Characterization of alkylguaiacol-degrading cytochromes P450 for the biocatalytic valorization of lignin

Significance

Upgrading lignin, an underutilized component of biomass, is essential for the sustainability of biorefineries. Biocatalysis has considerable potential for upgrading lignin, but our lack of knowledge of relevant enzymes and pathways has limited its application. Herein, we describe a microbial pathway responsible for catabolizing alkylguaiacols, a major component of several industrial lignin streams. Catabolism is initiated by a cytochrome P450, with related P450s catalyzing the O-demethylation of different lignin-derived guaiacols and subsequent catabolism depending on the substitution pattern of the guaiacol. Importantly, the alkylguaiacol catabolic pathway enables bacterial growth on corn stover lignin produced by reductive catalytic fractionation. Overall, these insights greatly facilitate the engineering of P450s and bacteria to biocatalytically upgrade lignin.

Abstract

Cytochrome P450 enzymes have tremendous potential as industrial biocatalysts, including in biological lignin valorization. Here, we describe P450s that catalyze the O-demethylation of lignin-derived guaiacols with different ring substitution patterns. Bacterial strains Rhodococcus rhodochrous EP4 and Rhodococcus jostii RHA1 both utilized alkylguaiacols as sole growth substrates. Transcriptomics of EP4 grown on 4-propylguaiacol (4PG) revealed the up-regulation of agcA, encoding a CYP255A1 family P450, and the aph genes, previously shown to encode a meta-cleavage pathway responsible for 4-alkylphenol catabolism. The function of the homologous pathway in RHA1 was confirmed: Deletion mutants of agcA and aphC, encoding the meta-cleavage alkylcatechol dioxygenase, grew on guaiacol but not 4PG. By contrast, deletion mutants of gcoA and pcaL, encoding a CYP255A2 family P450 and an ortho-cleavage pathway enzyme, respectively, grew on 4-propylguaiacol but not guaiacol. CYP255A1 from EP4 catalyzed the O-demethylation of 4-alkylguaiacols to 4-alkylcatechols with the following apparent specificities (k_cat/K_M): propyl > ethyl > methyl > guaiacol. This order largely reflected AgcA’s binding affinities for the different guaiacols and was the inverse of GcoA_EP4’s specificities. The biocatalytic potential of AgcA was demonstrated by the ability of EP4 to grow on lignin-derived products obtained from the reductive catalytic fractionation of corn stover, depleting alkylguaiacols and alkylphenols. By identifying related P450s with complementary specificities for lignin-relevant guaiacols, this study facilitates the design of these enzymes for biocatalytic applications. We further demonstrated that the metabolic fate of the guaiacol depends on its substitution pattern, a finding that has significant implications for engineering biocatalysts to valorize lignin.

Lignin is a complex aromatic polymer found in plant cell walls (1). Although lignin is the most abundant aromatic polymer on Earth, comprising up to 40% of lignocellulosic biomass, it is underutilized in biorefineries, typically being slated for combustion to power the extraction of carbohydrates (2–5). Technoeconomic analysis and life-cycle assessment have highlighted the importance of lignin valorization to the economic viability and sustainability of biorefineries (6, 7). To this end, considerable effort has been invested in developing lignin depolymerization technologies (2–5). Such technologies produce a mixture of aromatic compounds, depending on the chemistry and the proportions of p-hydroxyphenyl (H), guaiacyl (G), and syringyl (S) subunits that comprise the lignin (2). A major challenge of lignin valorization is converting these complex mixtures of compounds to single chemical species in high atom yield.

Bacteria offer an attractive approach for converting lignin depolymerization products to value-added bioproducts via biological funneling (8–11). This process exploits the capacity of some bacteria to catabolize a variety of aromatic compounds as well as the convergent nature of the pathways responsible for this catabolism. A large number of “upper” pathways transform a variety of aromatic compounds to a few key intermediates, which are further transformed to central metabolism by a limited number of “lower” pathways (8, 12). In lignin-relevant catabolism, the key intermediates are typically catechol, protocatechuate, and gallate, all of which bear hydroxyl groups on adjacent ring carbon atoms (8, 9). The aromatic ring is then cleaved either ortho (between) or meta (adjacent) to those hydroxyl substituents (13). The development of bacteria as lignin-valorizing biocatalysts requires the elucidation and optimization of pathways able to catabolize the major components of industrial lignin streams.

An essential reaction in the catabolism of aromatic compounds derived from G- and S-type lignin is the O-demethylation of the aromatic methoxy groups to produce a catecholic intermediate. In the cases where bacterial strains have been engineered and optimized to produce bioproducts from lignin-derived compounds from G- and S-type lignin, aromatic O-demethylation is a rate-limiting step (14–16). Accordingly, there has been significant interest in the mechanistic understanding of the three enzyme types known to catalyze aromatic O-demethylation: tetrahydrofolate-dependent O-demethylases (17, 18), Rieske-type nonheme iron oxygenases (19–21), and cytochromes P450 (22–24). A critical unknown for all three enzyme paradigms is their structure–activity relationship on the various G- and S-type lignin-derived compounds, which exhibit substantially different ring substitution patterns depending on the upstream chemistry.

One important group of lignin depolymerization products whose catabolism has not been elucidated are the 4-alkylguaiacols. Depolymerization processes that yield alkylguaiacols include reductive catalytic fractionation (RCF) (25, 26), which has taken on new importance as the basis of a profitable, sustainable biorefining strategy to convert biomass into a variety of product streams (27). RCF of corn stover yields up to 38% monoaromatic compounds, with the predominant ones being 4-ethylphenol, 4-propylguaiacol (4PG), 4-propanolguaiacol (4P_OHG), 4-propylsyringol (4PS), methyl ferulate, and methyl p-coumarate (26). This mixture lends itself to bacterial biological funneling as rhodococci naturally catabolize several of these compounds. For example, Rhodococcus jostii RHA1 (RHA1 hereafter) catabolizes ferulate and p-coumarate via the Cou and β-ketoadipate pathways (28). Both Rhodococcus rhodochrous EP4 (EP4 hereafter) and RHA1 catabolize 4-ethylphenol via the Aph pathway, in which alkylphenols are hydroxylated by AphAB to form alkylcatechols (29). The latter are subject to meta-cleavage by AphC, an alkylcatechol 2,3-dioxygenase, and eventually funneled into central metabolism via pyruvate and either acetyl- or propionyl-CoA. Elucidation of alkylguaiacol catabolism would afford a more complete catabolism of lignin depolymerization products.

A number of rhodococci and related strains grow efficiently on guaiacol (30, 31). In these strains, catabolism is initiated by a cytochrome P450 that belongs to the Cyp255A2 family which catalyzes the O-demethylation of guaiacol to catechol (30). The best characterized of these P450s is GcoA_Amyc from Amycolatopsis sp. ATCC 39116 which, together with its cognate reductase GcoB, transforms a range of aromatic ethers including guaiacol and anisole (22). In R. rhodochrous 116, the resulting catechol is subsequently catabolized via ortho-cleavage and presumably the β-ketoadipate pathway (30, 32). The potential of these P450s for valorizing lignin is highlighted by the heterologous expression of gcoAB from R. rhodochrous J3 in Pseudomonas putida EM42 to enable growth on guaiacol and the engineering of GcoA_Amyc to enable the O-demethylation of syringol (23, 24). More generally, bacterial P450s have been implicated in the O-demethylation of 4-methoxybenzoate, veratrate, and O-methylated carbohydrates (33, 34). Finally, wood-rotting basidiomycetes harbor large numbers of P450 isozymes, some of which have been implicated in the O-demethylation of resveratrol and related stilbenes (35).

Herein, we describe the catabolism of 4-alkylguaiacols by two strains of Rhodococcus: EP4 (29) and RHA1 (36). A combination of transcriptomics, targeted gene deletion, and enzymology was used to identify the catabolic pathway. Our studies established that this catabolism is initiated by AgcAB, a CYP255A1 family P450 that catalyzes the O-demethylation of 4-alkylguaiacols. This distinguishes AgcA from GcoA, which both strains also contain. The ability of AgcA to bind and transform a variety of guaiacols was characterized. Finally, we investigated the ability of the rhodococcal strains to grow on corn stover RCF oil, which contains a mixture of compounds, including 4PG. The results are discussed in terms of related catabolic enzymes and pathways, as well as the design and engineering of biocatalysts to valorize lignin.

Results

Transcriptomic Analysis Revealed Alkylguaiacol Catabolic Genes in EP4.

We found that EP4, previously characterized for its ability to catabolize alkylphenols (29), could also grow on a variety of alkylguaiacols. When grown on 1 mM substrate, the growth yield, as measured by the final optical density (36), reflected the length of the alkyl side chain, with yields on 4PG > 4-ethylguaiacol (4EG) > 4-methylguaiacol (4MG) > guaiacol (Fig. 1A). However, the lag phase of the culture was shortest on guaiacol and longest on 4MG. Growth on 4PG was also confirmed using colony-forming units (CFUs) (SI Appendix, Fig. S1A).

Fig. 1.

(A) Growth of R. rhodochrous EP4 on guaiacol, 4-methylguaiacol, 4-ethylguaiacol, and 4-propylguaiacol. Error bars represent SD (n = 3). (B) The gene cluster encoding the Aph meta-cleavage pathway of EP4 with transcript reads from cells growing on 4PG are plotted below the cluster. Blue and red highlight agcAB and aphC, respectively. Genes of unknown function are colored gray.

To identify genes potentially involved in the catabolism of 4-alkylguaiacols, the transcriptomes of cells growing exponentially on 4PG relative to a succinate control were compared using RNA sequencing (RNA-seq) (Sequence Read Archive accession nos. SRR6877528 to SRR6877536). The most highly up-regulated genes in 4PG-grown cells included the recently identified aph genes, encoding a meta-cleavage pathway for alkylphenols, together with two additional genes in a putative operon located immediately downstream of aphC (Fig. 1B). These genes were predicted to encode a P450 and its cognate reductase. Based on their involvement in the O-demethylation of 4-alkylguaiacols, established below, these genes were annotated as agcA and agcB (SI Appendix, Table S1). None of the cat and pca genes, encoding the ortho-cleavage of catechol and the β-ketoadipate pathway, respectively, were up-regulated during growth on 4PG. Finally, a gene encoding a putative mycothiol-dependent formaldehyde dehydrogenase was up-regulated ∼6.8-fold during growth on 4PG (SI Appendix, Table S2). Interestingly, EP4 does not appear to contain a mycothiol-dependent formaldehyde detoxification pathway homologous to that of RHA1 (28).

AgcA Belongs to the CYP255A1 Family.

Bioinformatic analysis revealed that AgcA belongs to the CYP255A1 family, and shares 65% amino acid sequence identity with GcoA_Amyc, a CYP255A2 that catalyzes the O-demethylation of guaiacol (Fig. 2A and SI Appendix, Table S1) (22). However, the reciprocal best hit of GcoA_Amyc in EP4 is encoded by C6369_RS07265, with whose gene product it shares 76% sequence identity (SI Appendix, Table S1). C6369_RS07265 and the adjacent C6369_RS07270, predicted to encode a P450 reductase, lie in close proximity to catA, encoding the catechol 1,2-dioxygenase associated with the β-ketoadipate pathway (SI Appendix, Fig. S2). These genes, annotated as gcoAB, were not up-regulated in EP4 during growth on 4PG.

Fig. 2.

Relationship between the rhodococcal CYP255As. (A) Reactions catalyzed by AgcAB and GcoAB, respectively, and the subsequent catabolism of their respective reaction products via meta- and ortho-cleavage, respectively. R = CH₃, CH₂CH₃, CH₂CH₂CH₃. (B) Phylogenetic tree of CYP255As. The tree includes characterized enzymes GcoA_Amyc from Amycolatopsis ATCC 39116 and GcoA_J3 from R. rhodochrous J3 as well as the homologs present in EP4 and RHA1. The tree scale represents substitutions per site.

Our analysis further revealed that RHA1 also harbors both agcAB and gcoAB. As in EP4, these genes lie in close proximity to the aph and cat genes, respectively (SI Appendix, Fig. S2). In a structure-based phylogenetic analysis, AgcA_RHA1 and AgcA_EP4 belong to the same clade (Fig. 2B) despite sharing only 58% amino acid sequence identity. The above gene organization and the phylogeny of AgcA and GcoA suggest that these enzymes have different substrate specificities.

Genetic Validation of the AgcA and GcoA in RHA1.

We used RHA1 for molecular genetic analysis of alkylguaiacol catabolism. This is because RHA1 is much more genetically tractable than EP4 (29), and the relevant genes (agcAB, aph, gcoAB, cat, and pca) have high sequence identity and synteny between the two strains. Consistent with the bioinformatic analyses, RHA1 grew on guaiacol, 4MG, and 4PG as sole growth substrates (Fig. 3A and SI Appendix, Fig. S1B). As observed in EP4, the growth yield on the different guaiacols reflected the length of the side chain. However, unlike EP4, the longest lag phase occurred during growth on 4PG: RHA1 only grew on 1 mM 4PG after a lag phase of 20 h, during which the CFUs decreased by ∼50%.

Fig. 3.

Growth of wild-type and mutant R. jostii RHA1 strains on various guaiacols and expression of selected genes. (A–E) Growth of wild type (A) and ΔagcA (B), ΔaphC (C), ΔpcaL (D), and ΔgcoA (E) mutants on guaiacol, 4-methylguaiacol, and 4-propylguaiacol. Growth curves are color-coded according to growth substrate as indicated in A. (F) qRT-PCR of agcA, aphC, gcoA, and catA in wild-type and ΔagcA cells grown on G or 4MG. Values are in log₁₀, represent the mean of two technical replicates, and are normalized to sigA expression. Error bars represent SD (n = 3).

To further evaluate the respective catabolic roles of the P450s, the expression of selected genes was assessed in RHA1 growing exponentially on 4PG and guaiacol. qRT-PCR analyses revealed that agcA and aphC were up-regulated during growth on 4PG while neither gcoA nor catA was up-regulated under these conditions (SI Appendix, Fig. S1D). By contrast, gcoA and catA were up-regulated during growth on guaiacol while agcA and aphC were not (Fig. 3F).

To validate the two guaiacol catabolic pathways of RHA1, we constructed a ΔagcA strain and tested the ability of this mutant and others in our collection to grow on different guaiacols. Neither the ΔagcA mutant nor our previously constructed ΔaphC mutant grew on 4PG (Fig. 3 B and C). Both mutants grew on guaiacol with similar kinetics and yields as wild-type RHA1 (Fig. 3). This is consistent with the bioinformatic predictions that 4PG but not guaiacol is catabolized by AgcAB and the Aph pathway. Importantly, the ΔagcA mutant but not ΔaphC was able to utilize 4-ethylphenol, establishing that the Aph pathway is intact in the ΔagcA mutant (SI Appendix, Fig. S3). In addition, the ΔgcoA and ΔpcaL mutants grew normally on 4PG but neither grew on guaiacol (Fig. 3 D and E). These data are consistent with the prediction that guaiacol is catabolized by GcoA, ortho-cleavage, and the β-ketoadipate pathway. This conclusion was substantiated by qRT-PCR data, which demonstrated that agcA and aphC transcripts were much more abundant than gcoA and catA transcripts in RHA1 growing on 4PG (SI Appendix, Fig. S1D). Similarly, gcoA and catA transcripts were much more abundant than agcA and aphC transcripts in RHA1 growing on guaiacol (Fig. 3F).

The data for 4MG were more complex. Thus, of the tested mutants, only ΔaphC was unable to grow on this compound (Fig. 3). This suggests that regardless of how 4MG is O-demethylated, the resulting 4-methylcatechol is only degraded via the Aph meta-cleavage pathway. Further, the growth of the ΔgcoA and ΔpcaL mutants on 4MG was similar to that of wild-type RHA1, suggesting that AgcAB is the primary 4MG O-demethylase. This was substantiated by qRT-PCR analyses, which revealed that agcA transcripts were over an order of magnitude more abundant than gcoA transcripts in cells growing on 4MG (Fig. 3F). Somewhat unexpectedly, the ΔagcA mutant was able to grow on 4MG, albeit with a lag phase of ∼20 h (Fig. 3B). qRT-PCR analyses revealed that gcoA transcripts were ∼13-fold more abundant in ΔagcA cells than in wild-type cells growing on 4MG (Fig. 3F). These results suggest that in the mutant, GcoA O-demethylates 4MG and partially compensates for the loss of AgcA.

AgcA Binds 4-Alkylguaiacols with High Affinity.

To characterize the P450s, we produced AgcA_RHA1 and AgcA_EP4 in Escherichia coli and purified them to apparent homogeneity (SI Appendix, Fig. S4 A and C). As isolated, neither AgcA contained a full complement of heme (Rz ∼0.78, where Rz = A₄₁₇/A₂₈₀). Reconstitution with hemin yielded preparations with Rz values of ∼1.09 (SI Appendix, Fig. S4A). The AgcAs had Soret peaks at 417 nm as well as α- and β-bands at 567 and 536 nm, respectively, typical of the low-spin state of the ferric ion in ligand-free P450s (37). Incubation of reconstituted AgcA_RHA1 with dithionite and CO caused the Soret peak to shift to 450 nm, indicating that the reconstituted heme was in the proper coordination environment with an axial cysteine ligand (SI Appendix, Fig. S4B). The affinity of the AgcAs for 4-alkylguaiacols was examined by monitoring the induction of a type I absorption spectrum (38). In this assay, ligand binding induces a shift in the Soret peak as the ferric ion transitions from a low-spin to a high-spin state (Fig. 4A). AgcA_EP4 and AgcA_RHA1 had similar affinities for the tested guaiacols, including markedly lower affinities for guaiacol and 4P_OHG (Table 1 and SI Appendix, Table S3). Nevertheless, AgcA_EP4 had highest affinity for 4EG (K_d ∼0.4 μM) while AgcA_RHA1 had highest affinity for 4PG (K_d ∼0.2 μM).

Fig. 4.

Binding and transformation of 4PG by AgcA. (A) Difference spectra obtained upon titration of 0.2 μM AgcA_RHA1 with 4PG (0.1 to 1 μM). (A, Inset) The resulting binding curve with error bars representing the SD (n = 3). The solid line represents the best fit of a quadratic binding equation to the data, with fitted parameters K_d = 0.20 ± 0.02 μM and ΔA_421Max = 0.0210 ± 0.0006 (20 mM Mops, 90 mM NaCl, pH 7.2 at 25.0 °C). (B) Transformation of 4PG by AgcA_RHA1. Reactions were performed in the same buffer as A at 30 °C containing 100 μM 4PG, 1 mM NADH, 2 μM AgcA, and 10 μM AgcB. HPLC traces (from back to front) are of the 4PG standard, reactions containing 4PG without or with AgcA_RHA1, and the 4-propylcatechol standard. 4PG and 4-propylcatechol were added to 100 μM. Reactions were incubated for 1 h.

Table 1.

Affinity and apparent specificity of AgcA_EP4 and GcoA_EP4 for guaiacols

	AgcAEP4				GcoAEP4
Substrate	Kd, μM	kcat, s−1	KM, μM	kcat/KM, mM−1⋅s−1	Kd, μM	kcat, s−1	KM, μM	kcat/KM, mM−1⋅s−1
Guaiacol	104 (9)	5.3 (0.2)	170 (20)	31 (2)	0.25 (0.02)	3.9 (0.1)	1.8 (0.1)	2,000 (100)
4-Methylguaiacol	0.94 (0.06)	2.2 (0.1)	2.7 (0.4)	800 (100)	1.1 (0.04)	2.2 (0.1)	7.4 (0.3)	300 (10)
4-Ethylguaiacol	0.39 (0.03)	2.7 (0.1)	2.5 (0.3)	1,000 (100)	1.4 (0.04)	1.5 (0.1)	12 (1)	128 (8)
4-Propylguaiacol	0.6 (0.1)	17.1 (0.8)	2.0 (0.2)	8,700 (600)	21 (3)	0.19 (0.01)	33 (5)	5.8 (0.8)
4-Propanolguaiacol	22 (1)	0.36 (0.01)	17 (1)	22 (1)	*	—	—	—
2-Ethoxyphenol	—	—	—	—	0.5 (0.02)	1.80 (0.04)	1.0 (0.1)	1,800 (200)

Affinities and apparent steady-state kinetic parameters were measured at 25 °C in 20 mM Mops, 90 mM NaCl (pH 7.2). SDs are in parentheses.

^*Not determined.

AgcAB Catalyzes the O-Demethylation of 4-Alkylguaiacols.

To elucidate the substrate specificity of AgcAB, we produced the reductase in E. coli. AgcB was anaerobically purified to apparent homogeneity (SI Appendix, Fig. S5A) as it was O₂-labile. We worked with AgcB_EP4 as it was more soluble than AgcB_RHA1. Preparations of AgcB_EP4 absorbed maximally at 454 nm, which corresponds to a flavin adenine dinucleotide cofactor. An additional peak at 425 nm together with shoulders around 475 and 550 nm is diagnostic of the predicted 2Fe–2S cluster (SI Appendix, Fig. S5) (39). Preparations of AgcB_EP4 reduced cytochrome c with a specific activity of 22 ± 4 μmol cytochrome c⋅min⁻¹⋅mg⁻¹. In contrast to what has been reported for GcoAB_Amyc (22), size-exclusion chromatography studies indicated that AgcAB_EP4 does not form a stable complex in solution (SI Appendix, Fig. S6).

To establish that AgcA catalyzes the O-demethylation of 4PG, we incubated each P450 and AgcB_EP4 for 1 h with 100 μM 4PG and NADH at 30 °C (20 mM Mops, 90 mM NaCl, pH 7.2). High-pressure liquid chromatography (HPLC) analyses of the reaction mixtures revealed that 4PG was converted to a compound that eluted with a retention time corresponding to that of an authentic standard of 4-propylcatechol (Fig. 4B). In the absence of AgcA, 4PG was not transformed. The ability of AgcA_EP4 to catalyze O-demethylation of 4PG, 4PS, and 4P_OHG was also examined using liquid chromatography-mass spectrometry. Reactions with 4PG and 4P_OHG resulted in the appearance of prominent peaks with m/z values of 151.077 and 167.073, respectively (SI Appendix, Fig. S7). These correspond to the expected values of 4-propylcatechol and 4-propanolcatechol. By contrast, 4PS was not detectably depleted and ions corresponding to the singly or doubly demethylated products were not observed (SI Appendix, Fig. S7).

To determine the apparent specificity (k_cat/K_M) of AgcA, we established a coupled assay using AphC_RHA1, the alkylcatechol 2,3-dioxygenase from the Aph pathway of RHA1. This enzyme cleaves the alkylcatechol resulting from the AgcAB reaction to a yellow-colored meta-cleavage product whose formation was continuously monitored at 400 nm. In this assay, the P450s displayed Michaelis–Menten kinetics (SI Appendix, Fig. S8). Consistent with their affinity profiles, both AgcAs had highest apparent specificity for 4PG and 4EG as well as specificities for guaiacol and 4P_OHG that were up to 400-fold lower (Table 1 and SI Appendix, Table S3). The measured specificities of AgcA_EP4 were generally a bit higher than AgcA_RHA1, perhaps due to the use of AgcB_EP4 in the assays.

AgcA and GcoA Have Complementary Specificities for Alkylguaiacols.

To further compare the specificities of the AgcAs, which belong to the CYP255A1 family, and the GcoAs, which together with GcoA_Amyc (22) belong to the CYP255A2 family, we produced and purified GcoA_EP4 and GcoB_EP4 as described above for the AgcAs and AgcB_EP4. GcoA_EP4 and GcoB_EP4 had essentially the same spectra as the AgcAs and AgcB_EP4, respectively (SI Appendix, Figs. S4 and S5). Interestingly, the affinities and apparent substrate specificities of GcoA_EP4 for guaiacols were essentially the inverse of those of AgcAs (Table 1). That is, GcoA_EP4 had highest affinity for guaiacol and its apparent specificity for guaiacol was >340-fold higher than for 4PG. GcoA_EP4 also transformed 2-ethoxyphenol with high apparent specificity, consistent with what has been reported for GcoA_Amyc (22) and a P450 in R. rhodochrous 116 (30).

GcoB_EP4 Has Higher Specificity for NADPH.

In characterizing GcoB_EP4 for the above assays, we unexpectedly observed that this reductase preferentially utilized NADPH. In steady-state kinetic assays using cytochrome c as an oxidant, GcoB_EP4’s specificity (k_cat/K_M) for NADPH was over three orders of magnitude higher than for NADH (SI Appendix, Fig. S9 and Table S4). By contrast, AgcB_EP4 had >200-fold greater specificity for NADH versus NADPH.

EP4 Grows on Corn Stover RCF Oil.

To investigate the potential of EP4 and RHA1 to transform an industrially relevant lignin stream, these strains were tested for growth on corn stover RCF oil. EP4 grew on media containing 250 parts per million (ppm) RCF oil, as measured by CFUs (Fig. 5A). Analysis of the culture supernatant by a gas chromatograph coupled to a mass-selective detector (GC-MS) revealed that of the six major components of the RCF oil, 4-ethylphenol, 4PG, and 4P_OHG were efficiently removed (Fig. 5B and SI Appendix, Fig. S10 A, C, and D). These compounds appear to be catabolized via AgcAB and the Aph pathway, as agcA, aphA, and aphC were up-regulated during growth on corn stover RCF oil relative to a succinate control (Fig. 5C). The other three RCF components, 4-propanolsyringol, methyl ferulate, and methyl coumarate, were transformed to compounds with retention times of 13.06, 12.93, and 14.22 min, respectively (SI Appendix, Fig. S11). Indeed, previous efforts to grow EP4 on the latter three RCF components were unsuccessful, consistent with the strain’s lack of the cou genes, responsible for p-coumarate and ferulate catabolism. Interestingly, gcoA and catA were slightly down-regulated during growth on RCF oil relative to the succinate control (Fig. 5C). Finally, RHA1 did not grow well on corn stover RCF oil, even though the strain carries the agc, aph, and cou genes (Fig. 5A).

Fig. 5.

Growth of R. rhodochrous EP4 and R. jostii RHA1 on corn stover RCF oil. (A) Growth of strains in M9 media containing 250 ppm RCF oil (n = 3 for EP4; n = 2 for RHA1). (B) Depletion of 4-ethylphenol (4EP), 4-propylguaiacol, and 4-propanolguaiacol from cultures of R. rhodochrous EP4. Plotted values are the percentage of peak area at t = 0 divided by the peak area of the internal standard (3-chlorobenzoate). (C) qRT-PCR of key genes in R. rhodochrous EP4 cells harvested at 32 h of growth on RCF oil (RCF) versus on succinate (S). All values are in log₁₀ and represent the mean of two technical replicates, and are normalized to sigA expression. Growth of R. rhodochrous EP4 on succinate is shown in SI Appendix, Fig. S1A. Error bars represent SD.

Discussion

This study describes AgcA, a P450 that catalyzes the O-demethylation of 4-alkylguaiacols to initiate their catabolism in Actinobacteria via the Aph meta-cleavage pathway. AgcAs belong to the CYP255A1 family and have a low specificity for guaiacol. By contrast, GcoAs belong to the CYP255A2 family and have a high specificity for guaiacol (22). The genetic analysis of agcA and gcoA in RHA1 is consistent with the different specificities of these enzymes and establishes the different physiological roles of these two P450s. Further, the genetic association and coregulation of the agcAB and gcoAB genes with meta- and ortho-cleavage pathways, respectively, are consistent with the tendency of alkylated aromatic compounds to be degraded by meta-cleavage pathways (29).

Recent interest in GcoA stems from its ability to transform lignin depolymerization products such as guaiacol and syringol (22, 23). The identification of an enzyme able to catalyze the O-demethylation of alkylguaiacols significantly expands these efforts, given the prevalence of such alkylated compounds in a variety of industrially relevant mixtures of lignin depolymerization products. More generally, the different substrate specificities of AgcA and GcoA, despite their high amino acid sequence identity, facilitates the engineering of these enzymes for lignin transformation. Inspection of the GcoA_Amyc–guaiacol (Protein Data Bank ID code 5NCB) binary complex reveals that C-4 of the guaiacol is within 4 Å of Ile81 and Thr296, and within 5 Å of Val241, Ile292, Ala295, and Phe395 (22). These residues are conserved in GcoA_EP4, GcoA_RHA1, AgcA_EP4, and AgcA_RHA1 with the exception of Ile81 and Thr296, which are leucine and alanine, respectively, in the AgcAs. Thr296 is particularly interesting as its side chain is orientated toward the C-4 of guaiacol in the GcoA_Amyc–guaiacol complex. The smaller side chain of alanine at this position is consistent with AgcA’s higher affinity for 4PG relative to guaiacol (Table 1). Currently, we are investigating the structure of AgcA–alkylguaiacol complexes to elucidate the basis of the enzyme’s specificity. Such studies could further inform the engineering of these enzymes to catalyze the O-demethylation of compounds such as 4PS, a major component of corn stover and hardwood RCF oils (25, 26).

GcoB_EP4’s high specificity for NADPH contrasts with the specificity of GcoB_Amyc (22) and AgcB_EP4 for NADH (SI Appendix, Table S4). Nevertheless, this specificity is consistent with the available structural data. Notably, Asp272 of GcoB_Amyc has been identified as a determinant of cofactor specificity in these reductases (22). In an NADPH-utilizing P450 reductase from rat (40), this residue corresponds to Arg454 and forms a salt bridge with the 2′ phosphate of NADPH. The aspartate in GcoB_Amyc disrupts a similar interaction at this position. An amino acid sequence alignment (SI Appendix, Fig. S12) revealed that GcoB_EP4 also has an arginine at this position. By contrast, AgcB_EP4 has a glutamate at this position, consistent with its high specificity for NADH, while GcoB_RHA1 has an aspartate. The physiological significance of GcoB_EP4’s specificity for NADPH, which appears to be relatively unique, is unclear. However, it is possible that cofactor utilization could affect the strain’s ability to transform RCF oil.

The Aph pathway represents another case of convergence in the catabolism of aromatic compounds, as both 4-alkylguaiacols and 4-alkylphenols are funneled into this pathway (29). This is highlighted by the growth of EP4 on corn stover RCF oil with concomitant removal of 4-ethylphenol and 4PG. More specifically, the biochemical and molecular genetic data indicate that the catabolism converges at 4-alkylcatechol. The ability of the Aph pathway to catabolize a variety of 4-alkylcatechols implies that AphC and the subsequent enzymes have a relatively relaxed substrate specificity. This is in contrast to the β-ketoadipate pathway, where compounds such as p-coumarate, ferulate, vanillin, and 4-hydroxybenzoate are converted to protocatechuate prior to ring opening (28, 41, 42).

The growth yields of EP4 on the various guaiacols correlated with the length of the alkyl side chain (Fig. 1A). However, the growth rates of EP4 do not correlate with the substrate specificity of AgcA, suggesting that these rates reflect other factors, such as the induction of the catabolic genes or the uptake of substrates. In this respect, the RHA1 ΔagcA mutant was unable to grow on 4PG but grew on 4MG more slowly than the wild type (Fig. 3 A and B), consistent with the ability of GcoA to transform 4MG. This is corroborated by the qRT-PCR data, which demonstrate that gcoA was up-regulated in the ΔagcA mutant growing on 4MG with respect to wild-type (WT) RHA1 (Fig. 3F). Nevertheless, the strong growth of the ΔgcoA strain on 4MG and qRT-PCR data further indicate that 4MG is exclusively catabolized by AgcA in RHA1. Irrespective of which P450 catalyzes the O-demethylation of 4MG, the inability of the ΔaphC strain to grow on 4MG indicates that 4-methylcatechol is exclusively catabolized by the Aph meta-cleavage pathway.

The catabolism of 4-methylcatechol via meta-cleavage in RHA1 is interesting in light of what has been reported in other systems. Generally, alkylaromatic compounds are catabolized by meta-cleavage pathways, with the ortho-cleavage of methylcatechols leading to dead-end products. For example, the ortho-cleavage of 4-methylcatechol by CatA in Bacillus pumilus and Pseudomonas desmolyticum yields 3-methylmuconate with subsequent accumulation of 4-methylmuconolactone as a dead-end product (43, 44). However, in RHA1 cells growing on 4MG, catA transcripts were present at ∼1% of the level as aphC transcripts (Fig. 3F). Thus, 4MG does not appear to repress the cat genes although some component of RCF apparently does (Fig. 5C). The specificities of CatA and AphC could help determine the fate of 4-methylcatechol in RHA1, but these k_cat/K_M values are not known. Interestingly, Rhodococcus opacus 1CP contains several CatAs, all of which cleave 4-methylcatechol relatively efficiently (45). Indeed, this strain harbors an ortho-cleavage pathway that is adapted to p-cresol. This pathway is characterized by a CatA that has highest specificity for 4-methylcatechol and a cycloisomerase that transforms the resulting 3-methylmuconate to prevent the accumulation of 4-methylmuconolactone. Further studies, including metabolomics, are required to determine the true extent of ortho-cleavage of 4-methylcatechol in RHA1 as well as the mechanism underlying its routing to meta-cleavage.

The characterization of an alkylguaiacol catabolic pathway greatly facilitates engineering bacterial strains to transform a broader range of lignin depolymerization products. Such biocatalysts have particular potential in RCF-based biorefining, where biological funneling could replace chemical funneling in the valorization of lignin monoaromatic compounds (27). Exploitation of the agc and aph genes for biocatalyst development will require improved understanding of the regulation of these genes and the physiology of the strains containing them. For example, it is unclear why EP4 is better able than RHA1 to grow on corn stover RCF oil. Indeed, this result was somewhat unexpected given that RHA1 can catabolize p-coumarate and ferulate but EP4 cannot (29). This may be related to the poorer growth of RHA1 relative to EP4 on 4PG. Of note, 4PG had a significant bactericidal effect on a fresh inoculum of RHA1 (SI Appendix, Fig. S1B). Moreover, there may be additional toxic compounds in RCF oil to which RHA1 is more sensitive than EP4. Finally, the cou genes, which encode p-coumarate and ferulate metabolism through ortho-cleavage (28), may be subject to catabolite repression, as discussed above. Clearly, understanding the basis of RHA1’s relatively poor growth on RCF oil is important to engineering strains to degrade mixtures of lignin depolymerization products.

Materials and Methods

Chemicals.

Chemicals and reagents were of analytical grade and used without further purification unless otherwise noted. 4-Propanolguaiacol and 4-propylcatechol were synthesized as described in SI Appendix. Corn stover RCF oil was produced using established methods (26). Water for buffers was purified to a resistance of at least 18 MΩ.

Bacterial Strains and Growth Conditions.

Strains used in this study are listed in SI Appendix. Strains were grown on Luria–Bertani (LB) agar, LB broth, or M9 minimal media supplemented with the appropriate growth substrate as described in SI Appendix. For growth on corn stover RCF oil (26), a stock of solution of 50,000 ppm oil was made in dimethyl sulfoxide and added to M9-Goodies (250 ppm final). The medium was incubated at 30 °C for 30 min to maximize solubilization of the oil, and inoculated with cells to a final OD₆₀₀ of ∼0.05.

Growth Substrate Depletion.

Culture supernatants were analyzed using GC-MS as described in SI Appendix.

Transcriptomic and qRT-PCR Analyses.

Cells were grown and RNA was extracted as previously described (29). RNA-seq and qRT-PCR were performed as described in SI Appendix.

Gene Cloning and Deletion.

DNA was purified, manipulated, and propagated using standard procedures (46). Specific genes were cloned and deleted as described in SI Appendix.

Protein Production and Purification.

AgcA, AgcB, and AphC were produced heterologously as N-terminal polyHis-tagged (Ht-) proteins in E. coli and purified using chromatography as described in SI Appendix. Upon purification, proteins were flash-frozen as beads in liquid N₂ and stored at −80 °C until use.

Characterization of P450 Ligand Binding.

CO binding and K_d values for various aromatic ligands were evaluated spectrophotometrically as described in SI Appendix.

Characterization of P450 Reaction and Kinetics.

The products of the AgcAB-catalyzed reaction were evaluated using HPLC and liquid chromatography quadrupole time-of-flight mass spectrometry as described in SI Appendix. Steady-state kinetic parameters for AgcAB were evaluated spectrophotometrically by coupling the O-demethylation of the alkylguaiacol to the meta-cleavage of the resulting alkylcatechol using AphC as described in SI Appendix.

Data Availability.

All relevant data, protocols, and material information are present in the manuscript or SI Appendix. All sequence data can be accessed at NCBI BioProject PRJNA445226 (SRR6877531–SRR6877536) and Gene Expression Omnibus (accession no. GSE112193).