Gene Regulation Shifts Shed Light on Fungal Adaption in Plant Biomass Decomposers

Fungi dominate the turnover of wood, Earth’s largest pool of aboveground terrestrial carbon. Fungi first evolved this capacity by degrading lignin to access and hydrolyze embedded carbohydrates (white rot). Multiple lineages, however, adapted faster reactive oxygen species (ROS) pretreatments to loosen lignocellulose and selectively extract sugars (brown rot). This brown rot “shortcut” often coincided with losses (>60%) of conventional lignocellulolytic genes, implying that ROS adaptations supplanted conventional pathways. We used comparative transcriptomics to further pursue brown rot adaptations, which illuminated the clear temporal expression shift of ROS genes, as well as the shift toward synthesizing more GHs in brown rot relative to white rot. These imply that gene regulatory shifts, not simply ROS innovations, were key to brown rot fungal evolution. These results not only reveal an important biological shift among these unique fungi, but they may also illuminate a trait that restricts brown rot fungi to certain ecological niches.

F ungi dominate the turnover of wood, Earth's largest aboveground terrestrial carbon pool, and they promise commercially relevant pathways for deconstructing lignocellulose. The mechanisms of these fungi vary, with some removing lignin to gain access to cellulose and hemicellulose (white rot type) and others selectively extracting carbohydrates (brown rot type). These nutritional modes represent a spectrum of carbohydrate selectivities (1), but there is a threshold of wood lignin loss that divides white rot (lignin loss is higher) from brown rot (2,3), and as names imply, color is a distinguishing feature. Furthermore, wood solubilities (3), lignin demethylation patterns (4), and strength loss rates (5) in brown-rotted wood all reflect fungal mechanisms that are distinct from the mechanisms of white rot fungi.
Despite these reliable physiochemical distinctions between brown-and whiterotted wood (phenotypes), it remains unclear what genetic adaptations or mechanism shifts enable these distinct pathways. One obstacle is that brown rot fungi have evolved independently from at least seven different white rot lineages (6,7). Model fungi from different lineages have been used to inform our working model for brown rot, but often in isolation and with mixed conclusions. Unfortunately, as genomes have become increasingly available, comparative genomics of wood-degrading fungi (8)(9)(10)(11) have not identified clear, unified features that distinguish rot types. Another obstacle involves a paradox between lignocellulolytic genes and decay rates. It is evident that the evolution of brown rot fungi involved shedding lignin-degrading peroxidases (PODs) (1,12,13) and contracting the number of genes coding carbohydrate-active enzymes (CAZYs), including glycoside hydrolases (GHs) (7,14). Having fewer genes to encode familiar decay enzymes, however, has not slowed brown rot relative to white rot. Brown rot is typically faster than white rot, and this is exemplified by a standard wood durability test that requires shorter incubation periods for brown rot fungi than for white rot fungi (15).
This contradiction between brown rot fungal gene losses and efficiency gains (the "more-with-less" paradox) could be explained by novel genes or regulatory mechanisms (explanation 1), higher expression of the few genes retained (explanation 2), or some combination of both (explanation 3). Most have suggested novel pathways (explanation 1) are the key to brown rot, specifically mechanisms to deploy reactive oxygen species (ROS) as a pretreatment ahead of CAZYs. This has been posited for decades (16) and repeatedly supported by biochemical evidence (17)(18)(19), with irondependent Fenton chemistry as a unifying feature (20,21). This ephemeral (Ͻ48 h) ROS pretreatment step has also been isolated, physically (22), to reveal a brief upregulation of lignocellulose oxidation (LOX) genes at the hyphal front of Postia placenta (23). These "genes of interest" include some important to white rot mechanisms (e.g., GMC oxidoreductases for H 2 O 2 production [24]) and some linked to Fenton chemistry (e.g., phenolate-derived iron reductants [25]), relevant for brown rot. The functions of most of these genes, however, are poorly described or unknown and would require significant time/risk to characterize one by one, given a lack of precedent to inform annotations. Gene-by-gene approaches would also inherently assume that novel pathways, specifically ROS-related genes, are the key to brown rot (explanation 1).
Our goal with this study was to harness an experimental setup that could narrow the pool of novel ROS pathway targets (per explanation 1), but that also addressed explanations 2 and 3 as alternate hypotheses. To do this, we grew brown and white rot fungi from distinct clades along the length of wood wafers, and assessed the whole transcriptome from sections removed at set distances behind the hyphal front. This provided a time series of gene expression for tracking differentially expressed genes and for overlaying with wood physiochemical data and enzyme activities. By using these data for functional genomics comparisons among species, we reduced 10-fold the pool of candidate genes unique to brown rot fungal ROS pathways, a major advance in its own right. By looking at the temporal expression, we found the clear shift of LOX genes toward functioning in early decay stages in brown rot compared to white rot. By binning transcripts into functional groups, we also observed a surprising distinction in energy investments between white and brown rot fungi. It was apparent that the brief investment by brown rot fungi in oxidative mechanisms enabled a much higher transcript investment in CAZYs. This supports hypothesis 3, explaining brown rot innovations as a combination of novel pathways and deeper expression investments, and it implies that the evolution of ROS pathways by brown rot fungi did not obsolesce hydrolytic enzymes-it enabled them.

RESULTS
Temporal sequence of wood decay among test species. Four late-diverging fungi (relative to the evolution of other brown rot clades from white rot ancestors) from distant clades were studied, representing brown rot (Postia placenta and Gloeophyllum trabeum) and white rot (Trametes versicolor and Pleurotus ostreatus) nutritional modes. All four fungi, when grown directionally, from one end of a wood wafer toward the other end (Fig. 1A), created spatial gradients of decay with uniform hyphal fronts that benchmarked the hyphal front section and enabled reconstruction of time series. These sections included "early decay" at the hyphal front (0 to 5 mm; 30-to 50-h growth period) where fungal biomass (via ergosterol) was abundant but where wood mass losses were negligible or there was apparent mass gain as fungal biomass accrued ahead of wood decay ( Fig. 1B and C). The dilute alkaline solubility (DAS) test could not distinguish 0-to 5-mm sections from nondecayed wood for any of the fungi tested, but as decay progressed, DAS increasingly distinguished brown from white rot wood modifications (Fig. 1D).
Transcriptomic data by transcriptome sequencing (RNA-seq) indicated that 77 to 87% of genes in the genomes of the fungi tested were expressed when these fungi decayed solid wood (see Fig. S1 in the supplemental material and Data Set S1 available at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?accϭGSE108189), and principalcomponent analyses (PCA) of transcriptomes clearly revealed the distinct upregulation patterns in early decay zones, relative to older sections for all fungi tested (Fig. 1E). These RNA-seq data were validated with quantitative reverse transcription-PCR (qRT-PCR) using 57 genes relevant for lignocellulose degradation (see Fig. S2 in the supplemental material and Data Set S2 available at https://www.ncbi.nlm.nih.gov/geo/ query/acc.cgi?accϭGSE108189) and further used to reannotate genomes (Fig. S1). Collectively, this enabled downstream analyses of differential gene regulation between early and later decay stages, and it will enable others going forward.
Differentially expressed genes over fungal decay sequences. "Decay-stagedependent" differentially expressed genes (DEGs) (fold changes [FC] of Ͼ4; falsediscovery rate [FDR] of Ͻ0.05; n ϭ 3) in early or late decay stages were assigned for each species (see detailed criteria in Text S1 in the supplemental material), by following the decreasing or increasing expression trends along with decay, respectively (see Fig. S3 in the supplemental material and Data Set S3 available at https://www.ncbi.nlm .nih.gov/geo/query/acc.cgi?accϭGSE108189). The global temporal gene expression trends were immediately compared among species using ortho-DEGs, which display two major clusters more likely distinguished by nutritional modes (i.e., brown rot and white rot), rather than by fungal taxonomy (1, 2, 7, 10) ( Fig. 2A).
Among these "decay-stage-dependent" genes, those with (hemi)cellulosedegrading O-glycosyl hydrolase functions (e.g., GO:0004553 and GO:0008810) were significantly overrepresented in late decay for all four fungi (FDR Ͻ0.05, Gene Ontology [GO] enrichment analysis by Fisher's exact test) (see Table S1 in the supplemental material). Notably, iron ion binding functions (GO:0005506) were significantly overrepresented in early decay for both brown rot fungi, but also enriched in late decay stages for white rot fungi. This indicated the divergences of the temporal gene regulations between two wood-decaying modes besides the similarities.
Brown rot-specific DEGs. By using orthologous genes (the ortho group can be found in Data Set S1 available at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc ϭGSE108189), we further pinpointed 52 "decay-stage-dependent" DEGs that distinguished brown from white rot species ( Fig. 2B; Table 1), which greatly reduced the pool of genes of interest (e.g., those potentially driving ROS production) by more than 10-fold, relative to our previous study (23).
Thirty-four of these DEGs showed functions unique to early brown rot, coincident with the timing of ROS-based pretreatments. These included the genes encoding ferric reductase (FER; MCL_3182) and heme-thiolate peroxidase/peroxygenase (HTP; MCL_2183) ( Table 1; see Fig. S4 and Fig. S5 in the supplemental material). This brown rot FER has a "NOX_Duox_like_FAD_NADP" domain that may catalyze the generation of ROS (e.g., H 2 O 2 ) (NCBI Conserved Domains, E value of Ͻ3.06eϪ38), making it plausible that brown rot fungi have coopted the NOX (NADPH oxidase) domain in similar fashion as some pathogenic fungi that infect plants and degrade cellulose (26,27). The brown FIG 1 A fine-scale time series of wood decay using directed fungal growth in wafers. (A) Wood decay stages during brown rot (by P. placenta and G. trabeum) or white rot (by T. versicolor and P. ostreatus) were manually sectioned in aspen wafers, including a section without fungal growth (no fungus) and three sections behind the hyphal front (0 to 5 mm, 15 to 20 mm, and 30 to 35 mm). A total of 24 independent wafers (n ϭ 8, each represented) were sectioned per fungus per experiment, using the hyphal front to benchmark progression. A fungal biomarker, ergosterol (B), is shown relative to late-decay-stage levels (as %) and indicates significant biomass in the 0-to 5-mm "early" section. Wood mass loss (C) and dilute alkaline solubility of wood (D [a sensitive test of early brown rot modifications]) were measured to assess the decay intensity, and neither could distinguish the 0-to 5-mm section from sound wood (ANOVA HSD test, P Ͼ 0.1). These results, combined, verified that 0-to 5-mm sections were well-colonized but lacked discernible decay, and principal-component analyses (E) clearly distinguished the whole-genome expression patterns in this "early" section (0 to 5 mm) from those in advanced-decay sections (15 to 20 mm and 30 to 35 mm) for all four test fungi. Three bioreplicates from each section (i.e., A, B, and C) were well clustered, indicating the reproducibility of the RNA-seq results. rot unique HTP might be involved in halogenating, hydroxylating aromatic molecules and cleaving lignin, as found in Caldariomyces fumago chloroperoxidases and Agrocybe aegerita aromatic peroxygenases (28-32) (see the Text S1 in the supplemental material for in silico functional analysis of FER and HTP). In addition to these two, we also found four "hypothetical proteins," and four transcriptional factors were uniquely coupregulated in early brown rot that offer targets to better characterize decaying functions and regulatory mechanisms ( Table 1).
Eighteen of these 52 DEGs showed functions unique to late brown rot, upregulated after oxidative modifications of the substrate (17,22,23). Specifically, genes for cellulose hydrolysis and methylglyoxal/aldehyde detoxification were upregulated (Table 1), including one endo-␤-1,4-glucanase gene (MCL_7542) and two ␤-glucosidase genes (MCL_2441 and MCL_2248), one carboxyl esterase gene (MCL_7264), and one sugar transporter gene (MCL_6643), along with many aldo keto reductase DEGs. The products of these genes may have evolved characteristics adapted to oxidatively reacted wood components such as carbonyl and carboxylic acid groups introduced on residual carbohydrates (21,33). In line with this, cellulases and hemicellulases of brown rot fungi were recently shown to better tolerate Fenton-generated oxidative radicals, relative to those enzymes in Trichoderma reesei (34). Overall, these unique late-stage dynamics may involve genes inherited from white rot ancestors (1, 10) with functions adapted to substrate conditions that characterize brown rot, including many unexplored options relevant to industrial processes. In addition, those genes that are "unique to white rot" and "common to all" are discussed in Text S1 in the supplemental material ( Fig. 2B; see Table S2 in the supplemental material).
Temporal regulation shift of CAZY genes in brown rot versus white rot. Regulatory patterns of lignocellulose-degrading gene families revealed staggered control of lignocellulose-oxidizing (LOX; upregulated early) and glycoside-hydrolyzing (GH; upregulated later) functions as fungi decayed wood wafers, especially during brown rot ( Fig. 3; see Table S3 in the supplemental material). Brown rot LOX genes were, collectively, 10ϫ more upregulated in early than in late decay stages, significantly higher than in white rot, where DEGs were slightly greater in later decay stages (Fisher's exact test, P Ͻ 0.001) (Fig. 4A). By further looking at the contrast of early and late DEGs number in each functional CAZY family, it is clear that brown rot LOX expression was The similarity of global temporal gene regulation among species clearly indicated two major clusters distinguished by nutritional modes of brown rot and white rot. The temporal gene expression trends of 2,093 orthologous DEGs were used to calculate the correlations among fungal samples by Spearman's method. "A," "B," and "C" represent triple bioreplicates of RNA-seq data in each species. (B) Unique and common differentially expressed genes (DEGs) as a function of species and wood decay stage. The "decay-stage-dependent" genes, i.e., genes upregulated "early" or "late," were compared among four species with the orthologous genes. The same MCL ID was assigned to genes from one ortho group, while species-specific ID was used for that without orthologues in other species. "Late" DEGs were combined from 15-to 20-mm and 30-to 35-mm sections. The brown rot fungi are P. placenta and G. trabeum, and the white rot fungi are T. versicolor and P. ostreatus.
Gene Regulatory Shifts Enable Efficient Wood Decay ® shifted to the early decay stage compared to that of white rot (two-tailed paired t test, P Ͻ 0.01 [ Fig. 4B]). Specifically, iron-associated genes glycopeptide, heme-thiolate peroxidase, iron reductase, and cytochrome P450 DEGs were more likely correlated with early ROS pretreatment of brown rot, not white rot (Fisher's exact test, P Ͻ 0.05) (Fig. 3), in line with the brown rot-specific "decay-stage-dependent" DEGs pinpointed in the Venn diagram (Fig. 2B) and Table 1. The main (hemi)cellulose chain-cleaving GHs (e.g., GH5, -6, -7, -10, and -12) and sugar transporters were generally upregulated during late decay in both brown rot and white rot (early versus late DEGs' number, two-tailed paired t test, P Ͻ 0.05 [ Fig. 4A and B]). These together indicated the distinctive temporal regulatory features in brown rot, which has led the formation of the staggered "two-step" regulation ("LOX-then-GH") (23).
Different from the more defined expression timing of brown rot LOX genes, those in white rot were expressed in both stages of decay. For example, laccases, several  aryl-alcohol oxidases, and class II peroxidases were early upregulated only in white rot, which may work for gentler enzymatic pretreatment, in contrast to the harsher ROS pretreatment strategy (e.g., Fenton reaction) in brown rot. On the other hand, many white rot LOX genes, including those encoding 22 class II peroxidases in both white rot species, 26 lytic polysaccharide monooxygenases (LPMO), two cellobiose dehydrogenases (CDH), 14 copper radical oxidases, and two aromatic dioxygenases, were also co-upregulated with GH families in later decay stages (Fig. 3). This coregulation is in line with the simultaneous oxidation of lignin and cellulose at later decay stages during white rot, as well as the saccharification by glycoside hydrolases. The contractions/ losses of these genes would likely have shed similar functions among brown rot fungi (2,7,9). Again, these distinctive gene expression features reflected the different wood-decaying mechanisms adopted by two classes of fungi (12). Gene Regulatory Shifts Enable Efficient Wood Decay ® Notable exceptions for GH temporal regulation were found, although they were mainly expressed at late decay stages. For example, several pectin-cleaving enzymes (1 to 6 genes in each species) (35), expansin-like proteins (1 or 2 genes) (36, 37) and some side-chain-cleaving GHs (38, 39) (e.g., GH31 ␣-xylosidase, GH35 ␤-galactosidase, GH51 ␣-L-arabinofuransidase, and GH79 ␤-glucuronidase) were upregulated early in brown or white rot fungi, suggesting a general requirement among wood decay fungi to loosen plant cell walls in order to facilitate ingress.
Shift of transcriptional investments toward GH expression in brown rot. Gene expression levels (as reads per kilobase per million [RPKM]) reflected DEG patterns but also revealed a deeper distinction in gene expression investments between brown rot and white rot fungi ( Fig. 5; see Data Set S4 available at https://www.ncbi.nlm.nih.gov/ geo/query/acc.cgi?accϭGSE108189). In general, the ratio of GH to LOX expression was Ͼ1 for brown rot but Ͻ1 for white rot irrespective of decay stage, and this ratio was nearly 30-fold higher in brown rot than in white rot in later decay stages (Fig. 5A), indicating brown rot GH expression has been strengthened compared to white rot expression.
Using PCA to analyze gene family expression levels of all decay stages/species further indicated that GH expression tended to be associated with late brown rot, while expression of some LOX and CBH genes was associated with late white rot (Fig. 5B). White rot was characterized by coding gene expansion in families of class II peroxidase, lytic polysaccharide monooxygenase, copper radical oxidase, and laccase (white rot versus brown rot, two-tailed paired t test, P Ͻ 0.005) (Table S3) (1,7,9) and had higher expression levels to match gene numbers (Fig. 5C). Brown rot fungi, however, expressed higher levels of GH despite having a contraction of GH-coding genes, such as in families of GH3, GH5/45, GH10/11, GH12/44/74, and GH27 (two-tailed paired t test, P Ͻ 0.001) (Table S3). These RPKM data reveal an energy trade-off that would not be apparent in a genome sequence analysis and that supports hypothesis/explanation 2 (fewer genes retained in genome, but those genes expressed more). Positive numbers in the y axis mean more early DEGs, and the gene functions are more likely associated with early decay, while negative numbers mean the gene functions are tending to be associated with late decay. Significant differences (P values) were calculated with a two-tailed paired t test. ND, no significant differences were found.
Extra supports from enzyme presence and specific activities. This distinction in GH investment between brown rot and white rot was supported by enzymatic activities, as was the timing (i.e., staggered "two-step") of the brown rot mechanism. Overall, GH enzyme-specific activities, accounting for the proportion of GH in the total secreted proteins, were clearly higher in brown rot than in white rot ( Fig. 6A; see Fig. S6 in the Relative expression levels of lignocellulose-degrading genes associated with white rot or brown rot, shown as a function of decay stage. QRD, benzoquinone reductase; QSYN, putative hydroquinone synthases, including phenylalanine ammonia-lyase and phenolic 3/4-O-methyltransferase; HTP, heme-thiolate peroxidase; DYP, dye-decoloring peroxidase; GMC, glucose-methanol-choline gene family, including AA3_2 GMC oxidoreductase, AA3_3 alcohol oxidase, AA3_4 pyranose oxidase, AA3_5 aryl-alcohol oxidase/pyranose dehydrogenase, AA4_vannilyl-alcohol oxidase, and AA7_glucooligosaccharide oxidase/chitooligosaccharide oxidase. Iron homeostasis includes iron reductase, iron permease, and AA1_2 ferroxidase; POD, class II peroxidase; CRO, copper radical oxidase; LPMO, lytic polysaccharide monooxygenase; CDH, cellobiose dehydrogenase. Gene Regulatory Shifts Enable Efficient Wood Decay ® supplemental material). In line with the late upregulation of GH, most of the GH enzymatic activities increased with increasing decay intensity for all fungi, with the exceptions of pectinase and several side chain-cleaving GHs ( Fig. 6A; see Fig. S6). As expected, laccase presence (via staining native PAGE gels) was detected only in early white rot (Fig. 6B). The H 2 O 2 production followed the decreasing trend during brown rot, but not in white rot, in line with the expression patterns of related ROS-producing genes (Fig. 6C).

DISCUSSION
Our results narrow the pool of candidate ROS pathway genes by more than 10-fold from earlier work isolating differentially expressed, early upregulated genes, and they confirm in another brown rot fungus G. trabeum the same "staggered" LOX-then-GH pattern of gene expression discovered recently in P. placenta (23). By comparing with the white rot ancestral expression patterns, this builds a case that the formation of this staggered gene regulatory mechanism in brown rot was to some extent due to the temporal expression shift of LOX genes. This suggests that regulatory rewiring may have contributed to brown rot adaption, including cis-regulatory elements and transacting factors. Studies have shown that the transcription factors have been extensively altered in Basidiomycetes that include most of the wood-decaying fungi, after divergence from Ascomycetes (11).
Differential expression across mycelia colonizing wood likely allows fungi to stage the release of plant cell wall components in a certain order, a sequence that distinguishes brown from white rot deconstruction patterns. A key barrier that both brown and white rot fungi face in degrading plant cell walls is the spatial heterogeneity of the cell wall structure, composed of cellulose microfibrils surrounded by hemicellulose and lignin, as well as pectin and glycoproteins that "gel" the matrix, particularly in the middle lamella (38,39). To surmount these barriers and release soluble sugars from The specific enzyme activities (IU/g extracellular proteins [hierarchically clustered]) were measured from sections with no fungus, 0 to 5 mm (i.e., hyphal front), 15 to 20 mm, and 30 to 35 mm, with three bioreplicates pooled from 24 wood wafers (i.e., n ϭ 3, each with 8 wafers) (see Text S1 for methods and substrates used for enzymatic assays). The activities were shown as Z-score scaled values: red colors indicate higher levels, while green colors represent lower levels. Hierarchical clustering was made by using the "euclidean" distance matrix and "complete" clustering. (B) The presence and temporal expression of the representative wood-decaying enzyme laccase were confirmed in white rot fungi (T. versicolor and P. ostreatus) by SDS-PAGE by active staining with 2,2=-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) (ABTS). (C) Measurements of the wood-oxidizing agent H 2 O 2 clearly indicated its decreasing trend during brown rot (P. placenta), but not white rot (T. versicolor). The H 2 O 2 (nmol/g wood [relative amount]) were measured from sections of 0 to 5 mm and 30 to 35 mm. The significant differences were calculated with two-tailed paired t test.
carbohydrate polymers, brown rot fungi apparently rely on nonenzymatic lowmolecular-weight ROS oxidants (e.g.,·OH) to "loosen" the cell wall matrix (40,41), while white rot fungi confront lignin more directly using ligninolytic peroxidases (42)(43)(44). This metabolic distinction reveals itself in the "signatures" we see in wood, where carbohydrate loss outpaces lignin loss during brown rot, and vice versa during white rot (3). Our transcriptomic patterns support this mechanistic distinction between brown and white rot fungi, and our data improve annotations and contribute functional information to enable comparative genomics.
By viewing the whole transcriptome profiles as gene functional clusters, rather as gene-by-gene patterns of differential expression, we found another important and broad distinction between the transcript and enzyme investments of brown and white rot fungi among functional gene classes. Brown rot fungi have been conventionally characterized by contractions in gene families, including GHs (1,10,14), yet we observed that the additive expression levels (via RPKM summation) for the small set of GHs in brown rot genomes were consistently and collectively higher than those in the larger set of GHs in white rot fungi ( Fig. 5A and C). This likely relates to the fact that white rot fungi must not only invest in more types of GHs but they must also divert energy from GH expression to express PODs, copper radical oxidases (CROs), and LPMOs. This may also explain another distinction between brown and white rot fungi-the loss of carbon catabolite repression (CCR) among brown rot fungi (45). White rot fungi, at least the types we tested here that degrade lignin and carbohydrates at similar rates (simultaneous white rot [3]), may proceed more slowly than brown rot on a strain-by-strain basis due to this requirement for sustained lignin degradation, demanding stricter induction/suppression control of gene expression to save energy. It is logical that a brown rot pathway would offer a fundamental advantage in terms of growth efficiency over a white rot pathway.
Given this advantage, one could argue this might explain why brown rot evolved multiple times from white rot ancestral clades. If, however, brown rot fungi alleviated the need for lignin degradation and gained an energetic advantage over their white rot ancestors, a key question arises: why is white rot more common than brown rot in nature (46)? White rot nutritional modes often dominate fungal communities and wood decay processes in tropical (47) and temperate studies (48), and a meta-analysis of polypore studies (2,428 species records, total) by Ryvarden (49) virtually excluded brown rot from low-elevation tropical and temperate systems below 35°N latitude. This range restriction for brown rot fungi has traditionally been attributed to a substrate preference for conifers (e.g., see reference 49); however, conifers (gymnosperms) are more ancient than angiosperms, while white rot fungi are more ancient than brown rot fungi (6), misaligning coevolution or comigration of brown rot fungi with their conifer tree hosts.
It is possible, instead, that the energetic advantages conferred by the brown rot mechanism are kept in check in nature by ecological, not biological constraints. Community factors, including "cheater" bacteria and fungi that pilfer extricated sugars before absorption by fungi (outlined by Travisano and Velicer [50]), may limit the advantages of a diffuse brown rot pretreatment to environments with less competition such as the higher stress environments found at higher altitudes and latitudes. A streamlining of energy investment in secreted enzymes may only pay off in metabolic "returns" when there is less risk that sugars will be acquired by other organisms before sugars are transported to within the fungal cell wall. Establishing the potential ecological and environmental pressures shaping evolution and limiting the occurrence of brown versus white rot would, in this light, seem an essential pursuit for future research-to link these fungal genomes to the key traits that control this balance of wood rot type on Earth.
Overall, this work implies that brown rot evolution involved shifts in energy investment strategies as much as shifts in gene portfolios or development of novel ROS pathways (supporting explanation 3, outlined in the introduction). This energy return on investment may be a useful framework to understand these fungi in nature, as Functional annotation of fungal genomes. The BRAKER1 annotated protein sequences in this work were queried against the nonredundant (nr) protein sequence database using BLASTP (v2.2.28) (59) (https://blast.ncbi.nlm.nih.gov/), and functional domains were searched with InterProScan (v5.17-56) (60) (http://www.ebi.ac.uk/interpro/). The search outputs of BLASTP and InterProScan were then input into Blast2GO (v4.0.7) (61) to create merged annotation files. More than 70% of genes could be assigned Gene Ontology terms or InterPro domain, indicating the improved genome annotation of this work compared to JGI's database (Fig. S1).
Focusing on functions related to plant cell wall decomposition, lignocellulose-degrading gene families were manually annotated according to the Carbohydrate-Active enZYmes database (http://www .cazy.org/) and relevant literature (Table S3) and then classified as LOX and GH genes in terms of their oxidative and hydrolyzing roles, respectively. Specifically, these gene families were searched in four tested fungi using blastp (v2.2.28) with well-characterized proteins or by functional domain of the InterPro protein signature database. Gene functions were then verified again via blastp in NCBI using the stringent UniprotKB/Swiss-Prot database, as well as by phylogenetic analyses in Jalview (v2.10.1) (62-64) (see Text S1 in the supplemental material for gene functional annotation). For class II peroxidases (i.e., lignin peroxidase, manganese peroxidase, and versatile peroxidase), the functions were further assigned according to the catalytically active tryptophan and three "Mn binding" acidic amino acid residues (two glutamic acid, one aspartic acid) (7). Expression levels (RPKM) of these gene families were analyzed by principal-component analysis in RStudio (v1.0.143) (http://www.rstudio.org).
Protein extraction and enzyme activities. Total extracellular proteins from individual wood sections were extracted in 0.5 M NaCl for 48 h with stirring at 4°C. The extracts were filtered to remove the wood/fungal debris, and protein concentrations were measured using the Bio-Rad protein assay kit (Bio-Rad, CA, USA).
TEXT S1, DOCX file, 0.1 MB.   The authors declare no competing financial interests.