VersaTile – a method for rapid docking enzyme and scaffoldin construction
VersaTile is a DNA assembly method that was developed to eliminate the current technical constraints to construct multimodular proteins in a combinatorial way. It is a Lego-like assembly method specifically for modular proteins that do not share DNA homology . VersaTile follows a two-step approach (Figure S1). First, a repository of all tiles is constructed. Second, any assembly with a selection of these tiles can be created in a one-step reaction with short hands-on time and high efficiency. A tile is defined as a coding sequence for a specific module that is made compatible with the VersaTile technique. The coding sequence is therefore flanked by six-nucleotide long position tags and BsaI recognition sites, and cloned in a dedicated entry vector, pVTE. BsaI cleavage within the position tags generates position-specific, single-stranded overhangs that are joined in a dedicated destination vector, pVTD. Hands-on time is limited to pipetting each tile along with an appropriate destination vector, BsaI and T4 DNA ligase, followed by a cyclic temperature protocol and a transformation step.
Designer cellulosomes are composed of two types of modular proteins: (1) The scaffoldin, which serves as the backbone molecule, is composed of multiple cohesins and in many cases a CBM. (2) Multiple docking enzymes , composed of a dockerin module and one or two enzymatically active modules, are selectively incorporated into the scaffoldin by virtue of the cohesin-dockerin interaction. For both multimodular proteins, a separate VersaTile assembly system was developed.
First, we designed a three-way system allowing the construction of docking enzymes composed of three modules (Figure S2). Table S1 shows the position tags employed in the three-way system. The position tags encode two amino acids that will act as linkers between the selected modules. Second, we developed a five-way system allowing the construction of scaffoldins composed of five modules (Figure S3). Table S2 shows the position tags employed in the five-way system. We note that BsaI cleavage within the first and last position tags generates single-stranded overhangs that are the same for both systems, therefore indicating that the same collection of destination vectors can be used for both docking enzyme and scaffoldin assembly systems. The docking enzyme assembly system allows the construction of docking enzymes composed of three modules. These could for example be an enzyme, a linker and a dockerin, or two enzymes and a dockerin. However, in some cases, it would be interesting to construct a docking enzyme composed of an enzyme, a dockerin and no linker. To compare the activity of constructed docking enzymes with the natural enzyme, the system should preferably also allow for the construction of proteins composed of a single enzymatic module without dockerin. For this purpose, we designed His-tag tiles flanked by different combinations of position tags. Use of these tiles allows the conversion of the three-way system to a two-way system (Figure S4). The same approach was followed to convert the five-way scaffoldin system into a four-, three- or two-way system (Figure S5). In addition to the twelve His-tag tiles, we also prepared tiles encoding a Glutathione S-transferase (GST) and StrepII tag. These tiles were constructed to be fitted at the first position of the scaffoldin assembly system.
Upon the development of the two VersaTile construction systems, two separate tile repositories were generated. The first tile repository contains building blocks needed for the construction of docking enzymes (dockerins, linkers and enzymatically active modules). The second tile repository contains the building blocks needed for the construction of scaffoldins (CBMs, cohesins and tags). First, we filled the tile repositories with the non-catalytic cellulosomal building blocks. Seven cohesins and eight dockerins were selected from naturally occurring cellulosome-producing organisms (Table S3). Additionally, the Clostridium thermocellum CipA CBM3a (Ct-CBM3) (Table S4) was added. Finally, nine linker tiles were added to the docking enzyme tile repository (Table S5). These tiles were designed specifically for the second position in docking enzyme constructs.
Cloning in the standard destination vector (pVTD2) results in the expression of a protein consisting solely of the selected tiles. Our repository contains multiple variants of this destination vector, allowing the N- or C-terminal addition of the CBM3a module, His-, StrepII- and/or GST-tag (Figure S6). Enhancing the variety of available destination vectors further improves the flexibility of the VersaTile platform. Whereas assembly in pVTD2 allows the construction of scaffoldins composed of five modules, this number is increased to seven when assembling in more complex destination vectors such as pVTD17. Since this vector already integrates a GST and CBM module in its backbone, five cohesin tiles can be selected for assembly. In some cases, the additional tags are cleavable.
VersaTile allows the production and optimisation of designer cellulosomes - Galactomannan case study
Based on the target lignocellulose component, a range of lignocellulosic enzymes can be added to the tile repository, and a designer cellulosome composed of these catalytic modules can be optimised. Here, we selected galactomannan (GM), a specific hemicellulose primarily found in softwoods. GM consists of a backbone composed of β-1,4-linked D-mannose residues and α-1,6-linked D-galactose side chains. β-Mannanase catalyses the random hydrolysis of the β-1,4-D-mannosidic linkages in the GM backbone. β-Mannosidase catalyses the hydrolysis of the glycosidic bond of the terminal, non-reducing β-D-mannose residues in β-D-manno-oligosaccharides (MOS). The enzyme α-galactosidase catalyses the hydrolysis of the α-1,6-linked galactose side chains.
Five GM-degrading enzymes were added to the docking enzyme tile repository. In previous research, many Thermobifida fusca enzymes have been successfully converted to the cellulosomal mode [6, 18, 19]. We therefore selected the T. fusca β-mannanase (GH5) and β-mannosidase (GH2) as the two backbone-degrading enzymes. Both enzymes have been characterised before [20-22]. Whereas the natural mannanase comprises a GH5 enzymatic module and a family 2 CBM, the mannosidase does not carry additional modules. Our preliminary activity analysis revealed that the preferred substrate of the mannanase is GM, followed by mannan and MOS. The mannosidase prefers p-nitrophenyl β-D-mannopyranoside (pNP-β-mannose), followed by MOS, mannan and GM. To date, no T. fusca α-galactosidase has been characterised. Three α-galactosidases from different origins were thus selected as putative candidates to work synergistically with the selected backbone-acting enzymes. Clostridium cellulolyticum α-galactosidase is a multimodular enzyme composed of an N-terminal GH27 enzymatic module, a family 6 CBM and a C-terminal dockerin. Bifidobacterium adolescentis α-galactosidase is a GH36 enzyme. Cellvibrio japonicus α-galactosidase is a GH27 enzyme. Each tile was amplified without its native signal peptide. An overview of the tiles encoding GM-degrading enzymes can be found in Table S6. Since the T. fusca β-mannanase and C. cellulolyticum α-galactosidase are multimodular in nature, we have constructed multiple tile variants for these enzymes. Whereas the shortest tile variant encodes the enzymatically active module, the longer variants encode (part of) the linker or the linker and the CBM module. Finally, we also added the Cj_CBM35 tile which originates from the Cellvibrio japonicus Man5C mannanase enzyme and has been described as mannan-specific [23, 24] (Table S4).
Table 1 and Table 2 show the number of tiles that were prepared at each position in the docking enzyme and scaffoldin assembly systems, respectively. These numbers allow us to calculate the theoretical number of docking enzymes and scaffoldins that can be constructed. For example, the tile repository allows the construction of 630 (=10×9×7) monovalent docking enzymes composed of an N-terminal enzymatic module, a C-terminal dockerin and a linker separating these two modules. A total of 720 (=8×10×9) bicatalytic docking enzymes composed of an N-terminal dockerin followed by two enzymatically active modules can be constructed. Regarding the scaffoldins, a total of two CBM tiles and seven cohesin tiles allows us to construct 28 (=(2×7)×2) monovalent (one CBM and one cohesin), 294 (=(2×7×7)×3) bivalent (one CBM and two cohesins), 2,744 (=(2×7×7×7)×4) trivalent (one CBM and three cohesins) and 24,010 (=(2×7×7×7×7)×5) tetravalent scaffoldins (one CBM and four cohesins). Note that these numbers are calculated with a fixed number of one CBM per scaffoldin, allowing all possible orders and assuming the use of a standard destination vector that adds no additional modules.
In Figure 1, the full combinatorial power of VersaTile is displayed. With this repository, 3,578 different docking enzymes and 27,076 scaffoldins can be constructed. With a total of 3,578 different docking enzymes and 2 CBMs that can be located at 5 different positions, a stunning number of 2x1015 (=3,5784×5×2) tetravalent designer cellulosomes with one CBM can be considered.
A total of 151 modular proteins (79 mono- and bicatalytic docking enzymes and 72 scaffoldins) were constructed in this study. To facilitate the interpretation of the results, we have numbered these constructs and given each of them a specific icon. A full list of the constructed docking enzymes and scaffoldins can be found in Table S7 and Table S8, respectively. We have constructed monocatalytic docking enzymes with C- and N-terminal dockerins. For some of these constructs, the enzymatically active module was directly fused to the dockerin. In other cases, these two modules were separated by a selected linker sequence. The bicatalytic docking enzymes, composed of two enzymatically active modules and one dockerin, carry the latter module C- or N-terminally or in between the two selected enzyme modules. The scaffoldins all contained one to five cohesins and one CBM.
Parameters influencing expression level, stability and activity of docking enzymes
Based on initial exploratory analyses of randomly designed docking enzymes, a total of 79 docking enzymes were constructed (Table S7). Docking enzymes were expressed and purified by immobilised metal affinity chromatography (IMAC). Subsequently, their activities were tested based on the release of reducing ends using the 3,5-dinitrosalicylic acid (DNS), and/or a para-nitrophenol (pNP) assay. This setup allowed us to perform systematic pairwise comparisons to navigate through the multiparametric landscape that determines the eventual outcome. Below, we discuss the main parameters influencing docking enzyme expression level, stability and enzymatic activity, supported by a selection of data obtained throughout this work.
Delineation of tiles
Carbohydrate active enzymes (CAZymes) are often multimodular in nature, comprising enzymatically active modules, CBMs and intervening linkers. Usually, only the enzymatically active module is incorporated in the synthetic docking enzyme. Therefore, careful delineation of the tile is crucial to allow efficient conversion to the cellulosomal mode. Delineation is often complicated due to the absence of an available crystal structure. In addition, it cannot be excluded that a difference of a single amino acid affects the outcome either at the expression, folding or activity level. Furthermore, one of the key choices is to include the natural linker in a tile or not. The most optimal delineation is thus an empirical process, but the throughput of the VersaTile technique allows to explore this aspect more extensively. Generally, we have delineated tiles based on domain (HMMER ) and secondary structure (Phyre2 ) analyses, taking into account that predicted secondary structure elements should not be interrupted.
This empirical optimisation process was performed for the natural T. fusca mannanase enzyme. An unstructured protein sequence intervenes the N-terminal GH5 enzymatic module and the C-terminal family 2 CBM. The C-terminal part of this unstructured region is a proline-rich sequence, which is typically associated with a linker function. Four different tile variants of this enzyme were prepared, increasingly truncating the C-terminal CBM and linker (Table S6). First, CBM truncation was done. In addition, we performed a stepwise truncation of the linker for the proline-rich region and the remaining linker moiety, respectively. We evaluated the effect of truncating the T. fusca mannanase before fusing it to a dockerin module. Since a negative influence of multiple CBMs in a designer cellulosome has been reported [27, 28], we started with testing the tile variants that excluded the CBM sequence. The shortest variant (construct nr. 1) was inactive. There was no significant difference in activity between the second (construct nr. 2) and third variant (construct nr. 3) (Figure 2A), indicating that the added linker does not disturb the activity. This pairwise comparison demonstrates the major impact of the delineation parameter.
Dockerin – enzyme combination
During this work, seven different dockerins were used to construct docking enzymes. Although there are sequence similarities between dockerins from the same type, switching the dockerin module will result in the construction of a substantially different docking enzyme with difficult-to-predict characteristics. In some cases, enzymes can only be expressed when fused to a specific dockerin. In other cases, the selected dockerin drastically influences the enzymatic activity of the docking enzymes.
For example, four different dockerins were C-terminally fused to Cj-Aga (Doc-CtI: construct nr. 56, Doc-CtII: construct nr. 57, Doc-Cc: construct nr. 58, Doc-Rf: construct nr. 59). Only constructs nr. 56 and 58 (incorporating Doc-CtI or Doc-Cc) could be expressed. In the case of Tf-Manna-Li, being the T. fusca mannanase including its natural linker, we evaluated several direct C-terminal fusions with different dockerins. The fusion protein composed of Tf-Manna-Li and Doc-Cc (construct nr. 7) could not be expressed. A fusion of the same enzymatically active module to Doc-CtII (construct nr. 6) resulted in an activity loss of 61% compared to the natural enzyme without a dockerin. A fusion to Doc-Ac (construct nr. 13) resulted in an activity loss of 43% (Figure 2B).
Docking enzyme architecture
Dockerins can be fused to the selected enzymes N- or C-terminally and this arrangement can greatly influence the overall efficiency of the constructed docking enzymes. When a selected enzyme is multimodular in nature, the natural CBM module is often replaced with the dockerin, as such preserving the natural modular order of the enzyme. However, several enzymes are composed of a single module. In these cases, it is difficult to predict the optimal position of the dockerin and this needs to be tested empirically. Although dockerins are usually positioned C-terminally in their native proteins, they have been found N-terminally  and internally  between two enzymatically active modules as well. Moreover, dockerins that are positioned C-terminally in their native protein have also been shown to be active when positioned N-terminally in a designer cellulosome [18, 31], leaving the protein engineer with a multitude of design choices when constructing a novel docking enzyme.
The influence of dockerin position on the activity of docking enzymes was analysed for Tf-Manna-Li, Tf-Manno and Cc-Aga. The C. thermocellum type I dockerin was used for this analysis. The purification yield of the mannanase docking enzyme with N-terminal dockerin (construct nr. 5) was drastically lower than the purification yield of the docking enzyme with C-terminal dockerin (construct nr. 4) (Figure S7). Additionally, the N-terminal variant was inactive. This indicates that the C-terminal CBM of the T. fusca mannase is best substituted by a C-terminal dockerin, thereby maintaining the natural linker function. The natural mannosidase does not contain a CBM or other modules. In contrast to Tf-Manna, purification yields were higher with the dockerin positioned at the N-terminus of Tf-Manno (constructs nr. 33 and 34) (Figure S8). The activity of the N-terminal and C-terminal variants on pNP-β-mannose was similar (Figure 3A). This confirms that dockerins that are in their native protein positioned at the C-terminus can also be positioned at the N-terminal side of enzymes without diminishing the enzymatic activity [18, 31]. In this particular case, the N-terminal dockerin even appears to enhance the docking enzyme expression level.
Cc-Aga was converted to the cellulosomal mode in three different ways. The natural enzyme including its C-terminal linker was fused to the C. thermocellum dockerin either C- (construct nr. 43) or N-terminally (construct nr. 42). A C-terminal dockerin was also fused to the full enzyme, including its naturally occurring C-terminal CBM (construct nr. 44). When the dockerin was placed N-terminally, a less stable galactosidase, prone to degradation (visible as additional smaller bands on SDS-PAGE), was observed after purification (Figure S9). Activity analysis on p-nitrophenyl-α-D-galactopyranoside (pNP-α-galactose) showed that the construct with a C-terminal dockerin, but without CBM had the highest activity (Figure 3B). The activity of the N-terminal docking enzyme variant had a significantly lower activity than the variant with C-terminal dockerin, indicating once again the importance of preserving the natural modularity of an enzyme when converting it to the cellulosomal mode. In addition, this example also demonstrated again the effect of tile delineation.
Monocatalytic vs bicatalytic docking enzymes
Some native cellulosomal enzymes are multicatalytic, bearing multiple copies of catalytic modules and a single dockerin in a single polypeptide chain . Some studies on designer cellulosomes have indeed shown promising results using bicatalytic docking enzymes [33, 34]. Their results have indicated that next to the intermolecular synergy achieved by colocalising multiple docking enzymes, intramolecular synergy by designing multicatalytic docking enzymes can also be pursued. Using VersaTile, we quickly constructed a range of sixteen bicatalytic docking enzyme variants (constructs nr. 64-79). Several of the bicatalytic docking enzymes had high purification yields. However, they proved to be unstable as visualized on SDS-PAGE and monocatalytic docking enzymes were therefore selected for incorporation in the final complex.
Linker inserted in between the enzymatically active module and dockerin
The linker that merges the dockerin with the enzymatically active module determines the distance between the two modules and consequently the overall architecture of the docking enzyme. The selected linker type and length can have a significant effect on the activity of the docking enzyme and as such of the whole complex. In general, little research has been performed on the role of docking enzyme linkers. However, in a recent study, Kahn and colleagues (2019) illustrated their importance . With VersaTile, we are now able to quickly construct docking enzyme variants comprising the same enzymatically active modules and dockerin module, but a different linker sequence.
The effect of different linkers inserted between the enzyme and the dockerin was analysed for Tf-Manna (Figure 4A) and Cj-Aga (Figure 4B). For the mannanase, the best performing dockerin (Doc-Ac) (Figure 2B) was combined with either Tf-Manna-Li or Tf-Manna using four different linkers (Li-E: constructs nr. 20 and 24, Li-F: constructs nr. 21 and 25, Li-G: constructs nr. 22 and 26, or Li-H: constructs nr. 23 and 27). When Tf-Manna and Tf-Manna-Li were directly fused to the dockerin (constructs nr. 19 and nr. 13, respectively), an activity drop of approximately 40% was detected compared to the unfused proteins. In the case of Li-G (constructs nr. 22 and nr. 26), the loss of enzymatic activity due to the dockerin fusion could be partially compensated, limiting the activity loss to 30% for Tf-Manna and 24% for Tf-Manna-Li, indicating that for this specific enzyme-dockerin combination, a long, rigid linker is the optimal choice. Cj-Aga_Doc-CtI (construct nr. 56) had low expression levels. The introduction of different linkers (Li-A: construct nr. 60, Li-B: construct nr. 61, Li-C: construct nr. 62, Li-D: construct nr. 63) improved purification yields. Again, activity analysis revealed that a direct dockerin fusion (construct nr. 56) significantly decreased the enzyme activity (56%). This was compensated to some extent by the introduction of an appropriate linker. For this specific dockerin-enzyme combination, a short proline-rich linker (Li-D: construct nr. 63) proved to be the optimal choice, resulting in a limited activity loss of 30%. Use of another linker, such as linker A (construct nr. 60), further decreased enzymatic activity to approximately 80% loss compared to the natural variant.
Notably, fusion of a dockerin module had a different effect on the enzyme activity depending on the substrate used. The negative effect of the dockerin fusion appears to become less prominent when more complex substrates are used. When comparing the activity of the natural mannosidase enzyme (construct nr. 32) to the docking enzyme (construct nr. 39) on the simple substrate, pNP-β-mannose, it is clear that dockerin fusion caused a severe loss of activity (82%). However, when using the more complex substrate, mannan, the activity loss is limited to 28%. For the most complex substrate, GM, no significant loss in activity was observed (Figure 5).
Construction, purification and analysis of a trivalent designer cellulosome
To set up a trivalent designer cellulosome able to degrade GM, the multiparametric landscape had to be reduced to three complementary docking enzymes. The major and essential criterion was to select three enzymes, each fused to a different dockerin. This renders the optimisation process of the different docking enzymes interdependent, since the selection of one docking enzyme excludes the use of its dockerin for the other docking enzymes.
Out of all docking enzyme types, the construction of an active α-galactosidase docking enzyme proved to be the most challenging. We were unable to convert the B. adolescentis α-galactosidase to the cellulosomal mode. The designed constructs were either enzymatically inactive or could not be expressed. Several of the C. cellulolyticum α-galactosidase docking enzymes were active on pNP-α-galactose but had minimal activity on the more complex substrate, GM. Even when both backbone-acting enzymes were added to the reaction mixture, galactosidase activity remained minimal. This left the C. japonicus α-galactosidase as the only suitable enzyme to be fitted in the envisioned designer cellulosome. With only the CtI- and Cc-dockerin fusions of this enzyme expressing well, Cj-Aga was particularly restrictive in the selection of its dockerin. Since the Cj-Aga_Doc-CtI fusion protein showed low expression levels, this fusion was further engineered by linker optimisation, resulting in Cj-Aga_Li-D_Doc-CtI (construct nr. 63) (Figure 4B) as the first selected docking enzyme. Tf-Manno_Doc-Bc (construct nr. 39), which bears Doc-Bc could easily be purified, was stable with no degradation or multimerisation and showed clear activity (Figure 5). Therefore, Tf-Manno_Doc-Bc was not further optimised. The delineation experiment indicated Tf-Manna-Li as the best delineation. Pairwise comparisons and linker optimisation finally yielded Tf-Manna-Li_Li-G_Doc-Ac (construct nr. 26) (Figure 4A).
A scaffoldin composed of cohesins Coh-Ac, Coh-Bc and Coh-CtI was constructed (construct nr. 135). This scaffoldin was also equipped with the C. thermocellum CBM3a, to allow binding to cellulose, and a GST-tag to allow pull-down of the multi-enzyme complex. The designer cellulosome was produced by co-incubation of the scaffoldin with the respective docking enzymes. Subsequently, a GST pull-down was performed. SDS-PAGE analysis of the eluted fractions revealed that each docking enzyme could interact with the scaffoldin (Figure S10 and Figure S11). Additionally, the pull-down assay allows the isolation of pure complexes that can be subjected to activity analysis. Test reactions consisted of a 0.5% GM solution, supplemented with 10 pmol of designer cellulosome. During the first hour of the assay, the release of reducing sugars was measured. After 2, 4, 6 and 8 hours, samples were taken to determine the release of the specific sugar monomers mannose and galactose. Figure 6 shows the results of the activity assay when the selected mannanase, mannosidase and/or galactosidase docking enzymes were incorporated in the complex.
The results indicate that the GM backbone is only degraded to MOS when mannanase is incorporated in the complex. When mannosidase and/or galactosidase are also included, we do not observe a significant increase in released reducing ends. In this case, mannose and/or galactose are present in the reaction mixture at the micromolar range. They are thus not detected when measuring the released reducing sugars. This signal is overwhelmed by the release of MOS and is therefore only linked to mannanase activity. When only mannanase is included in the complex, mannose is released at a rate of 5.6 µM/h. A complex only including mannosidase released no mannose. Combining these two enzymes results in a mannose release of 15.1 µM/h. Upon addition of galactosidase, a maximum rate of mannose release of 17.1 µM/h is achieved. The rate of mannose release is notably lower than the rate of reducing sugar release (0.57%), indicating that the majority of degradation products are MOS.
When only galactosidase is incorporated in the complex, galactose monomers are released at a rate of 26 µM/h. Upon addition of mannanase, galactose is released ten times faster (267 µM/h). Galactosidase benefits from mannanase addition because mannanase releases MOS, which are the preferred substrate for galactosidase. Upon incorporation of mannosidase, galactose release is enhanced an additional 1.4-fold, reaching a maximum galactose release rate of 383 µM/h. This suggests an additional synergistic action between the two exo-acting enzymes. Overall, this threefold activity analysis confirms that each of the incorporated enzymes remains active upon incorporation in the complex.
Optimising scaffoldin architecture results in an increase of released galactose monomers
With the VersaTile approach at our disposal, it is fairly straightforward to construct multiple scaffoldin variants. As such, we were able to quickly analyse different parameters and further optimise the complex.
To investigate if the location of the docking enzymes on the scaffoldin influences the activity of the enzymes, six scaffoldin variants were constructed. In each scaffoldin, Ct-CBM3 was fixed at the first position and Coh-Ct, Coh-Bc and Coh-Ac were shuffled for positions 2, 3 and 4 (construct nrs. 135-140). Each scaffoldin was combined with the optimised docking enzymes, and complexes were purified by GST pull-down (Figure S12 and Figure S13). Test reactions consisted of a 0.5% GM solution, supplemented with 10 pmol of GST-isolated designer cellulosome (Figure 7).
The results show that each enzyme remained active at every position of the scaffoldin. Release of reducing sugars, galactose and mannose is detected for each of the constructed multi-enzyme complexes. However, not all complexes degrade the GM substrate at the same rate, and a preferred modular arrangement of the scaffoldin, and therefore of incorporated docking enzymes (Mannanase – Mannosidase – Galactosidase) can be detected. Degradation of the GM-backbone to MOS is significantly slower (p < 0.05) when mannanase is located at the third position of the scaffoldin (construct nr. 139 and construct nr. 140). The minimal amount of reducing sugars is released when mannosidase is located at the first position and galactosidase at the second (construct nr. 139). We observe the same pattern when analysing the release of mannose and galactose monomers. Once again, this indicates the synergistic action between the backbone-acting mannanase and the other two exo-acting enzymes. When mannanase activity is at its maximum, more easily-degradable MOS are present to act as a substrate for the other two enzymes taking part in the complex.
In all previously described scaffoldins, we incorporated the cellulose-specific C. thermocellum CipA CBM3 module. When using cellulosomes to degrade complex lignocellulosic material, this CBM may interact with cellulose while allowing the attached enzymes to degrade surrounding GM, as such releasing the valuable cellulose for further degradation. However, one might also be interested in directing the complex straight to the (galacto)mannan substrate. In this case, replacing the CBM with a mannan-binding module (MBM) can prove useful. We thus constructed additional scaffoldins in which Ct-CBM3 was replaced with a mannan-specific CBM originating from C. japonicus, Cj-CBM35 (constructs nr. 141-146). The cellulosome incorporating Cj-CBM35 essentially released the same amount of reducing sugars from pure GM (Figure S14, p = 0.381) as the one including Ct-CBM3.
During GM degradation, the backbone-acting mannanase releases MOS. Subsequently, galactose side chains can be removed from MOS by galactosidase. In an attempt to accelerate galactose release, multiple copies of the galactosidase docking enzyme were incorporated in the complex (scaffoldin constructs nr. 147 and 149). However, no additional release of galactose was observed using this approach (data not shown). Conversely, the release of galactose monomers (and reducing ends from MOS) could be significantly enhanced by the incorporation of an additional copy of the backbone-acting mannanase docking enzyme (construct nr. 150) (GST pull-down: Figure S15) (Figure 8).
These results suggest that when using the original complex (incorporating one copy of each enzyme), galactose release is limited by the lack of available substrate. The addition of multiple mannanase copies allows the acceleration of the release of MOS which subsequently serve as a more accessible substrate for galactosidase. The incorporation of three mannanases further accelerated the release of reducing sugars, but did not have a significant impact on the release of galactose monomers. This indicates that two mannanase copies are capable of releasing an excess of MOS, as such fully saturating the present galactosidases and allowing this enzyme to release galactose at its maximal rate.