Rapid characterization of the activities of lignin-modifying enzymes based on nanostructure-initiator mass spectrometry (NIMS)

Lignin is a major component of lignocellulosic biomass [1] and has the potential to be a valuable starting material for producing biofuels or high value bioproducts [2]. Moreover, conversion of lignin to fuels and chemicals [3–5] is regarded as a key factor for making lignocellulosic biomass economically feasible. The viability of lignin degradation depends on the availability of optimal lignin-modifying enzymes [6–9] that can efficiently degrade lignin biopolymers into simpler aromatics.

Enzymes responsible for lignin breakdown target the common lignin linkages [10, 11]: β-O-4 (arylglycerol-β-aryl ether), 5–5′ (biphenyl), β-5 (phenylcoumaran), 4-O-5 (diaryl ether), etc. Among these linkages, β-O-4 linkages represent approximately 50% and 60% of the total bonds of softwood and hardwood, respectively; making them important targets for enzyme assays. Hence, several surrogate substrates containing chromophores have been developed for studying cleavage of β-O-4 linkages. For example, guaiacylglycol and glycerol- β-O-(β-methylumbelliferyl) ethers have been synthesized for the fluorometric assays for β-etherases [12, 13]. As an alternative to spectroscopic assays, mass spectrometry-based enzyme activity assays enable detection of multiple products from a single substrate. One such approach uses nanostructure-initiator mass spectrometry (NIMS) [14–16] and glycan substrates [17] containing a charged perfluorinated tag, which partitions onto the NIMS surface, greatly enhancing detection of substrates and products.

Here, we describe the use of two synthetic β-aryl ether substrates for studying lignin-modifying enzymes using NIMS, a recently developed surface-based desorption ionization technique [14]. The NIMS chip is especially fabricated by etching the surface of a porous silicon wafer and treating it with an initiator molecule (e.g. a fluorous disiloxane). Samples are spotted onto the NIMS surface and then laser desorption/ionization of the analytes from the surface is used for the MS analysis. One interesting application of NIMS is called nimzyme [15] in which a fluorous tagged substrate (e.g., tagged glycans) is immobilized onto an NIMS surface. After incubating the surface with an E. coli lysate or a microbial community lysate, the NIMS surface was then washed. The products from the enzyme reaction can be detected by direct MS readout. More recently, nimzyme has been expanded to the use of soluble model substrates [16] to provide the specificity and sensitivity required to study glycosylhydrolases. One big advantage of the nimzyme method is multiplexing, which is achieved by simultaneous measurement of enzyme activities on multiple substrates. For multiplexing, different fluorous tags are typically needed for each substrate. The fluorous tag for each substrate is chosen to enable the differentiation of similar or identical products [17] from different enzymatic reactions by their unique mass tag (i.e., enabling different m/z in mass spectrometry). Nimzyme has been recently integrated with acoustic printing to create an enzyme characterization platform capable of one second/sample throughput, while only requiring only 20 μL sample volumes [18]. Currently, this platform has been used for measuring enzyme activities of glycosylhydrolases against a panel of well-characterized substrates [19].

Since approximately one-third of plant biomass is lignin, it is desirable to complement these existing assays for cellulose and hemicellulose active enzymes with assays designed to characterize lignin active enzymes. In addition, since NIMS is matrix free, data interpretation is readily facilitated, in comparison to MALDI-based assays that often use monolignols for desorption/ionization resulting in extensive lignin-related background ions.

Small-angle neutron scattering (SANS) reveals that these amphiphilic molecules form micelles like previous NIMS glycan substrates [16]. Micelles formation is important, and we hypothesize that the structure of micelles may reflect more natural reaction conditions in comparison to completely soluble substrates. Finally, the application of our LMEs assay demonstrates that phenolic and nonphenolic β-aryl ether substrates adopt different reaction pathways.

Since the substrate molecule (Fig. 1) consists of a hydrophobic fluorinated segment and a hydrophilic arylglycerol, it is expected that the molecules will self-assemble in aqueous solutions as we have previously observed [16]. To examine this, small-angle neutron scattering (SANS) was performed with compound (I) in aqueous solution at 5 mM. The SANS data suggest the formation of micelles. As shown in Fig. 2, the SANS curve is fitted to a spherical model with a mean radius of 2.6 nm (polydispersity 11%). This dimension is consistent with the estimated size of these molecules, ~ 2.6 nm. A core–shell structure is not revealed by SANS data because the neutron scattering contrast between the hydrophobic and hydrophilic segments are similiar to each other [20]. These aggregates/micelles may reflect more natural reaction conditions in comparison to completely soluble substrate molecules often used in activity screening as many lignin-degrading enzymes act on insoluble solid lignin substrates.

Figure 3 illustrates the results from the reactions of phenolic β-aryl ether dimeric substrates with two different lignin active enzymes, a laccase from the polypore mushroom Trametes Versicolor, and a manganese peroxidase (MnP) from the white rot fungus Nematoloma frowardii. Both reactions were performed at room temperature for 18 h. These results revealed that both enzymes generate similar product profiles, with slight differences in individual product concentration, suggesting that both enzymatic reactions may proceed via a similar mechanism.

It is interesting to see the formation of a small amount of product (m/z 2299) from the dimerization of phenolic β-O-4 substrate (m/z 1151), and no higher order of polymerization than dimerization was observed. Presumably, the micellar structure of the phenolic substrates forbids the access to the dimeric substrate from other phenoxy radicals. Over time, the alcohol in the dimeric product (m/z 2299) was found to be oxidized by enzymes to the corresponding carbonyl products as shown in the mass spectra (peak with m/z 2297). It seems that only one hydroxyl group is being oxidized, because no peaks corresponding to the products with multiple oxidation were observed.

The formation of this dimeric product most likely occurred through a phenoxy radical intermediate generated by the oxidation of the phenol subunit. Dimerization of the phenolic β-O-4 substrate via intermolecular phenoxy radical cross-coupling would produce product with the construction of carbon–carbon or carbon–oxygen bonds like the 5–5 or 4-O-5 linkages in lignin [21]. According to the NMR studies by Butler et al. [21], the product from dimerization is very likely to be formed by the 5–5 linkage.

Previously, Rittstieg et al. [22] used guaiacylglycerol-β-guaiacyl ether, a compound with structure similar to substrate (I) but without the NIMS tag to study laccase activity. In their study, a polymeric precipitate was observed following treatment of the substrate with laccase, which was attributed to a free-radical initiated polymerization reaction. Interestingly, we did not observe any precipitation in the present study. We attribute this to a putative influence of the NIMS probe, which may inhibit polymerization, meaning that only monomeric compounds are formed. This finding is consistent with the work of Gold et al. [23] who added a methoxy group at the 5′ position to prevent polymerization. While indeed the NIMS probe may be advantageous for the prevention of polymerization, it may also introduce bias due to the interactions of the fluorous tag with the enzyme and this topic will be an important area for future investigation.

The potential mechanism for the formation of products from phenolic β-aryl ether substrate (I) is shown in Fig. 4. The phenoxy radical intermediate II, generated by the oxidation of phenolic substrate I by laccase through single electron transfer, can delocalize to form resonance structure III. The loss of another electron from this carbon radical by laccase oxidation affords an important cation intermediate IV. Two distinct pathways are probably operational here and depend on the phenoxy radical intermediate (II). Pathway 1 involves the C_α–C_β bond cleavage [23–25], followed by rapid cleavage of the C_β–O ether bond to form product IX. C_α–C_β bond cleavage can also proceed through the phenoxy radical intermediates of the phenolic β-aryl ether substrate I or its C_α-oxo-product VII. Alternatively, the carbon–carbon single bond between the C_α and the aryl carbon can be cleaved. Again, the phenoxy radical is the intermediate needed to initiate the entire reaction cascade to produce aldehyde product VIII. For both reactions of laccase and MnP with phenolic β-aryl ether substrate, aldehyde product VIII is consistently produced in higher amounts than product IX, which indicates that the pathway with the cleavage of the C_α-alkyl phenyl bond is predominant relative to the pathway with the C_α–C_β bond cleavage. Product X (with m/z 1045) is presumably the hydrate form of aldehyde VIII.

Interestingly, when methanol was added to quench the reaction, the peak corresponding to the hydrate of aldehyde (m/z 1045) was reduced significantly and the new peak appearing at m/z 1059 was consistent with the formation of a product from the methanolysis of aldehyde VIII. When deuterated methanol (CD₃OD) was used to quench the reaction, as expected, the mass peak with m/z 1062 appeared. These results confirmed the identity of aldehyde VIII.

The time course of an enzymatic reaction contains valuable information about the properties of the enzyme. By obtaining product distribution at different time points in the enzymatic reaction, we can determine the reaction pathways, bottleneck steps in the reaction and key enzyme parameters, etc. In our NIMS-based enzyme assay, substrate and products concentrations were measured by the relative intensity of the m/z signals of each compound. It is important to point out a few attributes of these assays. Since both substrates and products possess highly similar fluorous tails and, thus, the dimethyl arginine groups should enable different molecules with similar ionization efficiency in NIMS, so it is possible to directly compare ion intensities as an approximation of substrate and product concentrations. Also, since multiple products are measured simultaneously from a 0.2 μL sample (from a 10 μL reaction) at each time point the disturbance of the reaction is limited. These analyses also consume little of the precious synthetic substrates. Figure 5 shows the reaction time course of laccase with phenolic β-aryl ether substrate (I) at room temperature over a time period of 180 min. Over the course of the experiment, a continuous decrease of the phenolic β-aryl ether substrate is observed along with an increase in the product formation (products IX, VII, VIII) with the exception of the dimerization product which, after initial increase in relative abundance in the first h, was completely converted to oxidation product (with m/z 2297).

As for the reaction of laccase with nonphenolic β-aryl ether substrate (Fig. 6), temperatures of 37 °C and longer reaction times were needed to accurately measure reaction products, and products profiles are different from those obtained when using the phenolic β-aryl ether substrate. Since there is no phenoxy radical intermediate, no product from the cleavage of carbon–carbon single bond between the C_α and the adjacent aryl carbon was observed. C_α-oxo-product XIV is predominant compared to product XV, which comes from the C_α–C_β bond cleavage. In addition, HOBt was added to the reaction as a mediator compared to no HOBt being added to the reaction of laccase with phenolic substrate (I). It is known that laccase reactivity decreases with an increase of the steric encumbrance; the use of mediator, HOBt, can help to overcome the problems related to substrate accessibility [9]. In this case, the nonphenolic substrate (XII) may have greater difficulty entering the laccase active site compared to phenolic substrate (I). HOBt is a small molecule that can be oxidized by laccase to an intermediate with high redox potential, which can oxidize the nonphenolic substrate (XII) to the corresponding products. More interestingly, some products, likely from the oxidation of the phenyl ring [26], were observed as evidenced by m/z 1193, 1195, which correspond to the addition of a hydroxyl group on the aromatic ring [26] of substrates XII and C_α-oxo-product XIV. The location of the hydroxyl group on the ring is still unclear; however, a small peak with m/z 985 was detected which may indicate that the oxidation occurs on the phenyl ring linked to the C_β through oxygen. When using identical reaction conditions, but without HOBt, only the substrate was detected using mass spectrometry.

The structure of our model compounds is designed by coupling a fluorous tag with the β-O-4 dimer through a carbon chain. This design offers the opportunity to construct enantiomeric pure substrates with ease. The β-O-4 dimer itself has two chiral centers; therefore, there are four possible enantiomers. It is known that enzymes are chiral, and they can distinguish among enantiomers [27, 28]. To understand enzyme specificity, enantiomeric pure substrates are needed. Our approach (shown here) could easily be extended to prepare these important chiral substrates (e.g., each with different mass tags attached to ensure unique identity in mass spectrometry) by synthetic organic chemistry to study the enzyme specificity. Moreover, due to the multiplexing nature of our NIMS assays, multi-substrate/product systems can be utilized to thoroughly investigate the enzymatic activities/specificities. Progress in this area will be reported in the future. In addition to the approach using model compounds for the study of LMEs, development of post-reaction derivatization approach like those we have reported for analysis of glycosyl hydrolases [29] is an important direction for future research. These studies would have the advantages of minimizing experimental bias and allow analysis of enzymes against the most process relevant feedstocks.

In summary, we have synthesized two β-aryl ether substrates suitable for high-throughput assaying of lignin-modifying enzymes by NIMS. In solution, these amphiphilic molecules tend to form micelles. Use of these substrates to assay a laccase and a manganese peroxidase revealed that phenolic and nonphenolic β-aryl ether substrates demonstrate different primary reaction pathways due to the availability of the phenoxy radical intermediates. Further work will focus on preparing other model lignin compounds suitable for NIMS analysis to study catalysis of breaking lignin bond linkages. Together with the substrates already prepared for studying the activities of cellulases and hemicellulases, this pool of substrates forms the basis for a powerful mass spectrometry-based multiplexing assay, which has the ability to simultaneously detect multiple functions of enzymes (or enzyme cocktails) responsible for the deconstruction of lignocellulosic biomass. Ultimately, this approach may potentially aid in identifying more efficient, low cost enzyme cocktails useful for converting all biomass polymers into valuable bioproducts.