Pilot-scale production of xylo-oligosaccharides and fermentable sugars from Miscanthus using steam explosion pretreatment

This study investigated pilot-scale production of xylo-oligosaccharides (XOS) and fermentable sugars from Miscanthus using steam explosion (SE) pretreatment. SE conditions (200 °C; 15 bar; 10 min) led to XOS yields up to 52 % (w/w) of initial xylan in the hydrolysate. Liquid chromatography - mass spectrometry demonstrated that the solubilised XOS contained bound acetyl- and hydroxycinnamate residues, physicochemical properties known for high prebiotic effects and anti-oxidant activity in nutraceutical foods. Enzymatic hydrolysis of XOS-rich hydrolysate with commercial endo -xylanases resulted in xylobiose yields of 380 to 500 g/kg of initial xylan in the biomass after only 4 h, equivalent to ~74 to 90 % conversion of XOS into xylobiose. Fermentable glucose yields from enzymatic hydrolysis of solid residues were 8 to 9-fold higher than for untreated material. In view of an integrated bio-refinery, we demonstrate the potential for efficient utilisation of Miscanthus for the production of renewable sources, including biochemicals and biofuels.


Introduction
The European Union (EU) is committed to a reduction of >40 % in greenhouse gas (GHG) emissions and a 25 % increase in the total European transportation fuels from biofuels by 2030 to meet and potentially exceed the objectives of the Paris Agreement for long-term decarbonisation of the EU energy system. Such goals are driven in part by the transition from fossil-based fuels to carbon-neutral plant biomass-based renewable energy sources (European Commission, 2018). The dedicated biomass crop Miscanthus, closely related to the leading biofuel crops maize and sugarcane, is a high yielding C4 rhizomatous perennial grass that combines high photosynthetic, nutrient and water efficiency and is well adapted to a wide range of climates and soil types with good environmental credentials (Clifton-Brown et al., 2017). Field trials have demonstrated the potential for Miscanthus to produce up to 17 t/ha of harvestable dry matter (DM) biomass per year in the UK (Hastings et al., 2009). Miscanthus is an established biomass crop for co-firing in power stations within Europe (Terravesta, 2019) but also represents an abundant and renewable feedstock for refining into biofuels, bio-based materials and chemicals. Hence, more readily up-scalable high biomass yielding and seed-based hybrids of Miscanthus are being developed and are expected to be marketready by around 2022, to contribute more to renewable energy and GHG mitigation targets, and for expansion of the European bio-economy (Clifton-Brown et al., 2019).
Pretreatment is a crucial processing step to increase the porosity and reduce the recalcitrance of biomass to deconstruction for the production of biofuels and biochemicals (Chaturvedi and Verma, 2013). Depending on the type and severity of pretreatment, these processes can be costly and energy-intensive. Moreover, they can have distinct effects on the separation of the major components of lignocellulosic biomass, i.e. glucan (~25 to 40 %), xylan (~25 to 50 %) and lignin (~10 to 30 %) as well as result in sugar losses or by-products inhibitory to enzymes and microbial fermentation (Menon and Rao, 2012). Steam explosion (SE) is generally recognised as a scalable, cost-effective pretreatment technique with a low environmental impact (Chen and Liu, 2015). Industries such as the Brazilian biotechnology company GranBio have commercialised a SE pretreatment technology platform for converting sugarcane bagasse into biofuels and biochemicals (GranBio, 2014). During SE pretreatment, biomass is saturated with steam at high temperatures (160 °C to 240 °C) and pressures (7 to 48 bar) for several minutes (5 to 15 min) and then exposed to atmospheric pressure making the material undergo an explosive decompression (Chen and Liu, 2015).
Hydronium ions generated from the dissociation of water under high temperature and pressure act as a weak acid to cleave the acetyl-and uronyl-groups attached to the xylan backbone. This promotes the formation of acetic and uronic acid, which in turn catalyses (auto-hydrolysis) removal of xylan with the limited dissolution of glucan, while lignin undergoes fragmentation and recondensation reactions (Pu et al., 2015).
Xylo-oligosaccharides (XOS) are emergent value-added compounds due to their prebiotic properties for use in food and pharmaceuticals as well as a variety of other applications (Amorim et al., 2019;Kumar et al., 2012). The worldwide prebiotic market is expected to grow from ~3.6 € billion in 2017 to ~6.6 € billion by 2023 (MarketsandMarkets, 2018). A variety of XOS with a degree of polymerisation (DP) ranging from 2 to 10 xylose units can be recovered subject to SE process conditions and biomass substrates (Kumar et al., 2012). XOS are a hydrolysis product of xylan, which generally consists of a xylose backbone decorated with various substitutions including arabinofuranosyl, acetyl-groups, and ester-linked phenolic acids including ferulic and pcoumaric acid. In addition, the solubilisation and removal of xylan from lignocellulosic biomass via SE could help optimise biofuel production costs, as the pretreatment also causes disruption of the cellulose microfibrils and increases the accessibility of the This study evaluated the generation of XOS and fermentable sugars from Miscanthus hybrids involving SE pretreatment. Several SE process variables such as temperature, residence time and biomass particle sizes were tested using a pilot-scale SE unit, and their effects on sugar recovery and formation of fermentation inhibitors in SE hydrolysates were considered as evaluation criteria for pretreatment efficiency. The yields and DP of XOS were quantified, and the structural features of XOS were identified. Enzymatic digestion of SE hydrolysate with commercial endo-xylanases resulted in the depolymerisation of XOS into low-DP (2-3) XOS, while the SE pretreated solids showed improvements in enzymatically digested glucose yields for potential bioethanol co-production. We highlight a potential integrated bio-refinery process strategy for the efficient utilisation of the dedicated biomass crop Miscanthus.

Raw material
Mx2779, also known as GNT-14, is a leading novel seed-based interspecies hybrid representative portion of the biomass material was also hammer milled and sieved into an average size of 0.18 to 0.85 mm (-20/+80 mesh) for SE pretreatment, as biomass particle size represents one of the most important factors affecting SE performance.
Biomass moisture content was determined per technical report NREL/TP-510-42621 (Sluiter et al., 2008a) and biomass was used as received for SE pretreatment.

Steam explosion of Miscanthus
Miscanthus biomass (0.25 kg) was suspended at a water/solid ratio of 10:1 (g/g) and

SF = Log10
where t is the residence time (min), T is the temperature (°C) of pretreatment, 100 is the base temperature, i.e. 100 °C and 14.75 is the arbitrary constant which is an empirical parameter related with activation energy and temperature from first-order kinetics (Carvalheiro et al., 2009). In this study, SF was in the range of 3.4 to 4.7. It should be noted that the performance of SE pretreatment is also highly dependent on biomass particle sizes and moisture content, which were not considered in calculating SF.
Following SE pretreatment, the material was recovered in 10 L containers and cooled to room temperature. 0.5 L of deionised water was added to the pretreated slurry to recover water-soluble carbohydrates from the solids, and the slurry was homogenised and strained through a muslin cloth to separate the liquid and solid fractions. An aliquot of the liquid hydrolysate post-SE was retained for analysis of sugars as well as by-products and degradation products. Another aliquot of the hydrolysate was subjected to end-  as calibration standards using serial dilution concentration ranges of 20 μg/mL, 10 μg/mL, 5 μg/mL, 2.5 μg/mL and 1.25 μg/mL.

Phenolics quantification and physicochemical features of XOS by LC-MS
Partial purification of hydrolysates was carried out by solid-phase extraction (SPE) using Sep-Pak C18 cartridges (Waters Ltd, Elstree, UK) as described by Hauck

Enzymatic hydrolysis of hydrolysate and SE pretreated solids
Commercial endo-xylanases NS22083 and NS22002 (Novozymes) were used to produce XOS with a DP range of 2 to 3 xylose units from the hydrolysate. Endoxylanase activity was measured using arabinoxylan (wheat flour; medium viscosity ~ 30 cSt, 12.5 mg/mL Megazyme) as substrate as described by the Megazyme Somogyi reducing sugar assay. One unit of enzyme activity is the amount of enzyme required to release one µmole of reducing sugar equivalents (as xylose by the Somogyi reducingsugar method) from arabinoxylan per minute under standard assay conditions (40 °C and pH 4.7). The measured enzyme activity for NS22083 and NS22002 was 3610 U/mL and 690 U/mL, respectively. The SE hydrolysate was first centrifuged before enzymatic hydrolysis to remove suspended solids. 1 mL of the SE hydrolysate was then mixed with 42 µL of 1 M sodium citrate buffer (pH 5) and with 5.6 µl of 5 % sodium azide.
The total volume in each tube was brought to 1.4 mL with Milli-Q water after the endoxylanase NS22083, or NS22002 was added at a dosage of 36 U/mL of hydrolysate

Biomass composition of high yielding Miscanthus hybrids
The biomass composition of the two Miscanthus hybrids, Mx2779 and Mxg, used in this study are highlighted in Table 1

Steam explosion pretreatment for XOS production
Key factors affecting XOS production by SE pretreatment were investigated in this study, including temperature, residence time, as well as biomass particle sizes. The . This trend is likely due to the lower bulk density and higher porosity of Miscanthus biomass at larger particle size allowing for more efficient steam penetration and pretreatment reaction, than that of smaller biomass particle size (Liu et al., 2013).
These results infer that Miscanthus does not necessarily require extensive mechanical processing into smaller particle sizes to reduce biomass recalcitrance or to eliminate mass and heat transfer limitations during SE pretreatment. Hence, processing with The SE parameters (200 °C; 15 bar; 10 min) for Mx2779 resulted in XOS yields up to 52 % (w/w) and a low yield of xylose ~5 % (w/w) of initial xylan (Fig. 1a). While for Mxg, SE parameters (200 °C; 15 bar; 10 min) gave XOS yields up to 50 % (w/w) and a relatively high concentration of xylose ~22 % (w/w) of initial xylan (Fig. 1a), the latter i.e. xylose monomers usually regarded as an impurity and unfavourable for the production of valuable XOS compounds. In order to satisfy commercial XOS an auto-hydrolysis step using a lab-scale batch mini-reactor system. These findings would imply that XOS production from Miscanthus for commercialisation cannot be solely based on the XOS yields obtained in experimental laboratory trials and hence the need for pilot-scale studies. Nonetheless, the similar XOS yields (~50 % w/w of initial xylan) but different xylose yields (5 to 22 % w/w of initial xylan) under the same SE conditions for the two Miscanthus hybrids (Fig. 1a), are possibly related to distinctive structural properties of their plant cell walls. Likewise, HPAEC analysis provided information regarding the concentration and DP distribution of XOS produced from the SE pretreatment Miscanthus hybrids, with 52 vs 153 g/kg xylobiose, 49 vs 113 g/kg xylotriose, 42 vs 82 g/kg xylotetraose, 29 vs 50 g/kg xylopentaose, and 104 vs 345 g/kg DP5> of initial xylan when comparing Mx2779 with Mxg respectively (Fig. 1b). These results further suggest distinctive biomass features of the Miscanthus hybrids that may affect xylan hydrolysis into XOS and thus promote the concept of genetically engineering or breeding Miscanthus hybrids suited for specific end-use applications and bio-refining products (da Costa et al., 2019).

Composition of steam explosion pretreated solids, XOS and hydrolysate
The composition of Miscanthus solids after SE pretreatment was found to be in the range of ~40 to 45 % (w/w) for glucan, ~5 to 7 % (w/w) for xylan, ~1 to 2 % (w/w) for acetyl-residues and ~24 to 27 % (w/w) for lignin (Table 2a). Other minor components, including arabinan and galactan, were also detected in the residual solids (Table 2a) (Table 1 and Table 2a), while xylan and acetyl concentrations decreased as these latter fractions are readily released during the autohydrolysis step into the hydrolysate. Glucan recovery was high (~88 to 95 %) whereas xylan recovery was low (~21 to 27 %) (Table 2a), likely because xylan is less crystalline and more thermally labile than glucan.
Furthermore, LC-MS detected lower levels (~2-fold) of acetylated XOS in the hydrolysate for Mx2779 compared with Mxg after SE pretreatment, which is in accordance with the amount of acetic acid quantified for the oligomers in the hydrolysate (Table 2b). The LC-MS data also indicated that the majority of the XOS (~84 to 89 %) contained two or more attached acetyl-groups. Additionally, a series of XOS substituted with p-coumaric and ferulic acids, inherent components of xylanlignin complexes, were observed in the LC-MS mass spectra for the Miscanthus hybrids. These findings substantiate the prebiotic and antioxidant potential of XOS compounds, which are attributed to such substitutions as well as the DP of XOS (Singh et al., 2015).
In addition to assessing SE conditions for high xylan-to-XOS yield, characterising the composition of the XOS-rich hydrolysate for undesired monosaccharides and nonsaccharide compounds is important as these components may require removal for instance via downstream purification steps to obtain pure XOS. Different concentrations of by-products and sugar degradation products in the hydrolysate generated by SE conditions are highlighted in Table 2c. A broad range of compounds such as hydroxymethylfurfural (HMF) and furfural can be formed from the degradation of hexose and pentose sugars, respectively (Hu and Ragauskas, 2012), while acetic acid is produced by de-acetylation from xylan and formic acid is a product of HMF and furfural breakdown (Almeida et al., 2007). These products act as impurities and at elevated concentrations as inhibitory compounds to fermenting micro-organisms and thus can represent a major challenge for commercial production of bioethanol and other platform chemicals derived from lignocellulosic biomass (Zabed et al., 2016). Under the SE parameters which yielded the highest amount of XOS, Mx2779 and Mxg contained very low concentrations of the degradation compounds HMF (~1 to 4 g/kg) and furfural (~3 to 4 g/kg) of DM solids (Table 2c), equivalent to about 0.1 to 0.2 g/L of HMF and 0.3 g/L of furfural. In addition, other degradation and by-products measured were lactic acid (~5 to 12 g/kg), formic acid (~7 g/kg) and acetic acid (~20 g/kg) of DM solids (Table 2c), equivalent to ~0.6 g/L of lactic acid, ~0.6 g/L of formic acid and ~1.5 g/L of acetic acid. Moreover, very low concentrations of phenolic compounds were generated (Table 2c), such as p-coumaric and ferulic acid which are involved in crosslinking xylan and lignin in grasses (Carpita, 1996) (Table 2c) are likely a result of the initial amount of extractives and ash in the biomass causing a buffering effect. When SE temperature and residence time was increased to more harsh conditions, i.e. beyond 200 °C and 10 min, a lower hydrolysate pH and greater amount of inhibitory products such as HMF (~5 g/kg), furfural (~5 to 8 g/kg) and acetic acid (~25 g/kg) of DM solids were generated. Collectively, only minor degradative effects on glucan, lignin and xylan were exerted during SE pretreatment conditions (200 °C; 15 bar; 10 min), and the low concentrations of degradation and by-products obtained from SE pretreated Miscanthus indicate that a purification step for their removal prior to downstream enzymatic hydrolysis and fermentation may not be necessary.

Enzymatic hydrolysis of steam explosion hydrolysate
To further improve the selective release of xylobiose and xylotriose which represent the most valuable XOS compounds for health benefits (Amorim et al., 2019;Singh et al., 2015), enzymatic hydrolysis with commercial endo-xylanase preparations of the SE hydrolysate was performed. As shown in Fig. 2, high DP XOS was effectively cleaved by the commercially available Novozymes endo-xylanases NS22083 and NS22002 to fragments mainly of xylobiose and with the lower formation of xylotriose. Moreover, the XOS profile produced by the endo-xylanases were similar between the Miscanthus hybrids but yielding different amounts of predominantly xylose, xylobiose and xylotriose. For Mx2779, xylobiose increased from 52 to ~380 g/kg of initial xylan and from 52 to ~280 g/kg of initial xylan, while for Mxg xylobiose increased from 153 to ~500 g/kg of initial xylan and from 153 to ~370 g/kg of initial xylan using NS22083 and NS22002 respectively, within 4 h of enzymatic hydrolysis (Fig. 2). The proportion of xylobiose represents an increase of ~6-fold and ~3-fold for Mx2779 and Mxg, respectively, when compared to the SE hydrolysate without endo-xylanase treatment.

Enzymatic hydrolysis of steam explosion pretreated solid residues
As the xylan component is partly solubilised into the hydrolysate during SE pretreatment, the remaining pretreated solids represent a fraction rich in glucan and lignin that can be further converted into bioethanol and biochemicals. As can be deduced from the glucose yields obtained after enzymatic hydrolysis over 72 h with Cellic® CTec2, SE pretreatment improved the digestibility performance of the pretreated biomass and glucan available to cellulases as compared to untreated material by 8 to 9-fold (Fig. 3). For Mx2779 and Mxg, the glucose yields of the untreated material were only 8 and 5 % w/w of initial glucan, respectively. Whereas following SE pretreatment, glucose yields of the pretreated material increased to ~70 and 40 % w/w, respectively (Fig. 3). These findings suggest that xylan removal and perhaps deacetylation of xylan by SE pretreatment are amongst the factors that increase the accessibility of the pretreated residual solids glucan to cellulolytic enzymes (Oliveira et al., 2013;Zhang et al., 2018). While the SE conditions were similar for Mx2779 and Mxg, the different glucose yields from the two pretreated materials were suggestive of variable biomass recalcitrance between Miscanthus hybrids. Even though delignification was limited in the SE pretreated biomass, it was ~5-fold higher for Mxg (~20 %) compared to Mx2279 (~4 %) ( Table 2a), suggesting that delignification per se did not necessarily render the residual solids more susceptible to enzymatic attack (Fig.   3). However, it could be related to compositional and structural changes to biomass as a result of SE pretreatment, including rearrangements in the structure of glucan or lignin fragmentation and recondensation on pretreated solid surfaces, as well as the nature of the lignocellulosic matrix that lignin is present in (Oliveira et al., 2013;Pu et al., 2015).
Hence, delignification and lignin transformations/relocation that occurs during SE pretreatment and their effects on biomass recalcitrance represent factors to elucidate (Pu et al., 2015).

Mass balance
An overall mass balance of the SE process and enzymatic hydrolysis steps applied to Mx2279 is summarised in Fig. 4. Based on 1 kg of dry mass (DM) Miscanthus biomass, ~900 g (~90 %) of the original untreated solids were recovered after the pre-soaking step (Fig. 4). This is due to loss of the fine particles and biomass during the crudely conducted removal of excess liquid through a muslin cloth prior to SE pretreatment. For ~900 g DM of Mx2779 input into the SE process (200 °C; 15 bar; 10 min), ~90 % was obtained as solid fraction (~380 g glucan, ~96 g xylan and ~250 g lignin) and the remaining ~10 % was dissolved into the hydrolysate mostly comprising of solubilised XOS (~120 g) (Fig. 4). Enzymatic hydrolysis after only 4 h of the SE hydrolysate with commercially available Novozymes endo-xylanases improved xylobiose quantities from ~12 to ~88 g (Fig. 4), demonstrating that ~40 % of initial xylan from Mx2279 and ~44 % of initial DM xylan from Mxg could be converted into xylobiose. This is of particular industrial interest as the value-added compound xylobiose exhibits the strongest prebiotic activity amongst the XOS compounds and a higher sweetness potency than sucrose (Moura et al., 2007;Park et al., 2017). The less accessible ~40 % (w/w) of initial xylan remaining in the pretreated residual solids was also enzymatically released as XOS with the commercial Novozymes endo-xylanases NS22083 (Fig. 4). Hence, the overall process achieved a xylan (228 g) to XOS (170 g) conversion rate of ~75 %. To recover the maximum amount of XOS for an optimised biorefinery, the possibility to directly treat the whole SE slurry with endo-xylanases, prior to separating the slurry into SE pretreated solids and hydrolysate, as well as the potential to lower the enzyme dosage, requires further investigation. Furthermore, the glucan rich pretreated solids from Mx2779 was subjected to 72 h enzymatic hydrolysis for initial investigation with a low solids loading (1 % w\w) and high dosage of Cellic® CTec2 (30 % g enzyme/g glucan) to indicate maximum enzymatically accessible glucan content, rather than a high solids loading and low dosage which should provide a target for commercially feasible glucan hydrolysis. We noted ~70 % of the glucan was released as glucose (equivalent to ~258 g of glucan) (Fig. 4), which can be further processed to bioethanol or platform chemicals via fermentation. Future testings will encompass the effects of shorter hydrolysis time, total solids loadings, glucan conversion and enzyme trial dosage levels. The post-enzymatic hydrolysis solids left behind were not quantified, although it is anticipated that this solid fraction should contain the remaining ~250 g of lignin that could be further used to produce lignin-based materials or combusted to contribute towards energy requirements within the proposed bio-refinery process.

Conclusions
Pilot-scale production of XOS and fermentable sugars from Miscanthus was investigated using steam explosion (SE) pretreatment. Under the SE conditions studied (200 °C; 15 bar; 10 min), XOS yields up to 52 % (w/w) of initial xylan were obtained.
The main effect of commercial endo-xylanases on the XOS-rich hydrolysate was the production of xylobiose (380 to 500 g/kg of initial xylan), a low-DP XOS having the highest pre-biotic potential. SE pretreatment also improved the cellulolytic hydrolysis of pretreated solids to increase the production of fermentable sugars by 8 to 9-fold. In view of an integrated bio-refinery, SE represents a prospective pretreatment technology for the production of XOS and fermentable sugars from Miscanthus.

This work was financed by the Biotechnology and Biological Sciences Research
Council (BBSRC) (BB/P017460/1). The authors thank Paul Robson and the Miscanthus breeding team at IBERS Aberystwyth University for the hybrids (funded by BBSRC BB/CSP1730/1, BB/N016149/1, BB/K01711X/1, BBS/E/W/10963A01 as well as the GIANT-LINK project LK0863) and the BEACON Biorefining Centre of Excellence for supporting the work presented in this paper.

Supplementary data
Supplementary data of this work can be found in the online version of the paper. without additional chemicals. Bioresour. Technol. 265, 387-393. Table 1. Cell wall composition of two high biomass yielding Miscanthus hybrids on an as-received basis (a) and extractives free basis (b). Data are means ± standard error (n ≥ 3).     Table 2.

Figure and table captions
Biomass recovered (%) = g of DM residual solids recovered after SE pretreatment/100 g DM untreated biomass.    Table 2b