The liquid fraction from hydrothermal pretreatment of wheat straw provides lytic polysaccharide monooxygenases with both electrons and H2O2 co-substrate

Background Enzyme-aided valorization of lignocellulose represents a green and sustainable alternative to the traditional chemical industry. The recently discovered lytic polysaccharide monooxygenases (LPMOs) are important components of the state-of-the art enzyme cocktails for cellulose conversion. Yet, these monocopper enzymes are poorly characterized in terms of their kinetics, as exemplified by the growing evidence for that H2O2 may be a more efficient co-substrate for LPMOs than O2. LPMOs need external electron donors and one key question of relevance for bioprocess development is whether the required reducing power may be provided by the lignocellulosic substrate. Results Here, we show that the liquid fraction (LF) resulting from hydrothermal pretreatment of wheat straw supports LPMO activity on both chitin and cellulose. The initial, transient activity burst of the LPMO reaction was caused by the H2O2 present in the LF before addition of LPMO, while the steady-state rate of LPMO reaction was limited by the LPMO-independent production of H2O2 in the LF. H2O2 is an intermediate of LF oxidation as evidenced by a slow H2O2 accumulation in LF, despite high H2O2 production rates. This H2O2 scavenging ability of LF is important since high concentrations of H2O2 may lead to irreversible inactivation of LPMOs. Conclusions Our results support the growing understanding that fine-tuned control over the rates of H2O2 production and consumption in different, enzymatic and non-enzymatic reactions is essential for harnessing the full catalytic potential of LPMOs in lignocellulose valorization.


Background
Lignocellulosic biomass is the most abundant source of renewable carbon in Nature. Its enzyme-aided valorization to biofuels and building blocks for the chemical industry provides a green and sustainable alternative to the petroleum-based chemistry. Because of its inherent recalcitrance, the lignocellulose of plant cell walls requires mechano-chemical pretreatment to increase its susceptibility to enzymatic conversion. Hydrothermal pretreatment does not require use of chemicals and is a simple and environment friendly method that has proven to be efficient for different biomasses [1]. The most abundant soluble by-products from hydrothermal pretreatment, and from analogous dilute acid pretreatment [1], are hemicellulose-derived mono-and oligosaccharides, and various phenolic compounds [2][3][4].
Another breakthrough in the LPMO field was made in 2017 when it was shown that LPMOs can use H 2 O 2 instead of O 2 [46]. Although the nature of the true cosubstrate of LPMOs is a matter of scientific debate, the fact is that H 2 O 2 is used much more efficiently than O 2 [46][47][48][49]. The H 2 O 2 -based mechanism also depends on the presence of external electron donor, but here the reductant is only needed for the initial "priming" of the Cu(II) resting state of the LPMO to its catalytically active Cu(I) form [46,50]. Once in its active form, an LPMO can catalyze a number of oxidative cleavages until the active site copper happens to be re-oxidized, either by H 2 O 2 or O 2 [50]. The reductants that are required for LPMO activation are amenable to abiotic oxidation by O 2 and H 2 O 2 is often a product of these oxidations, complicating experimental assessment of LPMO action [51]. LPMOs are also a subject of irreversible inactivation by non-productive redox processes in the catalytic center [46,48,49,51]. Therefore, the fine-tuned control over the concentration of the oxygen co-substrate is of utmost importance in harnessing the full catalytic potential of LPMOs.
Here, we have studied to what extend liquid fractions (LFs) from hydrothermal pre-treatment of wheat straw support the degradation of cellulose by a Trichoderma reesei LPMO (TrLPMO9A) as well as the degradation of chitin by a Serratia marcescens LPMO (SmLPMO10A). Reducing power and LPMO-independent generation of H 2 O 2 in such liquid fractions were found to drive the LPMO activity, shedding new light on the possible interplay between biomass pretreatment and subsequent saccharification by LPMO-containing enzyme cocktails. 14 C-labeled chitin nanowhiskers (CNWs) were prepared by N-acetylation of non-labeled CNWs with 14 C-acetic anhydride exactly as described in Kuusk et al. [52]. 14 C-labeled bacterial cellulose was prepared by laboratory fermentation of Gluconobacter xylinum (ATCC 53582) in a medium supplied with uniformly 14 C-labeled glucose as described before [53,54]. 14 C-labeled microcrystalline cellulose (BMCC) was prepared by incubating 14 C-labeled bacterial cellulose (2 g L −1 ) with 1.0-M HCl at 100 °C for 3 h followed by extensive washing with water. The specific radioactivities of CNWs and BMCC were 4.18 × 10 6 and 6.4 × 10 5 dpm mg −1 , respectively. Before using in experiments, both CNWs and BMCC were treated with EDTA to remove divalent metal ions. For that the polysaccharide (2 g L −1 ) was incubated with 10-mM EDTA at room temperature overnight. The EDTA-treated polysaccharides were extensively washed with water and 50-mM sodium acetate (pH 5.0) through repetitive centrifugation and re-suspension steps. The water was Milli-Q ultrapure water that had been passed through a column with Chelex ® 100 resin (BioRad). The stock solution of sodium acetate buffer was stored over beads of Chelex ® 100 resin. Solutions of H 2 O 2 and ascorbic acid were made freshly before use.

Enzymes
SmLPMO10A was produced and purified as described before [55]. TrLPMO9A was produced as follows: the gene encoding TrLPMO9A was obtained by PCR from genomic DNA of T. reesei QM9414 using oligonucleotides SO1 (5′AAC CCA ATA GTC AAC CGC GGA CTG CGC ATC ATG ATC CAG AAG CTT TCC AA) and SO2 (5′ACC GGT GCG TCA GGC TTT CGC CAC GGA GCT CTA GTT AAG GCA CTG GGC GT). The expression vector was assembled with the yeast recombination cloning method using the PCR fragment and PacI (Fermentas) linearized pTTv248 vector backbone [56]. The final expression vector contained a targeting sequence for the cbh1 locus (tre123989), 2184 bp of cbh1 5′ region containing the cbh1 promoter and 1745 bp of cbh1 3′ region and the hphR selection marker [57,58]. After plasmid rescue and transformation into E. coli [56], the construct was verified by sequencing. The expression cassette was liberated from the vector backbone with PmeI (Fermentas) restriction enzyme digestion prior to transformation. T. reesei strain M362 (M124 Δtre72567, Δtre122081 and Δtre120312), which is deleted for three major cellulase genes (cbh2, egl1, egl2), was transformed with the expression cassette and grown on MM + hygromycin transformation plates [59]. Transformants were screened first by PCR for 5′ and 3′ flank integration into the cbh1 locus and absence of the open reading frame for cbh1 (for PCR primers see Additional file 1: Table S1). The generated strain, M1906, was cultivated for protein production in a BioFlo 510 15L reactor (New Brunswick Scientific, USA) with 10-L operating volume, using a culture medium containing lactose (40 g L −1 ), spent grain extract (30 g L −1 ) (Harbro Ltd, UK), KH 2 PO 4 (5 g L −1 ), (NH 4 ) 2 SO 4 (5 g L −1 ) MgSO 4 (2.4 mM), CaCl 2 (4.1 mM), CoCI 2 (3.7 mg L −1 ), FeSO 4 ·7H 2 O (5 mg L −1 ), ZnSO 4 ·7H 2 O (1.4 mg L −1 ) and MnSO 4 ·7H 2 O (1.6 mg L −1 ) and Struktol J647 Antifoam (1 mL L −1 ). The cultivation was carried out at 28 °C and pH 4.8-4.9. The pH was controlled by addition of base (5% NH 4 OH) or acid (10% H 3 PO 4 ) when necessary. The cultivation was done with constant aeration (10 L min −1 ) and mixing (150-500 rpm) was adjusted to keep the oxygen concentration at 30%. Lactose (20% (w/v)) feeding was initiated after 61-h cultivation and adjusted as described in [60]. The cultivation was terminated after 163 h. The culture supernatant was concentrated with Millipore Pellicon 2 filter (10 kDa membrane cutoff ). TrLPMO9A was purified with the following procedure: 0.5 L of the concentrated culture supernatant was exchanged to 10-mM sodium phosphate pH 7.0 using a Sephadex G25 column (column volume 3.5 L), applied to a DEAE Sepharose anion exchange column (column volume 1.0 L) and eluted using a 0-100 mM (30 column volumes) NaCl gradient. The fractions were analyzed with SDS-PAGE using 4-20% Stain-Free gradient gels and Imaging System (BioRad, Hercules, California, USA). The fractions containing TrLPMO9A were pooled and buffer was exchanged to 50 mM sodium acetate pH 5 using ultrafiltration (Prep/Scale-TFF 1ft 2 Cartridge, PTGC 10 k Polyethersulfone). Ammonium sulfate was added to the pooled sample to a final concentration 0.5 M after which the sample was applied to a Phenyl-Sepharose HIC column (column volume 0.14 L). TrLPMO9A was collected from the flow-through. The buffer of purified TrLP-MO9A was changed to 25-mM sodium acetate pH 5.0 and the enzyme was concentrated using ultrafiltration as described above. SDS-PAGE analysis of purified TrLP-MO9A is shown in the Additional file 1: Fig. S1. Contaminating endoglucanase [61], xylanase [62] and mannanase trace activities [63] of purified TrLPMO9A were 1.2, 3.2, and 2.6 nkat mg −1 protein, respectively.
The purified LPMOs (around 150 µM) were copper saturated by overnight incubation with CuSO 4 (threefold molar excess) and subsequent removal of free copper by ultrafiltration. The concentration of LPMOs was determined by measuring absorbance at 280 nm using molar extinction coefficients of 35,200 and 54,360 M −1 cm −1 for SmLPMO10A and TrLPMO9A, respectively. Horseradish peroxidase (HRP, Sigma) was used as purchased. The concentration of HRP was determined by measuring absorbance at 403 nm using molar extinction coefficient of 102,000 M −1 cm −1 .

Hydrothermal pretreatment of wheat straw
Chopped wheat straw from Finland was pre-soaked with water to 50% dry matter content and loaded into a 30-L pressure reactor with a batch size of 1.71-kg dry matter. The material was heated to 195 °C with direct steam injection and a pressurized water jacket, and the temperature was maintained for 15 min. After the treatment, the material was quickly cooled to 80 °C with the water jacket, and dissolved material was extracted by pumping 80 °C water through the material bed. The first 6 L of extract was collected and this material is hereafter referred to as liquid fraction (LF). The pretreated solids were collected manually. During the period of analysis (about 2 months) the LF was stored at 4 °C. After that, the LF was stored frozen as aliquots of appropriate volume.
The LF was analyzed for soluble carbohydrates by HPAEC with pulse amperometric detection (Dionex ICS 3000 equipped with CarboPac PA1 column). Analysis was performed before and after acid hydrolysis (3% H 2 SO 4 , 1 h at 120 °C), to determine both mono-and oligomeric sugars. Furfural, hydroxymethyl furfural and acetic acid were analyzed by HPLC, using a Bio-Rad Aminex HPX-87H column, with 5-mM H 2 SO 4 as eluent. Soluble phenolics were determined by UV-absorbance at 215 nm and 280 nm, according to the method for acid-soluble lignin determination described by Goldschmid [64].
Compositional analysis of the solids was performed according to Sluiter et al. [65]. The main components of the solid fraction were glucose (41.5%), xylose (7.3%), lignin (24.5%), and ash (5.8%). The solid fraction was kept frozen in plastic bags.

Degradation of CNWs by SmLPMO10A
Experiments were made in 50-mM sodium acetate pH 5.0 at 25 °C. Stirring was omitted but the reaction mixture was gently mixed with a pipet before each sampling. The concentration of CNWs was 1.0 g L −1 and the concentration of SmLPMO10A was varied between 0.05 and 0.25 µM. The SmLPMO10A was added to the CNWs followed by the addition of LF (pre-incubated at 25 °C for the indicated time) to start the reaction. The amount of added LF corresponded to 5%, 10% or 15% (v/v) of the final reaction volume. At selected time points, 0.1-mL aliquots were withdrawn and mixed with 0.025 mL of 1.0-M NaOH to stop the reaction. Non-labeled CNWs (to 3 g L −1 ) in 0.2-M NaOH were added to improve sedimentation [52] and solids were separated by centrifugation (5 min×10 4 g). SmLP-MO10A activity was calculated based on the concentration of radioactive soluble products (expressed in N-acetylglucosamine equivalents, NAG eq ) exactly as described in Kuusk et al. [48]. In this previous study, it was established that one SmLPMO10A oxidative cleavage on average leads to release of approximately four NAG eq and this 4 to 1 ratio takes into account that part of the oxidized sites remains in the insoluble substrate [48]. Therefore, the concentration of NAG eq corresponds to the total concentration of monosaccharide equivalents in the soluble fraction and is not dependent on the average degree of polymerization of soluble products (which is known to be around 8 for the SmLP-MO10A/CNWs system [48]). In the experiments with HRP, the LF (or solid fraction, see below) was mixed with CNWs. HRP (1.0-µM final concentration) was added to the mixture of CNWs and LF followed by the addition of SmLPMO10A (30 s after the addition of HRP) to start the reaction. In the control experiments without LF or SmLPMO10A, the experiments were made as described above but the LF or SmLPMO10A were replaced with corresponding amount of buffer.
In the experiments for measuring the concentration of H 2 O 2 in LF upon pre-incubation of LF at 50 °C, the aliquot of pre-incubated LF (to a final concentration of 10% v/v) was added to the mixture of CNWs and SmLPMO10A to start the reaction. The SmLPMO10A reaction was conducted at 25 °C as described above. In some cases, ascorbic acid (0.1-mM final concentration) was added (30 s before the addition of LF) to ensure efficient priming of SmLPMO10A [50]. The concentration of H 2 O 2 was calculated from the concentration of soluble NAG eq using a previously established stoichiometry of 4 NAG eq /H 2 O 2 [48].
In the experiments with the solid fraction from the hydrothermal pretreatment of wheat straw, 40-mL water (at 4 °C) was added to 5 g of frozen solid fraction. After 30 min, the solids were separated by centrifugation and the pellet was homogenized by grinding in a mortar (at 4 °C) until it was possible to handle the suspension with a pipet. The concentration of solids in homogenized material was measured by weighing (after drying in a rotary evaporator). The experiments with the solid fraction (10 g L −1 solids were added to SmLPMO10A reaction) were made exactly as described above for the experiments with LF.
In all cases, the sample for the zero-time point was withdrawn before the addition of LF and was treated as the other samples. The reading of the zero-time point was subtracted from the readings of all time points.

Degradation of BMCC by TrLPMO9A
Experiments were made in 50-mM sodium acetate pH 5.0 at 25 °C or 50 °C in 1.0-mL total volume. Stirring was omitted but the reaction mixture was gently mixed with a pipet before each sampling. The concentration of BMCC was 0.6 g L −1 or 1.2 g L −1 and that of TrLP-MO9A was varied between 0.1 and 0.5 µM. The LF (preincubated at 25 °C or 50 °C for the indicated time) was added to the BMCC and the reaction was started by the addition of TrLPMO9A. The amount of added LF corresponded to 10% or 20% (v/v) of the total reaction volume. At selected time points, 0.2-mL aliquots were withdrawn and the solids were immediately separated by centrifugation (2 min×10 4 g). Of note, the stopping by alkali was not suitable because of high and nonstable background readings of the LF BMCC mixtures in the alkaline conditions. The concentration of soluble products (expressed in Glc eq ) was calculated from the radioactivity readings in the supernatants. For this, the radioactivity readings in the supernatant were first converted into the degree of conversion of BMCC (using total radioactivity of BMCC in the sample) and the degree of conversion of BMCC was converted into the concentration of glucose equivalents (using total glucose in BMCC). In the experiments with HRP, the HRP (1.0-µM final concentration) was added to the mixture of BMCC and LF 1 min before starting the reaction by the addition of TrLPMO9A. In the control experiments without LF or TrLPMO9A, the experiments were made as described above but the LF or TrLPMO9A was replaced with corresponding amounts of buffer. In the experiments where reactions were supplied with H 2 O 2 , the H 2 O 2 (10-50-µM final concentration) was added to the mixture of BMCC and LF immediately before starting the reaction by adding TrLPMO9A.
In all cases, a sample for the zero-time point was withdrawn just before the addition of TrLPMO9A and was treated as the other samples. The reading of the zerotime point was subtracted from the readings of all time points.

Pre-incubation of LF before LPMO reaction
The 1.0-mL frozen aliquots of LF in 2.0-mL screw-cap vials were placed in a thermostat bath and incubated at 25 °C or 50 °C for durations ranging from 0.5 to 96 h. Time after time the vials were gently mixed by turning around and the caps were opened (at least once in a day) to allow equilibration with fresh air. Incubation was made in the dark without stirring. The zero time for pre-incubation is the time when the vial with frozen LF was placed (it took few min to melt the material) in the thermostat bath. A small amount of solids in the LF precipitated by gravity and these were not added to the LPMO reactions. After defined pre-incubation times, the LF was added to the LPMO reactions as detailed above.

The liquid fraction from hydrothermal pretreatment of wheat straw supports activity of a chitin-active LPMO
The kinetics of H 2 O 2 -driven degradation of chitin ( 14 C-labeled crystalline α-chitin nanowhiskers, CNWs) by SmLPMO10A has been characterized in detail before [48,50]. Provided with 0.1-mM AscA as reductant, the k cat value for oxidation of CNWs was 6.7 oxidative cleavages s −1 and the K m values for H 2 O 2 and CNWs are 2.8 µM and 0.58 g L −1 , respectively. One molecule of H 2 O 2 supports one oxidative cleavage with concomitant release of 4 soluble N-acetylglucosamine equivalents (NAG eq ) [48]. Different reducing agents like ascorbic acid, gallic acid and methylhydroquinone can support H 2 O 2 -driven oxidation of CNWs by SmLPMO10A [50]. Here, we show that the liquid fraction (LF) from hydrothermal pretreatment of wheat straw (for the composition see Table 1) can also support oxidation of CNWs by SmLPMO10A. Addition of LF to the premixed CNWs and SmLPMO10A resulted in the release of 14 C-labeled soluble products (expressed in NAG eq , Fig. 1a). There was no activity in the control experiments without SmLPMO10A or LF. In line with earlier observations [46,50], the release of NAG eq were not detected in the presence of horseradish peroxidase (HRP) indicating that the H 2 O 2 is responsible for the activity of SmLPMO10A under the conditions used. Since no external electron donor (reductant) nor H 2 O 2 was added, these results suggest that compounds present in LF support SmLPMO10A with both, electrons and the H 2 O 2 co-substrate.
Time curves of NAG eq formation were in accordance with the so-called "burst" kinetics, which is characterized by an initial transient burst of activity followed by slow and linear formation of NAG eq in time (Fig. 1a). Under our experimental conditions, the initial burst decayed within the first 10 min; whereas, the linear formation of products in time continued up to the longest time point tested (2 h). One may speculate that the initial activity burst is due to rapid consumption of H 2 O 2 already Table 1 Main components of the liquid fraction from hydrothermal pre-treatment of wheat straw a LF also contained low amounts of glucose (total 0.53 g L −1 ), arabinose (total 0.34 g L −1 ), fructose (total 0.27 g L −1 ), and galactose (total 0.26 g L −1 ) b The total amount of sugars was measured after acid hydrolysis of LF    [48] but also on the nature and concentration of the reductant [50]. The v (H 2 O 2 ) is the rate of LPMO-independent formation of H 2 O 2 in LF. Note that with Eq. 1 we assume that the rate of H 2 O 2 consumption in polysaccharide oxidation by LPMO is much higher than the rate of its formation/decomposition in LF, so that the v (H 2 O 2 ) is directly reflected in the formation of LPMO products (NAG eq ). This assumption is plausible since doubling the concentration of SmLPMO10A had no effect on the steady-state rate of the LPMO reaction (Fig. 1a).
The time curves of the release of NAG eq made at different LF concentrations were in accordance with Eq. 1 (Fig. 1a, solid lines). Nonlinear regression analysis was used to find the values of parameters and there was a linear correlation between the concentration of LF and both, n[H 2 O 2 ] (t=0) and nv (H 2 O 2 ) (Fig. 1b, c). On the other hand both, n[H 2 O 2 ] (t=0) and nv (H 2 O 2 ) were independent of the concentration of SmLPMO10A. These results are in accordance with the model whereby the kinetics of LFdriven degradation of CNWs by SmLPMO10A is governed by the H 2 O 2 initially present in LF and that formed in LF, in an LF-dependent but SmLPMO10A-independent manner.

H 2 O 2 is formed in LF but does not accumulate to high levels upon incubation of LF in aerobic conditions
Sensitive detection of H 2 O 2 above a background in a redox-active environment such as LF, with a myriad of compounds and ongoing reactions, is a challenging task. Therefore, we exploited the dependency of the kinetics of (1)  Fig. 1a were in the order of 0.6 min −1 . This translates to the half-life of [H 2 O 2 ] (t=0) in the LPMO reaction of around 1 min. After 10 min of SmLPMO10A reaction, the exponential term is close to zero (as visible in Fig. 1a) and Eq. 1 simplifies to: Provided that n and v (H 2 O 2 ) is time invariant, Eq. 2 can be analyzed using linear regression. This simplified approach is justified, as the n[H 2 O 2 ] (t=0) and nv (H 2 O 2 ) values found using full progress curves and analysis according to Eq. 1 were very similar to the values found when using Eq. 2 (Fig. 1b, c).  Fig. 2b and c, respectively. The rate of H 2 O 2 formation in LF slightly decreased with the pre-incubation time of LF and was in the order of 1.0-1.5 µM h −1 (Fig. 2c). The concentration of H 2 O 2 in LF increased upon pre-incubation of LF but seemed to level off around 3-5 µM in longer pre-incubations (Fig. 2b). Importantly, the levels of H 2 O 2 in LF were much lower than those expected based on the rate of its formation. As an example, with a rate around 10  Of note, the solid fraction from hydrothermal pretreatment of wheat straw also supported SmLPMO10A activity. An experiment with 10 g L −1 solid fraction at 25 °C, pH 5.0 (Additional file 1: Fig. S3) yielded an estimated H 2 O 2 production rate of 1.4 µM h −1 .

LF from hydrothermal pretreatment of wheat straw supports activity of a cellulose-active LPMO
Since LPMOs are important components of commercial cellulolytic cocktails, we were interested in whether the LF can support LPMOs also at 50 °C, a more relevant temperature for industrial applications. Unfortunately, the use of 50 °C was not compatible with SmLPMO10A. Therefore, we used a cellulose-active LPMO of Trichoderma reesei (TrLPMO9A, formerly TrCel61A) [17,[66][67][68][69]] and 14 C-labeled bacterial microcrystalline cellulose (BMCC), to measure the [H 2 O 2 ] (t=0) and v (H 2 O 2 ) at 50 °C. The LF indeed supported TrLPMO9A in releasing soluble products, the concentration of which was expressed in glucose equivalents (Glc eq ), from BMCC at 50 °C (Fig. 3). In all cases, the increase in [Glc eq ] was linear over time during the measurement period (from 0.5 to 2 h) (Fig. 3) and the results were analyzed using Eq. 2. Most importantly, the slopes ( nv (H 2 O 2 ) ) and intercepts (n[H 2 O 2 ] (t=0) ) were independent on the concentrations of TrLPMO9A and BMCC. On the other hand, both slope and intercept increased with increasing concentration of LF. Supplementation of the reactions with HRP (1.0 µM) totally abolished the release of soluble products and no radioactivity was released in experiments without TrLPMO9A or LF (Fig. 3). Supplementation of the TrLPMO9A/BMCC/ LF reactions with H 2 O 2 (20 µM) caused an activity burst that was reflected in an increased intercept but had no effect on the rate of further Glc eq formation (0.50 ± 0.03 versus 0.52 ± 0.05 µM Glc eq min −1 ) (Fig. 3). Collectively, these results suggest that the formation of H 2 O 2 governs the steady-state rate of soluble product formation without being dependent on TrLPMO9A or cellulose concentration, while the initial activity burst is caused by the H 2 O 2 present in the LF before the addition of LPMO.

Stoichiometry of TrLPMO9A reaction
To derive the values of [H 2 O 2 ] (t=0) and v (H 2 O 2 ) from the kinetics of Glc eq formation, one must know an average number of Glc eq released per one molecule of H 2 O 2 consumed (i.e., parameter n in Eq. 2). The n can be measured through detailed kinetic characterization of H 2 O 2 -driven degradation of polysaccharides as has been done for SmLPMO10A [48]. Unfortunately, the specific radioactivity of our BMCC preparation was not sufficiently high to permit detailed kinetic characterization of its H 2 O 2 -driven degradation by TrLPMO9A. Therefore, we estimated the value of n using alternative approaches. Comparison of the rates of NAG eq formation measured using the SmLPMO10A/CNWs system and Glc eq formation measured using the TrLPMO9A/BMCC system suggested n = 2.1 ± 0.3 for the TrLPMO9A/BMCC system (Additional file 1: Fig. S4) at 25 °C. An increase in intercept upon supplementation of TrLPMO9A/BMCC/ LF reactions with H 2 O 2 (Fig. 3) provides an alternative approach for measuring stoichiometry. Of note, the latter approach can also be used at 50 °C. Supplementation of TrLPMO9A/BMCC/LF reactions (before the experiment the LF was pre-incubated at 50 °C for 24 h) with H 2 O 2 (10-50 µM) caused an increase in the initial burst of Glc eq release with no influence on the later, linear release of Glc eq in time (Fig. 4a). Intercept values obtained from linear regression analysis of data in Fig. 4a scaled linearly with the concentration of added H 2 O 2 (Fig. 4b).
The slope of this linear dependency suggested the value of n = 1.32 ± 0.11 for the TrLPMO9A/BMCC system at 50 °C and this figure was used throughout this study. Note, that the n shall not be confused with the average degree of polymerization of soluble products since the latter depends on the probability of an oxidized group being in soluble fraction, which is 0.5 for SmLPMO10A/ CNW [48] but not known for the TrLPMO9A/BMCC system. Regarding the purposes of this study, it is important and enough to know that an average of 1.32 soluble Glc eq are released from BMCC per one molecule of H 2 O 2 consumed by TrLPMO9A.

Rate of H 2 O 2 formation and accumulation upon incubation of LF at 50 °C
To assess redox properties under typical industrial bioprocessing conditions, we pre-incubated LF at 50 °C in the dark without stirring for time-periods ranging from 0.5 to 96 h, aerobically. After pre-incubation, an aliquot of LF was added (10% or 20% v/v) to BMCC followed by the addition of TrLPMO9A to start the LPMO reaction. The formation of Glc eq was linear in time (Fig. 5a) and the rate of Glc eq formation was independent of the concentration of TrLPMO9A (Additional file 1: Fig. S5). The time curves were fitted to Eq. 2 and the values of slopes and intercepts were converted to the values of v (H 2 O 2 ) and [H 2 O 2 ] (t=0), respectively, using n = 1.32. The [H 2 O 2 ] (t=0) increased (Fig. 5b) while v (H 2 O 2 ) decreased ( Fig. 5c) with pre-incubation time of LF. Corresponding results obtained with 20% LF in the LPMO reaction are shown in Additional file 1: Fig. S6. Note that the rate of H 2 O 2 formation in LF is strongly dependent on temperature. The v (H 2 O 2 ) values derived from experiments with 0.5-h pre-incubation and 10% LF were 1.5 ± 0.2 µM h −1 (Fig. 2c) and 22.7 ± 1.2 µM h −1 (Fig. 5c) Fig. 5a as an example). Therefore, the [H 2 O 2 ] (t=0) values were also measured using an alternative approach where the LF was preincubated at 50 °C but the LPMO reaction was made at 25 °C using the SmLPMO10A/CNW system and very short reaction times (up to 10 min). In these conditions, the amount of H 2 O 2 produced during LPMO reaction is negligible compared to its initial amount and Eq. 1 simplifies to Eq. 3.
The time curves of NAG eq formation were analyzed using non-linear regression according to Eq. 3 (Additional file 1: Fig. S7) and the [H 2 O 2 ] (t=0) values were found from the plateau values of NAG eq formation using n = 4. Of note, the shape of the time curves suggested that the SmLPMO10A priming reduction efficiency of LF decreased with pre-incubation of LF and 100-µM ascorbic acid was added to ensure efficient priming (Additional file 1: Fig. S7). The [H 2 O 2 ] (t=0) found using short times of LPMO reaction and analysis according to Eq. 3 were similar to those found using longer LPMO reactions and analysis according to Eq. 2 (Fig. 5b). In both cases, the [H 2 O 2 ] (t=0) seemed to scale linearly with pre-incubation time with the slope (i.e. the rate of H 2 O 2 accumulation in LF during pre-incubation) about 0.15-µM H 2 O 2 h −1 (or 1.5-µM H 2 O 2 h −1 when extrapolated to 100% LF) (Fig. 5b). This rate of H 2 O 2 accumulation in LF is far lower than the rate of its formation in LF (Fig. 5c), suggesting that H 2 O 2 is an intermediate in LF oxidation, supporting the conclusion derived from the pre-incubation experiments at 25 °C with the SmLPMO10A/CNW system (Fig. 2). and v (H2O2) were found by fitting the data in a to Eq. 2 and using a stoichiometry of 1.32 Glc eq /H 2 O 2 . The values of [H 2 O 2 ] t=0 were also measured using an alternative approach (designated with SmLPMO10A on b) where LF was pre-incubated at 50 °C but LPMO reaction was done at 25 °C using the SmLPMO10/CNW system and short reaction times (Additional file 1: Fig S7). Solid lines in panel B show best-fits of the linear regression analysis. Error bars represent S.D. and are from three independent measurements each made at different concentration of TrLPMO9A (Additional file 1: Fig. S5)

Discussion
Since their discovery as oxidative enzymes [5], LPMOs have been a subject of intensive research. Still, the nature of the true co-substrate of LPMOs is a matter of scientific debate. The nature of the co-substrate (i.e., O 2 or H 2 O 2 ) that governs the LPMO kinetics under our experiment conditions is of utmost importance regarding the interpretation of the data presented here. Most importantly, we observed that LF-driven formation of soluble products from cellulose was independent of the concentration of TrLPMO9A (Fig. 3 and Additional file 1: Fig. S5) and the concentrations of cellulose substrate (Fig. 3) in the concentration range studied. This suggests that the ratelimiting step for the release of soluble products from cellulose is independent of the LPMO, a suggestion which is supported by the observation that the rate of chitin degradation by SmLPMO10A also was independent of the enzyme concentration.
A number of different scenarios have been proposed for O 2 -driven degradation of polysaccharides by LPMOs [27,28] but in all these cases, the rate is expected to depend on enzyme and/or polysaccharide concentration. Catalysis involving insoluble polysaccharides takes place at a solid-liquid interface and two saturation scenarios are possible, saturation of the enzyme with substrate (as in the conventional Michaelis-Menten mechanism) and saturation of substrate with enzyme (also known as the inversed Michaelis-Menten mechanism) [70]. Further increase of substrate concentration in the conditions where enzyme is already saturated with substrate will not increase the rate; however, increasing the enzyme concentration in these conditions should increase the rate. On the other hand, a further increase of the enzyme concentration in conditions where binding sites on the polymer surface are saturated with enzyme will not increase the rate; however, increase of the substrate concentration under such conditions should increase the rate. Thus, our observations cannot be ascribed to saturating conditions. Cannella et al. proposed that pigment-derived excited electrons are responsible for the boosting effect of light on the degradation of cellulose by LPMOs [71,72] Therefore, one may speculate that formation of "excited electrons" in LF is responsible for supporting LPMO activity as depicted in Fig. 6a. This scenario would be in accordance with the observed independency of the reaction rate on the concentration of LPMO and polysaccharide as the formation of such "excited electrons" could be a ratelimiting intrinsic property of the LF (note that reaction rates do depend on the amount of LF, Figs. 1, 3). Inhibition of LPMO by HRP can, in principle, be explained by the use of these electrons in the HRP reaction. However, such an "excited electron" scenario cannot explain the activity burst observed upon supplementation of an LPMO reaction with H 2 O 2 (Figs. 3, 4). Importantly, the observed increase in product formation upon addition of H 2 O 2 scaled linearly with the concentration of added H 2 O 2 (Fig. 4b). Further considering the results of the experiment with added H 2 O 2 , it is difficult to explain how a strong oxidant like H 2 O 2 can support the formation of strong reductants like "excited electrons".
On the other hand, our observations can readily be explained in the light of H 2 O 2 -driven LPMO catalysis (Fig. 6b). According to this scenario, the release of soluble LPMO products is governed by the H 2 O 2 present in LF before the addition of the LPMO (burst, c.f [H 2 O 2 ] (t=0) ) and by H 2 O 2 formed during the LPMO reaction (steadystate, c.f v (H 2 O 2 ) ). It is well known that polysaccharidefree LPMOs [46,73], including TrLPMO9A [67,74], can produce H 2 O 2 in a futile oxidase reaction with O 2 . Importantly, the contribution of this route must be insignificant under our experimental conditions, as in this case the rate is expected to increase with increasing concentration of LPMO. All in all, our results suggest that the LPMO kinetics measured here reflect the presence of H 2 O 2 and the rate of H 2 O 2 formation in LF.
Numerous reports have shown that process samples of lignocellulose refining support LPMO activity [7,30,31,[36][37][38]. The positive effect on LPMO activity has been assigned to the electron donating ability of lignin and lignin-derived, mostly phenolic compounds. Here, we show that the LPMO supporting activity of LF is related not only to electron transfer to the LPMO but also to production of the LPMO co-substrate, H 2  The rate of H 2 O 2 production in LF decreased with preincubation time of LF under aerobic conditions but did not approach zero within the time frame of the experiment ( Fig. 5c and Additional file 1: Fig. S6B). One may speculate that some compounds responsible for H 2 O 2 production were depleted during the pre-incubation of LF at 50 °C, whereas the concentration of other Fig. 6 Possible mechanisms of liquid fraction (LF)-driven degradation of polysaccharides by LPMO. a According to this mechanism, LPMO uses O 2 co-substrate and the LF (represented by a phenolic compound) drives the LPMO reaction by generating excited electrons that are used in a monooxygenase reaction. An analogous mechanism has been proposed by Cannela et al. to explain the LPMO activity-boosting effect of light in the presence of pigments [71]. The excited electrons generated in the LF are used for the initial reduction of the LPMO and subsequent catalysis via an LPMO/polysaccharide/O 2 ternary complex and including delivery of a second electron. The generation of excited electrons in LF may be stimulated (e.g., by light as shown in the scheme by hν) but stimulation is not a necessary assumption here. Note that many different mechanisms have been proposed for O 2 -driven catalysis but all these mechanisms assume the delivery of two electrons per one cleavage of glycosidic bond [27,28]. b According to this mechanism the LPMO uses H 2 O 2 as co-substrate and the LF drives LPMO reaction by generating H 2 O 2 . The slow, LPMO-independent, formation of H 2 O 2 in the reaction between O 2 and LF is rate-limiting for LPMO catalysis. This mechanism also assumes the delivery of electrons by LF but here the electrons are used only in the "priming reduction" of the LPMO (from Cu(II) to Cu(I) form). Primed LPMO can catalyze a number of oxidative cleavages of glycosidic bonds until the polysaccharide free LPMO happens to be re-oxidized, either by O 2 or H 2 O 2 [46,48,50]. Note that re-oxidation of the LPMO by H 2 O 2 may lead to irreversible inactivation. Re-oxidation of a reduced LPMO by O 2 may also generate H 2 O 2 [73]. However, in our experiments, the rates of the routes involving LPMO re-oxidation must have been insignificant compared to the rate of formation of soluble LPMO products since product formation was independent of the LPMO concentration compounds remained near-constant. A detailed analysis of the very initial phase of the LF-driven LPMO reaction (Additional file 1: Fig. S7) showed that the priming reduction of SmLPMO10A becomes less efficient upon preincubation (c.f oxidation) of LF at 50 °C. This indicates that the compounds in LF that are depleted during the pre-incubation may also be good priming reductants of LPMO. However, the precise chemical nature of the compounds in LF that are responsible for the LPMO priming reduction and H 2 O 2 production/depletion remains to be studied. Although we removed divalent metals from substrates and buffers using treatment with EDTA and Chelex ® 100 resin, metal ions possibly present in LF [77] may contribute to the rate of H 2 O 2 formation and H 2 O 2 stability in LF.
Of note, the liquid fractions of hydrothermal pretreatment can be inhibitory for glycoside hydrolases. The main components responsible for the inhibition are hemicelluloses-derived oligosaccharides [4,42,78,79], but the inhibition by phenolic compounds has also been demonstrated [4,42,43,45,79,80]. Therefore, the overall effect of liquid fraction on lignocellulose degradation depends on the relative contributions of its inhibiting and activating effects, which in turn depend on the composition of the enzyme cocktail and the type of pretreatment. The closest alternative to hydrothermal pretreatment is dilute acid treatment, and generally similar effects could be expected from the corresponding liquid fractions. Pretreatments that lead to extensive delignification may not support LPMO activity since the soluble lignin-containing liquor is usually removed before enzymatic hydrolysis. Based on the present findings, further studies of the relationships between the method of pretreatment and LPMO activity in subsequent enzymatic processing are of interest.
The rate of H 2 O 2 formation in LF was strongly dependent on temperature with an estimated Q 10 around 3.0. The Q 10 value found here seems to be in accordance with a recently reported effect of temperature on the half-life of O 2 in a slurry of hydrothermally pre-treated wheat straw [81]. From the data in Peciulyte et al., one can estimate a half-life of O 2 of about 1 h at 50 °C [81]. This translates to a rate constant of O 2 consumption of 0.69 h −1 , which in turn translates to a rate of O 2 consumption of 140 µM h −1 (assuming an ambient O 2 concentration of 0.2 mM). This value is in the same range as the rates of H 2 O 2 formation in LF at 50 °C shown in Fig. 5c (after extrapolation to 100% LF). Thus, the data on H 2 O 2 formation in LF from the hydrothermal pre-treatment of wheat straw are in general accordance with the data of O 2 consumption by the whole slurry of hydrothermally pre-treated wheat straw. Interestingly, Peciulyte et al. showed that the abiotic consumption of O 2 in slurry was much faster than the diffusion of O 2 into the slurry, suggesting that the availability of O 2 may be rate-limiting for the oxidation of biomass components [81]. Apparently, the availability of O 2 along with the amounts and chemical nature of the components in biomass slurries are key determinants of the rate of H 2 O 2 formation and, thus, of LPMO activity and stability in biomass processing.

Conclusions
In this study, we have demonstrated that soluble compounds in the liquid fraction (LF) from hydrothermal pre-treatment of wheat straw support LPMOs with both, electrons and the H 2 O 2 co-substrate under conditions that are commonly used in enzymatic biomass processing. Both, a bacterial chitin-active and a fungal celluloseactive LPMO were supported by LF and the H 2 O 2 was produced in the LF in an LPMO-independent manner. Our results point to that H 2 O 2 is an intermediate and not an end product in LF oxidation. This is important, since the further reduction of H 2 O 2 prevents its accumulation, thus diminishing the probability of enzyme inactivation. The most probable candidates responsible for H 2 O 2 production are phenolic compounds in LF but their exact chemical nature remains to be studied. Further studies shall also reveal the relationships between the LPMO supporting efficiency of liquid streams and the method of pre-treatment and type of biomass. The results presented here may also provide a basis for development of LPMObased methods for sensing H 2 O 2 in complex redox-active environments.
The effect of LF on LPMO activity and on H 2 O 2 levels may have a major effect on biomass conversion reactions, as illustrated by the studies of Müller et al. [76] on the enzymatic conversion of various types of biomasses with an LPMO-containing cellulase cocktail. For example, using H 2 O 2 feeding under anaerobic conditions, Müller et al. found that the incorporation of fed H 2 O 2 into LPMO products was almost stoichiometric when degrading Avicel, i.e., a relative clean substrate, whereas this stoichiometric relationship was not observed when using lignin-containing substrates [76]. Of note, these studies also led to the suggestion that current commercial cellulase cocktails may contain more LPMOs than needed to convert available H 2 O 2 [76,82], which is in accordance with the lack of enzyme concentration dependency found in the present study. Clearly, the interplay between biomass pretreatment and the efficiency of the subsequent enzymatic conversion process needs to be revisited in light of recent findings on LPMOs. It is important to note that phenolics and other soluble compounds in the LF, such as hemicellulose fragments, may inhibit the glycoside