Simultaneous Overexpression of FERULOYL‐CoA 6′‐HYDROXYLASE 1 and COUMARIN SYNTHASE Leads to Coumarin‐Enriched Lignin and Improved Saccharification in Greenhouse‐ and Field‐Grown Poplar

ABSTRACT

The urgent need for renewable resources has increased the interest in woody biomass to manufacture bio‐based products. However, lignin recalcitrance limits the enzymatic conversion of wood into fermentable sugars, posing a major challenge for biomass deconstruction. To address this problem, we aimed at incorporating the coumarin scopoletin into the lignin polymer of poplar ( Populus tremula × P . alba ) by expressing FERULOYL‐CoA 6′‐HYDROXYLASE 1 (F6′H1) and COUMARIN SYNTHASE (COSY) in lignifying cells. Three constructs were evaluated: two bicistronic constructs, SCOP1 (COSY followed by F6′H1) and SCOP2 (F6′H1 followed by COSY), and one monocistronic, SCOP3 (only F6′H1). SCOP1 poplars produced most free scopoletin without altering overall lignin, cellulose or hemicellulose content. SCOP2 poplars were overall less efficient in scopoletin production and most of these lines showed a severe biomass yield penalty, whereas SCOP3 caused plant lethality. NMR and metabolic analyses confirmed that scopoletin cross‐coupled with G and S monomers during lignification in SCOP1 lines. In addition to scopoletin, the detection of benzodioxane structures revealed the incorporation of dihydroxycoumarins. Overall coumarin incorporation in lignin amounted up to 2.3%. After alkaline pretreatment, wood from greenhouse‐grown SCOP1 poplars released up to 29% more glucose compared to the wild type upon limited saccharification. Field‐testing of three SCOP1 lines showed a 6 to 11% increase in saccharification efficiency, with the line containing the lowest scopoletin levels maintaining normal growth. These results demonstrate that engineering lignin composition in poplar can improve saccharification, and emphasize the importance of construct design, translational research and field validation.

1. Introduction

Transitioning from a fossil‐based economy to a circular bioeconomy demands a shift toward sustainable and renewable resources (Gomez et al. 2008; Güleç et al. 2024; Vanholme et al. 2013). Lignocellulosic biomass, including agricultural residues as well as biomass from dedicated crops like poplar grown under short rotation coppice (Baruah et al. 2018; Zoghlami and Paës 2019), is a promising feedstock for the production of biofuels, bioplastics and chemicals (Jing et al. 2019; Kawaguchi et al. 2022; Ojo 2023; Zheng et al. 2022). This biomass consists primarily of the carbohydrate polymers cellulose and hemicelluloses, alongside the aromatic polymer lignin. The latter plays a crucial role in plants by providing them strength to grow upright, by facilitating water and nutrient transport and by acting as a barrier against pathogens. However, lignin is one of the main factors contributing to the low processability of lignocellulosic biomass (Hendriks and Zeeman 2009; Vanholme et al. 2013; Yoshida et al. 2008). The polymer hinders chemical pulping (Li et al. 2025) and the enzymatic hydrolysis of cell wall polysaccharides into monosaccharides, a process called saccharification (Chen and Dixon 2007; Van Acker et al. 2013).

To improve industrial processing, biomass crops with reduced lignin levels have been engineered (De Meester, Vanholme, et al. 2022; Li et al. 2008), but because of its crucial role, reducing lignin content has its limitations. A reduction in lignin levels in the cell wall of vessels can result in vascular collapse and, consequently, a biomass yield penalty (Bonawitz and Chapple 2013; De Meester et al. 2020; De Meester, Vanholme, et al. 2022; Van Acker et al. 2014; Voelker et al. 2010). To prevent this, lignin reduction has been targeted only in the fibres (De Meester et al. 2018; Gui et al. 2020; Vargas et al. 2016). Alternatively, lignin composition and structure, rather than its quantity, have been altered to make lignin more prone to degradation in industrial processes, while maintaining its biological role (Marita et al. 2001; Mottiar et al. 2016; Oyarce et al. 2019; Ralph 2006; Vanholme et al. 2008; Wilkerson et al. 2014).

In dicots, lignin is mainly synthesized through the oxidative coupling of the monolignols p‐coumaryl, coniferyl and sinapyl alcohol, which give rise to the p‐hydroxyphenyl (H), guaiacyl (G) and syringyl (S) units of the lignin polymer, respectively (Figure 1A). Various other monomers are incorporated into the lignin polymer, albeit often at lower levels and depending on the species (del Río et al. 2020; Ralph et al. 2019; Vanholme, De Meester, et al. 2019). In poplar (Populus ssp.), monolignol p‐hydroxybenzoates, and to a lesser extent also monolignol acetates, benzoates, p‐coumarates and ferulates, serve as natural monomers, among others (De Meester, Vanholme, et al. 2022; de Vries et al. 2022; Zhao, Yu, Lam, et al. 2021). Lignin composition is highly flexible, and various modifications can be achieved through lignin engineering without affecting biomass yield (Dixon and Barros 2019). For instance, the S/G ratio has been increased by overexpressing FERULATE 5‐HYDROXYLASE (F5H) (Li et al. 2003; Stewart et al. 2009). In p‐HYDROXYBENZOYL‐CoA:MONOLIGNOL p‐HYDROXYBENZOYLTRANSFERASE 1 (PHBMT1)‐deficient trees, lignin contained no detectable amounts of p‐hydroxybenzoates compared to WT trees, whereas PHBMT1‐overexpressing poplars showed an increased incorporation of p‐hydroxybenzoates in their lignin (de Vries et al. 2022; Zhao, Yu, and Liu 2021). Neither of these lignin‐engineered trees had major developmental defects, at least not when they were grown in a greenhouse.

FIGURE 1.

Lignin and scopoletin biosynthesis and overview of constructs. (A) General phenylpropanoid and monolignol‐specific pathway for lignin monomer biosynthesis in poplar. Overexpression of F6′H1 (FERULOYL‐CoA 6′‐HYDROXYLASE 1) and COSY (COUMARIN SYNTHASE) is used to engineer the pathway toward the production of scopoletin. 4CL, 4‐COUMARATE:CoA LIGASE; C3′H, p‐COUMAROYL SHIKIMATE 3′‐HYDROXYLASE; C3H, p‐COUMARATE 3‐HYDROXYLASE; C4H, CINNAMATE 4‐HYDROXYLASE; CAD, CINNAMYL ALCOHOL DEHYDROGENASE; CCoAOMT, CAFFEOYL‐CoA O‐METHYLTRANSFERASE; CCR, CINNAMOYL‐CoA REDUCTASE; COMT, CAFFEIC ACID O‐METHYLTRANSFERASE; CSE, CAFFEOYL SHIKIMATE ESTERASE; F5H, FERULATE 5‐HYDROXYLASE; HCT, HYDROXYCINNAMOYL‐CoA SHIKIMATE/QUINATE HYDROXYCINNAMOYL TRANSFERASE; PAL, PHENYLALANINE AMMONIA‐LYASE; PHBMT1, p‐HYDROXYBENZOYL‐CoA:MONOLIGNOL p‐HYDROXYBENZOYLTRANSFERASE 1. (B) Different constructs tested to produce scopoletin in the lignifying cells of poplar. The coding sequences of COSY and F6′H1 were linked together by a GSG spacer and a T2A sequence for SCOP1 and SCOP2 and put under control of the CELLULOSE SYNTHASE 8‐B promoter (ProCesA8‐B).

The observation that lignin composition can be modified without adverse effects on plant growth has led to strategies to boost the incorporation of alternative monomers in lignin that are typically absent or only present in trace amounts in native lignin. Such monomers need to contain a phenolic function to allow the generation of a phenoxy radical by cell wall‐localized oxidative enzymes (del Río et al. 2020; Li et al. 2018; Ralph 2006; Ralph et al. 2019; Vanholme et al. 2012; Yoshida et al. 2024). For example, when FERULATE MONOLIGNOL TRANSFERASE (FMT) from Angelica sinensis was overexpressed in poplar under control of a secondary cell wall cellulose synthase promoter, monolignol‐ferulate conjugates were successfully incorporated into the lignin at higher levels, resulting in an enhanced saccharification efficiency upon mild alkaline pretreatment conditions compared to the WT (Wilkerson et al. 2014; Zhou et al. 2017). Similarly, overexpressing a grass p‐COUMAROYL‐CoA:MONOLIGNOL TRANSFERASE (BdPMT1) in poplar resulted in the p‐coumaroylation of lignin, increasing the saccharification efficiency of alkali‐pretreated poplar wood, without a growth penalty (Lapierre et al. 2021). The incorporation of curcumin into the lignin polymer of Arabidopsis by simultaneously overexpressing two turmeric ( Curcuma longa ) genes, DIKETIDE‐CoA SYNTHASE (DCS) and CURCUMIN SYNTHASE 2 (CURS2) (Oyarce et al. 2019), resulted in a 24% increase in saccharification efficiency compared to the WT without negative growth effects. However, when this strategy was translated into poplar, the transgenic trees had a biomass yield penalty, indicating that translation from model systems into crops is crucial to further evaluate the strategy (De Meester, Oyarce, et al. 2022).

In another case, scopoletin has been tested as an alternative monomer for lignin engineering in the model plant Arabidopsis (Hoengenaert et al. 2022). The incorporation of scopoletin into the lignin polymer through its phenolic group forms a β‐ether unit in which the aromatic unit has a conjugated carbonyl function. Such ether‐bonded units are more easily cleaved under alkaline conditions, enhancing the polymer's susceptibility to degradation (Criss et al. 1998; Gierer and Ljunggren 1979; Imai et al. 2007). In addition, alkaline hydrolysis of scopoletin's ester functionality results in the formation of a carboxylic acid and an additional free phenolic group. Hydrolysis would therefore increase the hydrophilicity and the alkaline solubility of the pretreated polymer, benefiting the extraction of lignin in aqueous solvents. By overexpressing two genes, FERULOYL‐CoA 6′‐HYDROXYLASE 1 (F6′H1) and COUMARIN SYNTHASE (COSY), in lignifying cells of Arabidopsis, 3.3% of the lignin units were derived from scopoletin without negatively affecting plant growth (Hoengenaert et al. 2022). This modification resulted in a 40% increased saccharification efficiency upon alkaline pretreatment of inflorescence stems compared to the WT. In another study, F6′H1 was overexpressed in lignifying cells of Arabidopsis plants, which also led to the accumulation of scopoletin without affecting plant growth (Sakamoto et al. 2020). The overexpression of F6′H1 in lignifying cells of poplar also resulted in the incorporation of scopoletin in lignin, but the growth of these plants was severely affected (Wang et al. 2025).

Here, we investigated whether scopoletin can be incorporated into the lignin polymer of poplar to facilitate lignocellulosic biomass processing without affecting plant growth. We tested three different constructs (Figure 1B). The first two constructs were bicistronic, containing F6′H1 and COSY fused by the T2A self‐cleaving peptide, allowing the production of two proteins from a single transcript. To improve cleavage efficiency, a sequence encoding a glycine‐serine‐glycine spacer (GSG) was added upstream of the T2A sequence (Wang et al. 2015). Because it has been shown that the protein of which the corresponding gene is positioned before the GSG‐T2A sequence is more abundantly produced than that of the gene positioned after that sequence (Liu et al. 2017; Wang et al. 2015; Zhu et al. 2023), we designed two constructs: in SCOP1, COSY is positioned before F6′H1, whereas in SCOP2, F6′H1 is positioned before COSY. Poplar was also transformed with a third construct, SCOP3, to overexpress only F6′H1. By testing three different constructs, we increased the chances of generating transgenic poplar lines with desired wood and growth characteristics. Following two greenhouse screenings, three SCOP1 lines incorporating scopoletin into their lignin, showing improved saccharification efficiency and no biomass yield penalty, were selected for field evaluation.

2. Results

2.1. Generation of Transgenic Lines

To produce scopoletin in lignifying tissues of Populus tremula x P. alba (poplar), the coding sequences of F6′H1 and COSY, originating from Arabidopsis, were cloned under the transcriptional control of the secondary cell wall‐specific CELLULOSE SYNTHASE 8‐B (CesA8‐B) promoter from Populus trichocarpa . Three different constructs were generated (Figure 1B). In the first construct, proCesA8‐B:COSY_F6′H1 (SCOP1), COSY is positioned before F6′H1 in a bicistronic expression system, and in the second, proCesA8‐B:F6′H1_COSY (SCOP2), F6′H1 is positioned before COSY. The order of F6′H1 and COSY in SCOP2 is the same as that in the construct successfully used to incorporate scopoletin into the lignin of Arabidopsis (Hoengenaert et al. 2022). The third construct, proCesA8‐B:F6′H1 (SCOP3), aimed at only expressing F6′H1. The three SCOP constructs were transferred into poplar via Agrobacterium‐mediated transformation. The SCOP lines were first screened in a greenhouse setting to evaluate the levels of scopoletin production in their stems via metabolomics and to assess potential effects on growth. From this initial screening, the best‐performing lines were selected for a second, more in‐depth greenhouse evaluation of biomass yield and cell wall composition. Subsequently, the three best‐performing lines were further evaluated in a field trial.

2.2. Screening of the Transgenic Lines Through Growth and Metabolite Profiling

To investigate the impact of F6′H1 and COSY expression in lignifying cells on the growth of the trees, a single plant of each of the 36 independent lines (16 SCOP1, 13 SCOP2 and 7 SCOP3 lines) was transferred from in vitro conditions to soil and grown in the greenhouse for 23 weeks until WT poplars reached an average height of about 2 m (Figure 2A). SCOP3 lines were severely dwarfed when grown in soil, ultimately resulting in the death of all lines already after seven weeks of growth in soil. At 23 weeks, SCOP2 lines showed an average height reduction of 44% compared to the WT, whereas SCOP1 poplars, on average, did not display a significant height difference compared to the WT (Figure 2B). However, 11 out of 16 SCOP1 lines had a reduced height compared to the WT with reductions ranging from 9% to 37% (Figure S1). These reductions were less severe than those observed for SCOP2 poplars, for which 10 out of 13 lines had height reductions ranging from 15% to 95% compared to the WT.

FIGURE 2.

Screening of the different SCOP lines based on growth in the greenhouse and metabolic analysis. (A) Height curves of WT and transgenic poplars grown in the greenhouse in soil. (B) Pictures of SCOP1 and SCOP2 poplars on the day of the harvest. From left to right, pictures show the WT, the lines with the most extreme growth phenotypes, and a representative line with growth comparable to the WT. (C‐D) Peak intensities of scopoletin (C) and scopolin (D) for each independent line of SCOP1 and SCOP2 poplars. (E) Differentially accumulating metabolites that were at least fourfold higher in SCOP lines (excluding SCOP1‐1, SCOP1‐3, SCOP2‐6, and SCOP2‐16) compared to the WT. (F) Chemical structures of coumarins and coumarinolignoids that accumulated in the transgenic lines. Numbers correspond to metabolites listed in Tables S1, S2, and S3, with metabolites accumulating in SCOP1 shown in blue, those accumulating in both SCOP1 and SCOP2 shown in purple, and those that were found via the targeted search shown in black.

To examine whether the introduction of SCOP1 and SCOP2 construct resulted in the production of scopoletin, phenolic compounds were extracted from the xylem of 23‐week‐old poplars and analysed via ultra‐high performance liquid chromatography‐mass spectrometry (UHPLC–MS). Upon targeted analysis, either scopoletin, its 4‐O‐hexoside (scopolin) or both, were found to be produced at significantly higher levels in all but four transgenic lines compared to the WT, based on the 95% confidence interval of the WT (Figure 2C,D). The highly increased production of these metabolites indicated that transgenes were expressed and that the intended metabolic pathway was effective in most transgenic lines.

Based on these results, the lines that did not produce significantly more scopoletin or scopolin compared to the WT (i.e., SCOP1‐1, SCOP1‐3, SCOP2‐6, and SCOP2‐16) were excluded from further analysis. Comparative metabolic analyses were performed on the remaining 14 SCOP1 and 11 SCOP2 lines. These independent lines were treated as repeats of a given construct given that in this screening phase we analysed only one plant per independent line. A total of 12 296 m/z features were initially detected by UHPLC–MS. To enrich for compounds with a relatively high abundance, we filtered the features with an average signal intensity larger than 2000 in at least one group (WT, SCOP1 or SCOP2). Note that the signal intensity depends not only on the abundance of the compound in the sample but also on its physico‐chemical properties and might be influenced by co‐eluting compounds through ion suppression. Signal intensities can generally be used to compare relative abundances of the same compound in different samples but are not suitable for comparing relative abundances of different compounds within the same sample. A one‐way ANOVA showed that 277 m/z features were significantly different in signal intensity in at least one of the three groups. Post hoc statistical tests on these 277 features showed that 181 and 132 features had a higher intensity in SCOP1 and SCOP2, respectively, compared to the WT. Among these features, 49 in SCOP1 and 50 in SCOP2 poplars had a peak intensity at least fourfold higher than that observed in WT. For SCOP1 poplars, the 49 features could be assigned to 35 compounds, of which 32 could be (partially) annotated based on their MS/MS fragmentation (Figure 2E,F, Table S1). Nine compounds were annotated as esculetin conjugates (1–9), three as scopoletin (10) or its conjugates (11, 12), seven fraxetin conjugates (13–19), one isofraxidin conjugate (20), five coumarinolignoids (21–25), two flavonoids (26, 27), two benzenoids (28, 29), and three 6‐hydroxyphenylpropanoids (30–32).

The 50 increased features in SCOP2 lines could be assigned to 46 compounds of which 36 could be (partially) annotated (Figure 2E,F, Table S2). We found five esculetin conjugates (1, 3, 5, 6, 8), one scopoletin conjugate (11), four fraxetin conjugates (13, 14, 18, 19), one isofraxidin conjugate (20), a 6‐hydroxyphenylpropanoid (30), sixteen benzenoids (29, 36–50), two salicinoids (51–52), five phenylpropanoids (53–57), and one monolignol coupling product decorated by mercaptolactic acid and hexose (58). The differential compounds in SCOP2 showed an enrichment of coumarin‐related features compared to the WT (11 out of the 36 identified compounds), although the enrichment was less pronounced than that observed in SCOP1 poplars (26 out of the 32 identified compounds) (Figure 2F). Scopoletin was not among the 46 top‐accumulating compounds in SCOP2 poplars, whereas it ranked 8th in signal intensity among the accumulating compounds in SCOP1 poplars. The accumulating metabolite with the highest peak intensity in SCOP1 and SCOP2 lines was scopolin, which increased 38‐fold and 13‐fold, respectively, compared to the WT. The scopolin‐to‐scopoletin ratio was generally higher in SCOP2 than in SCOP1 (Figure S2).

The finding of five coumarinolignoids (21–25) among the top‐accumulating compounds in SCOP1 lines suggested that additional compounds of this metabolic class could accumulate in these lines. Upon a targeted search for coumarinolignoids in SCOP1 and SCOP2 lines, combinations of either esculetin, scopoletin, fraxetin or isofraxidin with coniferyl or sinapyl alcohol(s) with and without hexose decoration were sought. This resulted in the characterization of five additional coumarinolignoids (69–73, Figure 2F and Table S3). Their respective signals had a higher intensity in the SCOP1 lines as compared to the WT, but because the average intensity was below 2000, they were not retrieved in the untargeted analysis. Except for compound 72, these coumarinolignoids did not accumulate in SCOP2 lines. Notably, in the SCOP1 lines we found coumarinolignoids with scopoletin (21, 70, 71) and fraxetin (22–25, 69, 72, 73) moieties, but not with esculetin or isofraxidin moieties.

The aim of this lignin engineering approach was to boost the production of scopoletin without compromising plant growth. Based on the screening of the transgenic plants, SCOP1 poplars were chosen for further evaluation as they produced on average more scopoletin than SCOP2 lines (Figure 2C) and had less to no reduction in growth compared with SCOP2 and SCOP3 poplars (Figure 2B). From the SCOP1 poplars, four lines were chosen for further in‐depth analysis: two lines that grew similarly to the WT (SCOP1‐7 and ‐15), the line with the highest scopoletin production (SCOP1‐2), and one line with reduced growth compared to the WT (SCOP1‐5) based on the 95% confidence interval (Figure S1). The latter line was randomly selected from the lines with impaired growth to evaluate whether this was a genuine phenotypic effect or an artefact of the transfer to soil. Additionally, four SCOP1 lines (SCOP1‐4, ‐8, ‐12, and ‐24) that became available later in the project were included for further analysis.

2.3. Evaluation of SCOP1 Poplars: Growth, Scopoletin Incorporation Into Lignin and Saccharification Efficiency

Five replicates per line of the eight selected SCOP1 poplar lines were grown in the greenhouse, and their height was monitored over a 23‐week period until the WT reached an average height of 2 m (Figure S3A). Six of the eight SCOP1 lines showed a height comparable to that of the WT, whereas SCOP1‐5 and SCOP1‐15 had an 18% and 34% reduced height, respectively (Figure S3B,C). When harvested after 23 weeks, five of the eight SCOP1 lines had a fresh biomass yield (with bark, without leaves) comparable to the WT, whereas three lines had significantly less fresh stem biomass, ranging from 48% for SCOP1‐12 to about 62% and 65% for SCOP1‐5 and SCOP1‐15, respectively (Table S4). The diameters of SCOP1 lines were similar to those of the WT, except for SCOP1‐15 that was reduced by about 50% (Table S4). Consequently, the biomass yield of SCOP1‐2, ‐4, ‐7, ‐8, and ‐24 lines was not significantly different from the WT.

Metabolite profiling had confirmed scopoletin production in SCOP1 poplars. To evaluate the impact of this metabolic alteration on lignin content, the basal segments of dried, debarked stems from 23‐week‐old WT and SCOP1 poplars were analysed using the acetyl bromide soluble lignin (ABSL) assay. Analysis of the absorbance spectra at 280 nm revealed no significant differences in ABSL content between WT and SCOP1 lines (Figure 3A). Additionally, we subtracted the absorbance spectra of WT ABSL from those derived from the SCOP1 poplar lines. In the resulting difference spectrum, a clear increase in absorption was observed in all SCOP1 lines around 340 nm. This corresponds to the wavelength of the absorbance maximum of scopoletin, suggesting that scopoletin is present in the lignin polymer of all SCOP1 lines, albeit with varying levels (Figure 3B). No correlation was noted between the relative absorption of the ABSLs at 340 nm and biomass yield parameters (Figure S4).

FIGURE 3.

Acetyl bromide soluble lignin (ABSL) and saccharification yields of SCOP1 lines. (A) Average ABSL‐determined lignin content determined at 280 nm for SCOP1 poplars. (B) ABSL absorption difference spectra of SCOP1 minus WT were obtained by subtracting the average absorption spectrum of WT lignin from each of the average spectra from the SCOP1 lines. These spectra were overlaid with an absorption spectrum of a scopoletin standard subjected to the ABSL procedure subtracted from the average ABSL absorption spectrum of WT lignin (dashed black line). (C) Bars represent the average glucose release per cell wall residue (CWR) after alkaline pretreatment (62.5 mM NaOH for 3 h at 90°C) and after 72 h of enzyme incubation. Other treatments are shown in Figure S5. Percentages represent the difference in glucose release/CWR between SCOP1 poplars and the WT. Error bars represent the standard error, and n indicates the number of biological repeats. Significant differences (p < 0.05) according to a one‐way ANOVA with Tukey's post hoc are indicated by different letters. (D) Correlation plot between glucose release per CWR and the absorption of SCOP1 lines and WT at wavelength 340 nm (maximum of scopoletin absorbance).

To assess the impact of scopoletin production on wood processing, a saccharification assay was performed. Besides no pretreatment, two pretreatment methods were applied: an acidic (1 M HCl, 2 h, 80°C) and an alkaline pretreatment (62.5 mM NaOH, 3 h, 90°C). After (no) pretreatment, the wood was incubated with a cellulase and hemicellulase enzyme mixture for 72 h. Upon either no or acid pretreatment, the saccharification efficiency of all tested SCOP1 lines was similar to that of the WT (Figure S5). In contrast, alkaline pretreatment resulted in a significantly increased saccharification efficiency of wood for six out of the eight SCOP1 lines compared to the WT (Figure 3C). Two SCOP1 lines had no increase in saccharification efficiency: SCOP1‐8 and ‐15. Notably, the saccharification efficiency of SCOP1‐8 was also not significantly different from the SCOP1 lines with the highest saccharification efficiency, placing SCOP1‐8 intermediate between WT and these best performing lines.

To investigate whether the saccharification efficiency correlated with the level of scopoletin incorporated into lignin, we plotted the absorbance difference between SCOP1 and WT at 340 nm against the glucose release per cell wall residue (CWR) after alkaline pretreatment. As shown in Figure 3D, a positive correlation (R ² = 0.855) was found, supporting the hypothesis that the abundance of scopoletin positively influences the saccharification efficiency. Lines SCOP1‐2, ‐7, and ‐8 were selected for further analysis, because they showed no biomass yield penalty while having an increased saccharification and distinct scopoletin incorporation relative to the WT.

2.4. Detailed Cell Wall Analysis of SCOP1‐2, ‐7, and ‐8

To gain a deeper understanding of the impact of scopoletin on cell wall composition, the CWR, lignin, cellulose and hemicellulose contents were determined for the three selected lines (Table 1). No significant differences in the amount of CWR per dry weight were observed between the SCOP1 lines and WT. Lignin content was analysed using the gravimetric Klason assay, considered as the gold standard for lignin quantification in wood. Unlike the ABSL method, which estimates lignin content based on absorbance, the Klason assay directly measures isolated lignin by weight, providing greater accuracy (Fukushima and Hatfield 2004). In this assay, total lignin is calculated as the sum of the gravimetrically determined acid‐insoluble (Klason lignin) and the acid‐soluble lignin (ASL) fraction measured by absorption spectroscopy. Compared to the WT, SCOP1‐2 showed a significant reduction of 4.68% in Klason lignin, whereas ASL was significantly increased (Table 1). However, when combining both fractions, no significant difference in total lignin content was observed between SCOP1‐2 and WT. No significant differences were observed in Klason lignin and ASL for SCOP1‐7 and ‐8. Crystalline cellulose and hemicellulose levels were not altered in any of the three SCOP1 lines compared to the WT. These results indicate that the incorporation of scopoletin selectively alters the lignin composition without affecting the overall cell wall polymer content.

TABLE 1.

Cell‐wall characteristics of greenhouse‐grown SCOP1 and WT poplars.

	WT	SCOP1‐2	SCOP1‐7	SCOP1‐8
CWR/DW (%)	88.15 ± 0.86	87.60 ± 1.21	88.85 ± 1.33	88.72 ± 0.53
Lignin content
Klason lignin/CWR (%)	20.26 ± 0.52	19.31 ± 0.31*	19.95 ± 0.59	20.06 ± 0.60
ASL/CWR (%)	2.55 ± 0.07	2.83 ± 0.20****	2.65 ± 0.08	2.52 ± 0.06
Total lignin/CWR (%)	22.80 ± 0.55	22.15 ± 0.28	22.60 ± 0.60	22.58 ± 0.61
Polysaccharides
Crystalline cellulose/CWR (%)	46.69 ± 1.11	47.95 ± 1.14	46.41 ± 1.67	46.48 ± 0.88
Matrix polysaccharides/CWR (%)	33.01 ± 1.40	32.39 ± 1.16	33.20 ± 1.91	32.29 ± 1.15

Note: The cell wall residue (CWR), expressed as mass percentage of dry weight (DW), was determined gravimetrically after a sequential extraction. Klason lignin content was gravimetrically determined using the Klason method and expressed as mass percentage of CWR. Acid‐soluble lignin (ASL) was spectrophotometrically determined via the Klason method. The sum of these two lignin fractions constitutes the total lignin content. Crystalline cellulose content was determined by the Updegraff method. Matrix polysaccharides were determined via the mass loss upon trifluoroacetic acid treatment. Values represent mean ± standard deviation. Significant differences were evaluated by a one‐way ANOVA with Dunnett's post hoc; n = 10 for WT and n = 5 for SCOP1 poplars. *0.01 < p ≤ 0.05, **0.001 < p ≤ 0.01, ***0.0001 < p ≤ 0.001, ****p ≤ 0.0001. The bold indicated the values that differ significantly from the WT.

2.5. Microscopic Visualization of Scopoletin and Cell Morphology of SCOP1‐2, ‐7, and ‐8

To better understand the localization of scopoletin in the cell wall and the effect of its production on cell morphology, we investigated the wood of SCOP1‐2, ‐7, and ‐8 via microscopy, either bright‐field or fluorescence, and after Wiesner staining (Figure 4). No obvious structural abnormalities were observed in xylem cells of the three SCOP1 lines with any of the three techniques. However, bright‐field microscopy showed yellow coloration of the SCOP1 wood, indicating scopoletin presence in the cell wall, as previously observed in Arabidopsis (Hoengenaert et al. 2022). Fluorescence microscopy with an excitation wavelength of 405 nm and a reduced detector sensitivity to correct for WT lignin autofluorescence revealed an intense fluorescence in the cell walls of xylary vessels, fibres and ray cells of SCOP1 lines. This likely reflects increased scopoletin or related compounds in the cell wall, rather than an increased lignin content, because Klason and ABSL measurements were similar between the SCOP1 lines and the WT. SCOP1‐8 showed a lower fluorescence than SCOP1‐2 and SCOP1‐7, consistent with its observed relatively lower scopoletin absorption in ABSL (Figure 3). Wiesner staining, specific for hydroxycinnamaldehyde lignin end‐groups, showed no differences between the SCOP1 lines and the WT, confirming the unchanged lignin content.

FIGURE 4.

Cell‐wall‐localized scopoletin and xylem morphology in stems of SCOP1 poplars. Bright‐field microscopy revealed a yellow coloration in the xylem of the SCOP1 lines compared to the WT. Autofluorescence of the cross‐section was visualized upon excitation at 405 nm. To correct for the autofluorescence of lignin, detector sensitivity was reduced until limited fluorescence was detectable in the WT. Wiesner stained cross‐sections were visualized using bright field. Bars = 100 μm. n = 3 biological repeats for WT and SCOP1 poplars, a representative image is shown.

2.6. Detailed Lignin Structural Analysis of SCOP1‐2, ‐7, and ‐8 via NMR

To unambiguously demonstrate the incorporation of scopoletin into the lignin polymer, enzyme lignin from the three SCOP1 lines and the WT was analysed using two‐dimensional heteronuclear single‐quantum coherence (2D HSQC) nuclear magnetic resonance (NMR) spectroscopy (Table S5). This technique allows us to detect lignin monomer contributions to the polymer independently of their linkage type and to determine the relative frequency of each interunit linkage type (Kim et al. 2008; Mansfield et al. 2012; Ralph, Lapierre, Marita, et al. 2001; Ralph et al. 1999).

Comparison of the NMR spectra of SCOP1 and WT lignin revealed a significant increase in the accumulation of coumarins in the SCOP1 lines compared to the WT (Figure 5A–C; Figure S6). The individual coumarins scopoletin (Sc) and fraxetin (Fr) were clearly evidenced with a good match between the lignin correlation peaks and model compound data. Esculetin was below the set detection threshold. There is a differential peak, Figure 5C, that fits well for Isofraxidin (If) but other peaks for such structures overlap with other lignin peaks, and it could not be conclusively validated. The 2D‐difference HSQC‐spectrum of SCOP1‐2 minus WT most clearly revealed signals derived from scopoletin units, via their Sc ₈ and Sc ₅ correlations, indicating that at least some of the coumarins in the SCOP1 lignin were derived from scopoletin. In fact, two isomers of the anticipated S/G(4–O–β)Sc units from scopoletin's incorporation into lignin appeared to be indicated and well‐separated (labelled a and b for the two isomers in Figure 5). Characteristic fraxetin units were also revealed with diagnostic signals matching those of model compounds, notably the Fr ₂ and Fr ₈ correlations. Correlations for Sc ₇ and Fr ₇ were also prevalent, but overlapping, except that an Sc _7G correlation, that is, from G(4–O–β)Sc deriving from the initial addition of the monolignol coniferyl alcohol to scopoletin, was somewhat separated.

FIGURE 5.

Lignin structural profiling by 2D‐¹H–¹³C‐correlation NMR spectroscopic analysis. (A, B) Partial HSQC spectrum of the aromatic and unsaturated region from (A) SCOP1‐2 and (B) WT. The colours of the contours correspond to the structures drawn. (C) A 2D‐difference spectrum (the spectrum of SCOP1‐2 minus the one from WT) of the same region using the best nulling of the S_2/6 signal. Purple contours represent components elevated in SCOP1‐2 vs. WT, and dark‐blue contours represent components depleted in SCOP1‐2 vs. WT relative to the syringyl peaks. (D) Partial HSQC‐TOCSY spectrum of the aromatic and unsaturated regions to authenticate scopoletin unit assignments; the main Sc ₇–Sc ₈ and Fr ₇–Fr ₈ correlations are indicated via the dashed‐line rectangles. (E) Partial HSQC spectrum of the aliphatic (sidechain) region from line SCOP1‐2. Standard designation of lignin inter‐units was used, A for β‐ethers (β–O–4), B for phenylcoumarans (β–5), C for resinols (β–β), D for dibenzodioxocins (5–5/4–O–β), and E for spirodienones (β–1/β–O–4) even though D peaks were not evident here and are therefore not labelled (and the structure not shown). A′ is used for β‐ether units derived from cross‐coupling of either coniferyl or sinapyl alcohol with the catecholic coumarin fraxetin (Fr); those peaks, as indicated by the dashed box, are plotted at an 8× lower contour level; A′ units appear prominently in the sidechain region of the difference spectrum (not shown). Values displayed in the yellow box are the average of SCOP1‐2 and WT; scopoletin levels are divided into values per independent line; n = 3 biological repeats per line; ppm, parts per million (Hz/MHz). NMR‐spectra from SCOP1‐7 and ‐8 showed similar differences to WT and the data can be found in Table S5.

Lignin from SCOP1‐2, ‐7, and ‐8 contained 1.93%, 2.32%, and 1.27% coumarin‐derived units (Figure 5A), respectively, whereas no coumarin‐related signals were detected in the WT (Figure 5B). These observed variations in coumarin levels in the three lines confirmed the variations observed by the ABSL assay (Figure 3) and via microscopy (Figure 4). No significant differences were observed between the WT and the SCOP1‐2 and ‐7 lines in the relative abundance of S, G, and p‐hydroxybenzoate ( pHB) units. In contrast, SCOP1‐8 showed a modest but significant shift, with a 2% decrease in S units and a 4% increase in G units relative to the WT. H units were detected in trace amounts but, because of their overlapping signals with those of phenylalanine, could not be reliably integrated (Figure 5A,B).

Analysis of the interunit linkage types in the aliphatic region of the NMR spectra revealed a significant reduction in the frequency of resinol units (β–β, C) from 8.93% in the WT to 7.50%, 7.40%, and 7.77% in SCOP1‐2, ‐7, and ‐8, respectively (Figure 5E, Table S5). The frequency of β‐aryl ether units (β–O–4, A), phenylcoumaran (β–5, B), and spirodienone (β–1, E) structures did not differ between the WT and any of the SCOP1 lines. Conversely, clear benzodioxane (A′, so‐denoted because they are a special kind of β–O–4 unit) signals were detected in the sidechain region of the SCOP1 lines, accounting for 1.1%, 1.2%, and 0.9% of the interunit linkages in SCOP1‐2, ‐7, and ‐8 poplars, respectively, whereas no such benzodioxane peaks were evident in the WT. The presence of these benzodioxane structures indicates the incorporation of ortho‐diphenolic (catechol) units, such as esculetin or fraxetin, into the lignin polymers of the transgenic poplars. We assume that these benzodioxane structures are from fraxetin given that fraxetin is readily identified in the aromatics region of the spectra, and esculetin is not; the corresponding benzodioxane peaks match those from fraxetin‐derived model compounds more closely (Figure S6). This incorporation likely occurs through cross‐coupling reactions forming S/G(β–O–5)Fr dimers, as supported by MS analysis.

2.7. Translation Into the Field

Because the three selected SCOP1 lines (SCOP1‐2, ‐7, and ‐8) showed promising greenhouse results, including an increased saccharification efficiency of 29%, 22%, and 15%, respectively, upon alkaline pretreatment, and no biomass yield penalties, a field trial was initiated in May 2024, in Wetteren, Belgium, aiming to evaluate their performance outdoor (Figure 6A). The field trial had a randomized complete block design, comprised of five blocks with eight poplars per genotype per block, resulting in a total of 160 poplars (Figure S7).

FIGURE 6.

Evaluation of SCOP1‐2, ‐7, and ‐8 in the field. (A) Picture taken on the 27th of August 2024. (B) Adjusted mean of height of poplars in the field over a period of 142 days since the start of the field trial. Significant differences were evaluated according to a linear mixed model (y = block + genotype + plot + error); ***p < 0.001; n = 40. (C) Adjusted mean of ABSL levels of field‐grown SCOP1 poplars. Same letters indicate no significant differences at the 5% significance level according to a linear mixed model (y = block + genotype + plot + error); n = 10. (D) Absorption difference spectra were obtained by subtracting the average absorption spectrum of WT ABSL from each of the average ABSL spectra of the SCOP1 lines and from that of scopoletin subjected to the ABSL procedure (dashed black line). (E) Means of saccharification assay on alkali‐pretreated (62.5 mM NaOH for 3 h at 90°C) wood. The wood was incubated with an enzyme mixture for 72 h. After 2, 6, 24, 48 and 72 h, samples were collected to determine the glucose release. Significant differences were evaluated by a one‐way ANOVA with Tukey's post hoc, n = 10. * 0.01 < p ≤ 0.05, ** 0.001 < p ≤ 0.01.

The height of the trees was recorded every two weeks over a period of 142 days (Figure 6B). Although all lines exhibited a similar height until 99 days (measured on 5th of September 2024), SCOP1‐2 and SCOP1‐7 showed a 5 to 6% reduction in the final height (measured on 18th of October 2024) compared to the WT, in contrast to SCOP1‐8 that did not show significant differences in the final height. For all poplars in the field, the stem diameter was determined in winter at 80 cm above soil (Table S6). The stem diameter of SCOP1‐2 and SCOP1‐7 was significantly reduced by 7% compared to the WT, again in contrast to SCOP1‐8 that did not show any differences. Furthermore, all lines were monitored for insect and rust infections (caused by fungi of the genus Melampsora), but no significant differences compared to the WT were observed (Table S6). In December 2024, two trees from each genotype in each block were harvested. The weight of each tree was measured, revealing a significant reduction in both fresh and dry weight for SCOP1‐2 and SCOP1‐7 of on average 20% and 30% compared to the WT, respectively (Table S6). SCOP1‐8, however, showed no significant differences.

To investigate the lignin content and the presence of scopoletin, the 30‐cm basal portion of dried and debarked stems of the field‐grown WT and SCOP1 poplars were analysed together with wood samples of 23‐week‐old greenhouse‐grown poplars. An ABSL assay showed no significant differences in the lignin content between the WT and SCOP1 trees (Figure 6C). Furthermore, no significant differences were observed between field‐ and greenhouse‐grown poplars (Figure S8). Additionally, we subtracted the absorbance spectra of WT ABSL from those of SCOP1 poplar lines, for both field and greenhouse‐grown trees (Figure 6D). At 340 nm, the absorbance maximum of scopoletin, a clear increase in absorption was observed in the SCOP1 lines as compared to the WT controls, similar to the included greenhouse samples, with SCOP1‐7 exhibiting nominally the highest absorption. In contrast to the greenhouse‐grown lines, the relative absorption of ABSL at 340 nm appeared inversely correlated with the biomass yield (Figure S9). However, given that the field experiment was performed with a limited number of lines, we cannot unambiguously confirm the generality of this observed inverse correlation.

To evaluate whether the increased saccharification efficiency observed in greenhouse‐grown SCOP1 poplars is maintained when grown in the field, a saccharification assay was performed on the wood derived from the basal portion of dried and debarked stems of the field‐grown poplars. Unpretreated SCOP1 wood resulted in similar saccharification values as the WT for all lines (Figure S10). Upon alkaline pretreatment, significant increases in sugar release of 11%, 7%, and 6% were observed for SCOP1‐2, SCOP1‐7, and SCOP1‐8, respectively, as compared to the WT (Figure 6E).

3. Discussion

Scopoletin production in Arabidopsis can be achieved by overexpressing F6′H1 alone or together with COSY (Hoengenaert et al. 2022; Sakamoto et al. 2020). To translate this strategy to poplar, three constructs (SCOP1, SCOP2 and SCOP3) were generated, resulting in distinct phenotypes. SCOP3 caused lethality, suggesting that a product of F6′H1, for example, 6′‐hydroxycaffeoyl‐CoA, 6′‐hydroxyferuloyl‐CoA or a derivative thereof, is toxic. This hypothesis is supported by the observation that plant growth was generally more affected in SCOP2 lines as compared to that of SCOP1 lines. It is well‐established that the first gene in a bicistronic construct is more efficiently translated into its corresponding protein as compared to the second gene (Liu et al. 2017; Wang et al. 2015; Zhu et al. 2023). SCOP2 lines are thus expected to produce more of these toxic intermediates than SCOP1 lines. Generally, COSY appeared a rate limiting step in SCOP2 lines, whereas it likely mitigates toxicity by channelling the F6′H1 product(s) into coumarins in SCOP1 lines. These findings highlight opportunities to optimize scopoletin pathway engineering by adjusting expression levels of F6′H1 and COSY, for example by using promoters of different strengths (e.g., Pro LAC2 or Pro DX15 for COSY; Pro CesA8‐B for F6′H1 [Ko et al. 2012]) or by using a tricistronic construct with multiple COSY copies (e.g., ProCesA8‐B:COSY_F6′H1_COSY [Liu et al. 2017]). Alternatively, the use of a fibre‐specific promoter may mitigate the dwarfed phenotype if it would be attributable to vessel malfunctioning (Guo et al. 2025).

Among the 32 identified metabolites accumulating in SCOP1 poplars, 25 were coumarins derived from scopoletin, esculetin, fraxetin or isofraxidin (1–25), similarly to those in the SCOP Arabidopsis lines (Hoengenaert et al. 2022). Their accumulation likely results directly from F6′H1 and COSY expression. In Escherichia coli , F6′H1 converts caffeoyl‐CoA into 6′‐hydroxycaffeoyl‐CoA and COSY has been shown to convert 6′‐hydroxycaffeoyl‐CoA into esculetin in vitro (Vanholme, Sundin, et al. 2019; Yang et al. 2015). As caffeoyl‐CoA is a monolignol precursor, part of its pool is diverted into coumarin biosynthesis in our transgenic lines. Additionally, scopoletin is metabolized into fraxetin by the cytochrome P450 monooxygenase SCOPOLETIN 8‐HYDROXYLASE (AtS8H) in Arabidopsis (Tsai et al. 2018). The poplar homologue, B9H637 (encoded by POPTR_005G097900v3), shares 80.7% amino acid similarity and has the residues essential for iron and 2‐oxoglutarate binding (Siwinska et al. 2018), suggesting that fraxetin formation is a direct consequence of scopoletin formation in our transgenic poplars. In addition, hexosylated coumarins were detected in SCOP1 and SCOP2 lines, most likely made through UDP‐glycosyltransferase (UGT) activity from their respective aglycones (Le Roy et al. 2016). In SCOP1, the low scopolin‐to‐scopoletin ratio could reflect that high scopoletin biosynthesis may surpass the glycosylation capacity, whereas in SCOP2, lower scopoletin production may allow glycosylation to keep pace, resulting in a higher scopolin‐to‐scopoletin ratio.

Besides the five coumarinolignoids (21–25) among the top‐accumulating compounds in SCOP1 lines, a targeted search led to the identification of five additional coumarinolignoids (69–73). These include coupling products of scopoletin or fraxetin with coniferyl or sinapyl alcohol(s), with and without hexose decorations. These coumarinolignoids are presumably made via non‐enzymatic oxidative cross‐coupling of coumarin and monolignol radicals. Fraxetin‐monolignol dimers, also known as cleomiscosins, are found in several plant species (e.g., in the seeds of sugar apple [ Annona squamosa ] or excreted by Arabidopsis roots) (Begum and Kumar 2018; Begum et al. 2010; Sisó‐Terraza et al. 2016). In SCOP1, we also detected trimeric coupling products consisting of a cleomiscosin coupled to a coniferyl alcohol (72, 73) that, to our knowledge, have not been previously described. These trimers are likely made in the apoplast by the oxidative coupling of a monolignol radical to a cleomiscosin radical, and may extend further to tetramers, pentamers, etc. to become part of the lignin polymer. Indeed, benzodioxane structures such as those of cleomiscosins were observed via NMR in SCOP1 lignin.

The majority of the accumulating coumarins (1–9, 11–20) and coumarinolignoids (21, 24, 25) in SCOP1 lines were hexosylated. These are most likely sequestered into the vacuole, preventing their participation in the lignification process (Dima et al. 2015) unless released during apoptosis (Pesquet et al. 2025, 2013). Moreover, in the specific case of the hexosylated scopoletins 11–20 and the coumarinolignoids 21, 24, and 25 (and possibly hexosylated esculetins 1–9), a hexose is linked to the phenolic functionalities, as such preventing the oxidation of these metabolites to their corresponding free radical and consequently making it impossible to couple into the lignin polymer. To further improve coumarin incorporation, hydrolyzing the hexosides to their respective aglycones by expressing a β‐glucosidase in SCOP1 lines would free their phenolic group to allow their translocation to the apoplast, their oxidation to radicals and their incorporation into the lignin polymer (Le Roy et al. 2016).

Although nine esculetin‐containing metabolites were found to accumulate in SCOP1 lines, no esculetin‐derived coumarinolignoids were detected among the soluble phenolics in any of these lines, and no esculetin was detected in the NMR spectra (Figure 5), reflecting that esculetin does not efficiently participate as a monomer in the lignification process. In this respect, coumarinolignoids of esculetin coupled to sinapyl alcohol (in a crossed dimer called moluccanin) were found in twigs of Eurycorymbus cavaleriei, hinting that esculetin has the ability to couple with monolignols (Cheng et al. 2009). Similarly, model coupling reactions with esculetin and either coniferyl or sinapyl alcohol produced such cross‐coupled dimers. We hypothesize that esculetin is highly efficiently glycosylated in the cytoplasm of SCOP1 lines – more efficiently as compared to scopoletin and fraxetin –, after which the esculetin glycosides are unable to diffuse across the plasma membrane into the apoplast (Vermaas et al. 2019), preventing the participation of esculetin (glycosides) in lignin polymerization.

Despite unchanged lignin, cellulose and hemicellulose levels, SCOP1 wood exhibited an up to 29% higher saccharification efficiency after alkaline pretreatment. Three major structural changes likely explain this improvement. First, the coumarinolignoid detection among the soluble phenolics, the increased absorption of the ABSL at 340 nm and the NMR data confirmed the coupling and incorporation of coumarins into lignin. The extended conjugated system of the scopoletin moieties introduces alkali‐labile β‐aryl ether linkages (Criss et al. 1998; Gierer and Ljunggren 1979; Imai et al. 2007). In addition, the alkaline pretreatment conditions open the lactone ring of coumarins, resulting in a carboxylic acid and a phenolic functionality that increase lignin hydrophilicity and solubility and facilitate its extraction (Lapierre et al. 1989). Second, the frequency of resinol‐type β–β linkages significantly decreased from 8.93% in the WT to 7.40%–7.77% in SCOP1 lines. These structures can only be made by dimerization of two monolignol radicals and thus always mark lignin initiation sites, unlike β‐aryl ether (β–O–4) and phenylcoumaran (β–5) structures that can form either via monomer–monomer or, more frequently, monomer–polymer coupling. Coumarins engineered in SCOP1 lines are not able to form resinol linkages. It is well‐known that the C–C linkages in resinols are associated with cell wall recalcitrance (Jeong et al. 2021; Pu et al. 2013), and their reduced frequency could therefore contribute to the observed improvement in saccharification efficiency. Third, cleomiscosin‐type benzodioxane linkages (metabolites 22, 23, 25, 69, 72–73), as detected by metabolite profiling and NMR, resemble structures that have been identified in plants producing lignin monomers with an ortho‐diphenolic (catechol) structure. Such units are most evident from lignification incorporating 5‐hydroxyconiferyl alcohol and 5‐hydroxyconiferaldehyde in COMT‐deficient plants (Ralph, Lapierre, Lu, et al. 2001; Vanholme et al. 2010), caffeyl alcohol in the seed coat of vanilla ( Vanilla planifolia ) and several members of the Cactaceae, Euphorbiaceae and Cleomaceae (Chen et al. 2012; Tobimatsu et al. 2013; Wang et al. 2020), and piceatannol in palm fruit endocarp (del Río et al. 2017). Such benzodioxane units are described to prevent lignin‐carbohydrate complex formation, thereby lowering biomass recalcitrance (Ralph et al. 2004).

Although Arabidopsis remains invaluable for lignification studies (Koornneef and Meinke 2010), translating findings into biomass crops such as poplar is an essential step in moving from lab results toward applications. Although some lignin engineering strategies, such as CCR downregulation, have had similar outcomes in both Arabidopsis and poplar, with reduced lignin content, improved saccharification and varying yield penalties (De Meester et al. 2020; Van Acker et al. 2014, 2013), contrasting results have been obtained for other engineering strategies. For example, although curcumin incorporation into Arabidopsis lignin resulted in increased saccharification efficiency with no growth penalties, this strategy caused severe growth defects in poplar without improving saccharification efficiency (De Meester, Oyarce, et al. 2022; Oyarce et al. 2019). Constructs similar to SCOP2 and SCOP3 were effective in Arabidopsis (Hoengenaert et al. 2022; Sakamoto et al. 2020), but yielded variable scopoletin levels and growth penalties in poplar (Wang et al. 2025). We were able to avoid the undesired effects on growth by using a different construct in poplar: the bicistronic proCesA8‐B:COSY_F6′H1 construct (SCOP1) that had not been tested in Arabidopsis. These contrasting outcomes emphasize the importance of validating results from Arabidopsis in a biomass crop like poplar.

Field testing is a further step toward application, as environmental conditions strongly influence plant phenotype (Boerjan and Strauss 2024). SCOP1‐2, ‐7, and ‐8 grew normally in the greenhouse while showing an increased saccharification efficiency. In the field, SCOP1‐2 and ‐7 displayed modest height and diameter reductions (5% and 7%, respectively), whereas SCOP1‐8 growth was indistinguishable from the WT. Notably, of the three tested SCOP1 lines in the field, SCOP1‐8 was the line with the lowest amount of scopoletin in its lignin. Cell wall properties were largely maintained: scopoletin levels were similar to those in greenhouse‐grown poplars and the saccharification efficiencies were still significantly higher compared to the WT, although the improvements were less than those obtained with greenhouse‐grown trees. Reduced improvements under field conditions as compared to greenhouse conditions, as previously observed (De Meester, Van Acker, et al. 2022; Van Acker et al. 2017), are likely due to the differences in wood structure caused by the exposure to abiotic and biotic stresses that are present in the field and not in the greenhouse, and the analysis of winter‐harvested dormant wood from the field trial versus actively growing wood from the greenhouse (Cesarino 2019; De Meester, Vanholme, et al. 2022).

SCOP1‐8, with no growth differences compared to WT in the field, is a promising candidate for evaluating chemical pulping efficiency. Strategies introducing labile bonds into lignin can enhance delignification and pulp yield, such as through the introduction of monolignol ferulate conjugates (Zhou et al. 2017). Beyond processing improvements, elevated scopoletin/scopolin levels may enhance stress tolerance. Scopoletin acts as a phytoalexin in several species such as wild tobacco ( Nicotiana attenuata ) and sweet potato ( Ipomoea batatas ), conferring resistance against Alternaria alternate and Fusarium oxysporum, respectively (Sun et al. 2014; Wang et al. 2024). In addition, accumulation of scopoletin through overexpression of F6′H1 resulted in a reduced susceptibility of Arabidopsis to the parasitic nematode Heterodera schachtii and of transgenic soybean ( Glycine max ) to the pathogenic fungus Fusarium virguliforme (Beesley et al. 2023), whereas co‐overexpression of IbF6′H2 and IbCOSY in tobacco ( Nicotiana tabacum ) resulted in increased resistance against F. oxysporum (Wang et al. 2024). In our field trial, however, we did not observe reduced infection with insects or rust, potentially because the infection pressure was modest, or because the scopoletin levels were only increased in lignifying cells, and not in parenchyma cells. Duan et al. (2025) demonstrated that elevated scopolin levels in Melilotus albus protected the plants from drought stress. This suggests that the increased scopolin levels in SCOP1 poplars could provide an advantage during drought conditions. Such dual benefits, improved processing and potential stress resilience, further support the promise of this strategy for developing climate‐adapted bioenergy feedstocks. More detailed research is needed to investigate these possibilities.

4. Conclusion and Future Perspectives

Transitioning from a fossil‐based to a bio‐based economy requires overcoming the challenges posed by lignin. In Arabidopsis, expression of F6′H1 and COSY in lignifying cells successfully led to the incorporation of scopoletin into lignin, increasing saccharification efficiency after alkaline pretreatment, without growth penalty (Hoengenaert et al. 2022; Sakamoto et al. 2020). Here, we translated this strategy to the biomass crop poplar, in which construct design proved decisive. SCOP1 lines generally produced more scopoletin with no or a milder biomass yield penalty as compared to SCOP2 lines. Following an in‐depth evaluation of eight SCOP1 lines grown in the greenhouse, three lines were advanced to a field trial. In the field, line SCOP1‐8 had a beneficial combination of improved saccharification and wild‐type‐like biomass yield, thereby providing a proof of concept for our lignin engineering strategy under real‐world growing conditions. The field‐trial experiment also showed effects on the phenotype of the transgenic lines that were not observed when grown under greenhouse conditions. Lines SCOP1‐2 and SCOP1‐7 had reduced biomass yield as compared to control lines. These observations spark ideas for further optimizing the lignin engineering strategy by further tweaking the level and tissue specificity of F6′H1 and COSY expression.

5. Materials and Methods

5.1. Plant Material and Vector Construction

Populus tremula × P. alba INRA clone 717‐1B4 (poplar) was used as control and for transformation. The three different constructs, SCOP1, SCOP2, and SCOP3, were designed, synthesized and cloned as described in Methods S1.

5.2. Growth Conditions in the Greenhouse

For the initial screening, 36 independent transgenic lines (one plant for each line; 16 SCOP1, 13 SCOP2, and 7 SCOP3) and seven WT controls were transferred from in vitro culture to 5.5‐cm pots filled with Saniflor 691 commercial soil (Van Israel). Plants were placed in a water‐filled tray and covered with a cage liner (Tecniplast APET 1291H) for acclimatization. One side of the liner was lifted after 1 week for 2 days, followed by the other side after 4 days, and the liner was fully removed thereafter. Plants grew for 2 weeks in growth chambers (16‐h light/8‐h dark, ~21°C) before being transferred to 3‐L pots in the greenhouse. After 4 weeks, they were moved to 10‐L pots with the same soil.

For detailed SCOP1 analysis, 49 plants were transferred from in vitro culture to soil: eight SCOP1 lines (four selected from the initial screen and four additional lines), and ten WT controls.

5.3. Harvesting of Plant Material

For the initial screening of the different lines and for the in‐depth analysis of SCOP1 lines, the WT reached heights of approximately 200 cm after 23 weeks of growth in soil. At this point, plants were harvested by cutting the stem 12 cm above the soil level leaving 2–3 axillary buds to allow development of new shoots. The bottom 10 cm of the harvested stems were debarked, the wood cut into pieces of about 1 cm, frozen, and stored at −70°C for phenolic profiling. The leftover stem piece was weighed, debarked, air‐dried, weighed and ground in a ball mill for cell wall analysis and saccharification.

After a second growth cycle of the new shoot, when the WT reached heights of approximately 100 cm, poplars were again harvested 12 cm above soil. The bottom 10 cm was stored in 70% EtOH and used for bright‐field and fluorescence microscopy.

5.4. Phenolic Profiling

Stem pieces stored at −70°C were ground into wood powder using a Grindomix GM200 using precooled stainless‐steel jars (Retsch). Soluble phenolics were extracted from approximately 15 mg of stem material with 1 mL methanol (HPLC grade) at room temperature for 15 min under 750 rpm shaking. After drying 800 μL of each methanol extract in a SpeedVac, a liquid–liquid extraction was performed with 100 μL cyclohexane and 100 μL H₂O. The lower water phase was injected. A UHPLC–MS analysis was performed following the procedure described by Vanholme, Sundin, et al. (2019). More details can be found in Methods S2.

5.5. Cell Wall Preparation

The preparation of CWR involved a series of sequential extraction steps (De Meester, Oyarce, et al. 2022). In summary, ground wood powder was sequentially washed for 30 min each with Milli‐Q water at 98°C, ethanol at 76°C, chloroform at 59°C and acetone at 54°C. The remaining CWR was dried under vacuum and was determined gravimetrically (expressed as mass percentage of dry weight).

5.6. Lignin Quantification

ABSL was determined on 4 mg of CWR according to Moreira‐Vilar et al. (2014), with modifications, which can be found in Methods S3. Klason lignin was determined on 50 mg of CWR according to the National Renewable Energy Laboratory protocol (Sluiter et al. 2012), as described in Methods S3.

5.7. Carbohydrate Analyses

Prepared CWRs were used to determine hemicellulose and cellulose content according to Foster et al. (2010) with some adjustments, as described in Methods S4.

5.8. Saccharification Assay

The saccharification assay was performed according to Van Acker et al. (2016), which can be found in Methods S5.

5.9. Light and Fluorescence Microscopy

Stem slices (25 μm) were made using a Reichers‐Jung 2040 Autocut Microtome (Leica). Sections were stained using the Wiesner reagent. Wiesner staining was performed by incubating the sections for 1 min in a phloroglucinol‐HCl solution consisting of one volume of concentrated HCl (37%) and two volumes of 3% phloroglucinol dissolved in ethanol. Stained sections were visualized using a Keyence VHX‐7000 microscope (300× objective). Unstained sections were visualized with a 500× objective. Lignin autofluorescence was visualized using a ZEISS LSM 710 confocal microscope with a Plan‐Apochromat 20×/0.8 M24 objective. The fluorescence signal of the lignin polymer was obtained using an excitation wavelength of 405 nm and an emission window between 415 and 735 nm.

5.10. Lignin Compositional Analysis via 2D HSQC NMR

Five biological repeats per independent SCOP1 line from 23‐week‐old poplars' stems were used to create three samples (containing two, two and one biological repeats per sample). Ten WT biological repeats were used to create three samples (containing three, three and four biological repeats per sample). The 2D HSQC NMR spectra were acquired on enzyme lignin dissolved in DMSO‐d₆ as previously described (Kim and Ralph 2010; Kim et al. 2008; Mansfield et al. 2012). More details can be found in Methods S6. Model compounds to authenticate the various coumarin products incorporated into lignins were produced by simple radical coupling of the monolignol and the coumarin via peroxidase/H₂O₂ as traditional (Kim et al. 2020; Kim et al. 2017). The various compounds will have their NMR data entered into the ‘NMR database of lignin and cell wall model compounds’ (Ralph et al. 2024), including: Sc, #369; S(β–O‐3)Es, #365; G(β–O‐3)Es, #366; S(β–O‐5)Fr, #363; G(β–O‐5)Fr, #364.

5.11. Field Trial Set‐Up and Statistics

Three transgenic lines (SCOP1‐2, SCOP1‐7, and SCOP1‐8) and the WT were propagated in soil and transferred when 18 weeks old to the field located in Wetteren, Belgium in May 2024, under a genetically modified field trial authorization (B/BE/24/V2) from the Belgian Biosafety Advisory Council. The plot of the field trial consisted of 40 clonal replicates of each line divided into five blocks (eight poplars per line) plus a WT border (56 trees, not included in the experimental analysis), totaling 216 trees. The distance between the trees inside a block and between the blocks was 0.75 and 1.5 m, respectively. The statistical analysis of the data related to the field‐grown poplars is described in Methods S7.

5.12. Harvest Procedure and Biomass Acquisition

The trees were grown in the field from the 29th of May until 16th of December 2024. From spring until dormancy in autumn, the height of the main stem was measured every 2 weeks. The stem diameter of all poplars was measured at 80 cm above soil when the trees were dormant, during winter (2nd of December 2024). Forty trees were harvested on the 16th of December (2 trees per genotype per block) by cutting 20 cm above the soil level. Fresh weight was measured, the bottom 30 cm of the stem was collected for cell wall analysis, and the remaining part of the poplars was ground and dried to determine final total dry weight. The bottom 30 cm was debarked and dried (10 days at 45°C) to determine the dry weight, and ground in a ball mill for cell wall analysis.

5.13. Biological Interactions

Rust infection caused by species of the genus Melampsora and herbivorous insect damage were evaluated in October 2024. Rust severity was scored by INBO according to Camino et al. (2025). Insect damage was scored based on the following incidence scores: (1) Mild damage situated mostly at the bottom; (2) Severe damage at the bottom; (3) Severe damage at the bottom and higher parts of the plant. The severity scores (scored based on the percentage of damage of the most affected leaf) were 0, 0%; 1, 1%–10%; 2, 11%–25%; 3, 26%–50%; 4, 51%–75%; 5, 76%–99%.

Author Contributions

N.D.R., L.H., R.V. and W.B. designed the experiments. N.D.R., J.V.D., W.D.C., G.G., S.L. and J.R. performed the experiments. N.D.R., R.V., F.L. and J.R. collected and analysed the data. N.D.R., R.V. and W.B. wrote the manuscript with the contributions from all authors.

Funding

This work was supported by the iBOF project NextBioRef. N.D.R. and L.H. are indebted to the Research Foundation Flanders (FWO) for a predoctoral fellowship 1SC6625N and project grant G011620N, respectively. J.R. and H.K. were funded by the Great Lakes Bioenergy Research Center, U.S. Department of Energy, Office of Science, Biological and Environmental Research Program, under Award Number DE‐SC0018409.

Conflicts of Interest

A patent application (PCT/EP2025/050854) stating N.D.R., L.H., R.V. and W.B. as inventors on the use of scopoletin as an alternative lignin monomer in woody biomass, resulting in increased saccharification, has been filed jointly by VIB and Ghent University.

Supporting information

Figure S1: Height of SCOP1 and SCOP2 transgenic poplars after 23 weeks of growth in the greenhouse.

Figure S2: Ratio of the peak intensities of scopolin/scopoletin as detected in extracts of wood from greenhouse‐grown SCOP1 and SCOP2 trees.

Figure S3: Growth of SCOP1 poplars in the greenhouse for in‐depth analysis.

Figure S4: Relationship between coumarin incorporation in the cell wall and plant height in the SCOP1 lines grown under greenhouse conditions.

Figure S5: Saccharification data of wood from SCOP1 poplars.

Figure S6: Selected regions of ¹H–¹³C NMR correlation spectra showing coumarin (Co) H₈/C₈ and H₇/C₇ correlations that indicate syringyl (S) and guaiacyl (G) cross‐coupling and provide unit assignments.

Figure S7: Plot of field trial with SCOP1 poplars.

Figure S8: ABSL of field and greenhouse‐grown poplars.

Figure S9: Relationship between coumarin incorporation in the cell wall and plant height of SCOP1 lines grown under field conditions.

Figure S10: Saccharification time course of field‐grown SCOP1 wood after no pretreatment.

Table S1: Metabolites with a significantly increased abundance in SCOP1 lines compared to the WT. The proposed chemical identities of the compounds are based on accurate mass and MS/MS fragmentation spectra (Data S1).

Table S2: Metabolites with a significantly increased abundance in SCOP2 lines compared to the WT. The proposed chemical identities of the compounds are based on accurate mass and MS/MS fragmentation spectra (Data S1).

Table S3: Targeted search for coumarinolignoids.

Table S4: Phenotypic analysis of greenhouse‐grown SCOP1 poplars.

Table S5: NMR analysis of lignin from greenhouse‐grown SCOP1 and WT poplars.

Table S6: Biomass and phenotype of field‐grown SCOP1 lines.

Table S7: List of primers used for genotyping.

Data S2: MS/MS based structural characterization.

De Ridder, N. , Hoengenaert L., Van Doorsselaere J., et al. 2026. “Simultaneous Overexpression of FERULOYL‐CoA 6′‐HYDROXYLASE 1 and COUMARIN SYNTHASE Leads to Coumarin‐Enriched Lignin and Improved Saccharification in Greenhouse‐ and Field‐Grown Poplar.” Plant Biotechnology Journal 1–19. 10.1111/pbi.70739.

Data Availability Statement

The data that supports the findings of this study are available in the Supporting Information of this article.