Optimization of pretreatment conditions
To obtain higher yield of sugar, optimization of pretreatment conditions is necessary to remove lignin to the largest extent. In this work, three major factors, namely, pretreatment time, ratio of H3PO4 to H2O2 (v/v), and temperature were explored. Based on Table 1, pretreatment time was firstly optimized under the condition of 10% (w/v) of solid loading, ratio of H3PO4 to H2O2 (v/v, 1:9) at 100 °C. As is shown in Fig. 1a, the contents of cellulose (74.0–83.0%), hemicellulose (4.0–5.5%), acid-insoluble lignin (0.5–1.0%), and the concentration of sugar (7.4–7.6 mg/mL) were similar after 60, 90, and 120 min pretreatment, all of which was more effective than 30 min pretreatment. However, a lower biomass loss was observed for 60 min (56.9%) pretreatment compared with 90 min (64.9%) and 120 min (77.4%) pretreatments. Considering the overall results, 60 min was regarded as the most effective pretreatment time among 0–120 min.
In addition to pretreatment time, other two conditions, ratio of H3PO4 to H2O2 (v/v) and pretreatment temperature, were also optimized according to Table 1. As shown in Fig. 1b, c, in contrast to the ratio of H3PO4 to H2O2 (v/v), the temperature posed a great influence on the changes of components, biomass, and sugar concentration during the pretreatment process. In this study, 2:8 was verified as the optimal ratio of H3PO4 to H2O2 (v/v) due to the results that higher cellulose content (86.4%), lower hemicellulose (5.5%), lower acid-insoluble lignin content (0.2%), and higher concentration of sugar (8.4 mg/mL) were observed in comparison with other ratios. Three kinds of components and sugar yield were changed dramatically with different pretreatment temperatures, especially from 90–100 °C. At 100 °C, cellulose content occupied 86.4% (w/w) of the pretreated PH, which was approximately 1.4-fold higher than that of the raw PH. And sugar concentration generated from the pretreated PH achieved 8.6 mg/mL, which was about 10.5 times as much as the raw PH observed. Meanwhile, hemicellulose and lignin content decreased about 85.0 and 98.0%, respectively (Fig. 1c). However, these parameters remained steady as the temperature continued to rise. Thus, 100 °C was recognized as the optimal pretreatment temperature due to its lower energy cost and higher overall pretreatment efficiency.
In conclusion, the optimal pretreatment conditions were pretreatment time for 60 min, pretreatment temperature at 100 °C, 10% (w/v) of solid loading, and ratio of H3PO4 to H2O2 at 2:8 (v/v) corresponding to the concentrations (w/w) of H3PO4 and H2O2 with 23.10 and 21.85%, respectively.
Comparisons with other pretreatment methods
Up to now, a huge number of pretreatment methods for lignocellulosic materials have been applied to improve the enzymatic hydrolysis efficiency with lower cost [26,27,28,29,30,31, 36]. Previous studies showed that dilute acid would be able to remove hemicellulose composition and H2O2 could reduce the lignin content under an appropriate condition [37, 38]. About 80–100% hemicellulose and 75% lignin of wheat straw were removed by the combination of concentrated H3PO4 (70–80%) and low concentration of H2O2 (1–5%) at about 40 °C for more than 2 h . In this study, the PH biomass was pretreated with a mixture of lower concentration of H3PO4 (23.10%) and higher concentration of H2O2 (21.85%) at 100 °C for 60 min. The results demonstrated that more lignin (98.0%) was removed. Higher cellulose–glucose conversion (95.0%) was achieved with lower cellulase consumption (6.9 FPU/g glucan) using the pretreatment method in this research compared with the previous reports (about 77–94%) with 20 FPU/g glucan [39, 40]. Probably, the lignin removal could result in great increase of the enzymatic hydrolysis efficiency. Moreover, the changes of composition contents for PH pretreated by H3PO4 and H2O2 solely were also explored in this research. As shown in Fig. 2a, the cellulose content using separate pretreatment methods was a little higher than raw PH, but reached only about half of the combined pretreatment method. Besides that, separate pretreatment methods were less efficient in removing hemicellulose and lignin in comparison with the combined pretreatment method. As shown in Fig. 2b, higher sugar yield (8.6 mg/mL) was achieved when the PH was pretreated under condition of combined H3PO4 and H2O2 at the volume ratio of 2:8, in contrast to the H3PO4 (1.89 mg/mL) and H2O2 (2.31 mg/mL) pretreatment methods solely. The results demonstrated that the combination of H3PO4 and H2O2 pretreatment could remove the hemicellulose and lignin better and obtain much higher sugar yield than the separate pretreated PH.
Lignin, a complex polymer, is formed by oxidative coupling of three major C6–C3 units, namely, guaiacyl alcohol (G), syringyl alcohol (S), and p-coumaryl alcohol (H) . The monolignols are linked by different ether bonds (such as aryl–aryl ether, alkyl–aryl ether, and biphenyl ether), C–C linkages (such as β–β and β-5), and the combination of ether bonds and C–C linkages (such as alkyl–aryl ether bond + β-5 carbon linkage and alkyl–alkyl ether bond + β–β carbon linkage). Among them, the aryl–aryl ether bond is the major interunit linkage [41, 42]. H2O2 could be decomposed into hydroperoxide anion (HOO−), hydroxyl radicals (HO·), and superoxide anion radicals (O2·−) by heating [41, 43]. Previous studies have reported that aryl ether bonds, lignin ring, ethylenic, carbonyl groups, and other linkages could be cleaved by HOO− (strong nucleophile) and HO·, O2·−· (active radicals) generated from decomposed H2O2 [41, 44, 45]. Hence, part of lignin could be degraded with the treatment of H2O2. On the other hand, hemicelluloses were heterogeneous polysaccharides which contained either xylan or glucomannan backbones with acetyl group, galactose, arabinose, and methyl glucuronic acid on the side chains . Different from the lignin and cellulose, hemicellulose could be ruined more easily by disrupting covalent bonds, hydrogen bonds, and van der Waals forces (such as C–O and C–H stretch in hemicellulose) during acidic pretreatment [46, 47]. Simultaneously, two main structures including aromatic ring and ring-conjugated C=C bonds of lignin were broken slightly, which resulted in a little lignin removal . Moreover, the site of C
would be formed with the cleavage of α-aryl ether under acidic environment. The strong nucleophile, such as HOO− from H2O2 decomposition, could attack the C
position to prevent the condensation with other lignin molecules again . As indicated above, the condensed structure of lignocellulosic material was broken by removing hemicellulose using H3PO4, along with disrupting the lignin–carbohydrate linkages, and exposing Cα+ site, which might make it easier to remove the lignin using H2O2.
All the results indicated that, compared with mechanical pretreatment (milling and popping) and other chemical pretreatments (sulfuric acid, alkali, and ammonia fiber explosion), lower energy consumption, higher security, and greater efficiency for enzymatic hydrolysis were attained by the combination of H3PO4 and H2O2 pretreatment method.
Surface morphology of raw and pretreated PH
The enzymatic hydrolysis efficiency was determined by structure of the lignocellulosic materials . The morphology changes of PH before and after being pretreated by H3PO4/H2O2 could provide direct information for explaining the rapid increase of the hydrolysis efficiency. Obviously, the surface of raw PH was smooth and integrated as shown in Fig. 3a. However, our previous study has proved that popping pretreatment led to the appearance of pores in the PH surface without thickness change (Fig. 3c). And the structure of the PH was still integrated though it became loosened and distortional after being pretreated by H2O2–acetic acid (HPAC) (Fig. 3d) . In this study, significant pores occurred along with the pretreatment process by the combination of H3PO4 and H2O2 on account of hemicellulose removal. This pretreatment also resulted in the appearance of wrinkle on the surface of PH (Fig. 3a, b). We can safely come to the conclusion that structure of the PH was broken down and became much looser after being pretreated by H3PO4/H2O2. As a result, the enzymatic hydrolysis efficiency of the pretreated PH (95.0% glucose yield) increased much more than that of the raw PH (19.1% glucose yield) as hemicellulose and lignin were removed.
Solid-state cross-polarization magic angle spining carbon-13 nuclear magnetic resonance (CP/MAS 13C NMR) analysis of PH.
Solid-state NMR spectroscopy has been widely employed for structure characterization analysis of lignocellulosic materials during the past decades, and has always been considered a very useful analytical tool for carbohydrates and lignin . Therefore, it was adopted to investigate the composition of cellulose, hemicellulose, and lignin before and after pretreating PH by the combination of H3PO4 and H2O2 in this work. As demonstrated in Fig. 4 and Table 2, various components were detected, whose chemical shifts were the same as the previous report [50, 51]. The two peaks at 173.8 and 21.42 ppm were assigned to hemicellulose. The peaks at 130–155 ppm, and 56.11 ppm were assigned to lignin. The peaks at 89.23 and 65.06 ppm represented crystalline cellulose, and the peaks at 84.00 and 63.20 ppm corresponded to amorphous cellulose. The peak at 105.53 ppm was from C1 of cellulose, while 72.89–75.38 ppm was from hemicellulose and cellulose. In comparison with the raw PH, the signal intensity of crystalline cellulose (89.23 and 65.06 ppm) increased while the amorphous cellulose (84.00 and 63.20 ppm) decreased. This might be attributed to the presence of water in the pretreatment procedure, leading to the thermodynamic instability of amorphous cellulose and its partial conversion into crystalline cellulose . Furthermore, Wei et al. have reported that during the pretreatment process with H3PO4, the amorphous cellulose would be hydrolyzed when the temperature was above 50 °C . Therefore, the cellulose crystallinity of pretreated PH increased with the pretreatment method conducted in this study. In addition, compared with raw PH, the signals of hemicellulose and lignin disappeared for pretreated PH, which indicated that the hemicellulose and lignin of PH have been degraded during the pretreatment process. Remarkably, the aromatic and methoxyl (OMe) groups of lignin at 130–155 and 56.11 ppm, respectively, were not detected in the pretreated PH either. All the results demonstrated that the complex network structure of the lignocellulose was destroyed through removing hemicellulose and lignin. Thus, it was deduced that the dramatic structure changes resulted from removal of hemicellulose and lignin might be the primary reason for the higher enzymatic hydrolysis efficiency and utilization of pretreated PH.
Cellulase adsorption of raw and pretreated PH
Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was conducted to detect the difference of cellulase adsorption before and after pretreating PH vividly. Previous studies reported that cellulase from Trichoderma reesei was a complex enzyme including at least five components (CBH I, CBH II, EG I, EG II, and EG III). Among them, CBH I, EG I, and EG II were regarded as the main components of cellulase [54,55,56]. In this research, four components of cellulase, CBH I, CBH II, EG I, and EG II, were detected to explore the cellulase adsorption of raw and pretreated PH. As exhibited in Fig. 5, the bands of two cellobiohydrolases (CBH I, 66 kDa and CBH II, 58 kDa) and two endoglucanases (EG I, 54 kDa and EG II, 48 KDa) in SDS-PAGE gel corresponding to the reported molecular mass were observed [54, 56, 57]. Moreover, the bands of four components in raw PH were significantly weaker than those in pretreated PH, showing that the adsorption of CBH I, CBH II, EG I, and EG II to pretreated PH was stronger than to raw PH. As reported, cellulose was surrounded by hemicellulose and lignin in lignocellulosic materials . Thus, it was difficult for cellulose of raw PH to contact with cellulase efficiently. On the contrary, since nearly all the hemicellulose and lignin were removed during the pretreatment process, it was easier for cellulase to contact with and adsorb to the cellulose surface of pretreated PH.
Various inhibitors would come up during the pretreatment process , especially when acetic acid, formic acid, levulinic acid, furfural, and 5-hydroxymethylfurfural (5-HMF) were generated during dilute acid pretreatment process at a high temperature . Unfortunately, these inhibitors could restrict the growth and metabolism of E. coli by breaking down the single-strand DNA, inactivating the intracellular enzymes and reducing the intracellular pH [34, 58, 59]. In this study, after pretreated PH was hydrolyzed with cellulase, xylanase, and β-glucosidase, the enzymatic hydrolysate was analyzed using high-performance liquid chromatography (HPLC) to determine the categories and concentrations of inhibitors. Only three kinds of inhibitors, formic acid (5.4 × 10−3 mg/mL), levulinic acid (2.3 × 10−2 mg/mL), and furfural (2.8 × 10−6 mg/mL), were formed during the PH pretreatment procedure. Since the concentrations of inhibitors in the fermentation medium were much lower than the minimal inhibitory concentration (MIC) [60, 61], it was assumed that there might be no inhibition on E. coli fermentation to produce isoprene. Accordingly, lignocellulose pretreated by the combination of H3PO4 and H2O2 could be utilized to produce isoprene without removing the inhibitors with extra detoxification process.
Isoprene production by different fermentation methods
To test the efficiency of pretreated PH to produce isoprene, the engineered E. coli was utilized with separate hydrolysis and fermentation (SHF) and simultaneous saccharification and fermentation (SSF). In our previous study, the engineered strain, YJM25 (E. coli BL21™ (DE3)/pYJM21, pYJM14), was proved to be the optimal strain to produce isoprene, and the optimum glucose concentration in the fermentation medium was 3 g/L [21, 22]. Figure 6 shows that the isoprene yield by pure glucose fermentation was 298 ± 9 mg/L. 249 ± 6.7 and 294 ± 8.3 mg/L of isoprene were produced by the engineered strain with SHF and SSF methods, respectively, using the pretreated PH as the carbon source. The isoprene production via SHF and SSF had 8.3 and 9.8% glucose–isoprene conversions, equivalent to 83.5 and 98.8% of isoprene production via pure glucose fermentation, respectively. As the results showed, the glucose–isoprene conversion (9.8%) were higher than that (7%) published in the previous paper . Obviously, SSF method had similar isoprene yield with pure glucose fermentation method, indicating that the inhibitors derived from pretreatment procedure had little effect on isoprene production. However, the isoprene yield of SHF method decreased compared with SSF method, which was consistent with the results shown in previous reports [62, 63]. All in all, the pretreatment strategy (H3PO4/H2O2) could be promisingly applied to pretreat the lignocellulose in isoprene production.
Overall mass balance
Ultimately, the overall mass balance for the operations including H3PO4/H2O2 pretreatment, enzymatic hydrolysis, and fermentation step was obtained. As indicated in Fig. 7, after being pretreated with H3PO4 and H2O2, the biomass loss of raw PH was about 62.5% containing about 39.5% lignin, 12.4% hemicellulose, 3.2% cellulose, and other 7.4% which were the organic solvent extractives (such as protein and fat) and ash. The remained solids of pretreated PH of 37.5%, including 32.4% cellulose, 2.2% hemicellulose, 0.8% lignin, and other 2.1% which were the organic solvent extractives (such as protein and fat) and ash, were collected. And then the remained cellulose was hydrolyzed into glucose for isoprene production. The glucose recovery was proximately 91.0%. Hence, hemicellulose and lignin were reduced by 85.0% (the remained hemicellulose 22 g, raw hemicellulose 146 g) and 98.0% (the remained lignin 8 g; raw lignin 403 g), respectively. The glucose content of pretreatment PH increased by about 2.4 times (86.4% of pretreated PH and 35.6% of raw PH) in contrast to that of the raw PH. Additionally, the glucose and xylose hydrolysis efficiency of pretreated PH were 95.0 and 94.8%, which increased 5 times and 11 times, respectively. On the other hand, compared with the isoprene yield fermented using pure glucose (298 ± 9 mg/L), 249 ± 6.7 and 294 ± 8.3 mg/L of isoprene were produced using the pretreated PH by SHF and SSF methods, respectively. The results demonstrated that the novel pretreatment method for lignocellulosic materials would have greater efficiency and advantages in isoprene production. Meanwhile, this strategy might also be widely used in other lignocellulosic materials to produce various biofuels.
In this study, high efficiency of isoprene production with PH using the newly developed pretreatment method was achieved. However, the mechanism of lignin removal for this method was unexplored because of the complexity of the lignin degradation products, which requires further study in the future.