The carbohydrate-binding module of TrCel7A aids in navigating hemicellulose barriers in plant cell walls

Abstract

Efficient enzymatic deconstruction of plant cell walls is critical for the utilization of lignocellulose biomass. Key enzymes in this process are cellobiohydrolases, a class of cellulases that processively degrade crystalline cellulose. Many cellobiohydrolases possess a carbohydrate-binding module (CBM), yet the specific roles of CBMs in substrate interaction remains unclear. Here, we use single-molecule fluorescence microscopy to investigate how CBM1 of Trichoderma reesei Cel7A influences enzyme binding and motility on cellulose substrates of varying complexity. We compare WT Cel7A with a truncated variant lacking CBM1 (Cel7A^ΔCBM) on bacterial cellulose, phosphoric acid swollen cellulose, delignified milkweed cellulose, and holocellulose nanofibrils (hCNFs). While both variants showed similar steady-state binding densities on bacterial cellulose and phosphoric acid swollen cellulose, Cel7A^ΔCBM exhibited reduced binding on milkweed cellulose and hCNF, with the greatest reduction on the hemicellulose-rich hCNF. Alkali treatment of hCNF to remove hemicellulose partially restored Cel7A^ΔCBM binding, suggesting a role for CBM1 in substrate navigation and productive binding site recognition. Kinetic analyses revealed that CBM1 enables a rapid binding mode absent in the truncated variant. Consistent with this, the isolated CBM1 domain had a faster substrate association rate than the isolated catalytic domain. These findings demonstrate that CBM1 enhances cellulase–substrate interactions by accelerating binding and enabling navigation of the complex environment of plant cell walls. Our results emphasize the importance of CBMs in natural cellobiohydrolase function and highlight their value in the design of improved cellulases for industrial biomass conversion.

The efficient deconstruction of plant cell walls is crucial for maximizing the utilization of plant biomass. This deconstruction process is primarily catalyzed by carbohydrate active enzymes (CAZymes) that collectively function to transform plant biomass, which consists mainly of cell walls, into its basic elements. The complex nature of plant cell walls, which are composed of tightly interacting networks of cellulose, hemicellulose, and lignin, requires an array of CAZymes, including cellulases, to act in concert and for each of these enzymes to navigate to, recognize, bind to, and react with its respective substrate.

In fungi, the cellulases that carry out the bulk of cellulose deconstruction, termed cellobiohydrolases, act via a processive exo-hydrolysis mechanism to cleave cellobiose subunits from either the reducing or nonreducing ends of glucan chains in cellulose. Cellulose degradation by cellobiohydrolases involves at least four different stages: (i) substrate binding, where the enzyme adsorbs to cellulose, (ii) engagement of the active site with a glucan chain, (iii) processive hydrolysis involving repeated cycles of bond cleavage and cellobiose release, and (iv) dissociation from the substrate (1, 2). However, most cellobiohydrolases exhibit relatively slow turnover rates, and single-molecule investigations have revealed that many enzyme molecules bind to cellulose without appearing to move processively, implying that nonproductive binding is a limitation for enzyme efficiency (3, 4, 5). These findings motivate more detailed study of the mechanisms by which cellobiohydrolases navigate to and bind with their substrates and open the possibility of engineering cellobiohydrolases for enhanced efficiency, with the caveat that tighter binding affinity appears to come at the cost of a reduced turnover rate (6).

One common feature of cellobiohydrolases is the presence of two main domains: a catalytic domain that contains an active site tunnel and a carbohydrate-binding module (CBM) that is attached to the catalytic domain by a flexible linker (7). Protein domains capable of interacting with carbohydrates are common and can be found in most living organisms (8), and when such domains are part of CAZymes, they are termed CBMs. Phylogenetic analyses indicate that CBMs likely evolved independently in fungi and bacteria, with family 1 (CBM1) being predominantly found in fungi (9, 10). However, while CBMs are a common feature in many CAZymes, not all cellobiohydrolases contain them, suggesting that cellulose hydrolysis is not entirely dependent on CBM-mediated interactions. Furthermore, the fact that the isolated catalytic domain of Cel7A is also capable of binding to and hydrolyzing cellulose (11, 12) begs the question of why CBM domains exist on this and other cellobiohydrolases.

CBM1 of Cel7A is a 37 amino acid (∼4 kDa) domain made up of a small wedge-shaped fold and a cellulose-binding face composed of three tyrosine residues that are thought to align with the pyranose rings on the hydrophobic surface of cellulose crystals. Several hypotheses have been put forward to explain the functions of CBM1. First, it might enable the specific recognition of polysaccharide substrates within the complex matrices of plant cell walls and other multipolymer networks (13, 14). Second, CBM1 has been hypothesized to promote tighter binding to cellulose, which enriches cellulase concentration on the cellulose surface (9, 15). (16, 17, 18). Finally, CBM1 might also function as a “safety tether” to keep the catalytic domain of Cel7A in close proximity to cellulose when the catalytic domain de-threads or unbinds from its substrate. However, the validity and/or relative importance of these potential functions is still unclear.

To better illuminate the functions of CBMs in cellulase binding and reactivity, here we used single molecule fluorescence microscopy to directly image Cel7A variants with and without CBM1 as they bound and hydrolyzed cellulose. We compared the binding kinetics and motility of these variants on a range of substrates that represent the varied degree of cellulose accessibility that exist in plant cell walls. We found that on more complex cellulose substrates that are derived from native cell walls, CBM1 strongly enhances the binding kinetics of Cel7A to cellulose.

Results

Deleting CBM1 reduces the binding rate of Cel7A to bacterial cellulose

To characterize the role of the CBM of Cel7A in interactions with different substrates, we compared the nano-scale dynamics of WT Cel7A with that of a truncated variant consisting of only the catalytic domain (Cel7A^ΔCBM) (Fig. 1A). Using our single-molecule imaging platform (3, 19, 20), we labeled the two enzyme variants with fluorescent Quantum dots (Q-dots) and tracked them as they bound to and hydrolyzed bacterial cellulose (BC). To quantify the cellulose binding kinetics, we flushed 1 nM of labeled enzyme into the flow cell and quantified the number of bound particles over time as the system reached steady state. Compared to WT Cel7A, binding of Cel7A^ΔCBM was slower, although both reached similar steady-state binding densities within ∼350 s (Fig. 1, B and C). Next, we used single-molecule tracking to measure the velocity and processive run length of labeled Cel7A on the surface. When comparing the motility of the two enzyme variants, there were no significant differences in the proportion of processive molecules mean, velocity, or run length (Fig. 1, D–F).

Figure 1.

Binding and motility of WT Cel7a and CBM-deleted Cel7A^ΔCBM on bacterial cellulose.A, cartoons of the two enzymes used in this work: intact Cel7A and Cel7A^ΔCBM which lacks the CBM and connecting linker region. B, binding time course of Cel7A (blue) and Cel7A^ΔCBM (red) on bacterial cellulose. Bold lines are the means and lighter colored error bands represent one SD of four independent experiments. C, expansion of the initial 50 s of binding (area highlighted by a gray rectangle in B). D, the mean fraction of processive molecules in each field of view. Error bars are SD. E, the mean velocity and (F) mean run length of processive particles. Differences between Cel7A and Cel7A^ΔCBM in (D), (E), and (F) are not statistically significant by unpaired t test (p > 0.05). Processive fractions were calculated from three independent experiments; velocity and run length were calculated from at least 230 molecules pooled together from three independent experiments.

The effect of CBM on binding rate varies between different cellulosic substrates

BC is a common substrate for studying the activity of cellulase enzymes. Its high crystallinity, homogeneous structure, and purity make it ideal for such biochemical assays. However, the naturally occurring and more industrially relevant substrates of these enzymes are derived from plant biomass and differ in both structure and composition. In contrast to bacterial cellulose, plant cell walls are complex, diversified with other noncellulosic components, and highly heterogeneous. This complexity may impose constraints on the dynamics of cellulase enzymes in a way that is not manifested during the hydrolysis of purely cellulosic substrates such as bacterial cellulose. We therefore hypothesized that for Cel7A, the function of the CBM becomes more evident when the enzyme encounters more complex substrates.

To test this hypothesis, we compared the binding of Cel7A and Cel7A^ΔCBM to substrates having diverse structures and composition. Treating cellulose with phosphoric acid is a common method for disturbing crystalline cellulose, making it more amorphous. Thus, we used BC as a substrate to generate phosphoric acid swollen cellulose (PASC) and analyzed the single-molecule binding kinetics. On PASC, Cel7A and Cel7A^ΔCBM reached similar steady state–binding densities, with Cel7A^ΔCBM having a slower initial binding rate than Cel7A, as seen in BC (Fig. 2, A–D). However, compared to BC, the binding densities were ∼50% lower on PASC than on BC.

Figure 2.

Binding densities of Cel7A (blue) and Cel7A^ΔCBM (red) over time on four different cellulosic substrates.A-B, bacterial cellulose (BC), (C-D) phosphoric acid swollen cellulose (PASC), (E-F) milkweed cellulose (MWC), and (G-H) holocellulose nanofibers (hCNF). Left panels (A, C, E, G) show binding kinetics over 500 s, while right panels (B, D, F, H) focus on initial binding during the first 50 s. The dashed lines are biexponential fits to the data and the plateau and k_obs values (in Particles/100 mm² and s⁻¹, respectively) are given for each of the datasets. k_obs is the weighted average of the fast and slow rate constants (given in Table S1). Micrographs in the rightmost column (I, J, K, L) are scanning electron microscopy images showing the morphological characteristics of each substrate. Binding curves are means of at least three separate experiments. Scale bars in (I-L) represent 200 nm.

In addition to BC and PASC, we experimented with two substrates extracted from plant biomass—delignified cellulose extracted from milkweed (Asclepias syriaca) silks (MWC) and delignified holocellulose nanofibrils (hCNFs) from rapeseed straw (Brassica napus) prepared by gentle lignin extraction (see Experimental procedures for details). On these two plant-derived substrates, the steady state binding density of Cel7A^ΔCBM was significantly reduced compared to Cel7A, with a ∼50% reduction for MWC (Fig. 2, E and F) and an even greater, ∼90% decrease in the case of hCNF (Fig. 2, G and H).

The observed differences in binding kinetics between the substrates could be the result of differences in their composition, structure, and/or density. Therefore, to evaluate the nanoscale structure of the four substrates, we imaged them by scanning electron microscopy (SEM) and found the expected fibrillar structures of cellulose in BC, PASC, and MWC, with no major differences in the apparent diameter or form of the smallest visible fibrils evident between them (Fig. 2, I–K). However, we observed more large bundles in the MWC (Fig. 2K), which might represent cellulose macrofibrils (7). In contrast, the fibrils in hCNF (Fig. 2L) appeared to be more compacted than the other substrates and interspersed with a highly reticulated substance, potentially hemicellulose. The overall density of the substrates did not appreciably vary. To quantify the differences in initial binding kinetics between Cel7A and Cel7A^ΔCBM on the four substrates, we fit a double exponential function to each of the binding curves and calculated an observed binding rate constant (k_obs) using the weighted average of the fast and slow rate constants (values given in Table S1). On all four substrates, k_obs was higher for Cel7A than for Cel7A^ΔCBM, consistent with CBM1 enhancing the cellulose on-rate of WT Cel7A (Fig. 2).

Hemicellulose suppresses Cel7A binding kinetics on hCNF

Of the four substrates, removal of the CBM1 domain of Cel7A had the largest effect on the kinetics of binding to hCNF, imposing a ∼10-fold reduction in k_obs (Fig. 2G). The major compositional difference between hCNF and BC and PASC is the presence of native hemicellulose in hCNF (Fig. S1). To quantify the composition of hCNF, we carried out a neutral sugar analysis and found that hCNF is composed of 79% glucose, 15% xylose, and 6% other sugars (fucose, arabinose, rhamnose, galactose, and mannose) (Table S2). To test whether hemicellulose in fact contributes to the inhibition of Cel7A^ΔCBM binding, we treated hCNF with 4M potassium hydroxide (KOH) at room temperature overnight followed by thorough washing with water. This alkali treatment has been shown to be an effective treatment for the removal of hemicellulose (21, 22). A neutral sugar analysis of the hCNF after KOH treatment revealed the presence of glucose, but no other sugar was detected (Table S2). Though we cannot determine the exact effects of this treatment on cellulose structure, SEM imaging of hCNF after KOH treatment did not show any evident structural differences in comparison to untreated hCNF (Fig. 3A) and anisotropy analysis of SEM micrographs (see Methods) yielded comparable values of 0.061 and 0.044, respectively. However, in single-molecule binding assays, the steady-state density of Cel7A on base-treated hCNF increased compared to Cel7A on untreated hCNF (Fig. 3B), suggesting that the alkali treatment was effective in removing hemicellulose from hCNF, possibly exposing previously obscured binding sites on the cellulose surface. More importantly, the steady-state binding of Cel7A^ΔCBM, which was negligible on untreated hCNF, was enhanced on base-treated hCNF to near the level of Cel7A on untreated hCNF (Fig. 3B), although the binding rate was still considerably slower (Fig. 3, B, C and E), preventing enzyme accumulation from reaching steady state binding within the measured time frame (Fig. 3B).

Figure 3.

Removal of hemicellulose from hCNF enhances the binding of Cel7A^ΔCBM.A, scanning electron micrographs showing the morphological structure of holocellulose nanofibers (hCNF) before and after potassium hydroxide (KOH) treatment. B, binding densities over time of Cel7A and Cel7A^ΔCBM to hCNF and KOH-treated hCNF (hCNF-KOH). C, enlarged area marked by a gray rectangle in (B). D, comparative binding of CBM1 to BC, hCNF, and hCNF-KOH. Binding curves are means of at least three separate experiments. E, k_obs (mean ± SD) of Cel7A, Cel7A^ΔCBM, and isolated CBM1 on BC, hCNF, and KOH-treated hCNF. The dashed lines in (B), (C), and (D) are biexponential fits used to determine k_obs. Scale bars in (A) represent 200 nm.

Based on our observation that the binding rate of Cel7A^ΔCBM was slower than Cel7A on all substrates tested, we hypothesized that the CBM domain has a faster on-rate for cellulose binding than the catalytic domain. To test this hypothesis, we also performed single molecule binding assays using an isolated, Qdot-labeled CBM1 domain on BC, hCNF, and KOH-treated hCNF substrates (4). Whereas CBM1 could reach similar binding densities on BC and KOH-treated hCNF, we observed reduced densities on untreated hCNF (Fig. 3, D and E). Additionally, we found that on all substrates, CBM1 had k_obs values that were higher than the isolated catalytic domain, but comparable to those of the intact Cel7A. The relative similarity of the observed association rates between CBM1 and Cel7A as opposed to the slower binding of the Cel7A^ΔCBM suggests that the inclusion of the CBM in the intact enzyme is vital for its ability to achieve adequate binding to its substrate. In contrast to the catalytic domain, the binding kinetics of CBM1 appeared to be less affected by the structure and composition of the different substrates, which supports the hypothesis that the function of the CBM is to enhance the initial encounter of Cel7A with its cellulose substrate.

Cel7A binding rates differ due to varying on-rates

In our single-molecule binding assays, the observed rise to steady-state binding is a function of both the on-rate and off-rate constants; specifically, k_obs = k_on∗[Cel7A] + k_off, where k_on is the bimolecular on-rate and k_off is the first-order off-rate (23). To separate the contributions of on- and off-rates, we measured the dissociation rates of Cel7A and Cel7A^ΔCBM on all substrates by fitting an exponential function to the distribution of binding durations. As shown in Figure 4, there were some differences in off-rates, but for any given substrate, the difference between Cel7A and Cel7A^ΔCBM was within a factor of two. We independently measured the bleaching rate of the Qdots and found a bleaching rate of 0.009 s⁻¹. Thus, we compensated for the inherent bleaching rate by modeling the observed off-rate as a sum of the actual off-rate and the bleaching rate and calculated a corrected off-rate: $k_{\text{off}}^{\text{corrected}}=k_{\text{off}}^{\text{measured}}-k_{\text{bleach}}$.

Figure 4.

Representative exponential fits of binding duration distribution for each enzyme-substrate combination. Dissociation rate constants (k_off) were determined by subtracting a Q-dot bleaching time constant of 0.009 s⁻¹ from the complementary cumulative distribution function values at mean binding duration times.

With the observed Cel7A binding rates, the corrected off-rates, and the experimental Cel7A concentrations in hand, we calculated the enzyme on-rates for the different substrates. As shown in Figure 5 Table S1 and Table S3, variations in the observed binding rate resulted predominantly from differences in the Cel7A on-rate, with only small contributions of the off-rate. Notably, for every substrate where we could determine it, k_on was higher for Cel7A than for Cel7A^ΔCBM. Cel7A^ΔCBM on hCNF did not have a sufficient number of particles to confidently calculate a k_off, but a low accumulation rate is expected from a very slow on-rate. Consistent with the isolated CBM results shown in Figure 4, these results suggest that the CBM domain has a higher inherent on-rate for cellulose than the catalytic domain and that in intact Cel7A, binding to the substrate is mediated by both the CBM and catalytic domain. The fact that the largest difference between the on-rates of Cel7A and Cel7A^ΔCBM was observed on the substrates derived from plant cell walls; MWC and hCNF suggests that the CBM domain serves a navigational function for cellulase action on plant cell walls.

Figure 5.

Measured binding rate constants (k_obs), measured off-rate constants (k_off), and calculated on-rate constants (k_on) for Cel7A and Cel7A^ΔCBM on different substrates. On-rate is calculated from k_obs = k_on∗[Cel7A] + k_off where [Cel7A] is 1 nM in all experiments. Off-rate is corrected for photobleaching. Substrates are as follows: bacterial cellulose (BC), phosphoric acid swollen cellulose (PASC), milkweed cellulose (MWC), holocellulose nanofibrils (hCNF), and KOH-treated hCNF (hCNF-KOH). At least three separate experiments were used to determine each k_obs and k_off. Due to low particle number, k_off could not be measured for hCNF. Error bars are one SD.

Discussion

While it is widely observed that CBMs improve the activity of cellulase enzymes, the precise mechanisms underlying this improvement and their full range of functions in cellulose degradation are still not completely understood. CBMs might assist hydrolysis through substrate recognition, enhancement of enzyme binding to the surface of cellulose, and/or by improving or prolonging processive movement along the substrate. Here, we used single-molecule tracking and a panel of substrates with different molecular compositions and structures to help reveal the mechanistic advantage(s) conferred by CBMs for cellulases.

In our experiments using BC and PASC, we found that both Cel7A and Cel7A^ΔCBM can effectively bind to purely cellulosic substrates, although PASC could sustain substantially lower enzyme densities at steady state. This difference in enzyme density could be due to either structural differences between the substrates or to differences in substrate density that are not evident from SEM imaging (Fig. 2). Relevant to the first factor, treatment with phosphoric acid is a method to disrupt the cellulose crystals that results in a more amorphous substrate (24, 25). As a cellobiohydrolase, Cel7A primarily binds to and hydrolyzes crystalline cellulose (26, 27); thus, it is likely that either the number of available binding sites for it on PASC or the accessibility of the cellulose is reduced. The fact that similar steady-state binding densities did not differ substantially between enzyme variants on either BC or PASC suggests that the CBM does not enhance the enzyme’s ability to maximally occupy binding sites on cellulose in either a crystalline or amorphous configuration. These findings are in accord with previous reports that compared the saccharification efficiency of Cel7A and Cel7A^ΔCBM on cellulose and found overall comparable kinetics and glucose yields (4, 12, 28), though others report lower binding of the isolated catalytic domain (11, 29).

In BC and PASC, cellulose is exposed and easily accessible for the enzyme to bind. Cel7A, however, evolved to function in the deconstruction of considerably more complex plant biomass. In plant cell walls, cellulose fibrils are covered with other polymers, mainly hemicellulose and lignin, that are known to inhibit the activity of cell wall–degrading enzymes (16, 30, 31). The plant-derived substrates used in this work, MWC and hCNF, were delignified but still retained some of the native hemicellulose associated with cellulose. Previously, we found that adding purified xylan during the synthesis of bacterial cellulose reduces the binding of Cel7A and the proportion of processive molecules on the resulting substrate (20). Here, we also observed a reduction in Cel7A binding to native hemicellulose-containing substrates. The reduction was evident in both the densities of enzyme molecules at later timepoints in the binding curve and a reduced binding rate that, in the case of MWC, prevented achievement of steady-state enzyme accumulation within the recorded time frame (Fig. 2E). Binding isotherms generated in previous studies show reduced binding of the isolated catalytic domain compared to the intact Cel7A, even on purely cellulosic substrates (32, 33); in contrast, we found similar steady-state binding of the two variants on BC and PASC. This discrepancy can potentially be explained by the lower enzyme concentrations used in our current work, where steady-state binding is reached by the equilibrium of k_on and k_off rather than by a lack of free binding sites. Since we did not observe significant differences in the k_off of the two enzyme variants, this suggests that the enhanced binding of the intact enzyme is the result of higher k_on facilitated by CBM1. The fact that Cel7A^ΔCBM binding relative to full-length Cel7A binding was most reduced on the MWC and hCNF substrates suggests that CBM1 specifically aids the association of Cel7A with cellulose in complex composite substrates that contain hemicellulose.

Removing the CBM domain of Cel7A had the most significant impact on binding to hCNF. SEM imaging showed that this substrate was also structurally distinct from the other substrates. However, although chemical removal of the hemicellulose by KOH partially restored binding, it did not significantly alter the appearance of the substrate, indicating that the inclusion of hemicellulose in hCNF, rather than its morphology, is the chief factor constraining the binding of Cel7A^ΔCBM to it.

To complement our experiments with the isolated catalytic domain of Cel7A, we also observed the binding kinetics of the isolated binding module. We found that it had comparable k_obs values and similar steady state binding densities on BC and hCNF-KOH but was slower to bind and achieved a lower binding density on untreated hCNF. Importantly, the k_obs values of the isolated CBM1 were generally similar to those of Cel7A and higher than for Cel7A^ΔCBM on hCNF and KOH-treated hCNF (Fig. 3E). This affinity to cellulose can explain the observation of a consistent lag in Cel7A^ΔCBM binding as compared to Cel7A and suggests that the CBM grants the intact enzyme a fast-binding capability that is absent in the truncated enzyme. While our findings showed no promotion of Cel7A motility by the CBM, suggesting that this domain is not directly involved in the enzyme's processivity, it is plausible that it assists in enzyme tethering, which would prevent its dissociation over extended periods. Our observations support a model that, in complex substrates, CBMs facilitate rapid binding and promote hydrolysis through a proximity effect, effectively enriching cellulase concentrations on the cellulose surface.

Single-molecule tracking of Cel7A and other cellobiohydrolases was previously performed by a range of techniques, including atomic force microscopy, optical tweezers, and fluorescence microscopy. However, most of these works focused on the motility of enzymes during the catalytic cycle (34, 35). While other works address the binding kinetics of intact Cel7A to isolated cellulose (36, 37), here we directly compared binding kinetics between WT Cel7A, CBM-deleted Cel7A, and isolated CBM1 on a range of substrates with varying complexity that included those derived from native plant cell walls, highlighting the role of CBM1 in mediating binding to the more complex substrates that exist in nature and are used in biorefineries. Our observed association rates for Cel7A to BC are well within the range of previously reported rate constants (36, 38). Our calculated k_off values are at the lower end of published values (11, 39), which we attribute to the high photo-stability of the quantum-dots we used to label the enzymes.

Successful lignocellulose deconstruction depends on the efficiency of cell wall–degrading enzymes. Although these enzymes have been refined through evolution over millions of years, the selective pressures that shaped them did not optimize them for the requirements of industrial biomass processing, which can involve high temperatures, chemical and physical pretreatment of the biomass, and ideally, very rapid deconstruction. To optimize the saccharification of plant biomass, we must improve and expand on naturally existing cellulase enzymes. Rational design of cellulases is challenging and demands an understanding of the different stages of enzymatic cellulose degradation, the first of which is the attachment of the enzymes onto insoluble cellulose. This work highlights the importance of CBMs for this initial step of enzyme-cellulose binding, especially in the context of more complex substrates, and stresses the essential role of CBMs in the enzymatic deconstruction of plant biomass as compared to model cellulose substrates. While the omission of the CBM1 domain has the potential to improve cellulose saccharification by reducing enzyme “traffic jams” and facilitating a higher turnover rate (34), its inclusion likely remains critical for providing adequate substrate-binding affinity when degrading complex substrates derived from plant cell walls. This trade-off between increased turnover rate and binding capacity is a key factor in designing novel enzymes and enzyme combinations and could help us tailor unique biocatalysts to specific substrates. Such customization could be achieved by leveraging naturally occuring or rationally designed CBMs with moderate substrate affinity that will allow for sufficient binding specificity without compromising processivity.

Experimental procedures

Substrate preparation

For preparing BC, Gluconacetobacter hansenii (ATCC 23769) was used to inoculate 50 ml of Schramm–Hestrin (SH) media, creating primary cultures that were then incubated at 30 °C with shaking (∼180 rpm) for 48 h. Secondary cultures were cultured in sterilized 2-L glass trays containing 500 ml of SH media along with 50 ml of the primary culture. Bacterial cultures were covered with aluminum foil and incubated in the dark at 30 °C for 5 days to allow cellulosic pellicles to form. Harvested pellicles were washed three times with distilled water, then with 70% ethanol, and again three times with distilled water. They were then incubated in 0.5 M sodium hydroxide at 80 °C for 30 min, followed by more washes with distilled water. To neutralize the pH, cellulose pellicles were washed three times with 50 mM sodium acetate buffer (pH 5.0) and then three more times with distilled. In further processing, the pellicles were cut into pieces with scissors, suspended in 30 ml of distilled water, and sonicated five times for 30 s each (intensity setting 9) with 1-min intervals using a sonic dismembrator (Thermo Fisher Scientific, model 100) on ice. Finally, the resulting cellulose suspension was passed through a 100 μm chamber (Microfluidics, H10Z) in a microfluidizer (Microfluidics, model LM20) at 15,000 psi for 10 min (approximately 20 cycles for the 30 ml samples).

PASC was prepared form a suspension of processed bacterial cellulose according to Zhang et al. (24) with some modifications: Three milliliters of microfluidized bacterial cellulose were placed into a 50 ml falcon tube and 10 ml of ice-cold H₃PO₄ (86.2% w/v) were added to the cellulose in aliquots of 4 ml, 4 ml, and 2 ml, agitating between each aliquot. Cellulose was precipitated with 40 ml of ice-cold ethanol at a rate of 10 ml per addition with vigorous stirring resulting in a white precipitate. The supernatant was removed and 0.5 ml of 50 mM ice-cold sodium acetate (0.05 M, pH 5) was added to neutralize the pH. The PASC pellet was resuspended with 45 ml of ice-cold distilled water.

MWC was extracted from dried milkweed pods collected at the Arboretum of Penn State University. Seed and silks from 1 pod (roughly 1.5 g per pod including seeds) were manually separated, and the silks were processed similarly to the method described above for bacterial cellulose. For delignification, 1 ml of microfluidized MWC was centrifuged and the supernatant was removed. The pellet was resuspended in a 50 ml solution of 0.1 M HCl and 10% NaClO₂ and incubated at room temperature overnight. After delignification, samples were washed repeatedly with distilled water until pH was neutralized.

To produce hCNFs (40), the following protocol was used. First, 3% peracetic acid (PAA; 0.35 g pure PAA/g dry material, pH 4.8) was used to delignify dried rapeseed straw four times. This process was carried out at 85 °C for 45 min without stirring. Between cycles, the used PAA was decanted, followed by a single water wash, and subsequently, fresh 3% PAA was added. Delignified material (holocellulose) was rinsed with water following the fourth cycle until the conductivity fell below 10 μS/cm. The holocellulose was blended for 2 min in a Vitamix A3500i blender to get a homogenous slurry. A 0.1 wt% holocellulose dispersion was blended for 30 min to produce nanofibrillated polydispersion. The hCNFs were isolated from the polydispersion as a supernatant using centrifugation at 4400g for 15 min.

Enzyme preparation

Isolated TrCel7A and TrCel7A^ΔCBM were kindly provided by Stephen R. Decker, National Renewable Energy Laboratory (28). Enzymes were buffer exchanged into 50 mM borate buffer (pH 8.5) using Bio-Spin P-30 Bio-Gel spin-columns (Bio Rad). Enzyme concentrations were determined by absorbance at 280 nm using extinction coefficients of 74,906 and 45,859 M⁻¹ cm⁻¹ for the intact Cel7A and Cel7A^ΔCBM, receptively. Biotinylation was carried out using EZ-Link NHS–LC–LC–biotin (Thermo Fisher Scientific catalog 21343), by combining the enzyme with biotin–NHS dissolved in anhydrous dimethylformamide at a biotin:enzyme molar ratio of 10:1 and incubated for 4 h in the dark at room temperature. Unbound biotin was removed by buffer exchanging the enzyme-biotin mixture into 50 mM sodium acetate using Bio-Spin P-30 Bio-Gel columns. Concentrations of the biotinylated enzymes were calculated using absorbance at 280 nm and biotin concentrations were determined using a Pierce Fluorescence Biotin Quantitation Kit (Thermo Fisher Scientific). Glycerol was added to biotinylated enzymes to 10% v/v and the enzyme concentration diluted to 5 μM with sodium acetate. Aliquots of 5 μl were flash frozen in liquid nitrogen and stored at −80 °C until use.

Cloning, expression, and purification of CBM1

The DNA sequence encoding CBM1 from Trichederma reesei (UniProt G0RVK1_HYPJQ) was amplified using the following primers: ttaactttaagaaggagatatacatatgcctacccagtctcactacggc and acgcggaacaagcatcaggcactgagagtagtaagggtt and cloned into an Ampicillin-resistant plasmid containing C-terminal EGFP, AviTag, and 6×His tag via Gibson assembly (NEBuilder HiFi DNA Assembly Cloning Kit) following the manufacturer’s protocol. Expression of CBM1-EGFP-AviTag-6 × His was induced by 0.75 mM IPTG (EMD Millipore, MA; Cas: 367-93-1) in Escherichia coli strain BL21(DE3). After overnight incubation at 19 to 21 °C with shaking at 180 RPM, cells were harvested by centrifugation at 5000g for 20 min at 4 °C and resuspended in protein lysis/wash buffer (phosphate buffer, 40 mM imidazole, 1 mM β-mercaptoethanol, 1 mM PMSF, and protease inhibitors (leupeptin and pepstatin)). The cells were sonicated on ice (duty cycle: 50%, output control: 6, time: 3 min, Branson SFX250 Sonifier) and the supernatant isolated using centrifugation at 184,000 g for 35 min at 4 °C (rotor type 50.2 TI, Beckman Optima LE-80K Ultracentrifuge). The supernatant was combined with pre-equilibrated Ni-NTA agarose resin (GoldBio) and incubated with gentle inversion at 4 °C for 1 h. After incubation, the resin was pelleted by centrifugation at 5000 g for 5 min at 4 °C. The resin was transferred to a gravity column pre-equilibrated with wash buffer and washed with lysis/wash buffer. CBM1 protein was eluted with an elution buffer (phosphate buffer, 500 mM imidazole, 3.6 mg/ml DTT) and quantified by Bradford assay.

Single molecule microscopy

Single-molecule imaging was done using a previously published SCATTIRSTORM microscope (3, 19). Experiments were performed in flow cells that allow minimal disruption of the microscope’s stage and were prepared in-house by sandwiching them between a glass microscope slide and a cover slip. Cellulose substrates were pipetted onto the cover slips (20 μL at 2 mg/ml of substrate per channel) and dried at 50 °C for at least 30 min. Identification of the substrates in the flow cells was achieved using the microscope’s interference reflection microscopy mode to select an appropriate field of view. Imaging was then performed by total internal reflection fluorescence microscopy using a 488 nm laser with an image acquisition rate of one frame/s for 500 or 1000 s. The final concentration of Cel7A, Cel7A^ΔCBM, and CBM1 during imaging was 1 nM.

SEM imaging

For SEM imaging, substrate suspensions (10 ml) were pipetted onto disks of 0.2 mm Millipore filter membrane. Subsequently, the samples underwent a sequence of 5 min ethanol washes at increasing concentrations: 25%, 50%, 60%, 70%, 85%, 95% (v/v), and finally 100% ethanol. A Leica EM CPD300 instrument was then used for critical point drying of the samples. After drying, the membrane disks with the samples were affixed to aluminum stubs using carbon tape and then coated with a 5 nm layer of iridium using a Leica EM ACE200 sputter coater. Imaging was performed using the secondary electron detector on a Zeiss SIGMA VP-FESEM instrument. Image anisotropy values for hCNF and KOH-treated hCNF were measured using the FibrilTool plug-in in ImageJ (41). At least three micrographs from each substrate were measured.

Fitting, kinetic parameters, and statistical analysis

Particle binding was measured using the “find maxima” function in ImageJ and their density was normalized to an area of 100 μm² (confirmed to contain substrate using interference reflection microscopy). All fittings and statistical analyses were done using MATLAB (R2022b) or Prism GraphPad (version 10.2.2). The observed association rate constants of k_obs were determined by fitting single-molecule curves with a double exponential for Cel7A and Cel7A^ΔCBM or a single exponential for CBM1. The dissociation rate constants k_off were calculated by fitting the distribution of binding durations (1– cumulative distribution function) to a first-order exponential. Measured off-rates were corrected by subtracting the Qdot photobleaching time constant of 0.009 s⁻¹, determined by measuring the photobleaching rate of Qdots immobilized on a glass slide.

Immunofluorescence labeling, staining, and imaging

Imaging of xylan immunofluorescence-labeled and cellulose S4B-stained samples was performed as described previously (20) and imaging was done on a Zeiss Cell Observer SD microscope equipped with a Yokogawa CSU-X1 spinning disk unit. For Alexa Fluor 488-conjugated secondary antibodies detection, a 488 nm excitation laser and a 525/50 nm emission filter were used, and a 561 nm excitation laser and a 617/73 nm emission filter were used for S4B signal visualization.

Neutral sugar analysis

The solubilized neutral monosaccharides from trifluoroacetic acid and two-stage sulfuric acid hydrolysis were quantified on a Dionex ICS3000 system equipped with a PA20 column, a PA20 guard column, and a borate trap (Dionex).

Data availability

All data will be shared upon request.

Supporting information

This article contains supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Reviewed by members of the JBC Editorial Board. Edited by Chris Whitfield