Introduction

Industrial bioprocesses often operate at high dry matter (HDM) conditions to improve environmental performance, energy efficiency, and volumetric productivity. In lignocellulosic biorefining, it is essential to reduce the minimum ethanol selling price MESP (Macrelli et al. 2014; Janssen et al. 2014; Xiros et al. 2017) and the costs associated with product recovery. It is hard to define a threshold value between low and high dry matter conditions, as it depends upon a combination of feedstock, pretreatment technology, enzymes, and fermentation strategies. A generic definition for HDM conditions is the absence of free-floating water, (i.e. all water present is soaked into the biomass), although this is far from being generally accepted (Xiros et al. 2017; da Silva et al. 2020). Arbitrarily, a value around 15–20% w/v has typically been associated with HDM regardless of the biomass and pretreatment type. However, when processing lignocellulosic biomass at HDM, a low substrate-to-product conversion has frequently been reported. Handling biomass at HDM requires new technological solutions: in the last decade, the first generation of industrial pilot scale or demo plants for HDM hydrolysis was released, relying often on a new generation of commercial enzymatic preparations able to function at up to 20% dry matter content (Jørgensen et al. 2007a, b; Cannella and Jørgensen 2014; Xu et al. 2019; da Silva et al. 2020).

Nevertheless, beyond this limit, the HDM condition continued to represent an obstacle, and its adverse impact had been coined into the negative definition of “high-solids effect” (Jørgensen et al. 2007a, b). The causes of this were thought to be essentially a mass transfer phenomenon until studies at the molecular level pointed towards more complex interactions between cellulose, water, and solutes (Felby et al. 2008; Kristensen et al. 2009). The topic was recently reviewed by da Silva et al. 2020, highlighting how several phenomena could all participate in generating the negative effect at HDM conditions: enzyme precipitation, unproductive binding, substrate diffusion, end-products inhibition, shear and thermal stresses, and water constraint. Among these phenomena, water availability is a key parameter, but its constraint at HDM conditions has received limited attention, and often with conflicting conclusions in the past.

During saccharification of cellulose, water molecules exist in dynamic pools representing the different interactions entertained with solvent (water itself, called free- or bulk-water), insoluble cellulose (surface-bound water), and cellulose cavities or inter-fiber lumens, an intermediate pool of water molecules often defined as lumen-water. At HDM conditions, free water is not apparent since water molecules are constrained into surface-bound and lumen-water pools. During the pretreatment of biomass and/or enzymatic hydrolysis, the cellulose ultrastructure and its surface are modified enlarging lumen cavities (Rahikainen et al. 2020), opening inter-fiber voids, and releasing quantities of water-soluble saccharides. The water molecules, therefore, migrate from the cellulose-constraining environment to the solvation spheres of water soluble saccharides. These changes contribute to the liquefaction of the lignocellulose, creating visible free-floating water and when studied using two-dimensional low field nuclear magnetic resonance relaxometry (2D-LFNMR), a “free water” peak become evident. It has been disputed whether water constraint has a positive or negative effect on enzymatic activity, and early conclusions indicated that the water on the cellulose surface was trapped in a sort of unproductive state (Felby et al. 2008; Roberts et al. 2011). However, as the concentration of soluble molecules increases (e.g. glucose), water molecules migrate away from the cellulose surface to coordinate the solvation of the soluble molecules, thereby reducing water availability for the cellulases and consequently retarding their kinetics (Roberts et al 2011; Hsieh et al. 2014; Liu and Chen 2016). Possibly, a certain degree of water constraint on the cellulose surface could be beneficial to counteract this putative loss of available water. Results indicate that biomasses able to retain most of the water thanks to physio-chemical pretreatment, gave higher enzymatic hydrolysis yields (Selig et al. 2012; Liu and Chen 2016; Weiss et al. 2019).

Concerning the negative effects of HDM content, a consensus was achieved regarding the inhibition of cellulases. Yet, modern commercial cellulase preparations contain various additional enzyme activities, among them oxidoreductase lytic polysaccharides monooxygenases (LPMO). These enzymes can oxidatively cleave cellulose chains, thus resulting in a synergetic effect when combined with traditional cellulases by creating new entry sites. Consequently, they are classified as auxiliary activity (AA) in the CAZymes database (Drula et al. 2021). Since the seminal discovery of LPMOs in 2010 by Vaaje-Kolstad et al. 2010, the community has attempted to elucidate their reaction mechanism(s). Recently, comprehensive reviews were published covering both mechanisms and functional characterization of LPMOs (Eijsink et al. 2019), and pointing to new possible applications, i.e., nanocellulose isolation and functionalization (Arantes et al. 2020; Magri et al. 2022).

In this paper, we focus on the possible products and structural consequences of the cellulose oxidation that can occur at the cellulose surface. LPMOs can oxidize the cellulose chains either at the C1 position to produce aldonic acids or at the C4 position to produce 4-ketoaldose (gemdiols). Either way, the oxidation causes a chain rupture with new chain termini or entry sites being generated. Thus, even though the mechanism is still debated (monooxygenase or peroxygenases), the end result for the cellulose surface is functionally the same: a net negative charge is created, either as the C1-oxidized aldonic product or a lone pair for the C4-oxidized keto-aldose (Beeson et al. 2015).

Studies on fiber structural integrity upon LPMO oxidation (Villares et al. 2017) and mathematical modeling (Vermaas et al. 2015; Trentin and Skaf 2019; Mudedla et al. 2021) have speculated that water could be strongly retained or bound in the proximity of the cellulose oxidation sites. The otherwise structured layer of water could also be disrupted around the structural “defects” introduced by polysaccharide chain breaking, affecting in some extent the binding of water. Therefore, in this work, we aimed at deciphering experimentally if LPMO enzymes could promote water constraint on cellulose surfaces, thereby partially counteracting the negative high-solids effect during enzymatic hydrolysis and thus resulting in synergy when employed together with cellulases. We employed two LPMO enzymes: a C1/C4-oxidizing variant from Thielavia terrestris and a C1-oxidizing variant from Thermothelomyces (Myceliophthora) thermophilus (TtAA9E and MtAA9A, respectively). For comparison, we also employed (2,2,6,6-Tetramethylpiperidin-1-yl)oxyl TEMPO-mediated oxidation (C6-oxidizing), to produce surfaces of oxidized cellulose without rupturing the polysaccharide chains (Fig. 1). These oxidized cellulose materials were then subjected to 2D LFNMR to investigate if water-cellulose interaction had been affected.

Fig. 1
figure 1

scheme of cellulose oxidation and hydrolysis. LPMO and GH enzymes cause chain’s rupture, while TEMPO oxidation preserve the polysaccharide chain

Full size image

Materials and experimental methods

Enzymes

Thielavia terrestris LPMO (TtAA9E) synthetic gene was obtained commercially (Eurofins) and cloned according to Kadowaki et al. 2021, having Pichia pastoris as the expression host. The coding sequence of MtAA9A from Thermothelomyces thermophilus M77 (MYCTH_85556, UniProt: KP901251, Sepulchro et al. 2021) was amplified from genomic DNA, including the native signal peptide using 5’-agcatcattacacctcagcaATGCTGACAACAACCTTCGC-3’(forward) and 5’-taaatcactagatatctctaTTAGCAACGGAAGACAGCCG-3’(reverse) primers, and cloned into the pEXPYR vector using ligation-independent assembly (Gibson et al. 2009). The expression plasmid was transformed into Aspergillus nidulans A773 (pyrG89; wA3; pyroA4) accordingly to Velasco et al. (2019) and the recombinant strain was selected using the pyrG auxotrophic maker as described earlier (Sepulchro et al. 2021). The purification followed the same steps for both enzymes using fast protein liquid chromatography-FPLC (Äkta purifier, Cytiva) equipped with a column for His-tag purification (HisTrap-HP 5 ml column, Cytiva); details on the elution protocol and further purification with Sepharose-gel column for protein purification had been previously reported from our group (Kadowaki et al. 2021). After copper saturation, a second purification step was performed to remove excess copper using centrifuge filtering vials (10KDa cut-off by Amicon, Millipore). The resulting MtAA9A and TtAA9E enzymes revealed no contamination by co-eluting proteins or peptides, nor copper (Kadowaki et al. 2021). The commercial cellulase mixtures Celluclast 1.5L, Novozym 188, and Cellic CTec2 (CT2) were obtained from Novozymes A/S, Denmark. CT2 had total protein content of 186 mg/g and 120 FPU/g cellulase activity.

Chemicals, materials and substrates

Avicel microcrystalline cellulose (Avicel), Whatman filter paper no. 1 (FP), and the oxidizing agent (2,2,6,6-tetramethylpiperidin-1-yl)oxyl also known as TEMPO were obtained from Sigma Aldrich, Saint Louis, USA.

Preparation of PASC Microcrystalline cellulose substrate

Avicel was swollen with phosphoric acid to generate phosphoric acid swollen cellulose (PASC) as described by Wood 1988 with a few modifications: 4 g of Avicel were suspended in 100 mL of phosphoric acid (85% w/v) at 40 °C and magnetically stirred for 1 h. The mixture was poured into 1.9 L of water at 40 °C and stirred for 1 h. The suspension was washed four times with 2 L H2O (MilliQ-quality), two times with 2 L of a 1% NaHCO3 solution to reduce acidity, and then three additional times with 2 L H2O (MilliQ-quality) and stored at 4 °C. These steps were executed in a 2 L cylinder to help decant the pulp solution via sedimentation. The final cellulose content of the PASC suspension, 1.5% w/v, was determined by enzymatic hydrolysis (24 h, 50 °C), with 75 FPU/g of Celluclast 1.5 L cellulolytic enzymes and Novozym 188 in a 5:1 ratio. The released glucose was measured with high-pressure liquid chromatography equipped with a refractive index detector (HPLC-RI), the details are described below in the dedicated section for chromatography. The average degree of polymerization (DP) was determined by measuring the total number of reducing ends and comparing this to the total amount of glucose, resulting in a DP of 52 as previously reported (Horn and Eijsink 2004; Westereng et al. 2015).

Preparation of TEMPO-oxidized cellulose fibers (TEMPO-FP)

Filter paper (FP) with varying degrees of oxidation was prepared by modifying a previously reported protocol by Tahiri and Vignon (2000). Briefly, 0.648 g of milled cellulose was dispersed in 80 mL of distilled water using magnetic stirring. Then 10 mg of (2,2,6,6-tetramethylpiperidin-1-yl)oxyl (TEMPO) (0.065 mM) and 0.20 g (1.9 mM) of NaBr were added to the samples until the reagents dissolved completely at room temperature. To reach different degrees of oxidation, 122 µl, 81.3 µl, or 61 µl of 8.8 mM sodium hypochlorite solution was added to each sample, respectively. Simultaneously, the pH of each sample was maintained at around 11.0 by dosing 0.5 M NaOH. When the pH became stable, the reaction was stopped by adding 10 mL of methanol, and the mixture was adjusted to pH 7.0 with 0.5 M HCl. Subsequently, the samples were washed with 800 mL of 95% ethanol and filtrated using a vacuum-filtration unit. Oxidized cellulose filtrates were dried overnight at 40○C and kept dry until further use. The final TEMPO-FP degree of oxidation of 0.5%, 1%, and 2.6% was quantified by determining amount of glucose oxidized into glucuronic acid (using the protocol as described in Cannella and Jørgensen 2014) relative to total carbohydrate mass (glucose monomers). High-performance anion-exchange chromatography equipped with a pulsed amperometry detector (HPAEC-PAD) was used for the quantification of glucuronic acid, while the glucose was quantified with HPLC-RI (the details of the chromatographic techniques are described below).

Preparation of LPMO-oxidized cellulose fibers (LPMO-FP)

Filter paper cellulose (FP) was oxidized with MtAA9A to produce batches of oxidized cellulose (LPMO-FP) for further LF-NMR analysis on water interaction and for calculating the water retention value (WRV). 1 g of milled Whatman no. 1, was incubated for 72 h at 10% w/v with 1 µM MtAA9A and 1 mM ascorbic acid in 20 mM citrate–phosphate buffer pH 6.2 in a total volume of 10 mL. Test tubes of 50 mL were used for the reaction and vigorously agitated with a Stuart SF1 flask shaker placed inside an incubator oven set at 50 °C.

Enzymatic hydrolysis of cellulose materials

Incubation of AA9 and cellulose at high dry matter and glucose at high concentration

A fixed amount of MtAA9A LPMO (1 µM) or TtAA9E (1 µM), and ascorbic acid (1 mM) were incubated at increasing concentrations of substrate (Avicel) 1%, 7.5%, and 15% w/v in a total reaction volume of 300 µl. The reaction mixtures contained 20 mM citrate–phosphate buffer pH 6.2 and were constantly mixed at 800 rpm and 50 °C for 24 h. The same conditions were also applied to have 1% of Avicel to test the effect of high molarity of soluble glucose on LPMO activity. In this case, glucose was added to give a resulting glucose concentration of 25, 50, and 75 g/L. After incubation for 24 h, samples were spun at 14,000 g and the oligosaccharides were analyzed by HPAEC-PAD.

Avicel hydrolysis with commercial cellulase cocktail CT2

Avicel cellulose was hydrolyzed with CT2 at a protein dosage of 12 mg/g cellulose or nearly 10 FPU/g cellulose. The hydrolysis was carried out in 2 mL tubes, with a total reaction volume of 1 mL and a substrate loading of 1%, 5%, 10%, 20%, and 30% of dry matter, containing 50 mM of citrate buffer at pH 5.0, and with or without the addition of 1 mM of ascorbic acid. The reaction was kept constantly mixed at 800 rpm and heated at 50 °C. The sampling was carried out accordingly with the methodology for high dry matter conditions described earlier (Kristensen et al. 2009) to not overestimate the yield caused by an increased concentration of the hydrolysate. 50 mg of the sample was weighed and diluted with 450 mg of 200 mM HCl to stop the enzymatic reaction. This dilution factor was shown to prevent any overestimation when applied to samples containing DM higher than 5%, while the 1% DM samples were analyzed undiluted. The samples were analyzed by HPLC equipped with refractive index detector (hence HPLC-RI) for glucose and cellobiose, and with HPAEC-PAD for the quantification of gluconic and cellobionic acid.

Water retention value – WRV measurements

The methods described by (Cheng et al. 2010; Weiss et al. 2018) were followed with a slight modification on the filtering membrane. Instead of using a cotton filter between the starting and the receiving vial, ultrafiltration vials with a filtering membrane made of cellulose and having a 3 kDa molecular mass cut-off were used (Amicon Ultra 0.5 ml, Merck). Superior reproducibility among replicates was obtained compared to the previous methods. 50 mg of sample oven-dried at 40 °C was weighed into filtering vials, and 450 µl of MilliQ-water was added and allowed to incubate overnight. The water was then removed by in a refrigerated centrifuge (4° C, Eppendorf 5810r, rotor F45-30–11) run at two different centrifugal forces and periods: 70 min at 1.800 g or 8 min at 10.000 g. Intermediate (every 10 min) and final weights were recorded (WWet) and finally the material was oven dried again at 40 °C and weighed (WDry). These values were then used for the calculation of WRV using the following formula: WRV = ((WWet– WDry)/( WWet)) × 100%. The samples used for this experiment were the following: LPMO-FP, and TEMPO-FP at 2.6% of surface oxidation (see respective sections above for details). The control experiment named FP contained Whatman no. 1 filter paper incubated for 72 h with ascorbate and buffer, i.e., identical to the LPMO-FP treatment except for the presence of the enzyme. All the materials were washed and kept dry at 40 °C until further use.

Chromatography and spectroscopy

Measurement of glucose and cellobiose by high-performance liquid chromatography (HPLC-RI)

Quantification of D-glucose and D-cellobiose was carried out using an Ultimate 3000 HPLC (Dionex, Germering, Germany) equipped with a refractive index (RI) detector (Shodex, Japan). The separation was performed in a Phenomenex Rezex ROA column (300 mm × 7.8 mm) heated at 80 °C with water/5 mM H2SO4 as eluent at a flow rate of 0.8 mL min−1.

Measurement of polysaccharide oligomers, glucuronic and gluconic acid by high-performance anion-exchange chromatography (HPAEC-PAD)

HPAEC was performed on an ICS-6000 system (Dionex, Sunnyvale, CA, USA), equipped with a pulsed amperometry detection (PAD) and with a CarboPac PA1 column (2 × 50 mm guard column followed by a 2 × 250 mm analytical column), operated at a flow of 0.25 mL min−1, at 30 °C. The gradient elution profile for the analysis of C1-oxidized oligomers, glucuronic and gluconic acid was conducted as described by Westereng et al. 2013. The peaks of the cellulose oligomers (native and C1-oxidized with degree of polymerization – DP up to 4) were assigned and quantified using standards from Megazymes ltd. Aldonic acid oligosaccharides were obtained by enzymatic oxidation of cellodextrines using cellobiose dehydrogenase (CDH). Oligosaccharides longer than DP5 were assigned according to Westereng 2013. All the chromatograms shown are averages of three independent experiments.

Low-field NMR: water-cellulose interaction

Measurements were made using a Bruker mq20 minispec with a 0.47 Tesla permanent magnet (equivalent to 20 MHz proton resonance frequency). The temperature inside the magnet was 40 °C. The samples were measured at several dry matter content ranging from 10 to 50%. The Carr-Purcell-Meiboom-Gill (CPMG) pulse sequence was used to determine the T2 relaxation times. Each sample was run with 32 scans and a 5-s recycle delay containing 5000 points and a pulse separation of 0.15 ms between each scan. Furthermore, the CPMG relaxation curves were analyzed using the inverse Laplace transformation method CONTIN. 2D T1/T2 relaxation times were determined using the protocol previously described by Jeoh et al. 2017. The sample types analyzed were LPMO-FP and TEMPO-FP oxidized filter paper (see above for details on the preparation). To be able to isolate the effect of the type of oxidation, the samples were analyzed against their respective control material lacking only the mediator agent (LPMO or TEMPO respectively) yet containing all the other components (ascorbate and buffer).

Results and discussions

AA9 activity at high dry matter

The high-solids effect, i.e. a decrease of enzyme activity at increasing solids concentration affecting cellulase preparations in general (Kristensen et al. 2009), was not observed for MtAA9A. Instead, a positive effect at increased substrate concentration or ratio vs the amount of AA9 was observed, indicating that an even higher concentration of substrate is needed to saturate this enzyme (Fig. 2A). Most of the cello-oligosaccharides (COS) were quantified, and results demonstrated that the distribution of the native COS relative to the oxidized COS remained unchanged irrespective of dry matter concentration (Fig. 2B). This indicate that the regioselectivity and substrate preference of the enzyme was not affected by the increasing substrate-water molar ratio, in agreement with Magri et al. 2022. Noteworthy is the increased amount of COS relative to the increment of the initial dry matter loading (from 1 to 15% dry matter) when keeping the dosage of AA9 fixed at 1 µM (Fig. 2C). To quantify the overall productivity of MtAA9A, an indirect method was employed. After the HPAEC-PAD analysis of the released COS, the samples (supernatant) were hydrolyzed overnight with beta-glucosidases enzyme (2 µM), and the major products of the hydrolysis, glucose, gluconic acid, and cellobionic acid were quantified again by HPAEC-PAD (Fig. 2C). At a fixed enzyme loading of 1 µM of MtAA9A, the productivity was calculated to be 1.4, 7.6 and 12.7 mg/L/h, respectively for 1%, 7.5% and 15% DM. The final concentration of COS ranged from 34 to 305 mg/L.

Fig. 2
figure 2

MtAA9A activity is increasing at high dry matter. In panel a HPAEC chromatograms of LPMO activity (24 h) on Avicel at 1% (light-blue), 7.5% (blue) and 15% (red) dry matter, 1 µMol of MtAA9A, 1 mM of ascorbic acid. A negative control lacking enzymes but having ascorbate is depicted as a grey line. Panel b reports the distribution of released oligosaccharides with DP2 to DP6 and C1-oxidized forms (blue-shaded samples represent native COS, while reds are oxidized COS, for both pools the darkest color represents the longest COS), against the sum of their signal intensity. c Quantification of COS of the same sample in previous panels

Full size image

The same experiment was also performed using another LPMO belonging to the same AA9 family, TtAA9E. The chromatograms obtained at 1, 7.5 and 15% DM revealed rather similar oligosaccharide distribution and concentration, suggesting a very early substrate saturation point at 1% DM when compared to MtAA9A (Supplementary data Fig. 1). The absolute amount of COS produced remained in a closer range (from 200 to 230 mg/L), revealing little increase when increasing the Avicel loading further. Instead, if considering the initial substrate loading, the overall percentage of cellulose released as COS dropped from 1.82 to 0.25% and 0.13% respectively for 1%, 7.5% and 15% DM loading of Avicel.

Despite a different substrate saturation and yield of substrate oxidation at comparable conditions, both LPMO enzymes were not subject to the negative high-solids effect, MtAA9A was even stimulated by higher DM contents. These data gain relevance if compared to cellulases, which are often severely inhibited (30–50% decrease of activity) already at DM close to 5% or 10% DM using molar dosages enzyme:substrate (Várnai et al. 2013). The apparent positive effect of HDM content on the activity of MtAA9A might have to do with the so-called futile cycle of the LPMOs, where the enzyme generates hydrogen peroxide when it is not productively bound to the carbohydrate substrate. The resulting accumulation of hydrogen peroxide can led to oxidative inactivation of the enzyme. One can therefore speculate that the low activity or low substrate saturation of the MtAA9A observed at 1% of DM could be the result of enzyme autooxidation. Along this line of thought, at HDM 15%, more MtAA9A would be productively bound to the substrate, and thus no or very limited amount of hydrogen peroxide is produced.

AA9 activity at high solute loadings

Water constraint causes inhibition of the activity of cellulolytic enzymes, and not only in the case of insoluble substrates. As previously demonstrated, high amounts of soluble saccharides could also constrain water and result in an inhibitory effect on cellulases (Selig et al. 2012; Hsieh et al. 2014). To test if AA9 suffered from this type of inhibition, MtAA9A was incubated with a fixed amount of Avicel cellulose and increasing concentrations of glucose. The results revealed that the enzyme was still active at glucose concentrations up to 75 g/kg (Fig. 3), in contrast to what has been reported for hydrolytic cellulases (Hsieh et al. 2014). Similar to the results reported above for increasing DM, the amount of COS released from Avicel by MtAA9A increased 46,4% (by total peak area) when comparing samples with 75 g/kg of glucose to the control experiment without glucose (Fig. 3b). Moreover, this is more pronounced for native COS (particularly cellotetraose and cellopentaose) than oxidized COS (whose amount remained unaltered at increasing glucose concentration). The differences might be due to the high amount of glucose molecules surrounding the active site of the enzyme and interfering with the catalytic cleavage or even the release of COS. Cellulases, particularly CBHs, are severely inhibited by the soluble end-products. Concentrations below 10 g/kg of cellobiose can lead to an 80% inhibition of TrCBH, whereas TrEGs could withstand concentrations up to 100 g/kg before reaching the same level of inhibition (Teugjas and Väljamäe 2013). The latter glucose concentration was considered prohibitive for earlier cellulolytic preparations, whereas it has been reported that modern preparations can stand very high concentrations of glucose, up to 150 g/kg, although severely inhibited (Cannella and Jørgensen 2014). Considering the enzymatic preparations, while cellulases would stop hydrolyzing cellulose at around 150 g/kg of glucose, LPMO enzymes could still be actively contributing to further oxidizing the remaining cellulose.

Fig. 3
figure 3

MtAA9A activity at high glucose concentration. Panel a HPAEC chromatograms of LPMO activity (16 h) on Avicel at 5% DM, 1 µMol of MtAA9A, 1 mM of ascorbic acid, and an increasing amount of glucose added, 25 g/l (green) 50 g/l (blue) and 75 g/l (red), control experiment (ctrl) is without glucose addition and is shown in black. Panel b relative intensities for every single pool of native or C-1 oxidized oligosaccharides: blue-shaded samples represent native COS and red-shaded samples are oxidized COS; for both pools, the darkest color represents the longest COS

Full size image

Contribution of LPMO to cellulose hydrolysis at high dry matter and high solutes concentration

The activity of LPMOs was also monitored while operating in concert with the complete hydrolytic machinery of the cellulolytic cocktail Cellic CTec2 (CT2) during hydrolysis of cellulose at several DM conditions. The results of the hydrolysis in the presence or not of ascorbic acid are shown in Fig. 4a and Supplementary Fig. 2, respectively. The production of gluconic and cellobionic acid was used as a marker of the C1-oxidizing activity of the LPMO present. The relative contribution of LPMO to overall cellulose conversion as shown in Fig. 4b is reported as the ratio between glucose oxidized (sum of gluconic and cellobionic acid) over the total amount of glucose released from cellulose quantified at each time point, consistent with earlier work (Cannella and Jørgensen 2014). The results show that at a high dry matter, the relative contribution of LPMO to the overall enzymatic hydrolysis yield was greater than at low DM content. Despite achieving the highest conversion of cellulose to glucose at 5% of DM, only 0.14% of the hydrolyzed cellulose was oxidized by LPMO at every time point tested. Analyzing the data relative to 30% DM conditions, the ratio of the LPMO’s activity over the activity of cellulases kept increasing throughout the time course of the hydrolysis: 0.32% of all released products were oxidized after 24 h, whereas this value increased to 1.64% after 96 h. The GHs instead were almost completely inhibited already at 48 h, as the overall hydrolysis yield (measured as the amount of glucose released relative to the possible theoretical amount) only increased by 3.6% at 96 h compared to the previous time point (48 h). Meanwhile, the gluconic acid/glucose ratio increased by nearly 30% (from 1.18 to 1.64 at 48 h and 96 h respectively) due mostly to the inhibition of the hydrolytic enzymes at glucose concentrations above 100 g/kg.

Fig. 4
figure 4

Avicel hydrolysis and LPMO contribution at HDM content. The enzymatic hydrolysis of Avicel was conducted with the cellulase cocktail CT2 dosed at 12 mg protein/g of cellulose together with 1 mM ascorbic acid for 96 h. Panel a final hydrolysis yield of cellulose calculated as the sum of glucose, gluconic and cellobionic acids released (y-axis). On the x-axis are reported the specific time points (shades of blue light 24 h, medium 48 h, darker 96 h) and the dry matter content. Panel b contribution of LPMO activity to cellulose degradation at each time point and dry matter content. Data are reported as the ratio between the sum of quantified amounts of gluconic and cellobionic acid over the amount of glucose released during the hydrolysis of Avicel. On the x-axis are reported the specific time points (shades of blue light 24 h, medium 48 h, darker 96 h) and the dry matter content. The experiments were carried out in triplicates, bars represent the mean value, while the error bars represent the highest and lowest values

Full size image

To summarize, when hydrolyzing cellulose at HDM with commercial cellulase preparations, the concentration of soluble sugars reaches an inhibitory level for the cellulases within a few hours. In addition, insoluble substrate gathers a conspicuous amount of water molecules all around the exposed surface (Matthews et al. 2006), likewise de novo solubilized glucose demand for high amounts of water molecules in their solvation spheres (Suzuki 2008). The water molecules are then constrained and less available for the functioning of cellulases, which are thus inhibited. This effect adds to the already mentioned end-product inhibition. This seemed not to be the case for AA9 enzymes, or at least, they are not inhibited to the same extent.

Surface-oxidized cellulose shows increased water retention

Avicel subjected to enzymatic oxidation with pure LPMOs at HDM (15% w/w) appeared to have an increased pellet volume compared to corresponding samples without the enzymes, indicating increased water retention (also noticed by Mudedla et al 2021 on processed woody cellulose). Although the swelling of cellulose upon defibrillation and/or loss of structural integrity is a known effect for carbohydrate-active enzymes (particularly endoglucanases, Berto et al. 2019, 2021), it is not known if LPMO-driven oxidation of cellulose surface alone can further improve swelling or increasing water/cellulose binding. This prompted us to assess if the cellulose surface oxidation of fibers could indeed be responsible for increased water retention.

In Fig. 5 are reported WRVs for Whatman No. 1 filter paper (FP) (native without oxidation), compared to its oxidized versions: either enzymatically oxidized using MtAA9A (LPMO-FP), or chemically using TEMPO (having a 2.6% of total oxidation, TEMPO-FP). The method is based on quantifying the amount of water retained by the fibers after centrifugation of the sample on a spin filter (Cheng et al. 2010; Weiss et al. 2018). Two different centrifugation regimes were applied, and both showed that considerably more water was retained by the oxidized fibers compared to the native ones, i.e. 30% more for LPMO-FP and 47% for TEMPO-FP. The oxidation introduced by LPMOs or TEMPO although differing in the carbon position might cause structural reassessments of the rod-shaped fibers, thus likely influencing the water:cellulose interaction. Similar conclusions were indicated by mathematical modeling studies when comparing C1-oxidized cellulose surfaces against pristine cellulose (Mudedla et al. 2021).

Fig. 5
figure 5

Water retention value (WRV) of filter paper fibers treated with MtAA9A and TEMPO. In panel a are WRVs at different time intervals (shades of blue, darker color for longer times) at 1800 g centrifugal force comparing filter paper (FP) treated with TEMPO at 2.6% of oxidation (TEMPO-FP), MtAA9A (LPMO-FP) and not treated FP (Control). In panel b are reported WRVs at two different centrifugal forces and time intervals (1800 g 70 min and 10,000 g 8 min) showing how the two methods yield statistically similar results

Full size image

An increased water absorption could also be attributed to the physical constraints of water caused by the formation of new chain termini as shown by previous mathematical modeling (Vermaas et al. 2015). The irregular cellulose surface resulting after the catalytic cut by LPMOs could create a positive steric hindrance for water molecule absorption. To unravel if the improved water-cellulose absorption was caused solely by an ionic interaction or by a physical constraint we used 2D LF-NMR (Fig. 6).

Fig. 6
figure 6

T1/T2 relaxation time calculation of native, TEMPO, and LPMO-oxidized cellulose. The sketch in panel a (adopted from Jeoh et al. 2017) visualizes the relative migration of the water pools from free to constraint. Panel b shows the constraint water pool relative to TEMPO-FP and in panel c LPMO-FP. In grayscale is reported the control material prior the oxidation. The intensity of the relative relaxation time is reported in the color scale from blue to red

Full size image

In contrast to one-dimensional T2 relaxation (supplementary data 3 and 4), two-dimensional T1/T2 LF-NMR has the potential to identify states of water that have the same T2, but different T1 values and vice versa. Tentatively, changes in T2 are more related to chemical constraints, while physical constraints tend to affect T1 relaxation more. In other terms, T1/T2 relaxation time could be used to indicate if the water constraint observed by the water retention values of Fig. 5 was due to the net negative charge of oxidation, or if it was due to a surface modification (oligosaccharide chains protruding from the surface). The 2D-LFNMR analysis in Fig. 6 was carried out on TEMPO-oxidized, and LPMO-oxidized FP (same samples as for the WRV calculation) and depicted in panels 6b and 6c, respectively. For TEMPO-FP an increased ability to constrain water was seen for the oxidized samples, as their T2 was reduced compared to native samples under-laid in grayscale. This was most likely caused by increased water interaction with oxidation sites on the FP surface, as shown by the shorter T2 relaxation times with the presence of C6-oxidation by TEMPO. Regarding LPMO-FP minimal differences were detected although a similar trend was observed regarding the T2: a net shorter relaxation time compared to an almost unaltered T1 relaxation time (blue halo stretching on shorter T2 values, compared to untreated material). Moreover, the relative amount of water in the more constrained environments increased with oxidation. This suggested that for both types of oxidation, water constraint was primarily related to changes in chemistry, i.e. due to carbon oxidation at C6 for TEMPO oxidation and at C1 for LPMO-oxidation. This is schematically illustrated in Fig. 7. However, this interpretation of the 2D-LFNMR data does not rule out the possibility that water is more constraint by physical interaction since it was predicted that water molecules could penetrate below the nascent oligosaccharide rising from C1-oxidation caused by LPMO (Uchiyama et al. 2022).

Fig. 7
figure 7

Model of cellulose water interaction. The model is a simplified representation of the water-cellulose interactions as depicted by Arola et al. (2022). The cellulose surface is shown in 3 chemical variants: native, enzymatically oxidized with LPMO, and chemically oxidized with TEMPO. The oxidations affect the surface of cellulose chemically and structurally: the new cellulose residues formed are colored in red, while the water molecules are depicted in blue for free water, orange represents the surface-bound water, and red represents the increasing tightness of contact with the net-negative charge molecules as a first layer of solvation

Full size image

Previous works (Selig et al. 2014; Weiss et al. 2018 and 2019) showed increased water constraint because of biomass pretreatment and this was correlated to improved enzymatic digestibility of the biomass (although not the only cause since dealing with a complex biomaterial). In the case of native cellulose fibers, water constraint is caused by the massive grid of hydroxyl groups present at the surface of cellulose and is liberated into the solvent during hydrolysis. When comparing high versus low dry matter conditions, the major difference can be attributed to the nominal amount of free water available to solubilize the nascent oligosaccharides and monomers liberated during enzymatic hydrolysis. At HDM conditions the availability of water molecules for enzymatic hydrolysis by cellulases could become the limiting factor: in fact, the deprivation of water molecules from the cellulose surface and transfer to the solvation sphere of oligos or monosaccharide leads to inhibition of the cellulases (Selig et al. 2012; Hsieh et al. 2014). Thus, introducing a net negative charge to the cellulose surface could be another mechanism for keeping water molecules available, reducing the overall water surface tension at the cellulose-solvent interface. In this way, LPMO enzymes could lead to higher wettability of the cellulose surface, and help alleviate the negative impact of HDM condition for cellulases, as suggested by the results obtained in this study and line with the mathematical simulations by Uchiyama et al. 2022. To date, only a handful of works investigated the role of water within the LPMO-GHs synergism, furthermore based on dynamic simulation. Our data on 2D-LFNMR depicting the activity of one only kind of C1-oxidizing LPMO should be enlarged with other enzymes with different oxidation mechanisms (C4 and C6) for describing a more general effect of cellulose surface dissolution.

Conclusions

The study showed that lytic polysaccharides monooxygenases (LPMO) of the AA9 family exhibit unaltered if not increased activity at HDM conditions (15% DM) when used in-solo, or when combined with a cellulase preparation up to 30% DM. Both of the tested LPMOs, MtAA9A and TtAA9E, functioned at glucose concentrations up to 75 g/l. In contrast, cellulases although still functioning are severely inhibited at such HDM conditions and solute concentrations as tested here. Considering that this is common in industrial conditions, it is an important finding. The results showed that in addition to the well-established belief that LPMO acts as a “generator-of-entry-sites” for CBHs or as a de-crystallizer of cellulose, yet another effect of LPMO activity could be helpful to the action of cellulases. The combination of a cellulose chain break and the introduction of a negative charge by C1-oxidation increase water constraint at the cellulose surface and whereby more water is available for hydrolysis. As shown in this study, LPMO enzymes increase the wettability of recalcitrant cellulose fibers. Future work will have to elucidate if this feature is common among LPMOs, as only two were studied here.