Chipped unbarked Norway spruce (Picea abies) was pretreated in the Biorefinery Demonstration Plant (BDP) (Örnsköldsvik, Sweden) by SEKAB E-Technology AB. The wood chips were impregnated with sulfur dioxide (approx. 2% based on the dry weight of the biomass), and subsequent pretreatment was performed using continuous steam explosion at 205°C for approx. 10 min. The pH of the generated slurry was approx. 1.8, and the dry-matter content was around 30%.

Prior to enzymatic saccharification experiments, the pH of the slurry was adjusted to 5.2 using a 10 M aqueous solution of sodium hydroxide, and the total solid (TS) content was adjusted to 12.5% (w/w) by dilution with ultra-pure water. The solid and liquid phases of the slurry were separated by centrifugation (Avanti J-26 XP, Beckman Coulter, United States). The solid phase was extensively washed with ultra-pure water until glucose was no longer detectable in the filtrate through analysis with a glucometer (Accu-Chek Aviva, Roche Diagnostics GmBH). The solids were then air-dried before further use. The dry-matter content of the washed and dried solids was 96.6% according to measurements using an HG63 moisture analyzer (Mettler-Toledo, Greifensee, Switzerland). Particles remaining in the liquid phase after centrifugation were removed by passing it through Nalgene™ Rapid-Flow™ sterile filters (0.2 µm, PES, Thermo Fisher Scientific, Waltham, MA, United States). The pretreated solids (PS) and the pretreatment liquid (PL) were then used in enzymatic saccharification experiments.

Two series of enzymatic saccharification experiments were performed. One consisted of pretreatment liquid (PL) with microcrystalline cellulose (Avicel^® PH-101 obtained from Sigma-Aldrich, St. Louis, MO, United States) and the other consisted of pretreated solids (PS) with 20 mM sodium acetate buffer, pH 5.2. Reaction mixtures consisted of 8.75 g (dry-weight) solids (PS or Avicel, used as substrate in the enzymatic reaction), 61.25 g liquid phase (buffer or PL), 540 µL Cellic^® CTec3 (Novozymes A/S, Bagsvaerd, Denmark) (1.13 g/mL), and 88 µL antifoaming agent (Tween-80). A separate experiment was performed to assure that inclusion of antifoaming agent did not affect the enzymatic reaction (data not shown).

Enzymatic saccharification experiments were carried out using 125 mL gas wash bottles (Quickfit^® Drechsel bottle, Sigma-Aldrich). Aerobic and anaerobic conditions were achieved by continuous supply of air or N₂ to the gas wash bottles through Polyamide 12 tubing (outer diameter 6 mm). The set-up could accommodate up to six gas wash bottles in parallel in each experiment, allowing for two sets of triplicates (one set with air and one set with N₂) and subsequent statistical analysis of the results. For each gas supply, a gas distributor divided the gas into three strands, and each gas strand was connected with a flow meter (Porter F65, Parker, United States). The purity of the N₂ was 99.996%. The gas flow was kept at one vvm (based on flow rate and reaction mixture volume). The gas wash bottles were positioned on a multipoint stirrer (Cimarec from Thermo Scientific) immersed in a water bath to allow for temperature control. The bottles were equipped with magnetic stirrer bars (cylindrical 40 × 8 mm), and the stirring speed was 215 rpm. Incubation was performed at 45°C for 72 h, and samples (0.5 mL each) were withdrawn after 24 h, 48 h, and 72 h by using an automatic pipette. Zero samples (0 h) were collected after adding enzyme. Reactions were conducted in triplicates, and the monosaccharides produced in the reactions were analyzed using HPAEC (Section 2.3.1).

After enzymatic saccharification experiments, the mixtures were centrifuged to separate the solid and the liquid phases. Solid phases were washed with abundant ultra-pure water until no glucose was detectable in the filtrate. Washed solids were air-dried until the dry-matter content was over 90%, and they were then sieved using sieve shakers (Retsch AS 200) with an aperture size of 100 μm to homogenize the solid fraction. Liquid phases were stored at 4°C until further analysis.

The liquid and solid fractions of pretreated softwood from the demonstration plant were analyzed to get an overview of lignin-related material and carbohydrates in the fractions. According to compositional analysis using TSSA (Table 1), the pretreated solids contained (DW) 44.5% glucan and 51.6% total lignin (Klason lignin and ASL). No other carbohydrates than glucan were detected. This indicates that the pretreatment conditions had been very harsh, as all hemicellulosic carbohydrates had been removed and as the lignin content was higher than the glucan content. From the beginning, it is the other way around, as untreated spruce wood contains 27.4% lignin and 41.7% cellulose (Sjöström, 1993). The aim with the pretreatment is to achieve an almost quantitative removal of hemicelluloses, without degrading cellulose, which is left for enzymatic saccharification. The fact that the total lignin content was higher than the glucan content indicates partial degradation of cellulose, formation of pseudo-lignin from hemicelluloses, or occurrence of both these events. Pseudo-lignin is a Klason-lignin-positive aromatic substance formed from carbohydrates during thermal treatment under harsh conditions (Shinde et al., 2018; Martín et al., 2022; Yao et al., 2022).

Analysis using Py-GC/MS (Table 1) suggested that the pretreated material predominantly (∼70%) consisted of carbohydrates and that around a quarter of the content was lignin. As expected for softwood (Ralph et al., 2004) such as spruce, the lignin predominantly consisted of G units (Table 1). With respect to the determined fractions, the discrepancy compared to the results from the TSSA analysis can be explained by pseudo-lignin being characterized as carbohydrates in the Py-GC/MS analysis, which has been observed previously (Wang et al., 2018a). Thus, compositional analysis using TSSA and Py-GC/MS indicated that the hemicelluloses had been quantitatively removed or degraded and that the pretreated material predominantly consisted of three substances: glucan, lignin, and pseudo-lignin.

The pretreated solids were analyzed further using HSQC NMR (Figure 1). Pretreated solids exhibited relatively poor solubility in DMSO compared to non-pretreated Norway spruce wood, and the signals were weaker. The pretreatment affected the aromatic region of the lignin, as indicated by the lower intensity of G2 and G6 after pretreatment. Signals from common interunit lignin linkages, such as β-O-4 and phenylcoumaran, were weak in the pretreated material (Figure 1). These changes are in agreement with previous studies of lignin after pretreatment under acidic conditions (Yao et al., 2022). An attempt to solubilize more of the pretreated material using a mixture of DMSO-d6 and pyridine-d5 (Kim and Ralph, 2010) to obtain stronger signals was unsuccessful.

SEM was used to further investigate the structure of the solid fractions. Micrographs of PS showed nano-sized particles or droplets on fibrillar surfaces (Figure 2A). There are previous reports on droplets on biomass surfaces caused by relocalization of lignin (Selig et al., 2007; Donohoe et al., 2008) and by pseudo-lignin formation (Sannigrahi et al., 2011) during pretreatment under acidic conditions. Such droplets could vary significantly in size, such as from 5 nm up to 10 μm (Donohoe et al., 2008) or 0.3 μm–8.0 μm (Sannigrahi et al., 2011). It is a possibility that harsh, acidic pretreatment conditions contributed to the formation of these structures (Figure 2A). After enzymatic saccharification with air, the surfaces appeared somewhat smoother (Figure 2B). This was less pronounced in reactions with N₂ (Figure 2C), where the enzymatic saccharification had reached less far.

As expected considering the relatively high content of galactoglucomannan in spruce wood (Sjöström, 1993), mannose was the predominant hemicellulosic monosaccharide in the PL (Table 2). The total concentration of hemicellulosic sugars (mannose, xylose, galactose, and arabinose) (Table 2) was 36 g/L, and the concentration of glucose was around 31 g/L (Table 3). Even if it is not intentional to degrade cellulose in the pretreatment, the glucose could come either from hemicelluloses or from partial degradation of cellulose.

The concentrations of substances in the pretreatment liquid (Tables 2, 3) can be compared to those found in a series of pretreatment liquids from Norway spruce pretreated with sulfur dioxide under low, medium, and high severity (Wang et al., 2018a; Martín et al., 2018). The TCAC content was just below 200 mM (Table 2), which is very high for a TS content of 12.5% considering that it was 203 mM after the most severe treatment when the TS was 25% (Wang et al., 2018a). The content of total phenolics was around 5 g/L, which again is very high compared to 1.6 g/L after the most severe treatment in a previous study of material with TS ∼30% (Martín et al., 2018). The combined concentration of furfural and HMF was around 59 mM (Table 2). Considering the TS, 12.5%, that is higher than the highest combined value for furfural and HMF reported by Martín et al. (2018), which was 87 mM at a TS of around 30%. The TAC value (1.2 with 1,000 × dilution) was also relatively high compared to previous studies, as Wang et al. (2018a) reported values from 0.59 up to 2.74 with TS 25% and 500 × dilution. That is consistent with the high values for total phenolics and furan aldehydes, as TAC covers both of these groups of substances.

The results of the analysis of the liquid phase are consistent with the results of the analysis of the solid phase, as both data sets point to very harsh pretreatment and, consequently, high contents of Klason lignin and phenolics. The content of lignin in the solid phase and the content of phenolics in the liquid phase would be relevant, as it may affect the ability of these fractions to serve as reductants in reactions with LPMO. In addition to that, substances in the PL might inhibit cellulolytic enzymes. Such substances include both sugars (causing feedback inhibition of cellulases) and aromatics, such as phenolics (Zhai et al., 2016; Martín et al., 2022).

The total monosaccharide concentration (∼67 g/L) of the PL used in this work was similar to the total monosaccharide concentrations (64 g/L–77 g/L) of the PLs studied by Wang et al. (2018a). As all PLs studied by Wang et al. (2018a) had total monosaccharide concentrations that were inhibitory to the cellulolytic enzyme preparations, it is safe to assume that the PL used in this work also caused sugar inhibition. All PLs studied by Wang et al. (2018a) were also clearly inhibitory to cellulolytic enzyme preparations with respect to other substances than monosaccharides, such as phenolics, despite that the concentrations of phenolics in the PLs [0.7 g/L–1.6 g/L according to data from Martín et al. (2018)] were much lower than in the PL used in this work (∼5 g/L). It is therefore likely that the PL used in this work also contained inhibitory concentrations of phenolics.

The enzymatic saccharification reactions contained the same amount of solids (8.75 g Avicel or PS), but different glucan content. Whereas Avicel consists almost exclusively of glucan (in the form of microcrystalline cellulose), the PS had a glucan content of 44.5%, and therefore the reaction mixtures with PS contained 3.89 g glucan rather than 8.75. Reactions with PS also contained lignin and pseudo-lignin, which have a negative effect on saccharification (Shinde et al., 2018; Martín et al., 2022; Yao et al., 2022). Avicel would therefore be a much better substrate than PS, but the reactions with Avicel also contained PL and therefore also sugars and phenolics that inhibit enzymatic saccharification. For these reasons, it was not self-evident which set of reactions that would result in the highest enzymatic saccharification yields.

Data on glucose concentrations and glucan conversion are shown in Table 3. As expected (Arantes and Saddler, 2011), the glucose concentrations increased rapidly in the beginning and leveled off at the end. The glucose concentrations (in g/L) generated between the zero sample and 72 h were: PS/air, 53.3; PS/N₂ 42.5; Avicel/air, 54.4; Avicel/N₂, 47.8. Thus, reactions with air always resulted in more glucose than the corresponding reactions with N₂ (25% more for PS and 14% more for Avicel). This can be attributed to the catalytic action of LPMO, and it was apparent both with PS and PL without any addition of external reductant. Although the glucose yields of enzymatic saccharification were rather similar for reactions with air (53 g/L–54 g/L) and for reactions with N₂ (42 g/L–48 g/L), the glucan conversion was higher for reactions with PS (68.6%–85.9%) than for reactions with Avicel (35.2%–40.1%). This phenomenon can be attributed to the presence of enzyme inhibitors in PL having a negative effect on the reactions with Avicel, and perhaps also to the enzyme:glucan ratio being more than twice as high in experiments with PS as in experiments with Avicel.

This work differs from previous studies of LPMO-catalyzed reactions with pretreated biomass in several different ways; as it (i) covers the separated solid and liquid phases from the pretreated biomass, (ii) involves controlled continuous gas addition using air or N₂, (iii) is based on hydrothermal pretreatment of spruce wood with sulfur dioxide in a demonstration plant, (iv) involves no redox-active chemicals except oxygen in air in additions after pretreatment, and (v) involves detailed characterization of the liquid and solid phases, for instance with respect to the occurrence of pseudo-lignin. In a series of experiments with steam-pretreated lodgepole pine with different lignin content, Hu et al. (2014) found that addition of an external reducing agent (gallate) had a positive effect on the LPMO reaction only with a completely delignified substrate, but not when the lignin content was 12% or higher. Rodríguez-Zúñiga et al. (2015) investigated washed pretreated solids from sugarcane bagasse, wheat straw, and corn stover. The gas phase was not controlled, but determination of gluconic acid, an oxidation product from the LPMO reaction, suggested that substrates with low (9.6% or less) Klason lignin content were associated with no or very small LPMO activity. Müller et al. (2015) showed improved saccharification of pretreated birch wood in flasks containing air in the headspace compared to reactions in flasks in which the headspace was flushed with N₂ and in which cysteine was added to the reaction mixtures to remove residual oxygen. Chylenski et al. (2017) investigated saccharification of sulfite-pulped softwood, and therefore their biomass contained much less lignin (4%–9% acid-insoluble lignin) than that used by Müller et al. (2015) or that used in the present study. As a result, improved saccharification with air in the headspace was observed only when an electron donor, such as ascorbic acid or lignosulfonates, was added to the reaction mixtures (Chylenski et al., 2017). These results agree with our findings, as the lignin content of the solid phase used in our experiments was clearly higher than the lignin contents of substrates for which no or very little LPMO activity has been detected without addition of an external reductant.

The liquid phases at the end of the reactions with Avicel (PL-Air and PL-N₂) were analyzed with regard to other substances than glucose (Table 2; Figure 3). Smaller increases (0.4 g/L–6.8 g/L) in the concentrations of sugars other than glucose (Table 2) can be attributed to enzymatic saccharification of oligosaccharides and disaccharides, and also to concentration of reaction mixtures at the end of the experiments due to evaporation caused by gas flow.

The pH decreased somewhat in PL-Air, and the TCAC value was higher than for PL-N₂. It is a possibility that oxidation products from LPMO-catalyzed reactions contribute to the high acid content in PL-Air. This is supported by data in Figure 3, which shows analysis of tentative C1 and C4 oxidation products. As judged from the elution time, the tentative C1 product [RT (retention time) 8.83] would correspond to gluconic acid (RT 8.84 for gluconic acid reference), and, at the end of the reaction (48 h), the peak area was much larger for reactions with air than for reactions with N₂ (Figure 3C). The area of an unidentified peak at RT 13.4 increased for reactions with air, which was not observed for reactions with N₂ (Figure 3B). This could tentatively reflect formation of a C4 oxidation product. Minor concentrations of the tentative C1 and C4 products were present also in reaction mixtures with N₂, but that could be due to that those reactions were not completely devoid of molecular oxygen or other oxidants. Also, N₂ contains molecular oxygen as contaminant, so small amounts of molecular oxygen would be added throughout the experiment even in reactions with N₂.

LPMO-catalyzed cleavage of cellulose leads to oxidation of carbons in the β-1,4-glycosidic bonds, i.e., oxidation of C1 or C4. Using Cellic CTec2, Cannella et al. (2012) studied C1 oxidation products, such as gluconic acid and cellobionic acid, in reactions with pretreated wheat straw. Müller et al. (2015) quantified C4-oxidized cellobiose in reaction mixtures with pretreated birch wood and Cellic CTec2.

With regard to aromatic and heteroaromatic substances, the concentrations of total phenolics and HMF did not change very much throughout the experiment (Table 2). The concentration of furfural decreased from 32 mM in PL to around 1 mM in PL-Air and PL-N₂. This is because furfural is much more volatile than HMF and the gas flow will remove it, as has also been observed in previous studies (Larsson et al., 1999). The TAC value decreased somewhat in both PL-Air and PL-N₂, which can be attributed to the removal of furfural, as both aromatics and heteroaromatics contribute to TAC. The total phenolic content was somewhat higher after the reactions than in the original PL (Table 2). One reason for this could be contribution to total phenolics by the enzyme preparation, which was not yet added to the PL when samples for analysis of total phenolics were withdrawn.

As expected on basis of the saccharification data in Table 3, the glucan (TSSA) and carbohydrate (Py-GC/MS) contents of PS-Air and PS-N₂ decreased compared to that of PS, while Klason lignin (TSSA) and total lignin (Py-GC/MS) increased (Table 1). The changes compared to PS were consistently larger for PS-Air than for PS-N₂. This can also be seen as shifts in the carbohydrate:lignin ratio according to the Py-GC/MS analysis (Table 1).

Spruce lignin contains around 20.5 phenolic hydroxyl groups per 100 C9 (phenylpropane) units (Adler, 1977). The content of phenolic hydroxyl groups on dry solids increased in PS-Air and PS-N₂ compared to PS (Table 1). This is expected, as the enzymatic saccharification of cellulose increases the fraction of lignin. When calculated as phenolic hydroxyl groups on total lignin (Klason lignin and ASL), the values for PS-Air and PS-N₂ were still slightly higher than for PS.

Solid residues from PS-Air and PS-N₂ reactions were analyzed using HSQC-NMR and FTIR. As for PS (Figure 1), the solubility was rather poor and the signals were relatively weak. Using FTIR, the solid fraction after enzymatic saccharification was compared to PS (Figure 4). Lower wavenumber regions, such as 1,030 cm⁻¹–1,060 cm⁻¹ and 1,108 cm⁻¹ (ring vibrations) were assigned to polysaccharides (Labbé et al., 2005; Fackler et al., 2010; Durmaz et al., 2016). Lower band intensities for the reaction with air are consistent with more efficient enzymatic saccharification. This is also supported by differences observed for the bands at ∼ 1,150 cm⁻¹ (Labbé et al., 2005) and 1,158 cm⁻¹ (Pandey and Nagveni, 2007), which are associated with C-O-C asymmetric vibrations. Bands from 1,300 cm⁻¹ to 1,450 cm⁻¹ were assigned to C-H vibrations attributed to cellulose (Labbé et al., 2005). Lignin exhibits characteristic bands at 1,510 cm⁻¹ and 1,600 cm⁻¹. These bands are assigned to aromatic carbon-carbon double bonds (Sarkanen and Ludwig, 1971) and the band at 1,510 cm⁻¹ is especially assigned for G units of lignin (Faix, 1991). In addition, an increase at 1,595 cm⁻¹ is associated with aromatic skeletal vibrations (Faix, 1991; Faix and Böttcher, 1992) and indicates formation of pseudo-lignin (Wang et al., 2018b). According to previous studies (Faix, 1991; Pandey, 1999; Pandey and Nagveni, 2007), the band at 1,268 cm⁻¹ is attributed to guaiacyl rings and carbonyl stretching in lignin. The band at 1,714 cm⁻¹ is related to C = O stretching in unconjugated ketones, associated with pseudo-lignin (Sannigrahi et al., 2011), whereas the band at 1,672 cm⁻¹ has been assigned to conjugated carbonyl groups (Gillgren et al., 2017).