepared using potato dextrose broth.The conidial concentration was adjusted to 1 × 10 5 conidia/mL, after which 200 μL of conidial suspension was mixed with 40 μL total f lavonoid, ferulic acid, phlorizin, or quercetin solution (initial concentration 10 mg/mL).A mixture comprising the conidial suspension and 40 μL methanol served as the control.The germination of F. oxysporum conidia after incubation at 28 • C for 24 h was observed using an inverted microscope (Leica, Germany).The conidial germination assay was repeated three times.
In addition, F. oxysporum mycelia were collected from PDA plates after the 4-day treatment with the f lavonoid, ferulic acid, phlorizin, quercetin, and methanol (control) solutions to determine the expression of pathogenicity-related genes in F. oxysporum by qPCR.The 2 −ΔΔCt method was employed to obtain the values of gene expression with an internal reference gene (F.oxysporum βactin, GenBank No. XM_031177127).The qPCR primers are listed in Supplementary Data Table S2, and this experiment included three biological replicates and three technical replicates.

Transcriptional regulatory network analysis
Unigene sequences were aligned with the TF sequences from O. sativa and A. thaliana in the PlantDB database to identify TFs.The DETFs (fold change >1.5 or <0.5 relative to control levels and P < 0.05) were classified using the PlantDB database.The OmicStudio tools were used to construct transcriptional tory networks for the DETFs and DAMs.Significant regulatory relationships between DETFs and DAMs were determined based on a ρ value >0.70 or <−0.70 and P < 0.05.

Gene function analysis of LrERF4
To analyze the expression patterns of ERF4 in L. regale and 'Siberia', their roots were collected after F. oxysporum (5 × 10 6 conidia/mL) inoculation for 2, 4, and 6 days.Total RNA was extracted using the TRIGene kit (Genstar, China), and cDNA was synthesized for qPCR analysis with ERF4-specific primers (Supplementary Data Tables S3 and S4).The expression values of ERF4 were calculated using the 2 −ΔΔCt method, with glyceraldehyde-3-phosphate dehydrogenase genes of L. regale and 'Siberia' as internal reference genes.
The LrERF4 open reading frame (supplementary material 'The cDNA sequence of L. regale and 'Siberia' lily ethylene-responsive transcription factor 4') was cloned and used to construct pCAMBIA2300S-LrERF4 overexpression vector and pHellsgate2-LrERF4 RNA interference vector according to the methods in Su et al. [54].The recombinant vectors were insterted into A. tumefaciens for genetic transformation of 'Siberia' and L. regale scales.
The scales and leaves of L. regale and 'Siberia' showed clear disease symptoms at 6 dpi.Meanwhile, the efficiency of transient RNAi/overexpression technology was high at 48-72 h after the A. tumefaciens-mediated transformation.Therefore, the scale samples were collected after pCAMBIA2300S-LrERF4 and pHellsgate2-LrERF4 were transiently expressed for 72 h to analyze the metabolite contents and gene expression levels as described in sections 'Analyses of lignin, total f lavonoid, ferulic acid, phlorizin, and quercetin contents' and 'Antifungal effects of total f lavonoids, luteolin, and quercetin extracted from L. regale', respectively.The primers for the qPCR analysis (Supplementary Data Tables S3  and S4) were designed according to the L. regale and 'Siberia' transcriptome data.Scales transformed with empty vectors were used as controls.In addition, the L. regale and 'Siberia' scales undergone transient RNAi/overexpression for 72 h were further inoculated with F. oxysporum, and at 6 dpi, all samples were harvested for evaluation of resistance and gene expression.The disease symptoms on the transgenic L. regale and 'Siberia' scales were recorded using a stereomicroscope (Olympus, Japan).These experiments included three biological replicates.

Analyses of lignin, total flavonoid, ferulic acid, phlorizin, and quercetin contents
Lily scales (sample collection is described in sections 'Inoculation and collection of L. regale and 'Siberia' scales for HPLC analysis of metabolite contents' and 'Gene function analysis of LrERF4') were dried to a constant weight.The content of lignin was determined as in the protocol described by Wadenbäck et al. [55].Flavonoids were extracted from 0.01 g dried scale samples using ethanol solution and then mixed with Al(NO 3 ) 3 solution under alkaline conditions.The content of total f lavonoid was determined using a multimicroplate reader (SpectraMax M2, USA) with absorbance at 470 nm.Rutin (Yuanye, China) was used to construct a standard curve for determination of total f lavonoids.These experiments included three biological replicates.
To determine the ferulic acid, phlorizin, and quercetin contents, 0.5 g dried scale samples were added to a 15-mL 60% CH 3 OH solution for a 90-min ultrasonic extraction at 60 • C. The extract was centrifuged and then concentrated using a rotary evaporator (N-1300, Eyela, Japan).The concentrated solution was used for HPLC analysis by an LC-40A system equipped with a Unitary C18 column (Shimadzu, Japan).There were three biological replicates included in this experiment.

Statistical analysis
Statistical differences in ERF4 expression level and contents of lignin, total f lavonoids, ferulic acid, phlorizin, and quercetin in L. regale and 'Siberia' after F. oxysporum inoculation were calculated by a multiple t-test.Student's t-test was used to analyze the antifungal effects of the total f lavonoids, ferulic acid, phlorizin, and quercetin on F. oxysporum growth.Student's t-test was also used to analyze the expression levels of pathogenicity-related genes after treatment with the different metabolites.The expression levels of genes related to LrERF4 and phenylpropane biosynthesis as well as the lignin, total f lavonoid, ferulic acid, phlorizin, and quercetin contents in L. regale after LrERF4 RNAi and in 'Siberia' after LrERF4 overexpression were also determined by Student's t-test.

Figure 1 .
Figure 1.Transcriptome analysis of L. regale roots after inoculation with F. oxysporum.A KEGG enrichment analysis of total unigenes.B, C, D KEGG enrichment of differentially expressed genes at 2, 4, and 6 dpi, respectively.Bubble size ref lects the number of genes annotated in a specific KEGG pathway, and color signifies the P-value.

Figure 2 .
Figure 2. Proteome analysis of L. regale roots after inoculation with F. oxysporum.A, B, C Volcano plots of DEPs at 2, 4, and 6 dpi, respectively.Proteins with up-and downregulated expression profiles are marked with red and blue dots, respectively.D, E, F KEGG enrichment analysis of differentially expressed proteins at 2, 4, and 6 dpi, respectively.Bubble size ref lects the number of genes annotated in a specific KEGG pathway, and color signifies the P-value.

Figure 3 .
Figure 3. Metabolome analysis of L. regale roots after inoculation with F. oxysporum.A KEGG phytochemical compound classification of DAMs.B, C, D KEGG enrichment analysis of differentially accumulated metabolites at 2, 4, and 6 dpi, respectively.Bubble size ref lects the number of genes annotated in a specific KEGG pathway, and color signifies the P-value.

Figure 4 .
Figure 4. Multi-osmic integrative analysis of phenylpropanoid biosynthesis in L. regale during F. oxysporum infection.The black line represents a common pathway for phenylpropanoid biosynthesis.The blue line represents synthesis of hydroxycinnamic acids.The light green line represents coumarin biosynthesis.The red line represents lignin/lignan biosynthesis.The dark green line represents stilbene biosynthesis.The yellow line represents f lavonoid biosynthesis.The full names of catalytic enzymes can be found in Supplementary Data TableS1.

Table 1 .Figure 5 .Figure 6 .
Figure5.Accumulation of lignin, total f lavonoids, ferulic acid, phlorizin, and quercetin in L. regale scales was greater and more rapid than in 'Siberia' during F. oxysporum infection.A L. regale had higher resistance to F. oxysporum than lily 'Siberia'.B HPLC chromatograms of ferulic acid, phlorizin, and quercetin in L. regale and 'Siberia' after inoculation with F. oxysporum.C Accumulation of ferulic acid, phlorizin, and quercetin in L. regale was higher than in 'Siberia'.D Lignin and total f lavonoid contents in L. regale were higher than in 'Siberia'.Different lowercase letters indicated statistically significant differences between two samples (P < 0.05).

Figure 7 .
Figure 7. Transcription regulatory network analysis of phenylpropanes responding to F. oxysporum.A Identification of differentially expressed transcription factors at 2, 4, and 6 dpi.B Transcription regulator network of phenylpropanes according to multi-omic integrative analysis.