In Vitro Enzymatic Depolymerization of Lignin with Release of Syringyl, Guaiacyl, and Tricin Units

ABSTRACT New environmentally sound technologies are needed to derive valuable compounds from renewable resources. Lignin, an abundant polymer in terrestrial plants comprised predominantly of guaiacyl and syringyl monoaromatic phenylpropanoid units, is a potential natural source of aromatic compounds. In addition, the plant secondary metabolite tricin is a recently discovered and moderately abundant flavonoid in grasses. The most prevalent interunit linkage between guaiacyl, syringyl, and tricin units is the β-ether linkage. Previous studies have shown that bacterial β-etherase pathway enzymes catalyze glutathione-dependent cleavage of β-ether bonds in dimeric β-ether lignin model compounds. To date, however, it remains unclear whether the known β-etherase enzymes are active on lignin polymers. Here we report on enzymes that catalyze β-ether cleavage from bona fide lignin, under conditions that recycle the cosubstrates NAD+ and glutathione. Guaiacyl, syringyl, and tricin derivatives were identified as reaction products when different model compounds or lignin fractions were used as substrates. These results demonstrate an in vitro enzymatic system that can recycle cosubstrates while releasing aromatic monomers from model compounds as well as natural and engineered lignin oligomers. These findings can improve the ability to produce valuable aromatic compounds from a renewable resource like lignin. IMPORTANCE Many bacteria are predicted to contain enzymes that could convert renewable carbon sources into substitutes for compounds that are derived from petroleum. The β-etherase pathway present in sphingomonad bacteria could cleave the abundant β–O–4-aryl ether bonds in plant lignin, releasing a biobased source of aromatic compounds for the chemical industry. However, the activity of these enzymes on the complex aromatic oligomers found in plant lignin is unknown. Here we demonstrate biodegradation of lignin polymers using a minimal set of β-etherase pathway enzymes, the ability to recycle needed cofactors (glutathione and NAD+) in vitro, and the release of guaiacyl, syringyl, and tricin as depolymerized products from lignin. These observations provide critical evidence for the use and future optimization of these bacterial β-etherase pathway enzymes for industrial-level biotechnological applications designed to derive high-value monomeric aromatic compounds from lignin.

GGE-ketone undergoes GSH-dependent ␤-ether cleavage by ␤-etherase enzymes LigE and LigF to yield guaiacol and GS-HPV as monoaromatic derivative products. GS-HPV undergoes GSH-dependent thioether cleavage by NaGST NU or LigG, producing GSSG and monoaromatic product HPV. As indicated by the dashed arrows, AvGR recycles cosubstrates GSH and NAD ϩ via NADH-dependent reduction of GSSG. For reactions involving an R-or S-configured epimer as the substrate, the isomer toward which each enzyme exhibits activity is shown in gray text. Panel C shows how ␤-etherase pathway enzymes degrade GTE through intermediate GTE-ketone to yield tricin and HPV. high yields. The formation of ␤-ether linkages during lignification generates a racemic lignin product containing both ␤(R)and ␤(S)-carbons that, after rearomatization of the quinone methide intermediate by proton-assisted water addition, are adjacent to either ␣(R)or ␣(S)-configured benzylic alcohols (11)(12)(13). Each unit therefore has 4 optical isomers and two "real" isomers-the so-designated threo and erythro (or syn and anti) isomers. Lignin depolymerization via ␤-ether bond cleavage has been demonstrated with chemical catalysis (14,15). In addition, cytoplasmic enzymes in a sphingomonad ␤-etherase pathway have been identified that oxidize and cleave model ␤-ether-linked aromatic dimers (16).
Despite what is known about the activity of individual ␤-etherase pathway enzymes with model diaromatic compounds, there is little information on their function with lignin oligomers. In vivo activity may be limited to aromatic dimers or small lignin oligomers due to restrictions in transporting large polymers into the bacterial cytoplasm, where the ␤-etherase pathway enzymes are found. Recently, modest recovery of low-molecular-mass aromatic compounds from lignin from a multistep enzymatic process that used a laccase mediator system, two ␤-etherases, and a glutathione lyase was reported (25). However, the size of lignin fragments that were subject to enzymatic cleavage was not determined, and it therefore remains unknown whether the ␤-etherases were active on only small or also large lignin oligomers.
To better understand the function of ␤-etherase pathway enzymes, we sought to use a minimal set of enzymes to develop a coupled in vitro assay capable of releasing G, S, and T aromatic monomers and recycling the cosubstrates NAD ϩ and GSH. Here we demonstrate complete conversion of GGE to guaiacol and HPV in a reaction including LigD, LigN, LigE, LigF, NaGST NU , and the Allochromatium vinosum DSM180 GSH reduc-tase (AvGR), which catalyzes NADH-dependent reduction of GSSG and recycles the cofactors needed in the reaction (Fig. 1B) (26). We also show that this combination of enzymes releases tricin from the model compound guaiacylglycerol-␤-tricin ether (GTE [ Fig. 1C]). In addition, we show that the same combination of enzymes releases G, S, and T units from bona fide lignin oligomers. We discuss new insights gained from this study and its implications for the future production of these and possibly other valuable products from lignin.

RESULTS
Design of a coupled in vitro assay for cleavage of ␤-ether-linked diaromatic compounds. As an initial substrate for this assay we used erythro-GGE, which is a mixture of enantiomers (␣R,␤S)-GGE and (␣S,␤R)-GGE that has been used extensively as a substrate with ␤-etherase pathway enzymes (17)(18)(19)(20)(21). We used recombinant preparations of LigD and LigN, as these dehydrogenases are reported to be sufficient for the NAD ϩ -dependent oxidation of R-and S-configured ␣ anomers of erythro-GGE in vitro (19). The assay also contained recombinant preparations of LigE and LigF that have been shown to separately catalyze the GSH-dependent conversion of a racemic mixture of GGE to guaiacol and the S-and R-epimers of GS-HPV (21). NaGST NU was present to catalyze the GSH-dependent cleavage of the GS-HPV epimers to HPV and GSSG. The properties of individual enzymes (Fig. 1B) predict that this coupled system will require equimolar concentrations of GGE and NAD ϩ and twice as much GSH for complete conversion of GGE to HPV and guaiacol.
In an attempt to reduce the amount of added NAD ϩ and GSH that would be needed for full conversion of diaromatic substrate to products, some reaction mixtures included recombinant AvGR, which catalyzes the NADH-dependent reduction of GSSG (26), thereby recycling the cosubstrates NAD ϩ and GSH for continued conversion of the ␤-ether substrates. This cosubstrate recycling system was tested with 6 mM erythro-GGE and limiting concentrations of NAD ϩ (2 mM) and GSH (4 mM) ( Fig. 2A). For a summary of all reactions carried out in this study, see Table S1 in the supplemental material. Using a mixture of LigD, LigN, LigE, LigF, and NaGST NU , without AvGR (reaction 1 in Table S1), we observed that erythro-GGE was partially converted to HPV and guaiacol (Fig. 2B). Quantification of this assay revealed that the erythro-GGE concentration decreased from 6.0 mM to 3.8 mM at the end of the assay, whereas the HPV and guaiacol concentrations were each 2.0 mM, the NAD ϩ levels were nondetectable, and the concentration of threo-GGE [a mixture of enantiomers (␣R,␤R)-GGE and (␣S,␤S)-GGE] increased to 0.1 mM, presumably due to the reported reversibility of the LigD/ LigN reactions (27). Thus, the final GGE concentration (3.9 mM, the sum of the erythro-GGE and threo-GGE concentrations) was consistent with consumption of 2.0 mM NAD ϩ . In addition, the production of 2.0 mM (each) HPV and guaiacol was consistent with the consumption of 4.0 mM GSH, where 2.0 mM GSH was consumed in the LigE/LigF reactions and an additional 2.0 mM GSH was consumed in the NaGST NU reaction.
To test the impact of AvGR on this assay, we added it to a parallel in vitro reaction (reaction 2 in Table S1). We found that in the presence of AvGR (Fig. 2C), GGE was completely consumed and equimolar amounts of HPV and guaiacol (6.0 mM each) were produced, without a detectable change in the NAD ϩ concentration or accumulation of any ␤-etherase pathway intermediates by the time of the assay's conclusion. To determine if any ␤-etherase pathway intermediates accumulated over the course of the assay, we tested for time-dependent changes in the concentrations of the substrate, known pathway intermediates, and products in a parallel reaction (Fig. 3). We found that as erythro-GGE degradation occurs there is a time-dependent accumulation and depletion of GGE-ketone and threo-GGE and, eventually, complete equimolar conversion of the substrate to HPV and guaiacol ( Fig. 2A to C). From these results, we conclude that the combination of LigD, LigN, LigE, LigF, NaGST NU , and AvGR is sufficient to process all of the chiral centers in a ␤-ether substrate such as erythro-GGE. In addition, we conclude that the presence of AvGR is sufficient to recycle the cosubstrates NAD ϩ and GSH that are needed for cleavage of ␤-ether bonds in a model diaromatic compound such as erythro-GGE.
From information available in the literature, it has remained unclear whether the GSH lyase from Sphingobium strain SYK-6, LigG, has a preference for ␤(R)-GS-HPV (18,21) or is capable of cleaving the thioether linkages in both ␤(R)-GS-HPV and ␤(S)-GS-HPV (24). As the presence of NaGST NU resulted in cleavage of both ␤(R)-GS-HPV and ␤(S)-GS-HPV in this coupled reaction system ( Fig. 2A to C), we sought to use this in vitro assay to test the activity of LigG under identical conditions (reaction 3 in Table S1). When we performed an assay using 6.0 mM erythro-GGE, 2.0 mM NAD ϩ , and 4.0 mM GSH, as well as the mixture of LigD, LigN, LigE, LigF, and LigG (without AvGR), we observed partial conversion of GGE to HPV and guaiacol ( Fig. 2D). At the end of this assay, the total GGE concentration (4.0 mM, the sum of the erythro-GGE and threo-GGE concentrations) was expected based on the consumption of 2.0 mM NAD ϩ . Further, the production of 2.0 mM (each) HPV and guaiacol was consistent with the consumption of 4.0 mM GSH (2.0 mM GSH consumed by each of the LigE/LigF and LigG reaction steps). When we added AvGR to a parallel reaction (reaction 4 in Table S1) mixture that contained GGE (6.0 mM), NAD ϩ (2 mM), GSH (4 mM), and a combination of LigD, LigN, LigE, LigF, and LigG, we did not observe complete conversion of GGE to HPV and guaiacol (Fig. 2E). Instead, we detected the diaromatic substrate (erythro-GGE), threo-GGE, GS-HPV, and GGE-ketone (0.7 mM). In contrast to what was found when NaGST NU was present under identical reaction conditions, the presence of LigG led to incomplete utilization of the diaromatic substrate and the accumulation of ␤-etherase pathway intermediates. From these results, we conclude that LigG is not able to completely cleave both ␤-epimers of GS-HPV in vitro. Consequently, all subsequent assays were performed using NaGST NU as a source of GSH lyase activity.

Production of tricin from GTE in vitro.
In grasses, the flavone tricin (T [ Fig. 1A]) is covalently linked to one end of lignin, via a ␤-ether bond (8,9,28). Although ␤-etherase pathway enzymes have been shown to cleave ␤-ether-linked diaromatic model compounds containing G and S monomers, to date there are no published data on their ability to remove the diaromatic flavonoid T units from any substrate. Thus, we sought to test the ability of the coupled assay to cleave GTE (Fig. 1C), a model compound containing a ␤-ether-linked tricin moiety. High-performance liquid chromatography (HPLC) analysis of the synthetic GTE ( When we incubated 1.0 mM GTE, 5.0 mM NAD ϩ , and 5.0 mM GSH with the combination of LigD, LigN, LigE, LigF and NaGST NU (reaction 5 in Table S1), we observed the complete conversion of GTE to tricin and HPV (Fig. 4B). This result predicts that LigD and LigN oxidize GTE to form GTE-ketone, LigE and LigF catalyze ␤-ether cleavage in GTE-ketone to form GS-HPV and tricin, and NaGST NU releases HPV from GS-HPV (Fig. 1C), suggesting that the larger ␤-ether-linked flavone model was able to access the active sites in LigD, LigN, LigE, and LigF. To test this hypothesis, we assayed for the presence of the expected ␤-etherase pathway intermediates, GS-HPV and GTE-ketone, from GTE. By performing a parallel reaction including the same substrates and only LigD, LigN, LigE, and LigF (reaction 6 in Table S1), we observed that GTE was degraded and tricin was produced (Fig. 4C). However, in this assay, there was no detectable production of HPV and we observed accumulation of GS-HPV. These findings indicate that the absence of NaGST NU prevented the conversion of GS-HPV to HPV (Fig. 1C). Finally, in an assay including only the enzymes LigD and LigN (reaction 7 in Table S1), we found that GTE was almost completely converted to GTE-ketone (Fig. 4D), as expected from the NAD ϩ -dependent ␣-oxidation of GTE. Together, the data show that T units can be derived from ␤-ether-linked model compounds in vitro using enzymes, cosubstrates, and intermediates that are known to be part of the ␤-etherase pathway (Fig. 1).

Release of G, S, and T units from lignin oligomers.
With the coupled enzymatic system in place, we tested it for activity with lignin oligomers. First, we tested if a mixture of LigD, LigN, LigE, LigF, NaGST NU , and AvGR (reaction 8 in Table S1) produced S units from a high-syringyl hybrid poplar (HP) lignin polymer (29,30). To ensure that the test was performed with lignin oligomers rather than low-MW material present in the lignin, we fractionated the HP lignin by gel permeation chromatography (GPC) and pooled the high-MW fractions ( Fig. 5; Table 1) for use as a substrate (Fig. 6). From 2.2 mg ml Ϫ1 of lignin oligomers having MWs between 9,000 and 12,000 (with 2 mM NAD ϩ and 4 mM GSH), we detected the production of 1.0 mM ␥-hydroxypropiosyringone (HPS), the HPV analog expected to be produced by cleavage of ␤-ether bonds from a syringyl unit at one end of the lignin chain. We also detected the formation of an unknown product in this reaction (Fig. 6B), which could be either a chemically modified S unit released from the HP lignin or a GS-linked intermediate product.  syringaresinol, a dimeric unit in the HP lignin polymer (29), was not detected as a product of the enzymatic reaction. Given the ability of the enzymatic assay to release HPS from HP lignin, we also tested for the release of aromatic monomers from a more complex lignin, such as the one derived from maize corn stover (MCS) (8,9). To generate substrates for these assays, we used preparative GPC to size fractionate MCS lignin ( Fig. 7; Table 2) and tested materials with different apparent MWs as the source of lignin oligomer substrates for enzyme assays (Fig. 8). To test for activity with these samples, we incubated LigD, LigN, LigE, LigF, NaGST NU , and AvGR (reactions 9 to 13 in Table S1) with MCS lignin oligomers (2.2 mg ml Ϫ1 ), 2.0 mM NAD ϩ , and 4.0 mM GSH (Fig. 8A). In these experiments, we detected release of HPV and HPS in assays using lignin oligomers with average MW ranging from 460 to 10,710 (Fig. 7). The highest concentrations of HPV (0.4 mM) and HPS (0.1 mM) were observed with lignin oligomers having an average MW of 1,390 (Fig. 8D) as the substrate. In general, larger lignin oligomers resulted in lower accumulation of HPV and HPS. In addition, similar to the observations with HP lignin, unknown products were detected in most of the enzymatic reactions with the different lignin fractions (Fig. 8). Tricin was observed as a reaction product only when using the lowest-MW fraction tested (MW ϭ 460 [ Fig. 8F]). In sum, we conclude from these experiments that a combination of LigD, LigN, LigE, LigF, NaGST NU , and AvGR can release some, but not all, G, S, and T units from MCS lignin oligomers.

DISCUSSION
In order to use a polymer like lignin as a source of valuable aromatics and other chemicals, it is necessary to develop new or improve on existing depolymerization strategies. Recently developed oxidative and hydrogenolytic depolymerization approaches are showing some promise for producing low-molecular-weight aromatics in near-theoretical yields (15,30), but there has also been considerable interest in exploring the biological production of aromatics from this renewable plant polymer (25). The use of the bacterial ␤-etherase pathway for biological depolymerization involves milder conditions that could result in the release of monomers, such as HPS and HPV, that are useful in their own right or extractable low-MW mixtures that could be further valorized (25, 31). There is now a large amount of information on the types of model diaromatic substrates recognized by individual ␤-etherase enzymes in vitro, the products of their activity, and their structural or functional relationships to other known enzymes (27,32). Despite this, information on their activity with lignin oligomers is lacking. In addition, as these are cytoplasmic enzymes, it is plausible that they evolved to break down ␤-ether links only in the smaller lignin oligomers that could be transported inside the cells. A recent study described an in vitro enzymatic treatment of lignin yielding 12% (by weight) of low-molecular-mass aromatic products, although it remains un-

Enzymatic Lignin Depolymerization
Applied and Environmental Microbiology known whether these aromatics were released from only small or also large oligomers (25).
In this work, we sought to develop a coupled in vitro system containing a set of ␤-etherase pathway enzymes that was capable of releasing monoaromatic compounds when incubated with different substrates. We reasoned that such a system would provide additional information on the ␤-etherase enzymes and aid in studies aimed at determining the requirements for release of valuable aromatics from bona fide lignin oligomers. We identified a minimum set of enzymes (LigD, LigN, LigE, LigF, NaGST NU , and AvGR) that, in a single procedure, cleaves ␤-ether linkages and completely converts model diaromatic compounds to aromatic monomers. We further showed that this coupled in vitro assay system is capable of stoichiometric production of monoaromatic products from model diaromatics in the presence of limiting amounts of the cosubstrates NAD ϩ and GSH. The ability to recycle NAD ϩ and GSH allows the use of small quantities of these expensive cofactors and increases the future utility of a coupled enzyme system for processing lignin oligomers in vitro. Finally, we showed that this coupled enzyme system has activity with fractionated lignin oligomers. Below we summarize the new information gained from using this assay with widely used or new model ␤-ether linked substrates as well as lignin oligomers of different sizes.
Insights gained from using the coupled assay with diaromatic compounds. Using GGE as a substrate, we demonstrated that the GSH reductase AvGR is capable of recycling the cosubstrates NAD ϩ and GSH, enabling the ␤-etherase enzymes to completely cleave GGE in the presence of substoichiometric amounts of these cofactors ( Fig. 2A to C). The A. vinosum DSM180 AvGR is well suited for this purpose, as most glutathione reductases described in the literature use NADPH instead of NADH as an electron donor (33). When AvGR was not present in an assay in which GGE concentrations were greater than those of NAD ϩ and GSH, there was incomplete hydrolysis of this diaromatic substrate, accumulation of ␤-etherase pathway intermediates, and depletion of NAD ϩ , as expected if the reaction was cofactor limited.
We were also able to detect the release of tricin when GTE was used as a substrate in this assay, showing that ␤-etherase pathway enzymes are capable of ␤-ether bond cleavage in a substrate bearing a large flavonoid moiety. This further shows that the ␤-etherase pathway enzymes are not limited to substrates containing only G and S monoaromatic units. In prior research, we had demonstrated the ability of LigE and LigF to cleave G-G, G-S, S-G, and S-S dimer models (17,21), so this result extends the knowledge of the diversity of substrates for these enzymes to the G-T dimers. Thus, although the ␤-etherase pathway enzymes are thought to be highly stereospecific, they are also capable of recognizing the many different configurations of ␤-ether-linked aromatics potentially present in lignin. With the results of these and previous findings combined (17,21), we conclude that the minimal set of enzymes used in this study is sufficient to enable the ␤-etherase pathway in vitro to release G, S, and T units from compounds modeling ␤-ether units in lignin.
This coupled assay also allowed us to directly compare the abilities of LigG and NaGST NU to function in the ␤-etherase pathway. We found that the presence of NaGST NU and AvGR, along with LigD, LigN, LigE, and LigF, was sufficient to allow complete conversion of GGE to HPV and guaiacol ( Fig. 2A to C). This is consistent with our prediction that NaGST NU can accommodate both GS-HPV epimers in its active site (Kontur et al., submitted) and the ability of this enzyme to produce stoichiometric amounts of HPV from GGE when added to this coupled assay. In contrast, when LigG replaced NaGST NU under otherwise identical assay conditions, there was incomplete hydrolysis of GGE to HPV and guaiacol, with significant accumulation of GGE-ketone and smaller amounts of GS-HPV ( Fig. 2A, D, and E). Thus, although it has been suggested that LigG can hydrolyze both ␤-epimers of GS-HPV (24), this result, along with those published previously (21), supports the hypothesis that LigG has a strong preference for ␤(R)-GS-HPV. This direct comparison of substrate conversion to products in assays that differ only in the addition of LigG or NaGST NU allows us to conclude that use of the latter enzyme has advantages owing to its greater ability to release HPV from both GS-HPV epimers under comparable conditions in vitro.

Release of aromatic monomers from lignin oligomers in vitro.
The features of this coupled ␤-etherase assay allowed us to begin testing the ability to remove monomer aromatics from bona fide lignin. Lignin is a heterogeneous, high-molecularweight polymer, with only limited solubility under the aqueous buffer conditions used for this assay. Consequently, to increase our chances of observing aromatic products under the conditions used for the coupled assay, we used several different lignin oligomers. We also fractionated these materials to test for release of aromatics from different-size lignin oligomers. This has provided several important new insights into the activity of ␤-etherase enzymes with lignin oligomers and identified opportunities for increasing our understanding of this pathway.
We tested the ability of this enzyme mixture to cleave lignin oligomers that were derived from HP lignin, an engineered poplar line comprised of as much as 97.5% S units (29). HPS was detected as a product when high-molecular-weight fractions of the HP lignin were used as the substrate. This provides direct proof that the enzyme mixture cleaves aromatic oligomers containing S units and that this set of ␤-etherase pathway enzymes is active with lignin oligomers. Given that the vast majority of the aromatic units in HP lignin are S units (29), we estimate that the oligomers used in the enzymatic assay had between 40 and 50 aromatic units (Table 1). With the concentration of lignin oligomer used in this assay (ϳ2.2 mg ml Ϫ1 ), complete substrate degradation would yield ϳ8 mM HPS. The measured HPS concentration in this assay was 1.0 mM, resulting in a 12.5% yield of HPS from HP lignin. Thus, it appears that the mixture of enzymes used in this study, while sufficient for complete cleavage of model diaromatic compounds and of some ␤-ether links in HP lignin, is not capable of complete cleavage of all the ␤-ether linkages in the HP lignin oligomers. It is possible that a heretofore-undescribed protein is required to further process these lignin oligomers or that inhibition of enzyme activity was caused by the presence of some of the high-MW oligomers. Although our findings with the model dimers, and previous research, indicate that LigD and LigN are sufficient for complete oxidation of diaromatic compounds (Fig. 2) (19,34), it is possible that the seemingly redundant dehydrogenases LigO and LigL have a higher affinity for higher-MW lignin oligomers. Similarly, LigP, a GSH-S-transferase with apparent redundant activity with LigE (20), may be of interest for the optimization of in vitro lignin depolymerization.
In the assays using HP lignin as a substrate, we did not detect syringaresinol as a product, even though this dimer is found in moderate abundance in this polymer (29). Existing models for the composition of HP lignin predict that syringaresinol is primarily internal to the polymer (29). Thus, it is possible that the failure to detect syringaresinol as a reaction product reflects the inability of the tested ␤-etherase enzymes to access and cleave ␤-ether bonds that are adjacent to a syringaresinol moiety or, perhaps, that the enzymes exhibited only limited exolytic activity, thus preventing the enzymes from ever reaching syringaresinol units in the polymer.
Having established that the coupled enzymatic assay exhibited ␤-etherase catalytic activity with high-MW fractions of the HP lignin oligomers, we tested a more complex lignin sample from corn stover as a substrate (MCS lignin). This lignin has been thoroughly characterized and shown to contain only tricin, which is covalently bonded to the lignin oligomers (9). Fractionation of this lignin was carried out and experiments with a wider array of lignin fractions were conducted to test for the release of the major aromatic monomers present in this material (G, S, and T units). Unfortunately, background absorbance was observed in the chromatograms, for all lignin fractions analyzed, at retention times that overlapped with tricin's. Although this background absorbance was present in all lignin fractions, fortunately, tricin was detectable as a defined absorbance peak with a retention time identical to that from the pure standard. The detection of HPV, HPS, and tricin from different MCS lignin fractions confirms the observations with the ␤-ether-linked models that the enzyme set used was active in the release of G, S, and T units from lignin. However, tricin was observed only with the lignin fraction having an average MW of 460 (Fig. 7). Using a crude assumption that the average aromatic unit in lignin has an MW of 210 and the known MW of tricin (330), this fraction represents mostly lignin dimers or a T unit with at most one or two other S or G units. Thus, the ability of the enzymes to cleave the ␤-ether linkage next to a flavonoid moiety appears to be restricted to lower-MW oligomers. In contrast, HPS and HPV were released from MCS lignin in assays using all of the fractions tested (Fig. 8), which we estimate to encompass a range of oligomers from dimers to 50-unit oligomers ( Table 2). The highest measured concentration of HPS and HPV corresponded to the lignin fraction with an average MW of 1,390, or ϳ7 aromatic units (Table 2). Using the same assumption of 210 as the average MW of an aromatic unit in lignin, and the mass of lignin used in the assay (2.2 mg ml Ϫ1 ), we estimate a yield of HPS plus HPV of ϳ5%, which is lower than the estimated HPS yield from HP lignin. This lower release of substrates from MCS than HP lignin likely reflects the more heterogeneous and complex structure of the MCS lignin sample and potential inability of the ␤-etherase pathway enzymes to access and cleave all ␤-ether bonds in the polymer. The low yield of low-MW aromatics in the multistep enzymatic study (25) is in agreement with the low release of HPS and HPV observed in this study, supporting the hypothesis that the currently known ␤-etherase enzymes are not sufficient for complete breakdown of ␤-ether bonds in polymeric lignins.
Taken together, the findings presented here reveal new and exciting features of the ␤-etherase pathway enzymes. We identified tricin as a valuable flavonoid that can be enzymatically cleaved from ␤-ether-linked models and from low-MW lignin fractions. We also demonstrated ␤-etherase activity with intact lignin oligomers of various sizes, some of which might even be too large to be transported into cells. These findings therefore provide a demonstration that partial in vitro depolymerization of lignin is possible with ␤-etherase enzymes, an important step toward the development of biotechnological applications designed to derive high-value monomeric compounds from bona fide lignin polymers. Improving in vitro lignin depolymerization may depend on future discoveries of novel enzymes active on ␤-ether bonds, such as the newly characterized NaGST NU (Kontur et al., submitted) used in this study, or the engineering of existing ␤-etherase enzymes to improve the range of ␤-ether-containing substrates they can utilize. Ultimately, the activity of the studied enzymes on oligomeric substrates provides an opportunity to develop and optimize conditions for aromatic release from lignin fractions derived from biomass deconstruction chemistries that are or will be used by industry.

MATERIALS AND METHODS
General. GGE was purchased from TCI America (Portland, OR). Tricin, GTE, GTE-ketone, HPV, ␥-hydroxypropiosyringone (HPS), and GGE-ketone were synthesized by previously described methods (7,9,35). All other chemicals were purchased from Sigma-Aldrich (St. Louis, MO). Methods to isolate and thoroughly characterize maize (Zea mays) corn stover (MCS) and high-syringyl transgenic hybrid poplar (HP) lignin samples, including gel permeation chromatography (GPC) and nuclear magnetic resonance (NMR) analyses and interpretation, were described previously (9,29,30). Manipulation of DNA and preparation of Escherichia coli transformant cultures were carried out according to previously described methods (36). All lig genes from Sphingobium sp. strain SYK-6, as well as the gene encoding AvGR from A. vinosum DSM180, were codon optimized for expression in E. coli and obtained from GeneArt (Life Technologies). NaGST NU was amplified and cloned from N. aromaticivorans DSM12444 genomic DNA (Kontur et al., submitted).
Plasmid and protein preparation. Procedures for cloning, recombinant expression, and purification of Tev protease, LigE, LigF, LigG, and NaGST NU are described elsewhere (21; Kontur et al., submitted). ligD, ligN, and the gene for AvGR, all codon optimized, were cloned into plasmid pVP302K (21) via the PCR overlap method (37)(38)(39)(40). Expression and purification of LigD, LigN, NaGST NU , and AvGR followed procedures similar to those used previously (21). Briefly, E. coli strain B834 cultures, transformed with expression plasmids, were grown aerobically overnight in 1 liter of autoinduction ZYM-5052 medium (41) supplemented with 100 g ml Ϫ1 of kanamycin. Cells were pelleted and extracts prepared via compression and sonication. Histidine-tagged proteins were purified from cell lysates via nickel-nitrilotriacetic acid (Ni-NTA) affinity chromatography with Qiagen Ni-NTA resin. His-tagged Tev protease was used to liberate N-terminal His tags and a second round of Ni-NTA affinity chromatography was used to remove the tag and Tev protease before separation by size exclusion chromatography. Protein preparations were concentrated and frozen with liquid N 2 .
Enzyme assays. In vitro enzyme assays with LigD, LigN, LigE, LigF, NaGST NU (or LigG), and AvGR (or a subset of those enzymes) were conducted in assay buffer (25 mM Tris, 2.0% dimethyl sulfoxide [DMSO] [pH 8.0]). The concentration of each enzyme was 50 g ml Ϫ1 in all assays. When GGE (6 mM) was the substrate, the initial cosubstrate concentrations were 2 mM NAD ϩ and 4 mM GSH. When GTE (1 mM) was the substrate, the initial cosubstrate concentrations were 5 mM NAD ϩ and 5 mM GSH. When an isolated lignin sample was used as the substrate (2.2 mg ml Ϫ1 ), the initial cosubstrate concentrations were 2 mM NAD ϩ and 4 mM GSH. Enzyme assays (volume, 1 ml or larger as needed) were carried out (in duplicate) in 2-ml vials that received the following in sequential order: (i) 20 l of substrate (GGE, GTE, or lignin) dissolved in DMSO (concentrated 50 times above the intended assay concentration), (ii) 880 l of 25.6 mM Tris (pH 11.5) (where the acidic effect of GSH drops the pH to 8.0 after addition of 5 mM GSH), (iii) 50 l of a stock solution in 25 mM Tris containing NAD ϩ and GSH (each is concentrated 20 times above the intended assay concentration), and (iv) 50 l of a 20-times-concentrated mixture of the desired enzymes. At the desired time points, 150-l samples were removed from an assay and enzymatic activity was abolished by pipetting into 5 l of 5 M phosphoric acid. GGE, guaiacol, HPV, and HPS concentrations were quantified for each time point (see below) using a linear regression of known standards for each compound.
GPC. (i) Preparative GPC. GPC of lignin samples was carried out using a Beckman 125NM solvent delivery module equipped with a Beckman 168 UV detector ( ϭ 280 nm) and a 30-ml Bio-Rad Bio Bead S-X3 column (a neutral, porous styrene-divinylbenzene copolymer). Dimethylformamide (DMF) was used as the mobile phase at a flow rate of 1.0 ml min Ϫ1 . Between 20 and 50 mg of lignin was dissolved in a minimal amount of DMF and injected into the mobile phase, and 1-ml fractions were collected until UV absorption decreased to baseline levels. Fractions were then subjected to analytical GPC to estimate their average molecular weight (MW). The DMF was evaporated in vacuo in order to recover material used for enzyme assays.
(ii) Analytical GPC. Analytical GPC of lignin samples was carried out with a Shimadzu Prominence ultrafast liquid chromatography system (LC-20AD pumps, SIL-20AC HT autosampler, CTO-20A column oven, and CBM-20A controller) and using two TSKgel Alpha-2500 (300 by 7.8 mm; Tosoh Bioscience) columns at 40°C. Samples (10-l injection volume) containing approximately 1 mg ml Ϫ1 of isolated or GPC-fractionated lignin were injected into a mobile phase (100 M LiBr in DMF) at a flow rate of 0.3 ml min Ϫ1 with a run length of 90 min. An SPD-M20A photodiode array detector ( ϭ 200 nm) was used for the determination of elution times, which were subsequently converted to MWs using regression analysis of ReadyCal kit polystyrene standards. C 18 column chromatography. C 18 column chromatographic separations were carried out using a Beckman 125NM solvent delivery module equipped with a Beckman 168 UV detector. Samples of 150 l were collected from enzyme assays and 20-l aliquots were injected into either a 4-by 120-mm Restek Ultra aqueous C 18 -reversed stationary-phase column or a 4.6-by 250-mm Phenomenex Luna 5u C 18 (2)-reversed stationary-phase column with a 1.0-ml min Ϫ1 mobile phase composed of a mixture of an aqueous buffer (5 mM formic acid in 95% H 2 O-5% acetonitrile) and methanol. Samples from enzyme assays using GTE as the substrate were analyzed on the Phenomenex column to improve separation of GTE and tricin. All other C 18 column chromatographic separations were carried out using the Restek column. For the Restek column, the methanol fraction of the buffer (with aqueous buffer as the remainder) was adjusted as follows: 0 to 6 min, 30% methanol; 6 to 15 min, gradient from 30 to 80% methanol; 15 to 27 min, 80% methanol; 27 to 28 min, gradient from 80 to 30% methanol; and 28 to 33 min, 30% methanol. For the Phenomenex column, the gradient system was as follows: 0 to 6 min, 10% methanol; 6 to 50 min, gradient from 10 to 90% methanol; 50 to 63 min, 90% methanol; 63 to 64 min, gradient from 90 to 10% methanol; and 64 to 70 min, 10% methanol.

SUPPLEMENTAL MATERIAL
Supplemental material for this article may be found at https://doi.org/10.1128/AEM .02076-17.
SUPPLEMENTAL FILE 1, PDF file, 0.2 MB. the manuscript. All authors approved of the final version of the manuscript and the data presented.