Intracellular pathways for lignin catabolism in white-rot fungi

Significance

White-rot fungi play an essential role in global carbon cycling because of their extraordinary ability to extracellularly deconstruct lignin, a recalcitrant plant biopolymer. Despite this, the intracellular metabolism of lignin-deconstruction products by this fungal group has been largely overlooked, potentially due to a lack of genetic tools and challenging growth-tracking protocols, which complicate studies that are considered routine in other organisms. Here, we definitively demonstrate that white-rot fungi are also able to modify and utilize aromatic lignin-deconstruction products as a carbon source. Our study elucidates a critical process occurring in soil ecosystems, encouraging further study of lignin’s catabolic fate in diverse environmental conditions toward an improved understanding of the role of these fungi in facilitating carbon sequestration in nature.

Abstract

Lignin is a biopolymer found in plant cell walls that accounts for 30% of the organic carbon in the biosphere. White-rot fungi (WRF) are considered the most efficient organisms at degrading lignin in nature. While lignin depolymerization by WRF has been extensively studied, the possibility that WRF are able to utilize lignin as a carbon source is still a matter of controversy. Here, we employ ¹³C-isotope labeling, systems biology approaches, and in vitro enzyme assays to demonstrate that two WRF, Trametes versicolor and Gelatoporia subvermispora, funnel carbon from lignin-derived aromatic compounds into central carbon metabolism via intracellular catabolic pathways. These results provide insights into global carbon cycling in soil ecosystems and furthermore establish a foundation for employing WRF in simultaneous lignin depolymerization and bioconversion to bioproducts—a key step toward enabling a sustainable bioeconomy.

Wood is a complex matrix of organic carbon that is recalcitrant to decay, mainly because of the presence of lignin. Lignin is a heterogeneous polyphenol and the second most abundant plant-based biopolymer (after cellulose) on Earth (1). From a utilization perspective, lignin represents up to 40% of the energy density of lignocellulosic biomass (2). Despite the fact that lignin is a massive natural carbon and energy reservoir, only a small group of filamentous basidiomycete fungi, namely white-rot fungi (WRF), have evolved the ability to efficiently depolymerize and mineralize lignin to CO₂ and H₂O (3, 4). Lignin depolymerization by WRF is mediated by the action of extracellular ligninolytic oxidoreductase enzymes, such as laccases and peroxidases, alongside other secreted metabolites (3). Considerable research efforts have been undertaken to understand how WRF depolymerize lignin (5–11), but the biochemical reactions that convert lignin into CO₂ have been largely neglected. Indeed, it is unclear if WRF intracellularly catabolize lignin degradation products to utilize them as a carbon and energy source or, rather, if lignin is depolymerized and mineralized extracellularly by the ligninolytic system (i.e., with peroxidases as previously described) (12–14) to facilitate access to cellulose and hemicellulose for use as a primary carbon source.

The catabolism of aromatic compounds derived from lignin depolymerization has been extensively studied in bacteria (15, 16) and, to a lesser extent, in yeasts (17, 18). Specifically, these microbes funnel monomeric aromatic compounds via “upper pathway” reactions into central aromatic intermediates, which are subsequently ring cleaved by dioxygenase enzymes (15, 16, 19) (Fig. 1, SI Appendix, Fig. S1, and Dataset S1). The ring-opened molecules are further catabolized via “lower pathway” reactions, where the compounds enter central carbon metabolism as an energy and carbon source. Conversely, very few studies have examined the transformation of aromatic compounds in WRF; studies primarily date from the 1970s to the 1990s and predominately report characterization of single enzymatic steps acting on the aromatic ring (20–23). To investigate the potential for WRF to catabolize aromatic compounds from lignin, here we employ ¹³C-isotopic labeling approaches to show that two model WRF, Trametes versicolor and Gelatoporia (= Ceriporiopsis) subvermispora, utilize poplar-derived aromatic compounds as a carbon source. In silico genome analysis led us to hypothesize a complete catabolic pathway for 4-hydroxybenzoic acid (4-HBA), which is an abundant ester-linked and primary aromatic compound commonly liberated from poplar lignin (24–26). Further metabolomic experiments confirmed several catabolic intermediates of the proposed pathway. Proteomic and transcriptomic analyses allowed us to select enzymes potentially involved in those pathways, which were validated for activity via in vitro enzyme assays. This work forms the foundation of a research area based on lignin catabolism by WRF which could be further exploited to convert the still undervalued biopolymer lignin (27, 28) into value-added compounds (19, 29).

Fig. 1.

Potential routes for lignin mineralization to CO₂ by WRF and analyses conducted in the current study. Metabolic pathways that are unclear or have not been described before in WRF are designated by red-dashed lines. Typical syringyl (S), hydroxyphenyl (H), and coniferyl (G) lignin-derived motifs from poplar are also depicted. See SI Appendix, Fig. S1 for additional details on aromatic catabolic pathways.

Results

T. versicolor and G. subvermispora Utilize 4-HBA and Vanillic Acid as Carbon Sources.

T. versicolor and G. subvermispora, both belonging to the same fungal class (Agaricomycetes) and order (Polyporales) (30), were selected for this study because of the availability of sequenced genomes (4, 9) and their distinct lignocellulose degradation mechanisms. T. versicolor performs simultaneous deconstruction of polysaccharides and lignin, whereas G. subvermispora preferentially deconstructs lignin over polysaccharides (3, 31, 32). We first confirmed that these fungi reduce lignin content, by 4.8 ± 0.9% and 11.8 ± 1.0%, respectively, within 14 d, in solid-state cultivations containing poplar as the sole carbon source. We also observed that the concentration of some of the most abundant soluble lignin-derived aromatic compounds decreased (SI Appendix, Fig. S2 and Table S1). These compounds include 4-HBA, vanillic acid, and syringic acid. Based on these findings, we initially tested 4-HBA (7.5 mM) as the sole carbon source to validate its utilization. 4-HBA content decreased over time, and the final conversion was 48% for T. versicolor and 76% for G. subvermispora at 14 d of cultivation (SI Appendix, Fig. S3). WRF form cell aggregates and grow as fungal mats under static conditions, which complicates carbon utilization studies based on growth. The difficult growth conditions, slow growth rates, and lack of genetic tools available for these WRF render these organisms more challenging to study than the faster-growing organisms more commonly used for aromatic catabolic pathway studies. Therefore, in order to determine whether this observed 4-HBA conversion was merely the result of extra- or intracellular modification as opposed to utilization, we used ¹³C-isotope–labeled aromatic compounds to ascertain its utilization as a carbon source. Then, we determined the level of ¹³C labeling in intracellular proteinogenic amino acids, which are metabolites derived from central carbon metabolism (Fig. 2A).

Fig. 2.

¹³C-labeling experiments demonstrate carbon flux of lignin-derived compounds to central metabolism. (A) Abbreviated map showing central carbon metabolic pathways and amino acid biosynthesis in WRF based on the current KEGG model for T. versicolor. (B–E) Substrate level (molar percent) of extracellular metabolites (cellobiose and 4-HBA) in (B) 6-d and (C) 17-d T. versicolor cultivations and (D) 8-d and (E) 16-d G. subvermispora cultivations. A value of 100% corresponds to the initial concentration (7.5 mM in all cases, excluding cellobiose concentration in the 17-d T. versicolor cultivations, which was 3.75 mM). Arrows indicate the sample collection time for ¹³C intracellular analysis. (F–G) Fractional labeling in detected proteinogenic amino acid fragments and other metabolites (acetate and succinate) in (F) T. versicolor and (G) G. subvermispora cultivations when providing unlabeled 4-HBA [negative control, CTL(-)] and ¹³C-ring–labeled 4-HBA. Amino acids fragments (i.e., [15], [57], [85], [159], and [302]) are the result of the derivatization and analysis as thoroughly detailed by Nalsen et al. (33). Individual points are connected with discontinuous lines to facilitate visualization. All results are the average of biological triplicates, and error bars represent the SD. Statistical significance (t test) is presented in SI Appendix, Figs. S4 and S5.

For the isotopic labeling experiments, both WRF were cultivated with either ¹³C-ring–labeled 4-HBA (¹³C-4-HBA) or ¹³C-ring–labeled vanillic acid (¹³C-vanillic acid), in addition to nonlabeled cellobiose to increase both the aromatic conversion rates (SI Appendix, Fig. S4) and the fungal biomass accumulated, thereby facilitating downstream intracellular analysis. Isotopes of syringic acid are not commercially available, and consequently, syringic acid was not included in this experimental campaign. Extracellular and intracellular samples were collected at different times: during 4-HBA conversion (Fig. 2 B and D and SI Appendix, Fig. S4), upon complete depletion of 4-HBA (Fig. 2 C and E and SI Appendix, Fig. S4), and upon complete depletion of vanillic acid (SI Appendix, Fig. S5, see rationale for sample collection in SI Appendix, Supplementary Methods). The two WRF exhibit unique conversion trends. T. versicolor simultaneously converted cellobiose and the aromatic substrate, whereas G. subvermispora converted the aromatic compounds more rapidly than cellobiose (Fig. 2 B–E and SI Appendix, Figs. S4 and S5), mirroring its preferential lignin degradation pattern. Intracellularly, both WRF exhibited enrichment of ¹³C-isotopes from labeled aromatic compounds into proteinogenic amino acids (compared to the unlabeled aromatic substrate) upon complete depletion of 4-HBA (Fig. 2 F and G and SI Appendix, Fig. S4) or vanillic acid (SI Appendix, Fig. S5). In the samples collected during 4-HBA conversion, G. subvermispora showed significant extents of ¹³C labeling (up to 12%, Fig. 2G) when 4-HBA was 60% converted from the media on day 8 (Fig. 2D); in contrast, T. versicolor exhibited lower extents of ¹³C labeling (up to 5%, Fig. 2F) when 4-HBA conversion was as high as 85% on day 6 (Fig. 2B). Because the fractional labeling derives from the incorporation of ¹³C-labeled (natural or exogenous) and nonlabeled carbon into proteinogenic amino acid fragments (33), the previous observation could be the consequence of preferential assimilation of carbon from nonlabeled cellobiose in T. versicolor, leading to a ¹³C-dilution effect, whereas G. subvermispora preferentially assimilates carbon from 4-HBA, leading to the higher fractional labeling. Overall, the significant ¹³C labeling increases observed when feeding ¹³C-4-HBA and ¹³C-vanillic acid compared to the control (unlabeled 4-HBA and vanillic acid) definitively demonstrate that both of these WRF use lignin-derived aromatic compounds as a carbon source.

In Silico Genome Analyses Predict Catabolic Pathways for the Conversion of 4-HBA.

To elucidate the catabolic pathways involved in the conversion of poplar lignin-derived aromatic compounds, we analyzed the genomes of T. versicolor and G. subvermispora by performing homology searches with known aromatic catabolic enzymes of bacterial origin and a selection of the few enzymes of fungal origin that have been biochemically validated to date [i.e., from the yeast Candida parapsilosis (34) and the ascomycete fungi Cochliobolus lunatus (35)] (SI Appendix, Fig. S1 and Dataset S1). We identified multiple homologous sequences for enzymes with putative oxidative decarboxylase activity [a reaction which replaces a carboxyl groups with a hydroxyl group (36)] and hydroxylase activity (Dataset S2), which are among the main biochemical reactions acting on aromatic compounds preceding aromatic ring cleavage (28) (SI Appendix, Fig. S1). We also identified homologous sequences for enzymes with predicted ring-opening dioxygenase activity and enzymes that perform reactions following aromatic ring cleavage, allowing us to hypothesize a complete pathway for the conversion of 4-HBA toward central carbon metabolism (Fig. 3 [see reactions 1 to 11], SI Appendix, Fig. S1). No homology matches were found for aromatic O-demethylases that could demethylate vanillate to protocatechuate, nor were homology matches found for known enzymes involved in the metabolism of syringic acid, which is catabolized via different pathways than 4-HBA and vanillic acid in bacteria (SI Appendix, Fig. S1) (19).

Fig. 3.

Proposed metabolic pathway in T. versicolor and G. subvermispora for the conversion of 4-HBA. Intracellular and extracellular metabolites detected in cellobiose, 4-HBA, lignin, and poplar cultivations (see Dataset S3 and SI Appendix, Figs. S7–S9 for quantitative and qualitative information for each metabolite and SI Appendix, Table S2 for the complete list of metabolites analyzed). As expected, based on previously described syringic acid catabolic pathways (SI Appendix, Fig. S1), none of these metabolites were detected in syringic acid cultivations and neither were the predicted metabolites included as standards in metabolomic analyses (i.e., 3-O-methylgallic acid, gallic acid, pyrogallol, and 2-pyrone-4,6-dicarboxylic acid; SI Appendix, Table S2). Fungal cultivations were conducted in triplicate. A metabolite is considered to be present in biological cultivations if it is detected in at least in two replicates. Media not inoculated with fungi (noninoculated) was also used as control for extracellular metabolomic analyses and intracellular when applicable (i.e., for pellets from lignin and poplar cultivations). Molecules without boxes next to the structure do not have commercially available standards. Continuous gray arrows indicate potential transport through the cell membrane. Continuous and discontinuous black lines correspond to validated (in this work) and proposed enzymatic steps, respectively. The steps are as follows: 1) oxidative decarboxylases GS_120062 and GS_90429, 2) hydroxylase TV_58730, 3) hydroxylation by cytochrome P450, 4) hydroxylases TV_58730 and GS_82057, 5) oxidative decarboxylases GS_120062 and TV_32834 and GS_90429, 6) dioxygenase, 7) ring-cleaving dioxygenases TV_28066 and GS_116134, 8) 4-hydroxymuconic semialdehyde dehydrogenase, 9) maleylacetate reductase, 10) ketoacid CoA transferase, 11) thiolase, 12) CARs, 13) aldehyde dehydrogenase,14) alcohol dehydrogenase, 15) alcohol oxidase, 16) aldehyde oxidase, 17) 4-O-methyl transferase, and 18) demethylase. Additional information for each of these putative enzymatic reactions is detailed in SI Appendix, Text S2.

Spatial Multiomic Analyses Unravel Aromatic Catabolic Pathways in T. versicolor and G. subvermispora.

To support the hypothesized in silico catabolic pathway and provide information about the spatial distribution of these biochemical reactions, we conducted extra- and intracellular proteomics and nuclear magnetic resonance (NMR) spectroscopy-based metabolomics as well as transcriptomics with differential substrate feeding in T. versicolor and G. subvermispora. Cellobiose-containing cultivations were utilized as controls with the following test cases: cultivations containing cellobiose supplemented with 4-HBA, syringic acid, or lignin isolated from poplar (SI Appendix, Fig. S6) and cultivations containing milled poplar chips as the sole carbon source (hereafter referred to 4-HBA, syringic acid, lignin, or poplar cultivations, respectively). Vanillic acid was not included in this set of experiments because, based on the in silico analysis, it was hypothesized that vanillic acid catabolism may share most of the downstream enzymes as those predicted in the pathway for 4-HBA (SI Appendix, Fig. S1). Overall metabolomics (SI Appendix, Text S1 and Figs. S7–S9 and Dataset S3), transcriptomics, and proteomics results (SI Appendix, Text S2 and Figs. S10–S12 and Datasets S4–S6) are shown in SI Appendix.

The multiomic analyses first focused on validating the presence of catabolic intermediates and enzymes from the proposed 4-HBA conversion pathway (Fig. 3 [see reactions 1 to 11], SI Appendix, Fig. S1). For this purpose, we conducted targeted NMR metabolomics, which included metabolites from the hypothetical pathway, when standards were available (SI Appendix, Table S2), and transcriptomics and proteomics, which focused on the enzymes identified in the in silico study (Dataset S2). 4-HBA was detected intracellularly in both WRF in 4-HBA cultivations and also in the intracellular fraction of G. subvermispora in lignin and poplar cultivations (Fig. 3 and Dataset S3). Hydroquinone, 1,2,4-benzenetriol, and protocatechuate were also detected in 4-HBA cultivations of G. subvermispora in the extra- and/or intracellular fraction (Fig. 3). As proposed in the in silico analyses, the putative enzymes involved in the conversion of 4-HBA to these compounds and further ring cleavage exhibit oxidative decarboxylase, hydroxylase, and ring-opening dioxygenase activities (SI Appendix, Fig. S1 and Dataset S1). The selected enzymes (Dataset S2) were organized into clusters using phylogenetic relationships to group putative homologous proteins from each WRF (Fig. 4A) and facilitate the visualization of protein expression and gene regulation patterns. Six putative oxidative decarboxylases (three in each WRF), two hydroxylases (one in each WRF), and six dioxygenases (two in T. versicolor and four in G. subvermispora) were found at significantly higher relative abundances or expression at the protein or transcript level, respectively, in 4-HBA cultivations in the intracellular fraction compared to the control. Some of these enzymes were also more abundant (i.e., oxidative decarboxylases GS_120062 and GS_90429 and the dioxygenases TV_28066, GS_111755, and GS_116559) or their corresponding genes showed higher expression (i.e., the hydroxylases TV_58730 and GS_82057) in lignin and/or poplar cultivations compared to the control (Fig. 4A). Conversely, none of these enzymes or the others selected from the in silico study were detected in the extracellular fraction, though it should be noted that some enzymes could potentially act extracellularly while being membrane bound. Accordingly, we analyzed the presence of predicted signal peptides and protein membrane domains in the targeted enzymes via bioinformatic analyses. Most of the selected dioxygenases and hydroxylases did not present either predicted signal peptides or transmembrane domains (Dataset S2), which strongly suggests that these enzymes act intracellularly. In contrast, many of the selected putative oxidative decarboxylase enzymes may function extracellularly, as they have either predicted signal peptides or transmembrane domains with long noncytoplasmic regions (Dataset S2).

Fig. 4.

Proteomic and trancriptomic analyses and in vitro biochemical validation. (A) Phylogenetic relationships with putative oxidative decarboxylases, hydroxylases, and dioxygenases selected from in silico analyses in T. versicolor (TV) and G. subvermispora (GS). The white/black matrix indicates the enzymes that were selected for an initial “screening” and for further in vitro activity “validation” with purified enzymes. The heat map shows proteomic (P) and transcriptomic (T) results for protein expression and gene regulation levels, respectively, in each growth media compared to the inoculated control (cellobiose-containing media) from biological triplicates. (B, C) Apparent specific activity in µmol NADH (Left) or NAD(P)H (Right) turnover per minute per milligram of enzyme of selected (B) oxidative decarboxylase and (C) hydroxylase candidates on diverse substrates. (D) Apparent specific activity in µmol O₂ consumed per minute per milligram enzyme of selected dioxygenase enzymes on diverse substrates. Substrates and products from these enzymatic reactions are depicted in SI Appendix, Fig. S17. BZT = 1,2,4-benzenetriol; CAT = catechol; HQ = hydroquinone; NS = nonsignificant differential expression compared to the control; CTL(-) = negative control, no substrate; PCA = protocatechuate; SA = syringic acid; U = unique. Enzymes assays were conducted in triplicate and error bars show the SD.

Based on the above observations, we propose that 4-HBA catabolism bifurcates to form hydroquinone and protocatechuate. Specifically, hydroquinone could be generated extra- or intracellularly via a membrane-bound oxidative decarboxylase. The resulting hydroquinone is hydroxylated intracellularly to generate 1,2,4-benzenetriol (Fig. 3). Concomitantly, protocatechuate is formed intracellularly via 4-HBA hydroxylation. Based on the in silico analysis (where we did not identify homologous enzyme candidates for protocatechuate ring cleavage, SI Appendix, Fig. S1), we hypothesize that protocatechuate converted to 1,2,4-benzenetriol via oxidative decarboxylation. To support evidence of the presence of these decarboxylation steps, we conducted an experiment in which we cultivated both WRF with 4-HBA containing a ¹³C-isotope label only in the carboxyl group (not in the carbons of the aromatic ring as in previous experiments). It is worth noting that 4-HBA conversion also involves decarboxylation steps in other alternative catabolic pathways, excluding, as far as we know, the route in which protocatechuate is cleaved via LigAB (4,5-cleavage, which would funnel the carbon from the carboxyl group to oxalacetate and central metabolism; SI Appendix, Fig. S1). No significant increase in the fractional labeling of the proteinogenic amino acids was detected (SI Appendix, Fig. S14) compared to the control (unlabeled 4-HBA) in the selected cultivation conditions, suggesting that the labeled carbon may be released via decarboxylation before entering central metabolism.

Another metabolite from the hypothesized pathway, and downstream ring cleavage, is β-ketoadipate (SI Appendix, Fig. S1). This molecule was not detected, but it can be abiotically decarboxylated, generating levulinic acid (19) (Fig. 3). Levulinate was detected in T. versicolor grown in cellobiose and 4-HBA cultivations, but the relative concentration was fivefold lower in cellobiose than in 4-HBA cultivations (Fig. 3 and Dataset S3). Levulinate was only detected in 4-HBA cultivations of G. subvermispora, which suggests that β-ketoadipate is a potential catabolic intermediate during 4-HBA conversion. It is noteworthy that 4-HBA was identified in the extracellular fraction of poplar cultivations with a concentration 28-fold lower than that observed in 4-HBA cultivations (Dataset S3). Therefore, certain metabolic intermediates, such as β-ketoadipate, may be below the detection limit or fully catabolized at the time of sampling, which could explain why they are generally absent in poplar cultivations.

These multiomic studies were also used to propose alternative catabolic steps. To that end, we analyzed other metabolites included in the NMR metabolomics library (e.g., aromatic substrates with different methoxy and hydroxyl substitutions and positions in the ring; SI Appendix, Table S2). In 4-HBA cultivations, 4-hydroxybenzaldehyde was present in the T. versicolor extracellular milieu, and 4-hydroxybenzyl alcohol was identified in both the intra- and extracellular milieu (Fig. 3 [see reactions 12 to 16]). The generation of 4-hydroxybenzaldehyde has previously been reported as a product of the activity of carboxylic acid reductases (CARs) (37). Three putative CARs were induced intracellularly in 4-HBA cultivations in T. versicolor (SI Appendix, Fig. S15), which supports our observation of 4-hydroxybenzaldehyde generation in this WRF. However, the relative abundance of these T. versicolor CARs was not significantly higher in poplar cultivations under our experimental conditions, whereas one CAR was more abundant in G. subvermispora in poplar cultivations as compared to the control. Methoxylated 4-HBA (4-methoxybenzoate) was also detected extracellularly in both WRF (Fig. 3 [see reactions 17 and 18]). This compound may be generated intracellularly from 4-HBA via 4–O-methyltransferases (38) (SI Appendix, Text S2) and secreted to the extracellular medium.

WRF are unique in their capacity to depolymerize lignin (39). Therefore, we also highlight the differential expression of extracellular oxidoreductases potentially involved in lignin depolymerization (SI Appendix, Tables S3 and S4). Interestingly, laccases were only detected in poplar cultivations, which suggests that the production of these enzymes is triggered by some compound(s) present in lignocellulose and not by monomeric aromatic compounds or oligomeric lignin. Manganese peroxidases (MnPs) were induced in every media compared to the cellobiose-medium control (e.g., excluding T. versicolor in the presence of 4-HBA, which did not express detectable levels of any MnPs); however, the MnP isoenzymes were, in general, different in each medium, corroborating previous observations of complex regulation of MnPs (40). The induction of other extracellular oxidoreductases, such as cytochromes P450 (CYP450s), were unique to cultivations containing lignin and poplar (Dataset S4). We note that different CYP450s were also detected intracellularly and that their abundance significantly increased in the diverse media based on proteomics analysis compared to the control, up to 24 CYP450s in G. subvermispora compared to only four in T. versicolor (SI Appendix, Tables S5 and S6). CYP450s exhibit diverse functionalities (41, 42), among them aromatic O-demethylation and hydroxylation. Understanding how CYP450s relate to aromatic catabolic processes is also a burgeoning area of study in aromatic catabolism. In general, these multiomics results demonstrate that T. versicolor and G. subvermispora, in addition to exhibiting different lignocellulose degradation patterns, also likely harbor different aromatic catabolic pathways and/or regulatory mechanisms.

Generation of 1,2,4-Benzenetriol and Subsequent Ring Cleavage by Dioxygenases Is Key for 4-HBA Utilization.

After using in silico and in vivo analyses to propose aromatic catabolic pathways, we next validated the activity of selected enzymes. Developing gene knockout strains for genetic validation of the proposed pathways would be ideal to confirm function; however, there are no available genetic tools for these WRF. Therefore, we focus on in vitro enzyme assays to validate some of the metabolic steps presented in Fig. 3. We screened for activity of seven putative decarboxylases and two putative hydroxylases selected from the proteomic and transcriptomic analysis (Fig. 4A and Dataset S7). In addition, we tested MNX1 and MNX3 as positive controls, which are previously characterized C. parapsilosis enzymes reported to perform aromatic oxidative decarboxylation of 4-HBA and hydroxylation of hydroquinone, respectively (34). Each enzyme candidate contains a well-defined FAD-binding domain (PF01494) and is predicted to be a NAD(P)H-dependent monooxygenase. As the first step, screening assays were performed using equal total protein concentration from clarified Escherichia coli lysate expressing the gene of interest, in the presence of different coenzyme combinations (NADH, NADH + FAD, NADPH, or NADPH + FAD), where FAD supplementation was included to ensure adequate occupancy of the flavin coenzyme during screening reactions. Enzyme activity was calculated from the rate of oxidation of NAD(P)H using absorbance at 340 nm, which is a commonly used method for measuring these enzyme activities (34, 43). In the screening experiment, all but one of the decarboxylase candidates presented significant turnover of NADH and NADPH in the presence of 4-HBA or protocatechuate, and both of the hydroxylase candidates showed significant turnover of NADH and NADPH in the presence of 4-HBA as substrate (SI Appendix, Fig. S16). The proposed reactions are described in SI Appendix, Fig. S17.

To further decipher 4-HBA catabolism, we selected enzymes that yielded significant activity in the screening and that showed significantly increased relative abundances in the presence of 4-HBA for further purification and characterization efforts. This included two pairs of decarboxylases (TV_32834 and GS_90429 and TV_175239 and GS_120062) and a pair of hydroxylases (TV_58730 and GS_82057). These enzyme pairs cluster together as homologous sequences from each of the two fungi in the phylogenetic tree (Fig. 4A). For oxidative decarboxylation reactions (Fig. 3, see reactions 1 and 5), TV_32834 preferentially decarboxylates protocatechuate, proposed to yield 1,2,4-benzenetriol using NADH (Fig. 4B). In contrast, its closely related homolog GS_90429 shows relatively lower apparent specific activity on all tested substrates, with highest apparent activity in the decarboxylation of 4-HBA to yield hydroquinone in the presence of NADPH. This suggests the presence of unique structural characteristics in these two closely related homologs that confer different substrate specificities to the proteins. Enzyme GS_120062 demonstrates oxidative decarboxylation activity on both 4-HBA and protocatechuate, preferentially in the presence of NADH (Fig. 4B), while its homolog TV_175239 was not soluble upon purification from the lysate. For the hydroxylation reactions (Fig. 3, see reactions 2 and 4), both TV_58730 and GS_82057 were active on hydroquinone and proposed to form 1,2,4-benzenetriol, and, to a lesser extent, TV_58730 was active on 4-HBA for the production of protocatechuate. Both enzymes show higher preference for NADPH in comparison to NADH and there was not turnover in the absence of substrate (Fig. 4C). Despite these positive results, it is worth mentioning that cofactor turnover observed at 340 nm may be also the result of futile cycling. Characterization of product formation will be required to understand the level of decoupling in these reactions.

Ring-opening dioxygenase activity was also evaluated (Fig. 3, see reactions 6 and 7). Thirteen dioxygenase sequences were identified through homology searches, of which seven sequences likely encode intradiol dioxygenases, two sequences likely encode extradiol dioxygenases, and four sequences resemble homogentisate dioxygenase (SI Appendix, Table S7) (44). Screening assays were attempted where product formation is measured spectrophotometrically, a common technique used in the study of these enzymatic pathways (45–47); however, these spectrophotometric signals were below the limit of quantification and subject to interference from abiotic oxidation of both 1,2,4-benzenetriol and hydroquinone (48), key substrates in the proposed in silico pathway (Fig. 3). As an alternative method, oxygen consumption measurements were performed on selected and purified enzymes to determine specific activity. Abiotic oxidation of 1,2,4-benzenetriol and hydroquinone was supressed via the addition of reducing agent Tris(2-carboxyethyl)phosphine. Based on proteomics and transcriptomics targets, five dioxygenase candidates (TV_115964, TV_28066, GS_116134, TV_116559, and GS_117547) as well as a positive control, the catechol-1,2-dioxygenase enzyme (CatA) from Pseudomonas putida KT2440, were selected for this study. Other than the control enzyme, only two WRF enzymes, TV_28066 and GS_116134, remained soluble upon purification. Both fungal proteins consumed molecular oxygen only in the presence of 1,2,4-benzenetriol (Fig. 4D). These two enzymes showed little to no oxygen consumption in the presence of catechol, hydroquinone, and protocatechuate, suggesting that 1,2,4-benzenetriol is the primary substrate. Overall, we have assigned a function to six fungal enzymes based on in vitro activity. Further characterization of these and additional aromatic catabolic enzyme candidates will be a high priority for continued studies.

Discussion

In response to a changing climate, there is growing interest in both gaining a better understanding of carbon cycling (49, 50) and developing technologies for a sustainable bioeconomy (19). While deciphering the catabolism of aromatic compounds by filamentous fungi is at an early stage (51, 52), it is highly relevant to these pursuits and is especially appealing for the valorization of lignin (27). Lignin is an undervalued substrate in lignocellulosic biorefineries because of lack of efficient depolymerization tools, either biological or chemical, which ultimately hampers the viability of plant-based biorefineries (27, 28). Aromatic catabolic bacteria have been demonstrated to be potential biocatalysts for upgrading monomeric aromatic compounds into value-added products (19). However, lignin depolymerization capabilities in this bacterial group is limited (53). Therefore, unraveling the mechanisms for aromatic catabolism in the most efficient lignin-degrading organisms in nature provides a step forward toward developing consolidated bioprocesses for lignin which involve simultaneous deconstruction and utilization of the degradation products for the generation of value-added compounds [similar to that described for cellulose (54)]. Furthermore, the insights gained in this study may also apply to other burgeoning industrial applications, including degradation and upcycling of recalcitrant plastics (55).

The ability of WRF to utilize lignin (and ultimately lignin-derived aromatic compounds) as a carbon and energy source is still a matter of debate in the scientific community. T. versicolor has previously been demonstrated to utilize the aromatic contaminant benzophenone-3 as a carbon source via ¹³C-labeling analyses (56), but catabolic pathway information in the field of xenobiotics detoxification is also limited. Our study employs a similar approach using ¹³C-labeled compounds and conclusively demonstrates that T. versicolor and G. subvermispora utilize lignin-derived 4-HBA and vanillic acid as carbon sources (Fig. 2). The high ¹³C-fractional labeling of certain amino acids—such as proline, glutamate, and aspartate, which are derived from metabolic intermediates in the glyoxylate and tricarboxylic acid cycles (Fig. 2 A, F, and G)—supports that carbon from aromatic compounds may enter central metabolism via acetyl-CoA and succinyl-CoA (Fig. 3). However, the levels of ¹³C-labeled alanine (derived from pyruvate, Fig. 2A) were also significantly higher when compared to the controls, which may be the result of carbon flux from malate to pyruvate via a maleylacetate dehydrogenase or an unknown pathway which involves 2,3- or 4,5- metacleavage of protocatechuate (although the latter is not supported by our in silico analysis) (SI Appendix, Fig. S1). Interestingly, the ¹³C-fractional labeling of certain amino acids, such as threonine and isoleucine, was significantly lower in T. versicolor as compared to G. subvermispora. According to the published Kyoto Encylopedia of Genes and Genomes (KEGG) model for T. versicolor (Fig. 1A; currently, there is not an available model for G. subvermispora), threonine can be only derived from glycine. However, there is an alternative pathway that could also generate threonine from aspartate, which is not complete in the KEGG model for T. versicolor; specifically, an aspartate-semialdehyde dehydrogenase (EC 1.2.1.11) is missing. Preliminary genome analyses predict the presence of putative aspartate-semialdehyde dehydrogenases in both WRF. However, based on these results, and before the activity of this specific enzyme is validated, it is suggested that the labeling disparities between these WRF may also originate from the presence of different metabolic pathways. Future fluxomic analyses will be key to understanding the carbon flux from aromatic compounds to central metabolism and to unraveling catabolic bottlenecks as well as for mapping more broadly central carbon metabolism in these fungi.

With regard to catabolic bottlenecks, 4-hydroxybenzaldehyde was detected extracellularly in the metabolomic analyses of T. versicolor in 4-HBA cultivations (Fig. 3) and in the extracellular milieu of T. versicolor in the ¹³C-labeling studies (up to 25% mol conversion from the initial 4-HBA concentration supplied) (SI Appendix, Figs. S3 and S4). In contrast, this metabolite was not detected in G. subvermispora cultivations in any case. These results suggest that downstream enzymatic conversion of 4-hydroxybenzaldehyde may be a potential bottleneck during 4-HBA catabolism in T. versicolor or that G. subvermispora does not generate the aldehyde (as supported by the lack of putative CARs in 4-HBA cultivations, SI Appendix, Fig. S15). The reason for the formation and accumulation of 4-hydroxybenzaldehyde in T. versicolor, which is an adenosine triphosphate–consuming step when catalyzed by CARs, is unclear, especially considering that aldehydes are in general more toxic than the corresponding acids (57).

Systems biology approaches and the use of enzyme phylogenetic studies (Fig. 4) were essential to map out a 4-HBA conversion pathway and downselect enzymes for further pathway validation. It is worth noting that transcript and protein abundance measured in transcriptomics and proteomics datasets, respectively, did not always correlate for the selected targets. This fact may reflect the different gene and protein dynamics at the time point in which the multiomics samples were taken. Interestingly, even though we selected homologous enzyme pairs from both WRF with similar omics results, in a few cases, only one of the studied fungi showed activity for the proposed substrate. That was the case for reaction one (4-HBA oxidative decarboxylation in G. subversmispora) and reaction two (4-HBA hydroxylation in T. versicolor). Based on these observations, we hypothesize that 4-HBA preferentially undergoes oxidative decarboxylation to hydroquinone and subsequent hydroxylation to benzenetriol in G. subvermispora before ring cleavage, whereas 4-HBA would preferentially undergo hydroxylation to protocatechuate and further oxidative decarboxylation to benzenetriol in T. versicolor (Fig. 3). Additional screening with other decarboxylases and hydroxylases (i.e., those showing significantly higher abundances in lignin or poplar) will be necessary to test this hypothesis. Furthermore, examining enzymes from other protein families that can perform the same or similar oxidative reactions, such as CYP450s with aromatic hydroxylation activity (Fig. 3 [reaction 3], SI Appendix, Text S2), will also be key for elucidating enzyme preferences for specific substrates.

In a broader context, one of the major questions raised in this study was the location of lignin mineralization processes for WRF. Based on our proposed pathway for the conversion of 4-HBA—which exhibits equivalent ring functionality to the p-coumaryl (H) motif in lignin—we hypothesize that the carbons from the aromatic ring could be sequestered intracellularly (and some of them incorporated as fungal biomass), whereas the carbon from the carboxyl group is released to the atmosphere as CO₂. Carbon storage in soils is at least three times higher than in the atmosphere or plants (58, 59), and decomposer fungi, among them brown-rot fungi (which do not depolymerize lignin) and WRF, are found to be abundant in the organic matter of shallow soils (49, 50). Considering the unique capabilities of WRF for breaking down lignin and utilizing lignin degradation products, this positions WRF as key players in the sequestration of lignin-derived carbon in soils. Furthermore, understanding the catabolism of coniferyl (G) and syringyl (S) motifs in lignin, such as vanillic acid and syringic acid-like aromatic compounds, will provide additional information on carbon sequestration from their methoxy groups. For instance, bacteria such as Sphingobium sp. SYK-6 also incorporate the methoxy group from vanillic acid into central metabolism and amino acids via C1 pathways, and SYK-6 utilizes an initial demethylation reaction for the catabolism of vanillic acid (60).

Multiomic analyses intended to elucidate syringate metabolism—which has not yet been described in filamentous fungi (51)—were also performed in this study. However, because of a lack of homologous enzyme candidates from in silico studies to target for in vitro characterization and few extracellular metabolites identified in the metabolomics study (Dataset S3 and SI Appendix, Text S3 and Fig. S13), the results were inconclusive. In fact, none of the metabolites predicted to be downstream syringic acid (SI Appendix, Fig. S1), such as 3-O-methylgallic acid, gallic acid, pyrogallol, and 2-pyrone-4,6-dicarboxylic acid, were detected. Nevertheless, the abundance of a series of unique decarboxylases and dioxygenases significantly increased in the presence of syringic acid relative to 4-HBA (Fig. 4), which provides a useful foundation for further studies. Chemical synthesis of ¹³C-labeled syringic acid for feeding to the WRF will be necessary to verify incorporation of carbon from this compound in central metabolism and to pursue pathway elucidation.

Going forward, WRF performance warrants further examination in modeled environmental conditions (e.g., solid-state cultivations instead of submerged cultivations) and in the presence of different abiotic factors in order to better understand regulatory processes and rates for lignin degradation and catabolism. We note that lignin degradation products are likely chemically diverse and difficult to track because of their inherent reactivity; therefore, time-course studies focused on common aromatic catabolic intermediates, upstream and downstream of ring cleavage events, will be key to unraveling carbon flux from lignin in these organisms. The utilization of dikaryon fungal strains (instead of monokaryons, as used in this study) will be advantageous to accelerate growth (61) and most likely, lignin utilization. Concomitantly, the development of efficient genetic tools in WRF will be necessary in order to ascertain gene-function relationships and to tune the metabolism for generation of value-added products. Lastly, the findings from this study imply that annotation, analysis, and inclusion of aromatic catabolic pathways in genomics and systems biology studies of lignin-degrading WRF is a worthwhile pursuit.

Methods

A brief summary of relevant materials and methods is presented below. Additional details are shown in SI Appendix, Supplementary Methods.

Lignin Extraction from Poplar.

Milled wood lignin was prepared from extractives-free poplar (Populus trichocarpa) according to the Björkman method with some modifications (62, 63). Compositional analyses and two-dimensional (2D) heteronuclear single-quantum correlation (HSQC) NMR spectroscopy were conducted in the original wood poplar and extracted lignin (SI Appendix, Figs. S2 and S6).

Fungal Strains and Media Preparation.

Monokaryon strains of Gelatoporia (Ceriporiopsis) subvermispora FP-105752 and Trametes versicolor FP-101664 were used in this study. G. subvermispora FP105752 RP-B was kindly provided by Daniel Cullen from the US Department of Agriculture Center for Forest Mycology Research (CFMR). T. versicolor (L.) Lloyd FP-101664-Sp was received from CFMR. Plates with fungal mycelia were maintained at 4 °C. Yeast-maltose-peptone-glucose (YMPG)-agar medium was used to maintain the strains and to prepare seed cultures. An adapted version of the minimal and defined Czapek–Dox medium (CDM) was developed for the corresponding experimental campaigns. Carbon sources and concentrations added to CDM for the submerged-state cultivations were specific to each experimental campaign, and the preparation is detailed in SI Appendix, Supplementary Methods.

Seed Culture Preparation.

To generate seed cultures in an equivalent metabolic state, WRF were subjected to a series of consecutive cultures in YMPG medium. An agar square with fungal mycelia was taken from agar plates maintained at 4 °C, deposited on an agar plate, and incubated for 7 d at 28 °C. Then, 4 to 5 agar squares with mycelia were inoculated in 50 mL of liquid YMPG and incubated at 220 rpm and 28 °C for 7 d (preseed). Resulting fungal pellets were homogenized, and the fungal suspension was reinoculated in YMPG under the same conditions than the preseed culture for 5 d. A ratio of 1:12.5 (v:v) of homogenized fungal suspension per media volume was used to inoculate the corresponding cultivations.

Fungal Cultivations.

All fungal cultivations were conducted in triplicate. All the experimental campaigns included controls which are detailed in SI Appendix, Supplementary Methods.

Solid-state cultivations on poplar.

Solid-state cultivations on poplar were performed in 125 mL flasks containing 2 g of milled poplar and 8 mL of H₂O. Cultivations were incubated in static conditions at 28 °C for 14 d. At the end of the cultivations, cultures were washed and filtered. The liquid fraction was analyzed using ultra high-pressure liquid chromatography–tandem mass spectrometer (UHPLC-MS/MS). The solid fraction was dried at 60 °C, weighed, and utilized for further compositional analysis.

Submerged cultivations with 4-HBA as the sole carbon source.

Cultivations with 4-HBA (7.5 mM) as the sole carbon source were conducted in 250 mL flasks with 25 mL of CDM and incubated at 28 °C in static conditions for 14 d.

Submerged cultivations with ¹³C-labeled aromatic compounds.

Submerged cultivations with ¹³C-labeled aromatic compounds were performed using ¹³C-ring–labeled 4-HBA, ¹³C-labeled 4-HBA in the carboxyl group, and ¹³C-ring–labeled vanillic acid in 125 mL flasks containing 12.5 mL of media and incubated at 28 °C in static conditions. Cultures were harvested at 8 and 16 d in G. subvermispora and at 6 and 17 d in T. versicolor (SI Appendix, Supplementary Methods). CDM containing 7.5 mM of cellobiose was supplemented with antioxidants and the corresponding ¹³C-labeled aromatic compound (7.5 mM), with the exception of 17-d T. versicolor cultures in which 3.75 mM cellobiose was utilized. Downstream analyses in the harvested fungal pellets were performed to detect ¹³C-labeled intracellular proteinogenic amino acids and other metabolites from central metabolism using gas chromatography (GC)–mass spectrometry. Samples were taken periodically from the extracellular milieu to analyze extracellular metabolites via UHPLC-MS/MS.

Submerged cultivations for multiomic experiments.

Cultivations for multiomic experiments were conducted in 250-mL flasks containing 25 mL CDM with cellobiose (5 g/L) and incubated at 28 °C. T. versicolor and G. subvermispora cultivations were supplemented with antioxidants and induced with different substrates at 5 d and 12 d, respectively. The substrates were 4-HBA (2 mM), syringic acid (2 mM), lignin isolated from poplar (25 mg), or poplar (2 g, added at time 0). Noninduced cultures (containing only cellobiose) were used as controls. Cultivations were harvested 48 h after the induction for further downstream metabolomic, proteomic, and transcriptomic analyses.

Bioinformatic Analyses.

DNA and protein sequences corresponding to G. subvermispora (9) and T. versicolor (4) strains were retrieved from the Joint Genome Institute (JGI) Mycocosm database (64). Genome analysis and in silico pathway discovery was performed via homology searches using biochemically validated aromatic catabolic enzymes (Dataset S1). A list of target enzyme candidates was compiled from all these analyses for putative hydroxylase, decarboxylase, and dioxygenase reactions in order to further analyze these proteins in the context of the multiomics experiments and to perform enzyme assays (Dataset S2). CARs were also identified. Protein nomenclature displayed in this work corresponds to JGI protein number preceded by prefixes “TV” and “GS” in the case of T. versicolor and G. subvermispora entries. Phylogenetic trees were constructed using the maximum likelihood method. Protein location was predicted using the InterPro integrated tools (65).

Multiomics Analyses.

Metabolomics.

Supernatants and pellets were collected from the fungal cultivations to generate extracellular and intracellular samples, respectively. Metabolites from supernatants and pellets [the latter subjected to the metabolite, protein, lipid extraction (MPLEx) protocol (66)] were analyzed via NMR.

Proteomics.

Supernatants and pellets were collected from the fungal cultivations to generate extracellular and intracellular samples, respectively. Pellets were subjected to the MPLEx protocol (66). All the samples were tryptically digested and analyzed using liquid chromatography with MS/MS. Data were analyzed and statistically assessed to compare protein abundances.

Transcriptomics.

Transcriptomics analyses were conducted on pellets. Total RNA was extracted and used to generate a complementary DNA library for the Illumina NextSeq 550 platform. Samples were sequenced and the obtained reads were mapped to genes. Differential expression analyses were performed.

Enzyme Production.

Plasmid construction.

pET11-based expression plasmids for WRF genes as well as control genes from P. putida KT2440 and C. parapsilosis were generated as further described in SI Appendix, Supplementary Methods.

Protein expression and purification.

E. coli-based protein expression and chromatographic purification is described in SI Appendix, Supplementary Methods.

Enzyme Activity Assays.

Screening assays were performed using lysate from E. coli containing heterologously expressed protein of interest, and reaction progress was measured by monitoring turnover of NAD(P)H via spectrophotometric signal at 340 nm. Enzyme kinetics were evaluated using purified protein. Hydroxylation and oxidative decarboxylation reactions were measured spectrophotometically to monitor NAD(P)H turnover using signal at 340 nm. Ring cleavage reactions mediated by dioxygenase enzymes were monitoried using a Clark-type polarographic O₂ electrode to monitor O₂ consumption as described further in SI Appendix, Supplementary Methods

Additional Analytical Work.

Compositional analysis (67) and 2D HSQC NMR spectroscopy analyses (68) on poplar and extracted lignin were conducted as described before. The analysis of aromatic compounds and cellobiose in the extracellular milieu in all the cultivations was conducted via UHPLC-MS/MS (53) and HPLC, respectively. Analysis of intracellular proteinogenic amino acids and other metabolites from central metabolism from ¹³C- labeling experiments were analyzed via GC as reported elsewhere (69).

Data Availability

Transcriptomics data have been deposited in Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi, study no. GSE151629) (70). Mass spectrometry proteomics data are deposited in the ProteomeXchange (71) Consortium via the PRIDE (72) partner repository with the dataset identifier PXD019839 (73). All other data are available in the article, SI Appendix, and Datasets S1–S7.