Phenolic compound-mediated covalent crosslinking as a universal mechanism for cellulase deactivation during lignocellulose saccharification Authors: Ya-Jun Liu, Chao Chen, Shenglei Wu, Guang-Lei Liu, Yingang Feng Published: 2026 ABSTRACT AbstractEfficient enzymatic saccharification is critical for lignocellulose bioconversion, yet it is often challenged by various inhibitors in lignocellulosic biomass that suppress cellulase activity. This study identifies the phenolic compounds (PCs), particularly phenolic aldehydes, as key mediators of cellulase crosslinking, significantly impacting lignocellulosic bio-saccharification. We demonstrate that PCs, originating not only from lignin derivatives but also from raw extractives, covalently bond with protein side chains primarily via an oxygen-dependent mechanism, leading to cellulase deactivation. While PC-mediated crosslinking broadly affects various cellulases, its impact can be mitigated by removing oxygen or introducing amino group-containing competitors. This study provides insights into the origins and inhibitory effects of PCs on cellulase activity, highlights the importance of optimizing pretreatment processes to reduce residual PCs in extractives and minimize PC generation from lignin, and offers practical strategies to alleviate their adverse impacts, thereby enhancing saccharification efficiency.Subject terms: Enzymes, Chemical modification, Enzyme mechanisms Introduction Plant biomass stands out as nature’s most abundant carbon feedstock, primarily composed of cellulose, hemicellulose, lignin, and other components such as non-structural extractives and ashes1. As a valuable sustainable carbon source, lignocellulose holds significant potential for producing bio-based fuels and chemicals that can replace fossil products2,3. This potential hinges on the ability to overcome the inherent recalcitrance of lignocellulosic structure and efficiently saccharify its polysaccharide matrix into fermentable sugars4,5. Cellulases, including endoglucanases, exoglucanases, and β-glucosidases (BGLs), as well as lytic polysaccharide monooxygenases (LPMOs)6, play a critical role in lignocellulose bio-saccharification7,8. These enzymes are essential for breaking down cellulose into glucose9, the most widely used monosugar carbon source for industrial fermentation processes. However, non-cellulosic components of lignocellulose often show inhibitory effects on the stability and activity of cellulases. The primary hemicellulose component, xylan, significantly affects cellulases by disrupting the cellulase-substrate interaction10. Xylan-derived xylo-oligosaccharides act as competitive inhibitors of cellulases by structurally mimicking cello-oligosaccharides11. Our previous study demonstrated that water-dissolved xylan can inhibit bacterial multi-enzyme complex cellulosomes12–14 by interacting with key cellulosomal exoglucanase components15. Lignin is known to inhibit cellulase activity via multiple mechanisms, although positive effects on enzymatic hydrolysis have also been suggested16. Macromolecule lignin inhibits enzymatic hydrolysis by introducing non-productive enzyme adsorption through hydrophobic, electrostatic, and hydrogen bonding interactions with enzymes17–19 and serving as a physical barrier that hinders enzymatic processing20. Water-soluble low molecular-weight lignin, rich in phenolic compounds (PCs), significantly influences enzymatic saccharification by causing cellulase precipitation. Unlike macromolecule lignin, this inhibitory effect cannot be significantly mitigated by increasing enzyme load or adding blocking proteins19,21,22. Previous studies have demonstrated that PCs can be oxidized to form semiquinones and quinones, which covalently interact with protein side chains containing free amino or thiol groups23–25. However, studies also suggested their potential to enhance cellulase activity under certain conditions26,27 and interact with LPMOs to improve their cellulose-degrading capacity28,29. PCs primarily originate from lignin during the pretreatment process, but are also naturally present in raw extractives of plant biomass as essential components of plant secondary metabolites24,26. The removal of raw extractives is generally associated with an enhancement in cellulose hydrolysis efficiency30–32, suggesting that these extractives interfere with the enzymatic saccharification process. Conversely, water-extractable compounds have been suggested to play promotional roles in the enzymatic hydrolysis of lignocellulose26,33. Among cellulases, BGL proteins are critical for mitigating pronounced feedback inhibition in both fungal and bacterial systems34,35. Previously, we developed whole-cell biocatalysts for lignocellulose saccharification by introducing a BGL protein from Caldicellulosiruptor sp. F32 (CaBglA) into Clostridium thermocellum34,36,37. C. thermocellum naturally produces extracellular multi-enzyme cellulosomes for efficient lignocellulose degradation38. In this work, we observe that the extracellularly produced BGL protein disappeared during the saccharification of pretreated wheat straw, thereby directly reducing saccharification efficiency. We further demonstrate that PCs, originating not only from lignin but also from raw extractives, contribute to the “disappearance” of CaBglA. Moreover, similar phenomena are observed across multiple cellulases, suggesting a universal inhibitory mechanism during enzymatic lignocellulose biosaccharification. Results CaBglA is unstable during lignocellulose consolidated bio-saccharification Previously engineered C. thermocellum-derived 2nd generation whole-cell biocatalyst ∆pyrF::Cel9K-CaBglA (WCB2)36 and 3rd generation whole-cell biocatalyst ∆pyrF::P2638-CaBglA (WCB3)34 (Supplementary Table S1) are designed to produce extracellular CaBglA proteins. WCB2 produces CaBglA as a fusion protein with the cellulosomal Cel9K36. WCB3 expressed CaBglA as a standalone protein34 (Supplementary Table S1, Supplementary Fig. S1). Both WCB2 and WCB3 demonstrated improved saccharification efficiency with Avicel as the substrate compared to the parent strain ∆pyrF34. However, during the saccharification of sulfite-pretreated wheat straw (SP) containing 59.37% of cellulose, 18.43% of xylan, and 9.97% of lignin, neither WCB2 nor WCB3 exhibited any advantage in reducing sugar production (Supplementary Fig. S2). Furthermore, compared to Avicel saccharification, WCB3 exhibited lower sugar yield (g/gsubstrate) during SP saccharification and greater cellobiose accumulation in SP hydrolysate (Fig. 1A, B), indicating CaBglA inactivation. The CaBglA protein exhibits instability during saccharification using SP as the substrate. A Relative sugar yields of WCB3 saccharification on SP and Avicel calculated by normalizing the supplemented substrate (40 g/L) to the amount of reducing sugar produced. The 3,5-dinitrosalicylic acid (DNS) method was used to quantify reducing sugar. B Relative contents of cellobiose and glucose in hydrolysates after 10-day saccharifications (as described in A) determined via HPLC analysis. Data are presented as mean values ± SD in (A) and (B) (n = 4 independent experiments). C SDS-PAGE analysis of SP and Avicel hydrolysates produced by WCB3. The result indicated the protein instability during SP saccharification. Results are representative of three independent replicates. D SDS-PAGE analysis of rCaBglA or BSA proteins incubated in SP and Avicel hydrolysates produced by WCB2. The rCaBglA protein was expressed from BL21(DE3)::pET28aNS-rCaBglA. All hydrolysates were collected after 3 days of saccharification. The protein incubation lasted for 3 days before analysis. Results are representative of three independent replicates in (C) and (D). Source data are provided as a Source Data file. In order to understand the fate of CaBglA produced by C. thermocellum during saccharification, we first chose the free CaBglA producer WCB3 and performed an SDS-PAGE analysis of its extracellular proteins after 3-day saccharification. Surprisingly, the extracellularly expressed CaBglA proteins (~53.4 kDa) were clearly detected in Avicel-derived hydrolysate and were rarely detectable in SP-derived hydrolysate (Fig. 1C). We further conducted incubations using recombinant CaBglA protein (rCaBglA) purified from Escherichia coli and Avicel- and SP-derived hydrolysates generated via WCB2 saccharification. Because WCB2 produced CaBglA in the form of a fusion protein with Cel9K, the bands of CaBglA-Cel9K (~150 kDa) and rCaBglA (~53.4 kDa) on SDS-PAGE do not overlap, and the CaBglA-Cel9K will not interfere with the observation of rCaBglA. As shown in Fig. 1D, rCaBglA remained detectable in Avicel-derived hydrolysate but was undetectable after incubation with SP-derived hydrolysate. BSA used as the control was stable and showed no detectable change post incubation. These results confirmed the instability of CaBglA during saccharification on pretreated biomass. CaBglA underwent covalent crosslinking during saccharification The depletion of CaBglA could arise from substrate adsorption, protein degradation, or covalent bonding. To investigate potential cellulosic adsorption, we incubated rCaBglA with Avicel and SP substrates for 3 days. No cellulosic hydrolysis could occur under such conditions due to the lack of cellulosomes. Subsequent separation into supernatants and pellets revealed that the majority of added CaBglA was detected in the supernatant rather than the pellet (Fig. 2A), suggesting that, unlike cellulosomes that strongly interact with cellulosic substrates39, CaBglA exhibits minimal or no interaction with cellulosic substrates. This result also indicated that CaBglA remained intact and active in solution when it was mixed with non-degraded cellulosic substrates. CaBglA is covalently crosslinked during saccharification. A SDS-PAGE analysis of rCaBglA incubated with cellulosic substrates in PBS buffer for 3 days. Different from Fig. 1C, cellulosomes are absent in this experiment. Both the supernatant and pellet fractions of the incubation mixtures were analyzed. Control, rCaBglA in GS-2 buffer. B SDS-PAGE analysis of rCaBglA following incubation with SP hydrolysate treated with EDTA or protease inhibitors. The hydrolysate was generated by WCB2. The sample at time 0 was analyzed as a control. Results are representative of three independent replicates in (A) and (B). C SDS-PAGE analysis of rCaBglA incubated with GS-2 buffer (Control) or SP- and Avicel-derived hydrolysates pre-treated by protease K. rCaBglA shown in (A) and (C) was expressed from BL21(DE3)::pET30a-rCaBglA, while the protein used in (B) was from BL21(DE3)::pET28aNS-rCaBglA. The bands referring to rCaBglA monomer and dimer are indicated by black and red arrows, respectively. Results are representative of two independent experiments. D BGL assay of incubation mixtures described in (C), with experiments conducted in triplicate independently, and the data are presented as mean values ± SD. Statistical analysis was performed using a two-sided Student’s t test, with significance defined as P < 0.05. Source data are provided as a Source Data file. To explore proteolytic hydrolysis as a potential factor, we incubated rCaBglA with SP hydrolysates treated with or without EDTA or protease inhibitors. The hydrolysates were generated through WCB2 saccharification. However, significant depletion of CaBglA was still observed (Fig. 2B), suggesting that proteolytic hydrolysis did not contribute to CaBglA instability during the saccharification process. To eliminate interference from endogenous proteins of C. thermocellum during SDS-PAGE analysis, we treated the hydrolysates with protease K to digest most WCB2-derived proteins. Protease K digestion was confirmed to be effective (Supplementary Fig. S3). After treatment, Avicel- and SP-hydrolysates were incubated with rCaBglA for 3 days. Only the SP hydrolysate led to a significant reduction in the CaBglA band intensity, accompanied by an additional band over 100 kDa (Fig. 2C). Mass spectrometry identified this band as a CaBglA dimer (Supplementary Fig. S4). Additionally, CaBglA activity was significantly reduced to 56.7% after incubation with SP hydrolysate compared to Avicel hydrolysate (Fig. 2D). Since the SDS-PAGE analysis disrupts non-covalent interactions, the observed formation of a CaBglA dimer indicated covalent crosslinking. Furthermore, the factors responsible for this crosslinking originate from non-cellulosic components in the hydrolysates. Given the pivotal role that BGLs play in alleviating cellobiose inhibition on the cellulosome34, the reduced SP saccharification efficiency observed in Supplementary Fig. S2 and Fig. 1 was likely due to the CaBglA crosslinking to a significant extent. The crosslinking factors are from lignin derivatives and extractives We performed alkaline pretreatments of rice straw (R) and wheat straw (W) under severe (160 °C, 1 h) or mild (60 °C, 2 h) conditions (Fig. 3A). The chemical compositions of these substrates were analyzed both pre- and post-pretreatment (R160, R60, W160, W60, WN, RN). Pretreated samples at severe conditions (R160 and W160) promote substantial lignin degradation, primarily via cleavage of acid-labile β-O-4 ether linkages accelerated by combined heat and alkali, while the mild alkali concentration (0.1 wt%) is insufficient to induce polymorphic conversion40,41. After water washing, the R160 and W160 samples exhibited reduced lignin and extractive contents compared to other samples, indicating effective lignin removal under such conditions. Additionally, extractives in R60 and W60 were greatly reduced compared to untreated raw material (Fig. 3B). CaBglA crosslinking is attributed to lignin derivatives and water-soluble extractives released into the saccharification system. A Schematic representation of the experimental workflow. Rice straw (R) and wheat straw (W) were subjected to alkaline pretreatment at 160 °C or 60 °C. Pretreated substrates were labeled as R160, R60, W160, and W60, while untreated substrates were designated RN and WN. All treated or untreated straw samples were analyzed for chemical composition and used as substrates for whole-cell saccharification using strain WCB2. Saccharification was conducted for 3 days, producing hydrolysate I. Straw samples were also subjected to water extraction to generate corresponding water extractives. The remaining solids underwent further 3-day saccharification by WCB2, yielding hydrolysate II. Hydrolysate I, water extractives, and hydrolysate II were used for subsequent incubation with rCaBglA, followed by SDS-PAGE and BGL assays. Hydrolysate I and II were treated with protease K prior to incubation. B Chemical composition analysis of treated or untreated straw samples. SDS-PAGE analysis (C) and relative BGL activity (D) after incubation with hydrolysate I. The relative quantities of CaBglA monomer and dimer bands, which correspond to the images in (C), are shown below the figure. E BGL activity after incubation with acid-soluble lignin solution and beechwood xylan solutions (xylan-1 and xylan-2 are from Sigma-Aldrich Co. Ltd. and Megazyme Ltd, respectively). SDS-PAGE analysis (F) and relative BGL activity (G) after incubation with water extractives. SDS-PAGE analysis (H) and relative BGL activity (I) after incubation with hydrolysate II. For (E–G), rCaBglA incubated in GS-2 buffer without lignin, xylan, or water extractives was determined as the control. For (C, D, H, I), rCaBglA incubated with Avicel hydrolysate was set to 100% for calculation. Data are presented as mean values ± SD in (B, D, G, I) (n = 3 independent experiments). Data in (E) are from three independent experiments. SDS-PAGE results are representative of two independent experiments in (C, F, H). Statistical analysis was performed using a two-sided Student’s t test in (D, G, I), with significance defined as P < 0.05. Source data are provided as a Source Data file. Subsequently, we conducted consolidated bio-saccharification (CBS) processes of all six substrates using WCB2. The resulting hydrolysate I underwent protease K treatment and was subsequently used for 3-day incubations with rCaBglA (Fig. 3A). Crosslinking was observed in all six conditions, but varied in extent (Fig. 3C). Incubation with hydrolysates from WN, RN, and R60 resulted in a significant increase in dimer formation, demonstrating pronounced crosslinking effects. In contrast, pretreatment at 160 °C led to weaker crosslinking. Consistent results were observed in enzyme activity assays: incubation with hydrolysates from substrates pretreated at 160 °C resulted in less inhibition of CaBglA activity (Fig. 3D). Combined with the chemical composition analysis, which showed that 160 °C pretreatment was more effective in lignin removal compared to 60 °C, we attribute the crosslinking factors to lignin derivatives and extractives. To investigate the role of lignin derivatives in CaBglA crosslinking, rCaBglA was incubated with acid-soluble lignin solution and two types of commercial beechwood xylan (Xylan-1 and Xylan-2). A distinct dimer band for CaBglA was observed in all mixtures, with the intensity of the band increasing over time during incubation (Supplementary Fig. S5A). Notably, the acid-soluble lignin solution appeared deep red, while the Xylan-2 solution showed a yellowish color. In contrast, the Xylan-1 solution remained nearly transparent (Fig. 3E, Supplementary Fig. S5A). The total phenolic contents (TPC) of the xylan solutions were determined. The TPC in Xylan-1 was negligible (~1.76 μg/mL), whereas Xylan-2 harbored a low but detectable level of phenolic compounds (~25 μg/mL) (Supplementary Fig. S5B). Correspondingly, incubation with Xylan-1 did not influence the CaBglA activity, whereas both lignin and Xylan-2 caused significant reductions in enzyme activity (Fig. 3E). Both the TPC analysis and the yellowish color of the xylan solution indicated the possible presence of residual lignin in the commercial product, suggesting that CaBglA crosslinking and decreased activity were caused by the lignin derivatives. After alkaline pretreatment, all six substrates underwent water extraction to isolate water-soluble extractives. The color of the extractives gradually lightened with increasing severity of the pretreatment conditions (Fig. 3A). Upon incubation with rCaBglA, we observed significant CaBglA dimer bands, accompanied by a substantial decrease in BGL activity by up to ~50% (Fig. 3F, G). Notably, RN and WN, which did not undergo alkaline pretreatment, also showed severe inhibition of CaBglA activity when water extractives were used. This observation suggested that, in addition to lignin derivatives generated during pretreatment, water-soluble extractives inherently contain compounds capable of inducing CaBglA crosslinking and subsequent loss of activity. Furthermore, the R160-derived extractives showed the least enzyme inhibition among all samples, indicating that severe pretreatment conditions effectively removed most of the extractives from rice straw. The severe pretreatment generated more degraded lignin and extractives, which were finally washed out from the substrates into black liquor. After water extraction, the remaining straw substrates were subjected to whole-cell saccharification using WCB2 to generate hydrolysate II (Fig. 3A). Incubation of rCaBglA with hydrolysate II did not result in detectable CaBglA crosslinking or inhibition of BGL activity (Fig. 3H, I). However, hydrolysates derived from RN and WN, which underwent only water extraction but not alkaline pretreatment, still caused significant CaBglA crosslinking inhibition. These findings further suggested that the combination of alkaline pretreatment and water extraction effectively removed nearly all inhibitors responsible for CaBglA crosslinking. Cellulase crosslinking is ubiquitous BGLs have been proven to play key roles in enhancing lignocellulose saccharification efficiency. CaBglA has been employed to construct whole-cell biocatalysts for CBS34,36,37. In addition to CaBglA, we selected other BGLs and cellulases from bacterial or fungal strains and determined the widespread occurrence of the crosslinking phenomenon. As shown in Fig. 4, all tested proteins exhibited smear bands and reduced activity after incubation with acid-soluble lignin solution for 2 h or 3 days, but with varying degrees of tolerance to acid-soluble lignin. Crosslinking of various cellulases from bacteria and fungi. A SDS-PAGE analysis of selected BGL proteins (CaBglA, CglT, TnBGL, TtBGL, Cel1A) and exo/endocellulases (Cel48S, Cel9K, Cel9D, Cel5A, Cel7B) following incubation in GS-2 buffer (B) or acid-soluble lignin solution (L) for 2 h or 3 days. Fungi-derived cellulases (Cel1A, Cel5A, Cel7B) are labeled in blue. Results are representative of two independent experiments. B Relative enzyme activity after the indicated incubations. Enzyme activity following incubation in GS-2 buffer without the addition of lignin solution was set to 100% as a reference for calculation. Values above columns show calculated results. Data are presented as mean values ± SD from three independent experiments. Statistical analysis was performed using a two-sided Student’s t test, with significance defined as P < 0.05. Source data are provided as a Source Data file. The BGL protein Cel1A from Trichoderma reesei QM941442 lost nearly all activity after just 2 h of incubation. Similarly, CglT from Thermoanaerobacter brockii DSM145743 lost over 90% of its activity under the same conditions. In contrast, TnBGL from Thermotoga naphthophila44 and TtBGL from Thermotoga thermarum45, which were reported to tolerate high glucose concentration and high temperatures, remained at more than 50% enzyme activity. The fungal endoglucanases Cel5A (accession number ABA64553.1 [https://www.ncbi.nlm.nih.gov/protein/ABA64553.1]) and Cel7B (accession number AAA34212.1 [https://www.ncbi.nlm.nih.gov/protein/AAA34212.1])46 resulted in ~30% and 70% loss of activity, respectively, after 2 h of incubation. In comparison, the exoglucanase Cel48S47 and endoglucanase Cel9D48 from C. thermocellum DSM1313 showed no significant decrease in activity after 2 h of incubation. When the incubation was extended to 3 days, protein crosslinking became evident, but there were still 36.7% and 76.9% of enzyme activity remaining for Cel48S and Cel9D, respectively. These findings demonstrated the prevalent protein crosslinking in both bacterial and fungal cellulolytic systems and indicated the high resistance of Cel48S and Cel9D compared to other tested cellulases. PCs are the major crosslinking factors To understand the major factors that induce cellulase crosslinking, we conducted GC-MS analysis of water extractives from both RN and WN. PCs p-hydroxybenzaldehyde, vanillin, p-hydroxybenzoic acid, vanillic acid, coniferyl aldehyde, syringic acid, ferulic acid, and p-coumaric acid were identified. Among these compounds, p-coumaric acid displayed the highest relative content (Supplementary Table S2). Alkaline pretreatment could generally decrease the levels of the compounds. Notably, p-hydroxybenzaldehyde, p-hydroxybenzoic acid, and coniferyl aldehyde were undetectable after pretreatment. In contrast, ferulic acid showed increased concentrations in the extractives from R60 and W60 than RN and WN, respectively, indicating the release of hemicellulose-derived ferulic acid under mild pretreatment conditions. For most detected PCs, fewer remained in the substrates with more severe pretreatment as expected (Supplementary Table S2). However, the vanillin content in the water extractives of R160 and W160 remained at 45.9% and 78.3%, respectively, of that in RN and WN (Supplementary Table S2), indicating that, of the detected compounds, water-extractable vanillin was particularly persistent to alkaline pretreatment. The residual vanillin could gradually be released into the saccharification system, contributing to enzyme instability. Additionally, high levels of vanillin and vanillic acid were detected, while p-hydroxybenzaldehyde, ferulic acid, and p-coumaric acid, identified in the water-soluble extractives of the straws, were not detected in acid-soluble lignin (Supplementary Table S3). With CaBglA as an example, the effects of the eight detected PCs on cellulase crosslinking were subsequently investigated via incubation experiments. All tested PCs resulted in a potential CaBglA dimer band based on SDS-PAGE analysis (Fig. 5A). Vanillin, p-hydroxybenzaldehyde, and coniferyl aldehyde led to more pronounced covalent interactions: rCaBglA exhibited blurry bands around 50 kDa and multiple high-molecular-weight bands (>100 kDa), indicating both modification of monomer protein and formation of polymeric proteins. The BGL activity assay further suggested that compared to phenolic acids, phenolic aldehydes had a stronger impact on CaBglA crosslinking and enzyme deactivation (Fig. 5B). It is noteworthy that the collective contribution of diverse PCs, including acids and aldehydes, present in hydrolysates and extractives must be emphasized due to their concentration-related inhibition of enzymatic activity. PCs in water extractives resulted in CaBglA crosslinking. A SDS-PAGE analysis of rCaBglA incubated with selected PCs. The band of about 50 kDa indicates the monomer protein. Multiple bands of over 100 kDa indicate the formation of polymeric proteins. Results are representative of two independent experiments. B Relative BGL activity of CaBglA after the incubation with PCs. Corresponding PC molecule structures are shown on the right. C Total phenolic content analysis. D Free amino content analysis. E Free thiol group analysis. Control, incubation with PBS buffer was set to 100% for calculation. Three independent replicates were conducted for every experiment, and the data are presented as mean values ± SD in (B–E). Statistical analysis was performed using a two-sided Student’s t test in (C–E), with significance defined as P < 0.05. Source data are provided as a Source Data file. Previous studies have demonstrated that PCs can oxidize to form quinones, which covalently interact with protein side chains containing free amino or thiol groups23–25. As depicted in Fig. 5C, the total phenolic content of rCaBglA increased by 1.5, 1.9, and 7.5-fold after incubation with p-hydroxybenzaldehyde, coniferyl aldehyde, and vanillin, respectively. Significant decreases in free amino groups (Fig. 5D) and thiol groups (Fig. 5E) were also observed in rCaBglA following these treatments. Notably, vanillin exhibited the most substantial effect on protein modification, aligning with its strongest inhibition of CaBglA activity (Fig. 5B). Circular dichroism (CD) spectra analysis revealed that phenolic aldehydes had minimal impact on the secondary structure of CaBglA (Supplementary Fig. S6). The rCaBglA protein incubated with p-hydroxybenzaldehyde was further analyzed using LC-MS/MS (Supplementary Fig. S7A). The analysis revealed the modifications at some lysine (K) and cysteine (C) residues of the proteins (Supplementary Data 1, Supplementary Fig. S7B), indicating that CaBglA crosslinking occurs via PC-mediated conjugation to free amino and thiol groups on these residues (Fig. 6), respectively, consistent with previous reports23–25. Notably, crosslinking was detected in both the dimer and monomer bands of rCaBglA on SDS-PAGE, suggesting two distinct modes of conjugation: one mediated by PCs between separate protein molecules, and another involving intramolecular conjugation between PCs and individual protein molecules. The monomer band exhibited a more extensive range of crosslinking modifications (Supplementary Fig. S7B) and constituted the dominant fraction in the sample (Supplementary Fig. S7A). This implies that crosslinking between PCs and individual proteins, rather than protein-protein polymerization, is likely the major factor contributing to the loss of enzymatic activity. Schematic representation of linkage formation between p-hydroxybenzaldehyde and free amino/ thiol groups in proteins. p-Hydroxybenzaldehyde is abbreviated as PHBA. The symbol ‘……’ represents the reaction process where cysteine (C) undergoes the ‘PHBA-mono’ type reaction with PHBA, followed by the subsequent ‘PHBA-linker’ and ‘PHBA-loop’ type reactions, which are the same as those involving lysine (K). PC-mediated crosslinking could be alleviated PCs have been considered to crosslink with protein through the oxidation to quinones23. To investigate the role of oxygen in cellulase crosslinking, we incubated rCaBglA with vanillin under aerobic and anaerobic conditions, with the addition of reducing agents (L-cysteine, Na2S) to further minimize oxidation. Under aerobic conditions, the rCaBglA protein exhibited a blurry band after 1-day incubation (Fig. 7A). After 2 days, the protein smear became more apparent, and a distinct dimer band was observed (Fig. 7A). BGL activity assay revealed that over 90% of enzyme activity was lost even after just 1 day of incubation (Fig. 7B). In contrast, no significant changes were detected under anaerobic conditions based on SDS-PAGE analysis. After 2 days of incubation, ~60% of enzyme activity remained. The addition of L-cysteine further protected the enzyme from PC-mediated inhibition. However, Na2S completely abolished enzyme activity, despite showing no crosslinking, indicating its strong inhibitory effect on CaBglA. These results indicated that excluding oxygen significantly alleviated PC-mediated crosslinking and preserved enzymatic activity. CaBglA crosslinking can be alleviated by oxygen removal and the addition of interaction competitors. A SDS-PAGE analysis of rCaBglA incubated with vanillin (+) in combination with L-lysine (Lys), ethylamine (EA), Tween 20 (T20), L-cysteine (Cys), and Na2S. Incubation was performed for 1 or 2 days. A control experiment with rCaBglA in PBS buffer without additives under aerobic conditions was included as a reference. Black and red arrows indicate the monomer and dimer bands, respectively. Results are representative of two independent experiments. B Relative BGL activity after incubation. The activity of rCaBglA in the control condition was set to 100% for calculation. Data are shown from three independent experiments. Source data are provided as a Source Data file. To investigate the role of amino-containing compounds in mitigating PC effects, L-lysine and ethylamine were added to the incubation system. Ethylamine proved particularly effective, as it protected rCaBglA such that 70% enzyme activity remained after 2 days of incubation (Fig. 7B). This confirmed that PCs induced protein crosslinking primarily through lysine modification and indicated that supplementing amino-containing compounds as competitive inhibitors could alleviate PC-mediated crosslinking. This inhibitory relief mechanism differs from the conventional approach of increasing enzyme load or adding blocking proteins, which function primarily by reducing non-productive adsorption (non-covalent interactions) between the enzyme and inhibitors such as lignin. The hydrophobicity of PCs has been proposed to contribute to their interaction with enzymes19. To explore this, we added the surfactant Tween 20 to the incubation system. Nevertheless, Tween 20 unexpectedly enhanced rCaBglA crosslinking and caused significant enzyme deactivation (Fig. 7). Previous reported enhancement of Tween 20 on cellulase activities was believed to be caused by blocking non-productive adsorption49,50, which is not the case for rCaBglA because it does not bind to cellulosic substrates (Fig. 2A). Instead, the rich hydroxyl groups in Tween20 may promote the cross-linking reaction and subsequent rCaBglA deactivation. Discussion Phenolic compounds (PCs) are known to significantly reduce enzymatic efficiency during lignocellulose saccharification22,51. Understanding the PC-mediated inhibition mechanism is crucial for enhancing enzymatic hydrolytic efficiency. Taking the BGL protein CaBglA as a model, we demonstrated that various PCs, particularly phenolic aldehydes, induce inactivation of cellulases primarily through covalent bonding. Most importantly, such PC-mediated inhibition was suggested to have a broad impact on lignocellulose bio-saccharification as the crosslinking was detected for BGLs, endoglucanases, and exoglucanases derived from bacterial and fungal strains (Fig. 4). Hydrogen bonding and hydrophobic interactions have been suggested as key factors in PC-mediated cellulase precipitation and inhibition19,52. Additionally, the formation of covalent bonds via the oxidation of phenolic compounds into semiquinones and quinones23 has been proposed as another key mechanism underlying PC-mediated enzyme inhibition52. Our findings demonstrate that oxygen availability plays a critical role in promoting the formation of PC-protein cross-links (Fig. 7). Dos Santos et al. reported the enhancement of nitrogen atmosphere on lignocellulose hydrolysis53. This further underscores the importance of maintaining anaerobic conditions during lignocellulosic saccharification to minimize enzyme deactivation caused by PC-protein interactions. Previous studies have identified lysine and cysteine residues as the primary interaction sites for PC-mediated covalent bonding under aerobic conditions24. In the case of CaBglA, which contains 31 lysine residues but only 2 cysteine residues, amino-mediated crosslinking was the dominant mechanism of PC-mediated crosslinking inactivation. This conclusion is supported by competition experiments using L-lysine and ethylamine to mitigate PC-mediated inhibition (Fig. 7) involving amino and thiol groups on protein side chains. It is noteworthy that, besides cysteine and lysine, other amino acid residues containing free amino and thiol groups could theoretically participate in cross-links. As lysine and cysteine are the most reactive nucleophiles under typical cross-linking conditions24, we primarily investigated the lysine and cysteine modifications as representatives to understand the PC-cellulase crosslinking during lignocellulose saccharification. Although a strong correlation between covalent cross-linking and the decrease of enzyme activity has been established, it should be noted that not all cross‑linking events necessarily lead to activity reduction. Modifications at some surface residues may not critically disrupt enzyme structure or function. In addition, the observed activity loss cannot be attributed exclusively to covalent cross-linking, as non-covalent interactions between PCs and the protein may also contribute to inhibition, a distinction not fully resolved in the present study. Future studies should therefore focus on identifying the specific residues whose cross-linking is most detrimental to activity, and on quantifying the relative contributions of covalent versus non-covalent interactions to the overall inhibitory effect, which would provide deeper mechanistic insights. Even under anaerobic conditions, we observed a significant loss of CaBglA activity exceeding 30% upon incubation with vanillin. The addition of L-cysteine provided additional protection to the enzyme from PC-mediated deactivation. However, it remains unclear whether this protective effect arises from the reductive properties of L-cysteine’s thiol group or competitive interactions via its free amino group. While PC-mediated protein crosslinking has been determined as an oxygen-dependent process54, our findings also suggest the possibility of additional covalent bonding mechanisms, such as Schiff-base formation55, that are independent of oxygen. It should be noted that the method used in this study to alleviate phenolic compound inhibition was primarily employed to demonstrate the mechanism of cross-linking inhibition. Its feasibility under high substrate loading conditions in practice requires further investigation. Additionally, given the stringent economic requirements of lignocellulosic conversion processes, the use of reagents such as lysine in our study lacks economic viability for industrial production. However, our research highlights the importance of addressing the removal of phenolic compounds during substrate pretreatment in various lignocellulosic bioconversion processes, as well as their impact on enzymes during conversion. More effective removal of phenolic compounds at the source, or engineering enzymes and strains to enhance their tolerance, may represent more promising strategies for mitigating inhibition compared to the addition of chemical reagents. In addition, our findings confirm that PCs originate not only from lignin derivatives generated during pretreatment56 but are also naturally present as raw extractives in lignocellulosic biomass, consistent with the previous reports26,33. While pretreatment is known to produce toxic lignin-derived phenolics that can impair enzymatic hydrolysis19,57,58, it also serves an important function in removing phenolic substances from the lignocellulose matrix (Fig. 3). It is noteworthy that the reductive phenolic aldehydes (e.g., vanillin) serve as electron donors for LPMOs to drive cellulose oxidation, during which PCs are oxidized to quinones23,24, which intermediate cellulase crosslinking25. Given the essential role of LPMOs in cellulose saccharification systems6, the involvement of LPMOs may consume PCs and partially alleviate phenolic inhibition on cellulases. However, potential inhibitory byproducts and kinetic constraints necessitate further optimization of enzyme cocktails for efficient saccharification. It is generally considered that the presence of cellulose protects cellulases from inactivation due to substrate binding59. We tested three cellulosomal enzymes (Cel9D, Cel9K, and Cel48S) by incubating them with Avicel first and adding acid-soluble lignin solution subsequently. The protein crosslinking was then investigated by SDS-PAGE analysis (Supplementary Fig. S8). The result revealed crosslinking effects on both supernatant (unbound) and pellet (substrate-bound) fractions of the mixtures containing lignin, suggesting that substrate binding cannot protect the selected cellulases from crosslinking to a great extent. Among all tested cellulases, Cel48S and Cel9D from C. thermocellum exhibited particularly high resistance to PCs (Fig. 4). These findings suggest that selecting enzymes inherently resistant to PCs could be a promising strategy for improving saccharification efficiency. Furthermore, comparative analyses of the structures of enzymes with differing levels of PC tolerance may provide valuable insights for engineered improvements in enzyme stability and performance during lignocellulosic saccharification. Our study highlights the widespread occurrence of PC-mediated covalent crosslinking as a key mechanism of enzyme deactivation and provides practical solutions to address this challenge. These include: optimizing the intensity of pretreatment, conducting saccharification under anaerobic conditions, incorporating competitive inhibitors (e.g., amino-containing compounds) to block crosslink formation, and developing robust enzymatic biocatalysts through screening or engineering for industrial use60. By addressing both pretreatment and enzymatic factors, we can significantly enhance the efficiency and economic viability of lignocellulose bioconversion. Methods Bacterial strains and chemicals The strains used in this study are listed in Supplementary Table S1. E. coli strains were cultivated at 37 °C in Luria Bertani (LB) liquid medium or on a solid LB medium containing 1.5% agar. C. thermocellum strains were cultivated at 55 °C with shaking at 200 rpm in GS-2 medium15 containing 5 g/L cellobiose or Avicel (microcrystalline cellulose) as the carbon source. Chloramphenicol (34 μg/mL), kanamycin (50 μg/mL), ampicillin (100 μg/mL), tetracycline (10 μg/mL), and thiamphenicol (Tm) (3 μg/mL) were supplemented to the medium when necessary. All the chemicals used in this study were of analytical grade. Gallic acid, ferulic acid, p-coumaric acid, p-hydroxybenzaldehyde, vanillin, vanillic acid, p-hydroxybenzoic acid, coniferyl aldehyde, syringic acid, caffeic acid, trimethylchlorosilane, and N, O-bis(trimethylsilyl)-trifluoroacetamide (BTTA) were purchased from Sigma-Aldrich Co. Ltd (St. Louis, MO, USA). Ethylamine, L-lysine, L-cysteine, and Na2S were purchased from Sinopharm Chemical Reagent Co., Ltd., while Tween-20 was obtained from Solarbio. Folin & Ciocalteu’s phenol reagent, 5,5’-Dithio-bis (2-nitrobenzoic acid) (DTNB), Di-Na-tetraborate decahydrate, sodium dodecyl sulfate (SDS), o-phthaldialdehyde (OPA), and dithiothreitol were obtained from Macklin Bio Inc. (Shanghai, China). Protease K and bovine serum albumin (BSA) were purchased from Tiangen Biotech (Beijing, China). Two types of beechwood xylan were purchased from Sigma-Aldrich Co. Ltd (St. Louis, MO, USA) and Megazyme Ltd (Ireland). Acid-soluble lignin was a gift from Shandong Sun Holdings Group Co., Ltd. The lignin was dissolved in a PBS buffer at pH 6.5, and the solution was adjusted to pH 6.5 before further experiments. Lignocellulose pretreatment and composition analysis Wheat straw and rice straw were collected from farmlands in Shouguang, Shandong Province, China, and were milled and sieved to obtain particles ranging from 20 to 80 mesh. The particles were thoroughly mixed and dried in an oven at 85 °C for 24 h before pretreatment. Sulfite pretreatment was performed to treat wheat straw using 20 wt% ammonium sulfite at 160 °C for 2 h (SP). Alkaline pretreatment was conducted for both wheat straw and rice straw using 0.5 wt% NaOH at 60 °C for 2 h or using 0.1 wt% NaOH at 160 °C for 1 h. The pretreated samples were washed with tap water to reach a neutral pH. In terms of composition analysis of raw and pretreated samples, the contents of extractives were first analyzed through sequential water extraction and ethanol extraction61. In brief, the samples underwent water extraction and ethanol extraction sequentially in a Soxhlet apparatus. The extracted non-structural components are dried to constant weight and quantified gravimetrically. Afterwards, the samples were subjected to a two-stage sulfuric acid hydrolysis to determine cellulose, hemicellulose, and lignin contents following the NREL (National Renewable Energy Laboratory) procedure62. This procedure involves primary hydrolysis with 72% (w/w) sulfuric acid at 30 °C for 1 h, followed by secondary hydrolysis at 121 °C for 1 h after dilution to 4% acid concentration, liberating monomeric sugars from structural carbohydrates. The resulting hydrolysates and saccharification filtrates were tested by Agilent 1200 high-performance liquid chromatography (HPLC) equipped with a Bio-Rad Aminex HPX-87H column and a refractive index detector (RID, Agilent 1260). The operating temperature was 55 °C with 8 mM sulfuric acid as a mobile phase with a flow rate of 0.6 mL/min. All experiments were carried out in three independent parallel experiments. Water-soluble extractives preparation 2.5 g of pretreated or untreated substrates were added to a Soxhlet extractor (Shuniu Glass Instrument Co., Sichuan, China) containing 200 mL of deionized water. The extraction process lasted for 10 h with a siphoning rate of 4–5 cycles per hour. The obtained extractives were concentrated to a final volume of 25 mL using an IKA RV10 rotary evaporator (IKA Works, Germany) and directly used for protein incubation. Plasmid construction The genes CabglA (accession number JX030398.1 [https://www.ncbi.nlm.nih.gov/nuccore/JX030398.1]) from Caldicellulosiruptor sp. F32, cel1A (accession number AB003110.1 [https://www.ncbi.nlm.nih.gov/nuccore/AB003110.1]) from Trichoderma reesei QM9414, and cel9D (accession number CP002416, region: 1634731-1636680 [https://www.ncbi.nlm.nih.gov/nuccore/CP002416.1?from=1634731&to=1636680]) and cel9K (accession number CP002416, region: 2111126-2113813 [https://www.ncbi.nlm.nih.gov/nuccore/CP002416.1?from=2111126&to=2113813]) from C. thermocellum DSM1313 were ligated into pET30a or pET28aNS (Supplementary Table S1) using the restriction endonuclease sites XhoⅠ and NdeⅠ or the ClonExpress® II One Step Cloning Kit (Vazyme Biotech, Nanjing, China). The primers used in this study are listed in Supplementary Table S4. Cellulase expression and purification The constructed plasmids were transformed into E. coli BL21 (DE3) or Rosetta (DE3) competent cells for protein expression. The obtained recombinant strains were grown at 37 °C in an LB medium. When the OD600 reached 0.8–1.0, the cultures were supplemented with 0.5 mM isopropyl-β-D-thiogalactopyranoside (IPTG) to induce the protein expression at 16 °C for 24 h. Afterward, the cells were harvested by centrifugation, resuspended in a buffer containing 10 mM imidazole and 500 mM NaCl, pH 8.0, lysed by ultrasonication, and centrifuged at 15,493 × g for 30 min. The supernatants were purified by Ni-NTA affinity chromatography and gel filtration using a Superdex 75 column (GE Healthcare). Cel48S purification was carried out using the engineered C. thermocellum strain ∆pyrF::Cel48S_CD-His1236. The strain was cultured for 40 h in GS-2 medium with 5 g/L Avicel (PH-101, Fluka), and the supernatant containing the His12-tagged target protein was collected. Purification was achieved sequentially through Ni²⁺-affinity chromatography, followed by gel filtration on a Superdex 200 column (GE Healthcare). The protein purity was determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Protein concentrations were determined following the Coomassie Brilliant Blue method63. Lignocellulose saccharification The CBS process was performed anaerobically at 55 °C with shaking at 170 rpm using 40 g/L of untreated and pretreated straw samples as well as Avicel (PH-101, Fluka) as the substrates and previously constructed C. thermocellum strains WCB236 or WCB334 as the biocatalysts (Supplementary Table S1). 100 μL of fermentation broth was collected per day to determine sugar production following the 3,5-dinitrosalicylic acid (DNS) method. The sugar-rich CBS hydrolysates were used for further incubation experiments. Protein incubation assay Purified BGLs and other cellulases (0.2 or 0.4 mg/mL) were incubated with acid-soluble lignin (16.7 mg/mL), xylan (16.7 mg/mL), PCs (2 g/L), water-soluble extractive, or CBS hydrolysates under aerobic or anaerobic conditions at 45 °C or 55 °C and 200 rpm for 1–3 days, by adding resazurin (1 µg/mL) as the indicator. When necessary, Tm was added at a final concentration of 10 µg/mL to avoid contamination. In general, incubations in 50 mM PBS buffer (pH 6.5) or GS-2 buffer (GS-2 medium without carbon sources, pH 5.5) were performed in parallel. BSA was employed as the control protein. Acid-soluble lignin, xylan, and PCs were dissolved in a PBS buffer, which was adjusted to pH 6.5 prior to further experiments. The precipitates, if any, were removed by centrifugation. When required, MES buffer (50 mM MES, 10 mM CaCl2, pH 6.0) was used instead of PBS buffer. CBS hydrolysates were first concentrated to obtain the supernatants, which were then pre-treated with protease K (Tiangen Biotech, Beijing, China) to minimize the interference of C. thermocellum intergenic proteins on observing incubated proteins. The protease K treatment was performed at 55 °C for 3 h and terminated at 99 °C for 15 min. To investigate the potential inhibition of phenolic-protein interactions, cysteine, lysine, Tween 20, Na2S, or ethylamine were added to the vanillin incubation system. Among these, Tween 20 was used at a final concentration of 1% (v/v), while the other chemicals were employed at molar concentrations equivalent to that of the supplemented phenolic compound vanillin. After incubation, the proteins were analyzed by SDS-PAGE and corresponding enzyme assays. The relative quantity of CaBglA monomer and dimer proteins in the samples was determined by analyzing the gel using the Quantity One Software (version 4.6.2; BioRad, Shanghai, China)64. The relative quantity calculation was performed automatically by defining the summed intensity of monomer and dimer bands equal to 100%. The proteins incubated with vanillin, p-hydroxybenzaldehyde, or coniferyl aldehyde were dialyzed overnight at 4 °C against 10 mM PBS buffer (pH 6.5) for further analyses of total phenolic content, free amino group content, free thiol group content, and circular dichroism (CD). Enzyme assay The BGL assay was conducted using p-nitrophenyl-β-D-glucopyranoside (pNPG) as the substrate in 100-μL GS-2 buffer at 55 °C for 10 min and terminated by adding 200 μL of 1 M Na2CO337. The final protein concentration was 4 μg/mL. The optical absorbance at 405 nm was monitored to determine the production of p-nitrophenol (pNP) based on a standard curve. One unit of BGL activity was defined as the amount of protein required to produce 1 µmol of pNP per minute. Cellulase activity was measured using 1% (w/v) Avicel for Cel48S, 1% (w/v) carboxymethyl cellulose-Na (CMC-Na) for Cel9D and Cel5A, 2 mM p-nitrophenyl β-D-cellobioside (pNPC) for Cel9K, and 2 mM p-nitrophenyl β-D-lactopyranoside (pNPL) for Cel7B as the substrate in 50- or 250-µL systems65–68. The reactions were initiated by adding 8, 16, or 64 mg/L purified protein into the MES buffer and maintained at 40 or 55 °C for 30 min or 12 h. The produced reducing sugars from Avicel and CMC-Na were assessed using the DNS method. The produced pNP from pNPC and pNPL was determined by monitoring the absorbance of the reaction mixtures at 405 nm. One unit of cellulase activity was defined as the amount of enzyme required to release 1 nmol (for Cel48S) or 1 µmol (for Cel9D and Cel5A) of reducing sugar (glucose equivalent) or 1 µmol of pNP per minute. GC-MS analysis The presence of PCs in extractives and acid-soluble lignin was determined by gas chromatography-mass spectrometry (TQ8040 NX, Shimadzu). In detail, the water-soluble extractives were adjusted to pH 1.0, and 1 mg/L of 4-isopropylphenol (dissolved in 1 g/L NaOH) was added as the internal standard. 4-mL mixtures were subjected to four rounds of extractions using 2 mL of ethyl acetate each. The collected ethyl acetate fractions (8 mL in total) were combined, treated with anhydrous sodium sulfate to remove any residual water, and vacuum-dried to 500 μL. Then 250-μL concentrated ethyl acetate fraction was mixed with 250-μL BTTA containing 1% trimethylchlorosilane and incubated at 70 °C for 30 min. The obtained sample was tested in split mode using a DB-5MS column (Agilent, J&W Scientific, 30 m × 0.25 mm × 0.25 μm). The analysis was performed with nitrogen as the carrier gas at a flow rate of 1 mL/min. The injector and transfer line temperature was 280 °C. The temperature program was 40 °C for 2 min, increased 6 °C/min to 300 °C, then maintained for 5 min. Ions were detected in full scan mode m/z 45–500. Three independent replicates were prepared for each experiment. LC-MS/MS analysis Selected protein bands in SDS-PAGE gel were cut and digested with trypsin for liquid chromatography-mass spectrometry analysis (Thermo Scientific Q Exactive™ HF-X) by Spectrum Biotech Co. (Suzhou, China). Briefly, the peptide mixtures were separated on the Ultimate 3000 (San Jose, Thermo Fisher) equipped with a custom C18 column (1.9 µm, 75 µm × 15 cm) at a 600 nl/min flow rate. A 75-min linear gradient was set up using 0.1% (v/v) formic acid in water as solvent A and 0.1% (v/v) formic acid in 95% ACN/H2O as solvent B: solvent B increased from 0% to 6% in 10 min, 6% to 10 % in 4 min, 10 % to 30% in 44 min, 30% to 40% in 8 min, 40% to 95% in 4 min and 95% to 6% in 5 min. For the data acquisition, we used a top 25 scan mode with an MS1 scan range of m/z 340-1100. Other parameters were as follows: MS1 and MS2 resolution was set to 120 K and 15 K; AGC for MS1 and MS2 was 3e6 and 2e4; isolation window was 1.6 Th; NCE was 27; dynamic exclusion time was 15 s. The experiment was performed once. For CaBglA identification, the MS spectra results were searched against the CaBglA protein sequence (accession number AFO70070.1 [https://www.ncbi.nlm.nih.gov/protein/AFO70070.1/]) by Beijing Protein Innovation Co. Ltd, using the Mascot (Version 2.3.01, Matrix Science) with target-decoy searching algorithm. Search type is MS/MS Ion, K, R was set as cleavage sites with a maximum miss cleavage of 2, Carbamidomethyl (C) was set as fixed modification, peptide and fragment mass tolerance is −/+15 ppm and −/+20 ppm69. For PC-CaBglA crosslinking identification, Byonic (Version 2.13, Protein Metrics, USA) was used for data analysis. The C-terminal of amino acid K, R was set as the cleavage site, allowing a maximum of 2 missed cleavages. Peptide and fragment ion tolerances were set to 10 ppm and 20 ppm, respectively. M (+15.994915 Da), N, Q (+0.984016 Da), protein N terminus (+42.010565 Da), and C (+57.021464 Da) together with crosslinking modification between K and C (1 PHBA: +104.026213 Da, 2 PHBA: +210.068074 Da) were allowed as variable modifications. A target-decoy searching algorithm was applied to ensure the lowest number of false positives. This operation allows the calculation of the false discovery rate (FDR) for each match so that all the proteins with an FDR range higher than 1% were rejected. An auto-cut of spectrum-level FDR was applied to remove low-confidence identifications. Total phenolic content (TPC) analysis The Folin-Ciocalteu method70,71 was used for TPC analysis of the BGL protein, with gallic acid as the standard. Briefly, 100 µL of protein samples were mixed with 500 µL of 0.1 M Folin reagent (pH 6.5) and incubated in the dark for 5 min. Subsequently, 600 µL of 700 mM Na2CO3 was added to the mixture, and the incubation lasted for another 2 h. The reaction solution was then used to determine the absorbance at 765 nm, which was used to calculate the TPC based on the gallic acid standard curve. Free amino group content analysis The OPA method was used to determine the free amino group content of the BGL protein25. The OPA reagent was prepared freshly by dissolving 7.62 g of sodium tetraborate decahydrate and 200 mg of sodium dodecyl sulfate (SDS) in 150 mL of deionized water. Separately, 160 mg of OPA was dissolved in 4 mL of absolute ethanol and then combined with the borate solution. Finally, 176 mg of dithiothreitol (DTT) was added, and the volume was adjusted to 200 mL with deionized water72. For the OPA method, 50 µL of protein samples were added to 500 µL of OPA reagent. The reaction lasted for 5 min at room temperature. The absorbance at 340 nm of the mixture was then determined to calculate the free amino group content based on the standard curve established using L-serine. Free thiol group content analysis The free thiol group content of the BGL protein was measured using Ellman’s reagent DTNB73,74 with modifications. The reaction mixtures containing 20 µL of BGL proteins, 20 µL of 0.1% DTNB (dissolved in 1% SDS), and 220 µL of 1% SDS were incubated at room temperature for 30 min. The absorbance at 412 nm was then measured based on the standard curve established using L-cystine. CD analysis Far-UV CD spectra of the BGL protein were recorded using a Chirascan CD spectrometer (Applied Photophysics, Leatherhead, United Kingdom)25. 0.1 mg/mL of protein samples were prepared in 10 mM phosphate buffer (pH 6.5). The CD spectra were monitored by scanning the wavelength range of 190–260 nm and were interpreted using the CDSSTR method after averaging and smoothing75,76. Triple scans were conducted for every sample. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Description of Additional Supplementary Files Supp Data 1 Reporting Summary Transparent Peer Review file Acknowledgements This work was supported by the National Key Research and Development Program of China (2023YFC3402300 to Y.F.); the National Natural Science Foundation of China (32370035 to Y.-J.L., 32470024 to Y.F.); International Partnership Program of Chinese Academy of Sciences (323GJHZ2022004MI to Y.-J.L.); the Natural Science Foundation of Shandong Province (ZR2025MS301 to C.C., ZR2023YQ026 and ZR2022ZD24 to G.L.); Qingdao Science and Technology Benefiting the People Demonstration Project (24-1-8-xdny-19-nsh to Y.-J.L.); Taishan Scholars Program (tsqn202306289 to Y.-J.L.); QIBEBT Strengthening Foundation Plan (QIBEBT/SEI/QNESL S202302 to Y.F.); QIBEBT International Cooperation Project (QIBEBT ICP202304 to Y.F.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Author contributions Conceptualization: Y.-J.L., Y.F.; Methodology: Y.-J.L., C.C., S.W.; Investigation: Y.-J.L., C.C., S.W.; Visualization: Y.-J.L., C.C., G.L., Y.F.; Funding acquisition: Y.-J.L., C.C., G.L., Y.F.; Supervision: Y.-J.L., G.L., Y.F.; Writing-original draft: Y.-J.L., C.C., S.W.; Writing-review & editing: Y.-J.L., G.L., Y.F. Peer review Peer review information Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available. Data availability All data are contained in the paper and its Supplementary Information files. All data underlying this study are available from the corresponding author upon request. The protein mass spectrometry data generated in this study have been deposited in the ProteomeXchange Consortium database (https://proteomecentral.proteomexchange.org) via the iProX partner repository (https://www.iprox.cn/page/project.html) under accession code PXD073102. Source data are provided with this paper. Competing interests A patent application has been submitted by the Qingdao Institute of Bioenergy and Bioprocess Technology with inventors Yingang Feng, Ya-Jun Liu, Chao Chen, and Shenglei Wu, partly based on the results in this paper. The method for inhibiting cellulase inactivation and its application in lignocellulose saccharification reported in this paper was covered in the patent application. The remaining authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Ya-Jun Liu, Chao Chen, Shenglei Wu. Contributor Information Ya-Jun Liu, Email: liuyj@qibebt.ac.cn. Guang-Lei Liu, Email: liugl@ouc.edu.cn. Yingang Feng, Email: fengyg@qibebt.ac.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-72972-x. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary Files Supp Data 1 Reporting Summary Transparent Peer Review file Data Availability Statement All data are contained in the paper and its Supplementary Information files. All data underlying this study are available from the corresponding author upon request. The protein mass spectrometry data generated in this study have been deposited in the ProteomeXchange Consortium database (https://proteomecentral.proteomexchange.org) via the iProX partner repository (https://www.iprox.cn/page/project.html) under accession code PXD073102. Source data are provided with this paper.