Deciphering How Clustered O‑Glycosylation Shapes Substrate-Binding Preferences in an Intrinsically Disordered Protein Region

Abstract

Post-translational modifications such as clustered O-glycosylation are frequently enriched in intrinsically disordered regions (IDRs), yet their roles in shaping molecular behavior and functional specificity remain poorly understood. Here, we addressed this challenge through an integrative, stepwise strategy that combines chemical synthesis, experimental characterization, and molecular dynamics (MD) simulations, employing the extensively O-mannosylated linker of Trichoderma reesei Family 7 cellobiohydrolase as a model IDR. Using synthetically accessible linker glycoforms bearing monomannose residues, we showed that unglycosylated and partially glycosylated linkers strongly prefer lignin over cellulose, driven by electrostatic interactions with the negatively charged lignin surface. Increasing glycan density or glycan size, such as through the introduction of dimannose, weakens lignin binding and partially shifts binding preference toward cellulose. Importantly, MD simulations informed and validated by these experimental data enabled prediction of currently synthetically inaccessible glycoforms, revealing that negatively charged glycans, such as phosphorylated mannose, effectively abolish lignin affinity while largely preserving cellulose binding. This work provides a generalizable and predictive framework for elucidating how complex glycosylation patterns regulate IDR-ligand interactions and identifies charge-mediated modulation as a key mechanism through which clustered O-glycosylation tunes IDR function.

1. Introduction

Proteins consist of diverse structural domains, including both well-defined ordered regions and intrinsically disordered regions (IDRs) (Figure ). − Although the structures and functions of many ordered domains have been extensively characterized, the study of IDRs has lagged behind despite their widespread presence in both prokaryotic and eukaryotic proteins. , This gap is particularly significant given that most IDRs are extensively modified by post-translational modifications (PTMs), , among which clustered O-linked glycosylation is one of the most complex and least understood forms. The essential roles of IDRs, including their glycosylation, in processes such as molecular recognition and functional regulation are now increasingly recognized. ,, Thus, to achieve a comprehensive understanding of protein behavior, it is critical to investigate IDRs and their clustered glycosylation. , Such an understanding requires clarifying how and why different clustered O-glycosylation patterns alter IDR functional properties, ultimately enabling the strategic use of glycosylation to enhance protein performance. ,

1.

Investigation of the impact of clustered O-glycosylation on IDR binding preference through analysis of the O-mannosylated linker domain of TrCel7A with cellulose or lignin. The domain architecture of TrCel7A includes a catalytic domain (CD), a carbohydrate-binding module (CBM), and a flexible, heavily glycosylated linker. R denotes potential attached glycans or other modification groups. The two types of O-glycans investigated in this study are highlighted in cyan.

Among the model systems for studying IDRs, the linker region of Trichoderma reesei Family 7 cellobiohydrolase (TrCel7A) has attracted particular attention. TrCel7A is a key enzyme in the cellulosic biofuels industry that hydrolyzes cellulose to release soluble cellobiose, whereas binding to lignin compromises its catalytic efficiency. , It comprises a catalytic domain (CD), a carbohydrate-binding module (CBM), and a heavily O-mannosylated intrinsically disordered linker (Figure ). , This clustered O-mannosylation has been shown to enhance the proteolytic stability of the linker region and has been implicated in regulating the binding preference between cellulose and lignin. , Despite this importance, the heterogeneous nature of previously available research samples has hindered precise evaluation. The specific effects of glycosylation patterns, including site occupancy and glycan structures, on the binding properties of the TrCel7A linker therefore remain poorly understood. − Addressing this knowledge gap requires access to homogeneously glycosylated isoforms (glycoforms). −

O-glycosylation in T. reesei features two distinct classes: common neutral glycans such as monomannose and dimannose, and negatively charged phosphorylated mannose, which is most likely phosphorylated at the 6-hydroxyl position (Figure ). Although linker glycoforms bearing neutral glycans have been shown to be accessible through chemical synthesis, those containing multiple phosphorylated glycans have remained an unsolved challenge due to the complexity of introducing such charged modifications. , To enable a comparative study of both types of glycans, we adopted an integrated experimental–computational strategy. We began by applying chemical synthesis routes to produce a series of glycosylated linker and CBM–linker glycoforms with systematically varied O-glycosylation patterns of neutral glycans. Access to these precisely defined glycoform collections allowed us, for the first time, to quantitatively evaluate how glycan composition and site-specific distribution influence cellulose and lignin binding. The experimental results provided a foundation for parametrizing and validating molecular dynamics (MD) simulations, which in turn offered a mechanistic rationale for the observed neutral glycan-mediated binding preferences. The validated computational model exhibited high predictive accuracy and was subsequently applied to phosphorylated mannose-containing glycoforms, revealing the remarkable ability of phosphorylated mannose to abolish lignin binding while retaining cellulose affinity.

2. Results

2.1. Design, Synthesis, and Characterization of Linker and CBM-Linker Glycoforms

Previous studies have established that systematic variation of glycosylation patterns across a panel of glycoforms is an effective strategy for elucidating how glycosylation influences protein properties. ,, To investigate how O-mannosylation modulates the binding preferences of the TrCel7A linker toward cellulose and lignin, we applied this strategy by generating and analyzing a series of linker and CBM-linker glycoforms (Figure ).

2.

Design, synthesis, and characterization of glycosylated TrCel7A linker and CBM-linker variants. (A) Structures, sequences, and glycosylation patterns of the seven linker variants and five CBM-linker variants analyzed in this study. (B) Direct comparison of the determined K _ads values for the synthesized glycoforms in their binding to cellulose and lignin. Error bars represent the standard deviation from three independent experiments. CBMS30 is the CBM bearing a single α-linked mannose at Ser30. Its established binding properties, reported previously, , serve as a control to isolate the effect of linker addition in the CBM–linker construct on binding properties.

The TrCel7A linker contains nine O-glycosylation sites. A fully combinatorial variation of glycosylation at each site would require preparing and analyzing an impractically large number of glycoforms. To balance diversity and feasibility, we developed a rational site-grouping strategy (Figure A). The nine sites were classified into three spatially contiguous groups: Group I (Ser23 and Ser24), Group II (Thr19, Thr20, and Thr21), and Group III (Thr11, Thr12, Thr13, and Thr14). This grouping was based on sequence adjacency and the working assumption that proximal residues would exert similar glycosylation effects on linker binding properties. Following this principle, we designed seven linker glycoforms: the unglycosylated form 1, four variants carrying naturally occurring α-linked monomannose residues at one or more site groups 2–5, and two variants in which all sites carried non-native α-linked galactose 6 or α-linked glucose 7.

Synthesis was achieved using a previously developed one-pot solid-phase peptide synthesis (SPPS) method coupled with an efficient deprotection protocol (Figure S1). The identity and homogeneity of each glycoform were confirmed as described in prior studies, using ultraperformance liquid chromatography–mass spectrometry (UPLC-MS) and circular dichroism (CD) (Figures S2–S8 and S15). , Binding affinities for cellulose (Avicel PH-101) and lignin (kraft lignin) were evaluated at 4 °C following a previously reported method, with binding strength expressed as adsorption equilibrium constants (K _ads) (Figure S18 and S19). , Comparison of K _ads values showed that the linker alone, regardless of glycosylation status, displayed no measurable affinity for cellulose (Figure B, 1–7). In contrast, most linker glycoforms bound lignin, although no clear correlation between glycosylation pattern and binding strength was observed. Glycoform 2 had the highest K _ads value (0.42 μM^–1), while glycoform 5 showed no detectable binding.

The lack of measurable cellulose binding rendered the linker variants unsuitable for quantitative evaluation of substrate selectivity. We therefore turned to CBM-linker constructs. Initial designs featured an unglycosylated CBM, but attempts to synthesize a completely unglycosylated version of variant 8 proved challenging, likely due to aggregation of the synthetic product. Based on prior evidence that O-glycosylation can mitigate aggregation and improve protein stability, we modified the design to include glycoforms 8–12, which combine linker variants 1–5 with the previously well-studied glycosylated CBM segment CBMS30 (Figure A). This CBM bears a single α-linked mannose at Ser30, a known glycosylation site, and resolved the aggregation issue when combined with the linker variants. Linker glycoforms 6 and 7 were excluded because their non-native glycans conferred no measurable advantage in the initial study.

The CBM-linker glycoforms were generated through the previously established SPPS and one-pot folding protocol. The synthesis and folding proceeded smoothly, yielding glycoforms 8–12 (Figure S1). , The identity and homogeneity of each glycoform were again confirmed by UPLC-MS and CD, including verification of correct disulfide bond formation and overall secondary structure (Figures S9–S13 and S15–S17). To our knowledge, glycoform 12 represents the first chemically synthesized O-linked glycoprotein of this length and structural complexity obtained in a single synthetic sequence. Comparison of the K _ads values for glycoform 8 with those of CBMS30 determined previously clearly revealed the impact of adding the linker in the CBM–linker construct, namely reduced cellulose binding but enhanced lignin binding (Figures B, S20 and S21). Analysis of glycoforms 9-12 showed that glycosylation of the linker increased cellulose binding, especially at the Group III sites. Fully occupying all the glycosylation sites did not further enhance cellulose binding. In contrast, glycosylation at any linker site group more significantly increased lignin binding, and again, glycosylation of all sites did not provide further benefit.

2.2. MD Simulation of Linker-Substrate Interactions

The markedly different affinities of the variants for cellulose and lignin prompted a mechanistic investigation into the molecular origins of this selectivity. MD simulations were employed for this purpose, as they are a well-established approach for probing molecular interactions between soluble proteins and insoluble substrates. , Linker 1, which exhibited a clear preference for lignin over cellulose, was selected as the initial representative model for mechanistic analysis.

Simulations were conducted following established protocols. , The cellulose model was derived from the cellulose Iβ crystal structure, while the lignin model was constructed from three representative phenylpropane unitsp-hydroxyphenylpropane, guaiacylpropane, and syringylpropaneinterconnected through canonical C–O and C–C linkages (Figure S23). − Initial linker–substrate complexes were built using VMD software, and each system was subjected to an independent 100 ns MD simulation. Binding free energies (E _total) were calculated from the final 10 ns of equilibrated trajectories using the GROMACS interaction energy group command.

The calculated E _total for linker 1 with lignin was −233.67 ± 58.50 kcal mol^–1, indicating a much stronger affinity than with cellulose (−20.56 ± 5.58 kcal mol^–1) (Figure A). This pronounced difference closely matched the experimental K _ads values, validating the simulation approach and its utility in deciphering binding selectivity. Decomposition of E _total into van der Waals (E _vdw) and electrostatic (E _elec) components revealed distinct interaction profiles for each substrate. E _vdw primarily reflected hydrophobic interactions (HI), weak π–π stacking, and steric packing effects, whereas E _elec encompassed charged hydrogen bonds (cHB), neutral hydrogen bonds (nHB), cation−π (Cπ) interactions, and other polar contacts. For cellulose, binding was dominated by E _vdw (−14.73 ± 5.31 kcal mol^–1), with a minor contribution from E _elec. In contrast, lignin binding was mainly electrostatic in nature, with E _elec reaching −187.55 ± 47.37 kcal mol^–1, accompanied by a smaller E _vdw contribution of –46.12 ± 13.54 kcal mol^–1 (Figure A and Tables S1 and S2).

3.

Structural basis for the differential binding affinities of linker 1 toward cellulose and lignin. (A) Computed binding free energies from MD simulations. (B) Key interactions between linker 1 and cellulose. (C) Key interactions between linker 1 and lignin.

Analysis of MD trajectories provided structural insights that rationalize these energetic differences. During cellulose binding, the linker transitioned from an extended to a curled conformation, limiting its surface contact with the substrate (Figures B and S24). Key interactions involved hydrophobic contacts between Pro4, Pro8, Pro9, and Ala18 with the hydrophobic triads (axial H1–H3–H5) of glucose rings, along with occasional nHBs between Thr20 and cellulose hydroxyl oxygens. In contrast, when bound to lignin, the linker retained an extended conformation, allowing a broader interface and stronger electrostatic interactions. Contacts included cHBs between the N-terminal amine and Arg15 with lignin carboxylates, Cπ interactions between Arg16 and aromatic lignin rings, and additional nHBs involving polar residues such as Ser24 and various lignin functional groups (Figures C and S24).

2.3. MD Simulations of CBM-Linker Interactions with Substrates

To elucidate how glycosylation modulates binding at the molecular level, we conducted MD simulations of the CBM-linker constructs (Figure S25 and S27). Cluster analysis of conformations from the 100 ns trajectories revealed that while the CBM domain of variant 8 consistently bound to cellulose, its linker region exhibited considerable conformational dynamics. Glycosylation significantly restrained this flexibility, with modifications at Group III sites producing the most stable conformation, whereas full glycosylation slightly increased linker motion again (Figure A, compare 11 and 12). This dynamic trend paralleled the experimentally determined binding affinities, supporting a close correlation between linker flexibility and substrate association. In contrast, all variants displayed lower conformational variability during lignin binding, consistent with their stronger experimental affinity for this substrate (Figure B).

4.

Differential binding behavior of CBM-linker variants toward cellulose and lignin. (A, C, E) MD trajectory analysis, binding energies, and interaction analysis for cellulose. (B, D, F) Corresponding analyses for lignin.

The computed E _total values reinforced these observations (Figures C and D). Variant 8 exhibited an E _total of −416.90 ± 46.83 kcal mol^–1 for lignin and −118.88 ± 8.98 kcal mol^–1 for cellulose (Tables S1 and S2), confirming a substantially stronger interaction with lignin. The E _total trends for 8–12 closely mirrored the experimental K _ads values, further validating the simulation model. Energy decomposition analysis showed that, for cellulose, glycosylation primarily affected E _vdw, whereas for lignin, glycosylation mainly modified E _elec. Analysis of the binding interfaces provided a structural rationale for these energetic patterns. For cellulose binding, 8 formed only a few nHBs and HIs within the linker region, resulting in weak substrate association (Figure E). Glycosylation in 11 enhanced van der Waals contacts, thereby strengthening binding. However, in glycoform 12, extensive glycosylation induced linker curling, reducing substrate contact relative to 11 and slightly lowering affinity. For lignin binding, the dominant interactions between 8 and the substrate were strong cHBs, which stabilized the complex (Figure F). Glycosylation in 11 introduced additional cHBs, further enhancing binding. In contrast, complete glycosylation in 12 reoriented the linker such that glycans at Group III sites were unfavorably positioned for lignin association, and concurrently, the monomannose at Thr14 disrupted the critical Arg16–CO₂ ^– contact, explaining the observed reduction in affinity.

2.4. Validation of the Computational Model by Investigating Glycoforms Bearing Dimannose

Having established the predictive capability of our computational model for the monomannosylated CBM-linker variants, we next sought to evaluate its performance on more complex dimannosylated systems. This extension was motivated by a key mechanistic insight: since increasing glycan density within the linker region attenuates electrostatic interactions that are critical for lignin binding, as shown by glycoform 12, installing larger glycans such as dimannose could potentially yield CBM-linker variants with enhanced cellulose selectivity, a long-standing objective in cellulosic biofuel research. To explore this hypothesis, we employed our model to characterize four new glycoforms, 13–16, designed with dimannose residues at Group I, Group II, Group III, and all linker glycosylation sites, respectively (Figures A, S25 and S27).

5.

Design and evaluation of new CBM-linker glycoforms with enhanced binding preference for cellulose. (A) Glycosylation patterns of the newly designed glycoforms. (B) Calculated binding energies with cellulose and lignin. (C, D) Binding conformations and experimentally determined K _ads of 16 when complexed with cellulose and lignin.

Simulation results indicated that glycoforms 15 and 16 maintained cellulose binding affinity comparable to the monomannosylated reference 12, while variants 13 and 14 showed reduced binding. Notably, the introduction of dimannose yielded two variants, 13 and 16, with markedly weaker lignin binding (Figure B; compare 13–16 with 12). Among these, glycoform 16 emerged as a particularly promising candidate, preserving strong cellulose affinity while simultaneously suppressing lignin interaction. MD trajectory analysis further confirmed that the dimannose residues in glycoform 16 significantly disrupted key electrostatic interactions with lignin (Figures C and S29).

To experimentally validate this prediction, we chemically synthesized glycoform 16 and measured its binding properties. The experimental data showed excellent agreement with the MD predictions: variant 16 retained a K _ads for cellulose similar to glycoform 12, while exhibiting a noticeably reduced affinity for lignin (Figures D and S22).

2.5. Unlocking a Selective Route to Eliminate Lignin Binding through Phosphorylation

Building upon our validated simulation framework, we next turned to the intriguing class of CBM–linker glycoforms bearing multiple negatively charged phosphorylated mannose residues, which remain beyond current synthetic reach. , Recognizing that the strong association between CBM–linker variants and lignin primarily arises from the formation of cHBs with lignin surface CO₂ ^– groups, we hypothesized that introducing negatively charged phosphate moieties could electrostatically repel these CO₂ ^– groups, thereby directly disrupting the primary mechanism of lignin association. Because cellulose lacks a similarly charged surface, this modification was expected to leave cellulose binding largely unaffected, thereby enhancing overall substrate selectivity. We further anticipated that strategically placed phosphorylation could even achieve the most desirable outcome: complete abolition of lignin binding through electrostatic repulsion while maintaining cellulose affinity.

To test this hypothesis, we performed molecular dynamics simulations using our verified computational model on a series of designed glycoforms 17–28, each systematically varying in the number and placement of mannose phosphate modifications across the linker (Figures A, S26 and S28). As expected, binding free energy analyses of the simulation results revealed that nearly all phosphate-modified variants exhibited substantially reduced lignin affinity compared to glycoform 12 (Figure B). Among them, glycoform 20, bearing phosphate modifications at all sites, displayed the most profound effect. Its interaction with lignin was essentially abolished, as indicated by an E _total of 8.52 ± 10.56 kcal mol^–1. Structural inspection further showed that the linker of glycoform 20 became highly solvated and failed to form stable contact with lignin (Figure C). Importantly, this dramatic suppression of lignin binding occurred with only a modest reduction in cellulose affinity, reflected by a binding E _total of –103.47 ± 11.72 kcal mol^–1. Together, these results reveal that mannose phosphorylation offers a surprisingly elegant and effective means of selectively eliminating lignin association.

6.

Computational screening of CBM–linker glycoforms bearing multiple negatively charged phosphorylated mannose residues to selectively eliminate lignin binding. (A) Glycosylation patterns of glycoforms with phosphorylated mannoses. (B) Calculated binding energies with cellulose and lignin. (C) Binding conformations of 20 when complexed with cellulose and lignin.

3. Discussion

The increasing recognition of IDRs in protein science has underscored a fundamental knowledge gap: how PTMs, particularly clustered O-glycosylation, regulate their molecular behavior and functional specificity. ,, To address this long-standing challenge, we developed an integrative, stepwise research strategy that unites chemical biology with MD simulations, using the linker region of the industrially important biofuel enzyme TrCel7A as a model system. , This approach represents a conceptual advance in glycoscienceleveraging chemically accessible glycoforms to establish and calibrate a computational framework, validating it with more complex but still synthetically attainable structures, and finally applying the validated model to investigate naturally occurring yet synthetically inaccessible glycoforms. Through this progressive strategy, synthetic limitations were converted into opportunities for predictive exploration, effectively extending the accessible scope of glycosylation research.

In the first stage, a series of TrCel7A linker glycoforms bearing systematically varied monomannosylation patterns was synthesized to examine how O-mannosylation influences substrate binding. The experimental data revealed a pronounced preference for lignin over cellulose, regardless of glycosylation, a property detrimental to biomass hydrolysis. MD simulations provided insight into this result, attributing it to differences in the molecular surfaces of cellulose and lignin. The cellulose surface, presenting hydrophilic hydroxyl groups and hydrophobic triads, allows only weak nHBs and HIs with the linker, resulting in negligible binding. In contrast, the lignin surface offers a more complex array of functional groups, including hydroxyl, hydrophobic, and carboxylate groups. The negatively charged CO₂ ^–moieties facilitate the formation of stronger cHBs and Cπ interactions, yielding less dynamic conformations and tighter binding.

Within the CBM–linker construct, the overall binding affinity reflects the combined contributions of both domains. Partial glycosylation of the linker introduced glycans that established additional HBs and HIs with both cellulose and lignin, thereby enhancing CBM-linker binding to both substrates. However, the stronger cHBs formed with lignin amplified this enhancement disproportionately, leading to a greater relative increase in lignin binding. When all glycosylation sites were fully occupied, steric interference from the glycans disrupted cHBs involved in lignin binding, leading to a pronounced decrease in lignin affinity while only modestly reducing cellulose binding.

In the next stage, we leveraged this mechanistic understanding to design and simulate CBM–linker variants bearing dimannose residues, a bulkier glycan that poses greater synthetic challenge but remains within experimental reach. The simulations predicted that dimannose would further disrupt lignin interactions while largely preserving cellulose contacts, producing a stronger cellulose preference. Subsequent experimental characterization confirmed this prediction, establishing the predictive power of the computational approach.

The results from these glycoforms bearing neutral mono- and dimannose residues, however, revealed an intrinsic limitation: while they partially shifted the binding preference toward cellulose, they could not fully suppress lignin binding. This limitation stems from their inability to completely disrupt the strong interactions with the negatively charged CO₂ ^– groups on lignin surfaces. Consequently, the overall binding preference for lignin persisted.

The final and most significant application of our strategy was the computational investigation of CBM–linker glycoforms bearing phosphorylated mannoses, which are naturally occurring but currently beyond the reach of chemical synthesis. The results were striking. Simulations revealed that these negatively charged glycans produced the most desirable outcomeselectively preserving cellulose binding while effectively abolishing lignin binding. This stark contrast to the behavior of neutral glycans provides a compelling rationale for the natural occurrence of phosphorylation on TrCel7A O-glycans.

4. Conclusion

This study establishes an integrated chemical biology–simulation framework as a powerful approach to elucidate how clustered O-glycosylation modulates the functional behavior of IDRs. Using the O-mannosylated linker of TrCel7A as a model system, we revealed that both the unmodified linker and its glycoforms bearing neutral glycans preferentially bind lignin over cellulose, driven by favorable electrostatic interactions with negatively charged carboxylate groups on the lignin surface. Increasing glycan density or size selectively weakens lignin binding by sterically disrupting these contacts. Remarkably, introducing phosphorylated mannose abolished lignin affinity through electrostatic repulsion while preserving cellulose binding. This discovery, consistent with our recent findings that negatively charged glycans exert particularly strong effects on protein properties, , highlights that charge-mediated modulation could act as an important regulatory mechanism in O-glycosylation , and underscores its significance for future glycosylation research and glycoengineering applications.

5. Experimental Section

5.1. Chemical Synthesis

The synthesis of TrCel7A linker variants followed the previously established procedures. For CBM–linker variants, the synthetic route originally developed for glycosylated CBM variants was adapted, , with one modification: after completing the CBM sequence via SPPS, elongation was continued to incorporate the full linker region. The final products of both linker and CBM–linker variants were purified by reverse-phase HPLC and characterized by UPLC–MS analysis to confirm their purity, molecular identity, and homogeneity.

5.2. Determination of Binding Affinity

Binding affinities were determined according to the previously reported procedures. , Linker or CBM–linker variants were dissolved and serially diluted in 50 mM sodium acetate buffer containing 50 mM NaCl (pH 5.0). Equal volumes of the variant solution and substrate suspension (2.4 mg/mL in the same buffer) were mixed in microcentrifuge tubes (total volume 100 μL) equipped with magnetic stir bars. Samples were gently stirred at 1,100 rpm and 4 °C for 2 h to reach equilibrium, then centrifuged at 14,000 g for 10 min. Two 10 μL aliquots of the supernatant were analyzed to determine the unbound variant concentration. Each measurement was performed in triplicate. The adsorption data were fitted to a single-site Langmuir isotherm model using OriginPro software to yield K _ads.

5.3. Molecular Dynamics Simulations

The crystal structure of CBM1 (PDB ID: 1CBH) served as the foundation for model construction. Linker and CBM–linker variants were built using the GLYCAM-Web Builder, subjected to energy minimization to obtain the initial conformation, and parametrized with the CHARMM36m force field for proteins and the CHARMM carbohydrate force field for glycans and their interactions. , Lignin was described using an optimized CHARMM-compatible model that accurately represents its aromaticity and charged surface properties. Water molecules were represented with the TIP3P model. Each variant–substrate complex was solvated in a rectangular water box and neutralized with sodium ions, resulting in systems of approximately 200,000 atoms. Periodic boundary conditions were applied, with a 14 Å cutoff for nonbonded interactions. Electrostatic interactions were treated using the particle–mesh Ewald algorithm, and all hydrogen bonds were constrained with the SHAKE algorithm. Simulations were performed using GROMACS 2022.5. Each system was subjected to energy minimization using the steepest descent method, followed by heating to 300 K and equilibration at 1 atm pressure. Production runs of 100 ns were then carried out in the NVT ensemble. Interaction energies between each variant and substrate were computed using the GROMACS interaction energy group command, separating electrostatic and van der Waals contributions.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.6c00079.

• Materials and methods, UPLC-MS analysis of linker and CBM-linker glycoforms, determination of the binding affinity of linker and CBM-linker glycoforms to cellulose and lignin, MD simulations (PDF)

#.

R.W., W.L., X.L., and Y.Z. contributed equally to this work. R.W., W.L., D.X., S.S., and Z.T. designed the research; R.W., W.L., X.L., Y.Z., J.T., H.Z., Z.S., C.J., and Y.L. performed the research; R.W., W.L., X.L., Y.Z., and Z.T. analyzed the data; R.W., W.L., X.L., S.S., and Z.T. wrote the paper. All authors reviewed and approved the final version of the manuscript.

This work was supported by the National Natural Science Foundation of China (Grant Nos. 92578127, 22577159, and 22507149), the CAMS Innovation Fund for Medical Sciences (Grant Nos. CIFMS, 2025-I2M-KJ-014 and 2021-I2M-1-026), the Postdoctoral Fellowship Program of the China Postdoctoral Science Foundation (Grant No. 2024M750237), and the Beijing Natural Science Foundation Youth Project (Grant No. 2264097).

The authors declare no competing financial interest.