Protection of naringenin chalcone by a pathogenesis‐related 10 protein promotes flavonoid biosynthesis in Marchantia polymorpha

Summary

• Pathogenesis‐related (PR) proteins are diverse stress‐ or pathogen‐induced proteins. Some are associated with specialised metabolism, including proposed functions for anthocyanin biosynthesis. However, data are limited to a few angiosperm species, and the mode(s) of action are uncertain. Using the liverwort Marchantia polymorpha (Marchantia), we examined whether pathogenesis‐related 10 (PR10) contributes to flavonoid biosynthesis in other land plant lineages and investigated its mode of action.

• Marchantia produces two major flavonoid types: flavones and the pigment auronidin. MpPR10.5 is a target of the auronidin regulator MpMYB14; therefore, Mppr10.5 mutants were generated using CRISPR/Cas9 and analysed for transcript abundance (via RNA sequencing) and for metabolite content. Recombinant MpPR10.5 protein was used for metabolite binding and stabilisation assays.

• Mppr10.5 mutants had reduced auronidin and flavone content, demonstrating that MpPR10.5 promotes flavonoid biosynthesis. Flavone and auronidin biosynthesis share a single flavonoid intermediate, naringenin chalcone (NC), suggesting MpPR10.5 acts on this compound. MpPR10.5 protein binds NC strongly (micromolar affinity), preventing spontaneous self‐cyclisation in vitro. Several phenylpropanoid and flavonoid genes were downregulated in Mppr10.5 and Mpchalcone isomerase‐like plants.

• This suggests PR10 proteins promote flavonoid biosynthesis by selectively binding unstable intermediates (NC), protecting them from degradation or undesirable nonenzymatic conversions and facilitating their transport to subsequent pathway steps.

Introduction

Flavonoid biosynthesis, a branch of the phenylpropanoid pathway, is a central specialised metabolite pathway of land plants. Flavonoids are key to how plants interact with the terrestrial environment, helping in tolerance of diverse abiotic stresses and attacks from pests and pathogens and providing pigmentation to flowers, fruit, seeds and vegetative tissues (Agati & Tattini, 2010; Cheynier et al., 2013; Landi et al., 2015; Davies et al., 2018). The flavonoid pathway commences with the production of chalcones by the polyketide synthase (PKS) enzyme CHALCONE SYNTHASE (CHS). The subsequent pathway steps lead to diverse, distinct flavonoid classes, including flavones, flavonols, isoflavonoids, aurones, auronidins, anthocyanins and proanthocyanidins (condensed tannin). The core biosynthesis genes for the major flavonoid classes have been extensively characterized across many plant species (Yonekura‐Sakakibara et al., 2019; Ferreyra et al., 2021; Davies et al., 2022). However, recent studies have brought to light possible important roles in flavonoid production for additional nonenzymatic proteins, the modes of action of which are generally not understood. The best studied is CHALCONE ISOMERASE‐LIKE (CHIL), which is necessary for efficient flavonoid biosynthesis in diverse land plant lineages (Morita et al., 2014; Berland et al., 2019). CHIL can bind with and alter the specificity of CHS and at least some other PKS enzymes, notably STILBENE SYNTHASE (Waki et al., 2020). As the PKSs direct substrate flow into the alternative phenylpropanoid pathway sections, such as flavonoids, stilbenes, dihydrochalcones and bibenzyls, CHIL acts at a key biosynthetic step where enzyme specificity and efficacy are particularly important.

Relatively unexamined candidates for other nonenzymatic proteins involved in flavonoid biosynthesis are some members of the pathogenesis‐related 10 (PR10) sub‐class of pathogenesis‐related proteins (PR proteins). Pathogenesis‐related proteins represent diverse protein families, often with unknown functions, whose corresponding genes are induced in response to pathogen attacks, environmental stresses or certain physiological processes (van Loon, 1985). Following identification of PR10 in parsley (PcPR10 of Petroselinum crispum; Somssich et al., 1986) and as the major allergen present in birch pollen (Bet v1 of Betula spp.; Breiteneder et al., 1989), PR10 genes have been identified in various gymnosperm and angiosperm species (Liu & Ekramoddoullah, 2006). Yet, no conserved role for PR10 family members has been identified, although they have commonly been linked to defence against pathogens (Park et al., 2004; Andrade et al., 2010; Fan et al., 2015; Longsaward et al., 2023). An association with specialized metabolite pathways has also often been noted, including for flavonoids (Hjernø et al., 2006; Muñoz et al., 2010), sporopollenin (Huang et al., 1997), benzylisoquinoline alkaloids (Hagel & Facchini, 2013), Amaryllidaceae alkaloids (Singh et al., 2018), iridoids (Lichman et al., 2020) and thebaine (Chen et al., 2018).

The mode of action of PR10 proteins is unresolved. Some PR10 proteins have been found to directly catalyse specialized metabolite biosynthetic reactions (Hagel & Facchini, 2013; Chen et al., 2018; Singh et al., 2018), while other PR10 proteins may guide pathway stereochemistry and product selectivity (Lichman et al., 2020). However, the ability to bind small metabolites may be key to all PR10 functions. Pathogenesis‐related 10 proteins have a compact, stable conformation, characterised by a seven‐stranded antiparallel β‐sheet wrapped around an α‐helix, along with two additional short helices (Hoffmann‐Sommergruber et al., 1997; Wen et al., 1997; Handschuh et al., 2007; Lebel et al., 2010; Fernandes et al., 2013). These elements collectively form a hydrophobic cavity that is the site of ligand binding (Aglas et al., 2020; Morris et al., 2021). A glycine‐rich loop L4 is preserved even in distant homologues among seed plants, and thus recognised as a signature motif (Fernandes et al., 2013). A diverse array of molecules has been shown to be able to be bound by different PR10 proteins, ranging from invading viral RNAs to varied specialized metabolites (Park et al., 2004; Chadha & Das, 2006; Sliwiak et al., 2016; Aglas et al., 2020). Notably, the PR10 of strawberry (Fra a of Fragaria × ananassa) can bind flavonoids, including quercetin‐3‐O‐glucuronide, myricetin and (+)‐catechin (Casañal et al., 2013). Additionally, Fra a genes are downregulated in an anthocyanin‐lacking strawberry line (Hjernø et al., 2006), and transient RNAi‐mediated silencing of the Fra a genes in strawberry fruits reduced amounts of anthocyanins and upstream metabolites and the transcript abundance for PHENYLALANINE AMMONIA LYASE (PAL) and CHS. Based on the available data, there are various proposals for how PR10 might contribute to flavonoid biosynthesis: chemical chaperones assist intercellular transportation of compounds (Casañal et al., 2013), participation in multi‐protein flavonoid biosynthetic complexes (metabolons) to restrict diffusion and protect unstable substrates (Morris et al., 2021), finely adjusting metabolic flux by differential binding of intermediates and channelling them between enzymes (Casañal et al., 2013) and transcriptional regulation of flavonoid biosynthesis genes (Muñoz et al., 2010). Thus, the mode of action of PR10 in the flavonoid pathway remains equivocal.

The models used for elucidating the genetics and biochemistry of the flavonoid pathway have traditionally been a relatively small number of angiosperm species. However, recently Marchantia polymorpha (hereafter, Marchantia), a member of the liverwort lineage of bryophytes, has emerged as a powerful system for addressing questions of flavonoid genetics and for identifying which aspects of flavonoid biosynthesis and regulation are conserved across land plants (Albert et al., 2018; Clayton et al., 2018; Kubo et al., 2018; Berland et al., 2019; Bowman et al., 2022; Zhu et al., 2023; Zhou et al., 2024). The two main flavonoid classes produced in Marchantia are the colorless flavones, which assist with tolerance of UVB light exposure, and the red pigment auronidin (Albert et al., 2018; Clayton et al., 2018; Kubo et al., 2018; Berland et al., 2019). The phenylpropanoid enzymes that convert phenylalanine to the flavonoid precursor p‐coumaroyl CoA are conserved across all land plants, including Marchantia, being PAL, CINNAMATE 4‐HYDROXYLASE (C4H) and 4‐COUMAROYL CoA LIGASE (4CL). From there, the pathway in Marchantia branches into two trajectories: one leading to flavonoids and the other to bibenzyls, a group of antimicrobial compounds for which > 125 different structures have been reported from liverworts (Asakawa et al., 2022). The initial bibenzyl‐specific step is catalysed by the PKS STILBENECARBOXYLATE SYNTHASE and a POLYKETIDE REDUCTASE. For flavonoids, the naringenin chalcone (NC) formed by CHS/CHIL is subsequently transformed into flavones by the actions of CHALCONE ISOMERASE (CHI) and FLAVONE SYNTHASE (FNS) or auronidins by AUREUSIDIN SYNTHASE (AUS) and uncharacterized enzymes (Berland et al., 2019; Davies et al., 2020) (Fig. 1). The biosynthesis of auronidin is transcriptionally activated by the R2R3MYB transcription factor MpMYB14 (Albert et al., 2018; Kubo et al., 2018).

Fig. 1.

Schematic of the phenylpropanoid biosynthetic pathway in Marchantia polymorpha. 4CL, 4‐COUMAROYL COA LIGASE; AUS, AUREUSIDIN SYNTHASE; C4H, CINNAMATE 4‐HYDROXYLASE; CHI, CHALCONE ISOMERASE; CHIL, CHALCONE ISOMERASE‐LIKE; CHS, CHALCONE SYNTHASE; DBR, double bond reductase; FNS, FLAVONE SYNTHASE; PAL, PHENYLALANINE AMMONIA LYASE; PKR, POLYKETIDE REDUCTASE; STCS, STILBENECARBOXYLATE SYNTHASE.

Previous studies on auronidin biosynthesis in Marchantia identified MpPATHOGENESIS‐RELATED10.5 (MpPR10.5/Mp8g00860) as being tightly associated with MpMYB14 expression. MpPR10.5 transcript levels increased over 30‐fold in 35S:MYB14 plants and decreased over sevenfold in Mpmyb14 mutants compared to wild‐type(WT) (Albert et al., 2018; Berland et al., 2019). Furthermore, in an independent study conducted by Kubo et al. (2018), MpPR10.5 was significantly upregulated (> 50‐fold) in their MpMYB14 overexpression line, compared to the WT control (full dataset available in the Marpolbase Expression database: https://mbex.marchantia.info/diffexp). This suggested that MpPR10.5 may have a role in auronidin biosynthesis and that Marchantia could be an excellent model system to elucidate the mode of action of PR10 proteins in the flavonoid pathway.

In this study, using a multidisciplinary combination of reverse genetics, RNAseq analysis and in vitro protein assays, we establish that MpPR10.5 promotes flavonoid biosynthesis in Marchantia, identify a novel feedback regulation mechanism that downregulates phenylpropanoid pathway gene expression and propose a mode of action through binding of MpPR10.5 protein to specific flavonoid pathway intermediates.

Materials and Methods

Plant lines and growth conditions

The WT Marchantia polymorpha L. subsp. ruderalis accessions Sey‐1 (male ♂) and Aud‐2 (female ♀) and general plant growth conditions were as reported in Albert et al. (2018). Plants were propagated through at least two generations of gemmae (> G2), grown from clonal gemmae from each line. Gemmae were germinated on mixed cellulose ester (MCE) membrane filter discs (pore size 0.8 μm, diameter 25 mm; Membrane Solutions, Auburn, AL, USA) on complete media (0.5 × Gamborg's B5 medium (Duchefa Biochemie, Haarlem, the Netherlands) 1% (w/v) sucrose, 1% (w/v) agar) under standard culture conditions of 25°C, 16 h photoperiod and 30 μmol m⁻² s⁻¹ light intensity provided by cool fluorescent tubes.

Stress induction conditions

Gemmae were germinated and maintained on complete medium under the standard conditions described above for 3 wk and then transferred onto either fresh complete medium (control plants) or the nutrient deficiency medium (1% (w/v) sucrose, 1% (w/v) agar) for a further 2 wk. Plants were grown under standard culture conditions. Four clonal plants/biological replicates were used in each treatment.

Identification and characterisation of Marchantia PR10 proteins

PR10‐encoding genes were identified in the Marchantia genome (MpTak v.6.1; https://marchantia.info/) using a Blastp search with the birch Bet v 1 protein sequence (GenBank ID: CAB02159.1) as a query (E‐value threshold < 1 × 10⁻³). Candidate genes were further validated by confirming the presence of the Bet v 1 domain (PF00407) in Pfam. The identified protein sequences were analysed using InterPro (https://www.ebi.ac.uk/interpro/) and Smart (http://smart.embl‐heidelberg.de/) tools to confirm the presence of the Bet v 1 domain. The subcellular localization of the proteins was predicted using the DeepLoc 2.0 subcellular localization predictor (https://services.healthtech.dtu.dk/services/DeepLoc‐2.0/) (Thumuluri et al., 2022).

Mutagenesis and transgenic overexpression

Overexpression constructs (CaMV35S promoter) were generated via LR recombination of the MpPR10.5 coding sequence (synthesised by Genscript, Piscataway, NJ, USA) into the pNWA101 vector and transformed into spores with 8 mg l⁻¹ G418 selection (Thermo Fisher Scientific, Auckland, New Zealand) (Albert et al., 2018).

CRISPR/Cas9 constructs were generated by synthesising a gRNA‐tRNA array targeting exon 1 of Mp8g00860/Mapoly0064s0111 and transformed into spores (Jibran et al., 2024). Mutants were screened using gene‐specific primers (ZYF146 and ZYF147; Supporting Information Table S1), propagated through gemmae (G1 stage), and re‐screened to confirm nonchimerism. The Marchantia chalcone isomerase (Mpchi; Mp6g05290.1), chalcone isomerase‐like (Mpchil; Mp5g01410.1) and the double mutants (Mpchi‐chil) were from previous studies (Clayton et al., 2018; Berland et al., 2019).

RNAseq differential gene expression analysis

WT♂, Mpchi, Mpchil, Mpchi‐chil, Mppr10.5, MpPR10.5‐OE (35S _pro : PR10) and empty vector (EV) control lines grown under control or stress conditions and sampled for RNA analysis (three biological replicates) (Albert et al., 2018). Libraries (150 bp pair‐ended) were prepared and sequenced with a NovaSeq6000 platform at Novogene (HK) Company Limited (Hong Kong). The raw sequence reads were cleaned and trimmed using SortMeRNA 2.1 and Trimmomatic 0.36, respectively and then mapped to the MpTak1v.6.1 genome using Salmon 0.9.1. Comparative analysis on differentially expressed genes was performed using DESeq2, with the transcript abundance expressed as the baseMean value generated by the DESeq2 R package. Nomenclature for gene family members (e.g. MpPAL1, MpPAL2) follows that of MarpolBase, as registered in the Marchantia Nomenclature Database (https://marchantia.info/nomenclature/).

Quantitative RT‐PCR

RNA samples used in RNAseq were analysed by quantitative reverse transcription polymerase chain reaction (qRT‐PCR) (three biological replicates), using gene‐specific primers (Table S1) normalized to the reference gene Mp1g01300.1, as described (Albert, 2015). Mp1g01300.1 is stably expressed under both nutrient‐stressed and nutrient‐unstressed conditions, as well as across different tissues, according to our RNAseq data (Table S2).

Chemical analysis

WT♂, Mppr10.5, MpPR10.5‐OE and EV control lines grown under control or stress conditions were sampled for metabolite analysis (four biological replicates). Powdered freeze‐dried samples (10 mg) were mixed with 1 ml of MeOH : water : formic acid (80 : 19 : 1, v/v/v) and shaken for 2 h at 20°C. After centrifugation at 10 000  g for 10 min, the supernatant was collected and evaporated to dryness. Phenolic compounds were resuspended in 1 ml of MeOH : water (20 : 80) and filtered (0.45 μm) before analyses. For analysis of auronidins, an aliquot of each sample was blown down to dryness and reconstituted in 5 : 3 : 92 (v/v/v) of acetonitrile : formic acid : water.

Identification of phenolic compounds was first performed via LC‐MS/MS using an LTQ linear ion trap mass spectrometer fitted with an electrospray ionization (ESI) interface (Thermo Fisher Scientific, San Jose, CA, USA) coupled to an Ultimate 3000 UHPLC and photodiode array (PDA) detector (Dionex, Sunnyvale, CA, USA). Phenolic compounds (3 μl) were separated on a Poroshell 120 SB‐C18 column (75 × 2.21 mm i.d.; 2.7 μm particle size) (Agilent, Torrance, CA, USA), maintained at 35°C. Solvents were (A) water with 0.1% formic acid and (B) acetonitrile with 0.1% formic acid, and the flow rate was 350 μl min⁻¹. The gradient was as follows: 0–2 min, 10% B; 5 min, 30% B; 15 min, 60% B; 16–20 min, 100% B; 20 min, 10% B. MS data were acquired in the negative ion mode for flavones and bibenzyls. The ESI voltage, capillary temperature, sheath gas pressure and sweep gas were set at −10 V, 200°C, 40 psi, and 10 psi, respectively. MS analysis of auronidins was acquired in the positive ion mode. The ESI voltage, capillary temperature, sheath gas pressure and sweep gas were set at 5 V, 275°C, 35 psi and 10 psi, respectively. Relative quantification of the phenolic compounds was performed as described in Andre et al. (2012) with simultaneous monitoring set at 254, 280, 320 and 520 nm for quantification.

Recombinant protein expression and purification

The nucleotide sequences for MpPR10.5 and PiPR10‐like proteins were codon‐optimised for bacterial expression, with a TEV protease cleavage site added at the 5′‐end and gateway recombination sites at both ends. These constructs were synthesised and cloned into pUC57‐Kan by GenScript (NJ, USA). For expression, MpPR10.5 was cloned into a modified p300‐NT‐DEST vector with a His‐SUMO tag, while PiPR10‐like was cloned into pDEST566 with a His‐MBP tag. Both proteins were expressed in E. coli ‘Rosetta‐Gami 2’ cells induced with 0.5 mM isopropyl‐β‐thiogalactoside in Terrific Broth at 20°C.

Cells were resuspended in 20 mM Tris–HCl, pH 8.5, 500 mM NaCl, 10 mM imidazole and homogenized twice at 15 000 psi using an Emusiflex C3 (Avestin, Ottawa, Canada). Soluble fractions were purified by IMAC using a His‐Trap HP 5 ml column (Cytiva, Marlborough, MA, USA). His‐SUMO‐MpPR10.5 was further purified by size exclusion chromatography on a Superdex 200 HiLoad column (Cytiva), while His‐MBP‐PiPR10‐like was dialysed, cleaved with TEV protease and purified by reverse IMAC and size exclusion on a Superdex 75 HiLoad column (Cytiva). Elution buffers were 20 mM Tris pH 8.5, 200 mM NaCl for His‐SUMO‐MpPR10.5 and 20 mM Tris pH 8.5, 150 mM NaCl, 2 mM DTT for PiPR10‐like. Proteins were concentrated, aliquoted and stored at −80°C. Purity was confirmed by SDS‐PAGE and size exclusion chromatography (Fig. S1). Concentrations were determined by UV absorption at 280 nm using extinction coefficients of 0.654 M⁻¹ cm⁻¹ for His‐SUMO‐MpPR10.5 and 0.852 M⁻¹ cm⁻¹ for PiPR10‐like.

Intrinsic tryptophan fluorescence assays

MpPR10.5 harbors two tryptophan residues (Trp18 and Trp97), both predicted to point towards the internal binding cavity of the protein (Fig. S2B). The His‐SUMO‐MpPR10.5 fusion protein was used for the experiments, having the advantage that the His‐SUMO N‐terminal tag does not carry any tryptophan that could interfere with the assay. Experiments were performed on a FlexStation 3 plate reader (Molecular Devices, San Jose, CA, USA). The excitation wavelength was 280 nm (with cut‐off at 325 nm), and fluorescence spectra were recorded over the 300–500 nm range at 10‐nm intervals, using six flashes/read and the photomultiplier tube set on AUTO. Experiments were done at 22 ± 1°C. Compound stocks at eight concentrations (either 0, 0.4, 1, 2, 4, 8, 14 and 20 mM or 0, 0.1, 0.2, 0.4, 0.8, 1.4, 2 and 4 mM) were prepared in DMSO. After a 10× dilution in the assay buffer (20 mM Tris, pH 8.5, 200 mM NaCl), 10 μl of compounds were dispensed in flat‐bottomed, black 96‐well plates (Greiner Bio‐One, Kremsmünster, Austria), and 90 μl of either His‐SUMO‐MdPR10 at 1.1 μM in assay buffer or assay buffer alone were added to the compounds. Each plate contained six replicates for the protein samples (in columns 2–7) and two replicates for buffer controls (in columns 1 and 8). Plates were incubated for 5 min at 22°C before measurements. After buffer control subtraction, binding curves were obtained by plotting the relative fluorescence measured at 340 nm (|ΔF|/F0) vs compound concentration, where F0 is the fluorescence of the protein without compound (0 μM) and |ΔF| = |F – F0|. GraphPad Prism was used to perform non‐linear regressions and determine dissociation constant (K _d) values using the One site ‐specific binding model.

Naringenin chalcone oxidation assay

Experiments were performed with a NanoDrop 2000c spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Measurements were performed over 30 min at 1 min intervals. Spectra were recorded in the UV–Vis range (190–840 nm). Peaks for naringenin and NC were monitored at 322 and 377 nm, respectively. A 150 μl sample (either assay buffer alone, consisting of 20 mM Tris pH 7.5 or 8.5, 200 mM NaCl, or His‐SUMO‐MdPR10 15 μM in assay buffer) was placed in a UVette cuvette (Eppendorf, Hamburg, Germany), and blank measurements were recorded. An aliquot (1.07 μl) of compound at 1.4 mM in DMSO (i.e. final compound concentration of 10 μM) was added to the cuvette and mixed by pipetting. The cuvette (path length = 1 cm) was promptly placed back in the spectrophotometer before starting the kinetic measurements.

Naringenin chalcone self‐cyclisation assay

Time course experiment was performed at 20°C on NC (150 μM) in buffer (20 mM Tris pH 7.5, 200 mM NaCl) in the presence or absence of His‐SUMO‐MdPR10.5 protein (15 μM). Compound detection was performed using a Waters Acquity UPLC system (Milford, MA, USA) equipped with a detector PDA detector and a mass single quadrupole detector (QDa; Waters, Milford, MA, USA). An aliquot of 2 μl was injected onto an Acquity UPLC BEH C18 column (2.1 × 50 mm, 1.7 μm particle size, Waters). The mobile phases were (A) H₂O with 0.1% formic acid and (B) acetonitrile with 0.1% formic acid. The flow rate was 0.60 ml min⁻¹, and the column temperature was 50°C. The 10 min gradient was as follows: 0–1 min, 10% B; 1–5 min, 10–95% B; 5–7.5 min, 95% B constant; 7.6–10 min, 10% B re‐equilibration time.

Data were recorded in negative ion mode between 100 and 1000 atomic mass units (amu) (cone voltage 5 V) and between 200 and 600 nm on the PDA detector. Naringenin chalcone and naringenin were quantified using selected ion recording at m/z 271 and compared with authentic standards.

Statistical analysis

The data were analysed by a one‐way ANOVA using Genstat v.22 (VSN International, Hemel Hempstead, UK). Comparisons were made among treatment means using Fisher's least significant differences (LSD) test and specifying α = 0.05. Different letters were assigned to significantly different means.

Results

Mp8g00860 is a pathogenesis‐related 10 protein

Previous studies found Mp8g00860 (Mapoly ID: Mapoly0064s0111), among the Marchantia genes most closely associated with auronidin production (Albert et al., 2018; Kubo et al., 2018; Berland et al., 2019). The Mp8g00860 gene contains an open reading frame of 450 bp, from two exons separated by a 261‐bp intron, that encodes a polypeptide of 149 amino acids with a calculated molecular weight of 16.57 kDa. It has a Bet v1/Major latex protein domain (PF00407) and 31.6% overall amino acid identity with the birch Bet V1 protein (CAB02156.1). Based on a predicted structure, Mp8g00860 has the prototypical structure of the Bet v1 protein (Radauer et al., 2008), with a glycine‐rich loop 4, and a secondary structure arrangement of β‐α2‐β6‐α with an antiparallel β‐sheet wrapped around a long C‐terminal α‐helix and a substantial central cavity (Fig. S2). The predicted 3D structure of MpPR10.5 closely resembles the Strawberry Fra a protein, known to be involved in flavonoid biosynthesis (Muñoz et al., 2010) (Fig. S3). Based on its structural features, and with numbering of gene family members reflecting ordering on the Marchantia chromosomes, Mp8g00860 was classified within the PR10 family and is henceforth referred to as MpPR10.5 (Table S3).

Pathogenesis‐related 10 genes appear to be enriched in the Marchantia genome, with a total of 16 candidate PR10 family genes identified (hereafter referred to as MpPR10s; Table S3). While the transcript abundance of MpPR10.5 is closely correlated with MpMYB14 activity and auronidin production, the transcript abundances of other MpPR10 genes were not altered in transgenics with altered MpMYB14 activity (Table S4), which suggests they might contribute to other biosynthetic pathways. Notably, MpPR10.3 and MpPR10.9 have gene expression patterns correlated with the production of flavones, being upregulated in response to UVB (Clayton et al., 2018) and in response to nutrient deficiency (Albert et al., 2018) (Table S5), with the response not being changed in myb14 mutant lines (Table S4).

Loss of MpPR10.5 function in Marchantia reduces flavonoid content

To directly establish whether MpPR10.5 functions in flavonoid biosynthesis, Marchantia lines were generated with mutations in the MpPR10.5 gene using CRISPR/Cas9‐mediated mutagenesis. Four lines were identified with mutations ranging from a 25‐bp insertion to deletions of up to 182 bp (Fig. 2a), which significantly disrupted the protein coding sequence and would thus result in loss of function.

Fig. 2.

Mppr10.5 mutants have reduced flavonoid content. (a) CRISPR/Cas9‐generated MpPR10.5 Marchantia polymorpha mutants. DNA sequence of CRISPR/Cas9‐induced mutations in the MpPR10.5 gene (Mp8g00860). Green text corresponds to the guide RNA sequence, and red text indicates the mutations generated. (b) Phenotype of Marchantia polymorpha plants under nutrient deprivation stress. (c) Quantification of total auronidins, flavones and bibenzyl compounds. Wild type (WT) and mutant MpPR10.5 (Mppr10.5‐1, ‐2, ‐3, ‐4) Marchantia lines were analysed for flavonoid content. Means ± SE, n = 4 biological replicates are shown; means that do not share a letter are significantly different (one‐way ANOVA, P < 0.05). Mppr10.5, MpPR10.5 CRISPR mutant lines.

WT and Mppr10.5 mutants were grown under nutrient deficiency stress conditions that induce auronidin and flavone biosynthesis (Albert et al., 2018; Zhou et al., 2024). Under stress conditions, WT plants had strong red thallus pigmentation while Mppr10.5 mutant plants were considerably less red (Fig. 2b). Quantification by LC‐MS and HPLC found that Mppr10.5 mutants had significantly reduced auronidin content (c. 84%) compared to WT (Fig. 2c). Additionally, the total flavone concentration in Mppr10.5 mutants was decreased by c. 60% compared to WT. By contrast, the concentration of bibenzyls, the major nonflavonoid phenylpropanoid compounds of Marchantia, was not altered (Fig. 2c).

MpPR10.5 overexpression (MpPR10.5‐OE) lines were generated that had a > 10‐fold increase in MpPR10.5 transcript abundance compared to the EV control lines (Fig. 3a). No visible increase in red pigmentation was observed in MpPR10.5‐OE lines compared to EV lines under nutrient stress conditions (Fig. 3b). Chemical analysis found no significant changes in total auronidins or bibenzyls concentrations, though total flavone concentrations were slightly increased in MpPR10.5‐OE lines (Fig. 3c).

Fig. 3.

Overexpression of MpPR10.5 did not alter auronidins and bibenzyls production in Marchantia polymorpha. (a) Expression level of MpPR10.5 in different transgenic Marchantia lines; means that do not share a letter are significantly different (log₂FoldChange > 1, P < 0.0001); (b) phenotype of Marchantia plants under nutrient deprivation stress; (c) quantification of total auronidin, flavone and bibenzyl compounds. Means ± SE, n = 4 biological replicates are shown; means that do not share a letter are significantly different (one‐way ANOVA, P < 0.05). EV, empty vector control; MpPR10.5‐OE, MpPR10.5 overexpression line.

Gene expression changes in response to loss of function of MpPR10.5

To gain insight into the overall effect of loss of function of MpPR10.5 in Marchantia, RNAseq analysis was conducted to identify genes with altered transcript abundance under nutrient stress conditions in the Mppr10.5 or MpPR10.5‐OE lines compared to WT.

There were 1940 differentially expressed genes (DEGs) in the Mppr10.5 lines and 58 DEGs in the MpPR10.5‐OE lines compared to WT at a log₂FC > 1 P _adj < 0.0001 cut‐off (Table S6), and no common DEGs were observed between these lines (Fig. S4). Transcript abundance for phenylpropanoid and flavonoid biosynthetic genes was altered markedly in Mppr10.5 plants compared to the WT, with a significant reduction for the candidate flavone and auronidin‐specific genes. Notable genes downregulated in Mppr10.5 included two PAL (MpPAL1 and MpPAL‐like2), one C4H (MpC4H3) and two CHS (MpCHS2 and 13). However, transcript abundances of the PAL (MpPAL‐like, PAL2, and PAL10), C4H (MpC4H1) and CHS (MpCHS17) genes with the highest transcript abundance were unchanged (Table S7). Additionally, all flavonoid pathway genes downstream of CHS, such as CHI, CHIL and AUS, were also significantly downregulated in Mppr10.5, as was MpMYB14 (Table S7). However, transcript abundances for three candidate bis‐bibenzyl biosynthetic genes, DOUBLE BOND REDUCTASE (MpDBR), STILBENECARBOXYLATE SYNTHASE (MpSTCS1A) and POLYKETIDE REDUCTASE (MpPKR), were unaffected in Mppr10.5 (Table S7). Overexpression of MpPR10.5 did not significantly affect the expression of any flavonoid biosynthetic genes or MpMYB14 (Table S7).

To further examine the basis for reduced transcript abundance for phenylpropanoid biosynthetic genes in the Mppr10.5 mutants, we generated RNAseq data for Marchantia mutant lines Mpchi, Mpchil and Mpchi‐chil (Berland et al., 2019), when grown under nutrient stress conditions, which induce auronidin biosynthesis (Table S8). Compared to WT plants grown under the same conditions, the only phenylpropanoid or flavonoid biosynthetic gene that had a significant change in transcript abundance in the Mpchi line was MpCHI itself, resulting from the CRISPR‐Cas9 deletion generated in that gene. By contrast, in the Mpchil and Mpchi‐chil double mutant, four genes showed significant differential expression. In addition to the changes in Mpchi and Mpchil transcript abundance resulting from the associated mutations, expression of MpPAL1 and MpCHS2 was significantly lower than for WT. These two specific PAL and CHS genes were among those downregulated in the Mppr10.5 mutants (Table S7). The RNAseq results were validated by qRT‐PCR (Fig. S5).

A possible explanation for changes in the expression of phenylpropanoid pathway genes resulting from the loss of MpPR10.5 gene activity is feedback regulation triggered by changes in the abundance of pathway metabolites. Therefore, we examined whether the MpPR10.5 protein can interact with relevant metabolites.

MpPR10.5 differentially binds flavonoids, with a high affinity for naringenin chalcone

Previous studies indicated that Bet v1 type PR10 proteins can bind to diverse metabolites, potentially for substrate channelling or stabilisation (Aglas et al., 2020). The Mppr10.5 mutants had reduced amounts of auronidins and flavones but not bibenzyls (Fig. 2c). This suggested the loss of function of MpPR10.5 did not affect the production of p‐coumaroyl CoA, the common substrate for both flavonoids and bibenzyls, but rather led to a reduction in the production or stability of NC, the only substrate shared by auronidins and flavones following the general phenylpropanoid pathway steps (Fig. 1). Thus, MpPR10.5 could act in the biosynthesis, stabilisation or metabolism of NC, and we developed an intrinsic fluorescence assay for recombinant MpPR10.5 protein to test whether it could bind NC and measure the comparative affinity of any binding to that with other potentially relevant metabolites.

Naringenin chalcone had a remarkable affinity for MpPR10.5, evident from the steep increase in|ΔF|/F0 values observed at very low substrate concentrations (Fig. S6) and the associated dissociation constant (K _d = 1.3 μM) (Table 1). Structurally related phenolics – chalcones (isoliquiritigenin), aurones (butein) and dihydrochalcones (phloretin and dehydroxy phloretin) – also had a high affinity for MpPR10.5, although significantly lower than that for NC. Flavones (apigenin and chrysin), flavonols (kaempferol) and flavanones (naringenin and eriodictyol), which are downstream flavonoids of quite distinct structures (Fig. 1), had a much weaker affinity for MpPR10.5, with K _d values ranging from 14 to 69 μM (Table 1; Fig. S6). Two phenylpropanoid precursors of flavonoids were assayed: cinnamic acid and p‐coumaric acid. Cinnamic acid had a very low affinity for MpPR10.5, with a K _d value over 119 μM that was more than > 50 times higher than that of chalcones, and p‐coumaric acid had no affinity (Table 1). Two bibenzyls tested, lunularic acid and a decarboxylated derivative, also had no affinity for MpPR10.5 (Table 1; Fig. S6). Thus, MpPR10.5 protein selectively bound a variety of flavonoids with preference over other tested phenolics. The highest affinity among all the tested compounds was for chalcones, particularly NC, which is a key biosynthetic intermediate for auronidin and flavone biosynthesis in Marchantia.

Table 1.

Binding affinity of different flavonoid and bibenzyl compounds for MpPR10.5 protein.

The binding affinity was measured using an intrinsic fluorescence assay. The K _d value indicates the dissociation constant. nd, not detected.

MpPR10.5 prevents NC from self‐cyclisation in vitro

The affinity data, with preference for NC and low affinity for precursor phenylpropanoids or downstream flavonoids, and the phenotypes of the Mppr10.5 loss‐of‐function mutants suggested a function for MpPR10.5 based on binding of NC. Naringenin chalcone is known to be unstable in solution, converting to naringenin through spontaneous self‐cyclisation (Mol et al., 1985). Thus, we used two independent experimental approaches to test whether MpPR10.5 binding could reduce NC self‐cyclisation into naringenin and protect it from other reactions that would cause NC degradation.

In the first approach, we took advantage of the distinct absorption peaks of NC (377 nm) and naringenin (322 nm) to assay for self‐cyclisation in the absence or presence of MpPR10.5. When NC was incubated in a buffer at pH 7.5 without MpPR10.5, the observed absorption peak shifted over time from 377 to 322 nm, indicating a conversion to naringenin, with completion almost reached after the 30‐min period of the experiment (Fig. 4a). By contrast, when an excess of MpPR10.5 protein was added to the reaction mixture, no decrease in the 377 nm absorption peak was observed, indicating stabilisation of the NC (Fig. 4b). Self‐cyclisation of NC to naringenin is known to be pH dependent (Mol et al., 1985). When the experiment was repeated at pH 8.5, the self‐cyclisation was slightly faster than at pH 7.5 (Fig. 4c). Addition of MpPR10.5 still stabilized the NC at pH 8.5, although at this higher pH some conversion to naringenin was observed over the course of the experiment (Fig. 4d). Naringenin itself remained completely stable under the experimental conditions (Fig. 4e). To validate this assay, we tested a PR10‐like protein from Petunia inflata, PiPR10‐like, which lacks key PR10 features including the glycine‐rich loop 4, the signature PR10 motif and the β2 and β3 sheets, compared to Strawberry Fra a and MpPR10.5 (Fig. S3). These structural differences result in an incomplete, loose cavity suggesting that PiPR10‐like would not function similarly to these proteins. PiPR10‐like failed to stabilise NC, with a complete absorption peak shift from 377 to 322 nm within 30 min (Fig. S7), confirming the robustness of this assay.

Fig. 4.

MpPR10.5 protein stabilised naringenin chalcone (NC) in vitro. The solid blue and red lines indicate the absorption peaks of naringenin and NC at 322 nm and 377 nm, respectively. The dotted blue and red curves (inserts) depict the absorption intensity of naringenin and NC over time at 322 and 377 nm, respectively. The observed shift of NC's absorption peak at pH 7.5 without (a) or with (b) MpPR10.5. The observed shift of NC's absorption peak at pH 8.5 without (c) or with (d) MpPR10.5. (e) The absorption peak of naringenin at pH 7.5. In all tests, measurements were conducted over 30 min, with readings taken at 1‐min intervals. Absorption curves at each time point are shown in different colors from blue (0 min) to red (30 min).

In the second approach, we directly quantified the amount of NC and naringenin in solution using LC‐MS, in the absence and presence of MpPR10.5 (Fig. 5). In the absence of MpPR10.5, NC rapidly converted into naringenin over time, with nearly complete conversion occurring within 80 min. In the presence of MpPR10.5, however, NC exhibited significant stability, with less than 10% self‐cyclising into naringenin after 160 min. Furthermore, apart from naringenin, no other degradation products were detected in the UV–Vis or MS analyses in the absence of MpPR10.5 (Fig. 5). Thus, MpPR10.5 had the strongest affinity for NC of all compounds tested and effectively protected it from self‐cyclisation, at least in vitro.

Fig. 5.

Selected ion chromatograms at m/z 271 representing the evolution of naringenin chalcone (NC) in the presence or absence of the MpPR10.5 enzyme. MpPR10.5 protein prevented NC from self‐cyclisation into naringenin in vitro.

Discussion

MpPR10.5 is a key component of flavonoid biosynthesis in Marchantia

Pathogenesis‐related protein is a general term often given to genes that are strongly induced by stress or pathogen attack, but for which the predicted amino acid sequence gives little clue to biochemical function. Many classified ‘PR proteins’ do not have direct antipathogen activity, with some, such as PR10, contributing to specialized metabolite pathways with varied functions (Ferreira et al., 2007; Muñoz et al., 2010). We identified 16 candidate PR10 genes in the Marchantia genome. Given the relatively small total number of gene models in the Marchantia genome (Bowman et al., 2017), this represents a significant gene family with potential key functions in Marchantia's biosynthetic pathways. Similarly, the expansion of other gene families, such as polyphenol oxidase, glutathione‐S‐transferase, dirigent protein and PKS, in the Marchantia genome relative to other land plant lineages has also been reported (Davies & Andre, 2023).

We demonstrate here that MpPR10.5 promotes flavonoid production in the liverwort model species Marchantia, with loss of MpPR10.5 activity causing a significant decrease in amounts of the two major flavonoid types, auronidin and flavones (Fig. 2). The production of auronidin and flavones was not completely abolished in Mppr10.5 mutants, which could be from functional redundancy among Marchantia PR10s (Table S3). There are a small number of reports showing PR10 promotes flavonoid biosynthesis in angiosperms; for instance, RNAi silencing of Fra a 1 and Fra a 3 (homologues of Bet v1) in strawberry reduced concentrations of anthocyanins, flavanols and related upstream metabolites (Muñoz et al., 2010). Similarly, overexpressing a cotton (Gossypium hirsutum) PR10 gene in Arabidopsis thaliana caused a twofold increase in total flavonoid content (Chen & Dai, 2010). The separation of bryophytes and tracheophytes is the deepest divergence in the land plant phylogeny, and the last common ancestor of liverworts and angiosperms is thought to date to > 400 million years ago. A conserved function of PR10 for flavonoid biosynthesis in both liverworts and angiosperms indicates that this function of PR10 may have been inherited from the last common ancestor and so is an ancestral feature of all land plant lineages. A priority for future work to address this proposal is to determine whether the PR10 orthologues found in the genome sequences of other land plant lineages, such as mosses, lycophytes, ferns and gymnosperms, also contribute to flavonoid biosynthesis.

MpPR10.5 protects NC

The exact role of PR10 proteins in secondary metabolite biosynthesis remains unclear, although several mechanisms have been proposed, including chemical chaperoning, intercellular shuttling, catalysis and sensing and regulation (Morris et al., 2021). We found that the loss of function of MpPR10.5 reduced the total amount of auronidins and flavones but not bibenzyls (Fig. 2), suggesting that MpPR10.5 likely acts on a common precursor for auronidins and flavones but not bibenzyls, that is NC, the product of CHS (Fig. 1). Our study suggests that MpPR10.5 preferentially protects NC, which is unstable and can readily undergo self‐cyclisation into naringenin (Mol et al., 1985). Naringenin chalcone was bound by MpPR10.5 with the highest affinity of any metabolite tested, preventing it from self‐cyclisation in vitro (Figs 4, 5). Thus, in an in vivo context, MpPR10.5 likely binds NC, stabilising it and preventing degradation or the formation of other reaction products, before delivery of NC to subsequent biosynthetic steps.

A precedent for nonenzymatic proteins directing specific flavonoid biosynthesis is the recent analysis of CHIL (Waki et al., 2020), which found CHIL may rectify the promiscuous reactions of CHS such that 2′,4,4′,6′‐tetrahydroxychalcone (the immediate NC precursor) is produced rather than other polyketides such as p‐coumaroyltriacetic acid lactone. Similarly, our in vitro data suggest that MpPR10.5 may stabilise NC and preserve its bioavailability for downstream enzymatic reactions in vivo.

The phenylpropanoid classes of flavonoids (flavones and auronidins) and bibenzyls of Marchantia are derived from the same common precursor molecule, p‐coumaroyl CoA (Fig. 1). The amounts of auronidins and flavones were significantly reduced in the Mppr10.5 mutants, but total bibenzyl content was not changed (Fig. 2). This is consistent with MpPR10.5 having a specific function along with CHS, as CHS is required for auronidin and flavone production but not bibenzyls. Strikingly, this relationship was also reflected in the RNAseq results, as several flavonoid biosynthesis genes were downregulated in Mppr10.5 mutants (e.g. MpCHI and MpAUS), but the expression of the bibenzyl biosynthesis genes MpDBR, MpSTCS and MpPKR was not altered (Table S7). Thus, the observed specificities of MpPR10.5 metabolite binding are reflected in the changes in gene expression, suggesting a metabolite‐mediated feedback signalling mechanism. In contrast to the strong affinity binding with flavonoids, especially NC, MpPR10.5 had either very weak or no affinity towards upstream phenylpropanoid precursors and no binding affinity to bibenzyls (Table 1; Fig. S6).

Naringenin chalcone concentration may mediate transcriptional feedback regulation of phenylpropanoid pathway gene activity

Several genes in the flavonoid biosynthetic pathway, and the key activator gene MpMYB14, were downregulated in the Mppr10.5 mutant (Table S7). This pleiotropic effect from the lack of PR10 could suggest a regulatory function. However, the protein type of PR10 argues strongly against a direct transcription factor function, and overexpression of MpPR10.5 in Marchantia did not alter the total auronidins or bibenzyls concentrations (Fig. 3). This is consistent with a lack of anthocyanin content increase from overexpression of Fra a1 and Fra a3 in strawberries (Muñoz et al., 2010), but contrasts with the twofold increase in total flavonoid content in the GhPR10 overexpression A. thaliana transgenics (Chen & Dai, 2010). Overexpression of MpPR10.5 did not affect the expression of any flavonoid biosynthetic genes or related transcription factors, suggesting that under normal conditions, MpPR10.5 expression does not limit flavonoid biosynthesis (Table S6; Fig. S4). It is possible that MpPR10.5 works together with other co‐factors, like Fra a 1‐Associated Protein (FaAP) in strawberries (Franz‐Oberdorf et al., 2017) and Mal d 1‐Associated Protein (MdAP) in apples (Puehringer et al., 2003), which may be limiting in the MpPR10.5‐OE lines. Together, these results indicate that PR10 does not have a direct regulatory role in the control of the flavonoid biosynthetic pathway. Rather, the downregulation of genes, such as MpPAL, MpC4H and MpCHS, in the Mppr10.5 and Mpchil mutants indicates the presence of metabolite‐mediated feedback regulation in phenylpropanoid biosynthesis in Marchantia. This may also be present in angiosperms, as silencing Fra a in strawberry reduced FaPAL and FaCHS transcript abundance (Muñoz et al., 2010).

Our data suggest that feedback regulation in phenylpropanoid biosynthesis may result from changes in the concentration and availability of NC or its unexpected by‐product, potentially mediated by MpPR10.5. We propose that the loss of MpPR10.5 disrupts the protective binding of NC, leading to an initial increase in free NC. This free NC may undergo uncontrolled oxidation or other reactions, potentially generating downstream products that could exert feedback inhibition on key genes, such as MpPAL, MpCHS and others. While it remains unclear whether NC itself or its oxidation by‐products are responsible for this inhibitory effect, previous studies on CHIL and related proteins suggest that NC may play a regulatory role. However, distinguishing between the effects of NC and its rapidly formed byproducts remains a challenge. The downregulation of MpPR10.5 or MpCHIL in Marchantia and PR10 in strawberry caused decreased PAL gene expression, whereas downregulating MpCHI in Marchantia or CHS in strawberry did not (Table S7; Muñoz et al., 2010). This supports the idea that NC itself is an important signaling metabolite, helping to sense flux within the flavonoid pathway. Indeed, recent studies have shown that NC enhances the thermostability of the CHIL protein, improving the efficacy of the CHS reaction (Waki et al., 2020; Wolf‐Saxon et al., 2023). Although CHILs are noncatalytic proteins and lack key catalytic or substrate‐binding residues present in CHI enzymes, they are conserved in all land plant genomes examined to date. The loss of CHIL function has been shown to greatly reduce amounts of flavonols and anthocyanins (Morita et al., 2014) or flavonols or proanthocyanidins (Jiang et al., 2015) in angiosperms and flavones and auronidins in Marchantia (Berland et al., 2019).

The increase in stability of CHIL and binding between CHS and CHIL proteins in the presence of NC led Wolf‐Saxon et al. (2023) to propose that NC provides feedback regulation upon flavonoid production. Specifically, CHIL could act as a receptor that senses and responds to downstream metabolite concentrations, enabling ligand‐mediated pathway feedback by influencing CHS activity and thus controlling flux through the flavonoid pathway. Our metabolite binding and RNAseq data for MpPR10.5 have striking similarities with those for CHIL, such as the strong affinity for NC and downregulation of the same flavonoid biosynthesis genes, which suggests the proteins may have related functionalities. It is possible that without MpPR10.5′s protection, NC could rapidly degrade in the cytosol. Our results also highlight a potential role for a PR10‐(CHIL/CHS) complex in regulating pathway flux, not only through enzyme activity but also via feedback regulation of gene transcription. Such feedback control based on metabolite flux is an important part of the regulatory mechanism of some specialized metabolite pathways, notably that for carotenoids (Cazzonelli & Pogson, 2010; Pereira, 2023), but it is not a well‐characterized aspect of the flavonoid pathway.

Bringing together the findings from this study and data from other Marchantia mutants, we propose a model for MpPR10.5 activity in flavonoid biosynthesis (Fig. 6). Our findings suggest that MpPR10.5 functions in flavonoid biosynthesis in Marchantia by selectively binding to NC and preventing it from degradation or untargeted nonenzymatic conversions. This may form part of a flavonoid metabolon (to be discussed later) or intercellular metabolite carrier mechanism (Fig. 6a). In Mppr10.5 mutants (Fig. 6b), flavonoid production (auronidins, flavones) is reduced but not abolished, likely because of partial redundancy between MpPR10.5 and other MpPR10 members (e.g. MpPR10.3) ①. The reduced protection of NC in Mppr10.5 mutants makes NC vulnerable to degradation/conversion into unknown products ②. Spontaneous conversion of NC into naringenin is possible ③, based on in vitro data (Mol et al., 1985). However, this contribution is expected to be minor in planta, because Mpchi mutants in Marchantia or other species contain only trace flavonoid concentrations (Forkmann & Dangelmayr, 1980; van Tunen et al., 1988; Li et al., 1993; Kang et al., 2014; Clayton et al., 2018; Berland et al., 2019). An alternative mechanism (to redundant PR10 genes) for the small amount of flavonoid production in Mppr10.5 mutants is NC freely diffusing from CHS to AUS or FNS or the involvement of a different carrier. This seems unlikely, given the instability of NC, although it cannot be excluded without generating mutants in multiple PR10 genes. Additionally, our results indicate a potential feedback regulation (inhibition) of phenylpropanoid genes mediated by NC or its by‐products ④, which may be sensed by CHIL, as suggested by Wolf‐Saxon et al. (2023). We emphasise that this is a proposed model, and several aspects, including the exact molecular mechanisms of NC protection, metabolite channelling and feedback regulation, require further validation through in vivo studies.

Fig. 6.

Proposed model of MpPR10.5 activity for flavonoid biosynthesis in Marchantia polymorpha. (a) Naringenin chalcone (NC) is synthesised by CHALCONE SYNTHASE (CHS), with assistance from CHALCONE ISOMERASE‐LIKE (CHIL). MpPR10.5 binds NC with high affinity, protecting it from degradation or from forming NC‐derived products (e.g. oxidation). This would also prevent spontaneous cyclisation into naringenin, requiring MpCHI to facilitate naringenin formation for flavone biosynthesis. NC is tightly bound by MpPR10.5, suggesting NC must be transferred between MpCHS/MpCHIL and MpPR10.5 and then subsequently offloaded to MpAUS and MpCHI, forming a metabolite channel. This may occur through direct or indirect protein–protein interactions. (b) Marchantia Mppr10.5 mutants lack the protection of the NC formed by MpCHS/MpCHIL. (1) Loss of MpPR10.5 activity may be partially compensated by other MpPR10 proteins (e.g. MpPR10.3), allowing some biosynthesis of auronidins and flavones to occur. Alternatively, free NC may diffuse to MpAUS and MpCHI for flavone and auronidin biosynthesis, respectively. (2) Free NC is unstable and degrades/converts into unknown reaction products. (3) Spontaneous conversion of free NC into naringenin. (4) Accumulation of NC or degradation/derived products provides feedback inhibition upon phenylpropanoid genes (e.g. MpPAL), indicated by the red blunt arrow. Dotted arrow indicates weak expression from PAL gene. The chemo‐sensing mechanism and mode(s) of transcriptional regulation are not known. AUS, AUREUSIDIN SYNTHASE; CHI, CHALCONE ISOMERASE; FNS, FLAVONE SYNTHASE; PAL, PHENYLALANINE AMMONIA LYASE. Arrow width indicates metabolite flux.

Metabolons are biosynthetic complexes formed of a mixture of proteins and metabolic intermediates so that the metabolite product of one enzyme may be passed directly as the substrate of the next enzyme in the pathway (‘substrate channeling’), preventing substrates from diffusing into the bulk solvent, enhancing biosynthesis efficacy (Wheeldon et al., 2016; Sweetlove & Fernie, 2018). A metabolon for flavonoid biosynthesis has been proposed for many years (Winkel, 2004), with membrane‐bound cytochrome P450 enzymes anchored within the endoplasmic reticulum and strings of soluble enzymes attached to those. This model has remained somewhat equivocal because protein–protein interactions between some enzymes have not been detected or are inconsistent between species (Nakayama et al., 2019). The recent discoveries on the importance of nonenzymatic proteins, such as CHIL and PR10, raise the possibility that these have been missing components of previous studies, and their inclusion could enable reconstitution of a functional flavonoid metabolon. Sugimoto et al., (2024) recently demonstrated that CHI variants lacking enzymatic activity can fully complement chi mutants. The authors suggest these disarmed CHI proteins may accept NC from CHS, provide an environment where NC can spontaneously isomerize to form naringenin and channel this to the next biosynthetic enzyme (in angiosperms, this is commonly FLAVANONE 3‐HYDROXYLASE). Our findings for MpPR10.5 share many similarities, including accepting NC with high affinity, albeit MpPR10.5 must act before MpCHI in Marchantia because it affects both auronidin and flavone production. Based on our current findings, we hypothesise that PR10 protein members in angiosperms may similarly function in metabolite protection, as observed for MpPR10.5, and could potentially contribute to metabolite channel formation in conjunction with MpCHI. However, further experimental validation is needed to confirm this proposed mechanism.

In summary, we have demonstrated the requirement of MpPR10.5 for WT levels of flavonoid biosynthesis in Marchantia, establishing the conservation of this function between nontracheophytes and angiosperms and showing a contribution to the production of the flavone and auronidin flavonoid groups for the first time. We also propose a possible mechanism of action for PR10 proteins that can be tested in future studies.

Competing interests

None declared.

Author contributions

YZ, KMD, NWA, JLB, CMA and CH contributed to funding acquisition; YZ, NWA, KMD, KES and JLB contributed to project conceptualisation; YZ, CH, NWA, JMC and CMA conducted experiments; YZ, CH, JMC and CMA analysed data; JWK prepared testing compounds and was involved in experiment design; YZ wrote the original draft; all authors contributed to editing the draft. YZ and CH contributed equally to this work and share co‐first authorship.

Disclaimer

The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.

Supporting information

Fig S1 His‐SUMO‐MpPR10.5 and His‐MBP‐PiPR10‐like proteins purification.

Fig. S2 MpPR10.5 (Mp8g00860) displays a typical PR10 structure.

Fig. S3 Structural comparison of MpPR10.3, MpPR10.5, MpPR10.9, and PiPR10‐like proteins.

Fig. S4 Venn diagrams illustrating the number of differentially expressed genes induced by knocking out or overexpressing MpPR10.5 in Marchantia polymorpha under nutrient‐deprivation stress.

Fig. S5 qRT‐PCR validation of RNAseq results presented in Table S7.

Fig. S6 MpPR10.5 binds to flavonoids with different affinities.

Fig. S7 PiPR10‐like protein did not stabilised naringenin chalcone (NC) in vitro.

Table S1 List of primers used in this study.

Table S2 Expression of reference genes across different tissues and conditions.

Table S3 Candidate genes encoding PR10 proteins identified in the Marchantia genome and their predicted subcellular localisation.

Table S4 BaseMean, log₂FoldChange, adjusted P‐value of MpPR10 genes in RNAseq analysis.

Table S5 BaseMean, log₂FoldChange, adjusted P‐value of Marchantia polymorpha PR10 genes in previous RNAseq analysis.

Table S6 Differentially expressed genes (DEGs) in Mppr10.5 and MpPR10.5‐OE lines at a threshold of log₂FC > 1 and P _adj < 0.0001.

Table S7 Transcript abundance changes of phenylpropanoid phenylpropionate pathway biosynthetic and regulatory genes in Marchantia polymorpha plants in response to nutrient stress.

Table S8 Summary of the comparative analysis of flavonoid and bibenzyl biosynthesis‐related transcripts between Mpchi, Mpchil, and Mpchi‐chil mutants, and WT.

Please note: Wiley is not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

Data availability

Raw sequencing data from all samples have been deposited in the SRA database (BioProject ID: PRJNA1055307, https://dataview.ncbi.nlm.nih.gov/object/PRJNA1055307). All Marchantia polymorpha sequence accession numbers given are those for the genome sequence at MarpolBase (https://marchantia.info), as registered in the Marchantia Nomenclature Database (https://marchantia.info/nomenclature/).