In a previous study, we reported on strains of L. lactis that successfully displayed type 1 cohesins on their surface, and demonstrated their ability to bind the β-glucuronidase-dockerin fusion protein UidA-dock1 . In this study, chimeric scaffold proteins consisting of cohesins from CipA and OlpB or SdbA were successfully displayed on the surface of L. lactis, however only CipA-SdbA chimeric scaffolds were capable of binding both UidA-dock1 and UidA-dock2, suggesting that either improper folding or inaccessibility of coh2O2 may have prevented its association with UidA-dock2. Previous studies have demonstrated that scaffold proteins derived from bacteria that anchor their cellulosome to the cell surface such as C. thermocellum, Ruminococcus flavifaciens, and Acetivibrio cellulolyticus, contain long inter-cohesin linkers (50–550 residues) compared to cellulosomes from organisms which do not anchor their cellulosomes such as Clostridium cellulolyticum (10 residues) [12, 48, 49]. It has also been proposed that linkers joining cohesins within CipA may increase the protein’s conformational flexibility . With the goal of improving coh2O2 accessibility for dockerin binding, scaffold-derived linkers were engineered in our synthetic scaffolds (Figure 1A), however no significant difference in enzyme binding at either cohesin was observed (Figure 2). Since the scaffolds were successfully displayed on the cell surface, we hypothesize that either improper folding of the scaffold protein may have resulted from unfavorable ionic interactions among amino acid residues, or that the coh2O2 domain remained buried within protein aggregates, ultimately inhibiting this cohesin’s ability to bind corresponding dockerin [51, 52]. In addition, deletion of the HtrA housekeeping protease in our strain may account for the misfolded proteins remaining associated with the cell surface . It has also been previously demonstrated that targeting recombinant fusion proteins to the cell wall of L. lactis can cause problems with secretion, anchoring, and/or folding .
Since the inclusion of linkers exterior to the coh2O2 domain did not result in binding of UidA-dock2 to the chimeric scaffolds, we replaced coh2O2 with coh2S1 and found that the resulting scaffold could bind UidA-dock1 and UidA-dock2 demonstrating that both cohesin domains were accessible and functional. SdbA differs from OlpB in that it contains one rather than four cohesins, as well as a lysine-rich region downstream of coh2S1 that shares a high degree of homology to a similar lysine-rich region of streptococcal M proteins located in our cwaM6, just upstream of the LPXTG sequence . We postulate that incorporating coh2S1 adjacent to the anchor motif of streptococcal M6 protein may emulate some structural characteristics found in the native SdbA anchor protein of the C. thermocellum cellulosome, resulting in improved accessibility for UidA-dock2 binding. A total of four variant scaffolds (Figure 1B) containing both a type 1 and type 2 cohesin (CBD-coh1C3-coh2S1), only a type 1 cohesin (CBD-coh1C3), only a type 2 cohesin (CBD-coh2S1) or no cohesin (CBD alone) were tested for their ability to bind UidA-dock1 and/or UidA-dock2. Cells displaying CBD-coh1C3 were successful in binding UidA-dock1 but failed to bind UidA-dock2, while cells displaying CBD-coh2S1 successfully bound UidA-dock2 but failed to bind UidA-dock1, demonstrating the specificity of the interaction (Figure 3). Cells displaying the larger trimodular scaffold CBD-coh1C3-coh2S1 were capable of binding both UidA-dock1 and UidA-dock2. Interestingly, in the case of these larger scaffolds, the amounts of UidA-dock1 and UidA-dock2 molecules bound was greater when compared with cells displaying the smaller scaffolds CBD-coh1C3 and CBD-coh2S1, respectively (Figure 3). One possible explanation is that CBD-coh1C3-coh2S1 is secreted or displayed with increased efficiency, as in a previous study, we also demonstrated that increased scaffold protein size did not reduce the efficiency of scaffold display or functionality . It also remains possible that better folding of each respective cohesin domain within CBD-coh1C3-coh2S1, when compared with the other constructs, may account for its ability to bind more UidA-dockerin fusion proteins.
Having determined the number of each UidA-dockerin fusion bound to displayed scaffold CBD-coh1C3-coh2S1, we analyzed their relative abundance within the assembled complexes, since protein ratios can ultimately have an effect on enzyme synergy and substrate-channeling [6, 26]. Assuming a 1:1 cohesin to dockerin binding ratio, it would be expected that CBD-coh1C3-coh2S1 should bind equimolar amounts of UidA-dock1 and UidA-dock2. The resulting ratio deviated from this prediction, since the UidA-dock1 / UidA-dock2 ratio approached 4:1 (Figure 3). In a previous study, the assembly of chimeric scaffold-derived enzyme complexes on the surface of Saccharomyces cerevisiae also resulted in deviations from expected ratios of enzymes, as cellobiohydrolase CBHII associated with scaffolds at lower levels than other enzymes . We therefore suggest that variability in the proper folding and/or accessibility of individual cohesin domains within a chimeric scaffold may affect binding of the enzymes to the scaffold.
To gain further insight into factors affecting protein binding to our synthetic scaffold proteins, we “docked” individual enzymes simultaneously or sequentially onto the chimeric CBD-coh1C3-coh2S1 protein. When simultaneously binding UidA-dock1 and LacZ-dock2 to the scaffold, an approximate five-fold decrease in UidA activity was observed compared to the binding of UidA-dock1 alone whereas no significant decrease in LacZ activity was observed in these assays (Figure 4). We hypothesize that the different effects on UidA and LacZ binding and/or activity may be due to either the location of the cohesin within the scaffold, to the size of each enzyme relative to the other, or differences in binding affinities between the two recombinant cohesin-dockerin interactions. Therefore, a similar binding assay was performed where the location of the cohesins on the scaffold protein was reversed. Similarly, UidA activity was two-fold lower when incorporated in the presence of LacZ-dock1, and once again, no significant change in LacZ activity was observed when incorporated in the presence of UidA-dock2 (Figure 4). Since LacZ is significantly larger than UidA (480 kDa vs 280 kDa), this suggests that enzyme size may result in steric factors inhibiting the binding of one enzyme partner, and that the relative location of each enzyme did not seem to play a role in the resulting activities when enzymes were incorporated simultaneously.
Sequential enzyme binding assays gave similar results as simultaneous binding assays where more than a two-fold decrease in UidA activity resulted when LacZ-dock2 was bound to the scaffold prior to UidA-dock1 addition. Contrarily, although LacZ activity decreased significantly when UidA-dock1 was bound to the scaffold protein prior to LacZ-dock2, reversing this order resulted in the same LacZ activity as when LacZ-dock2 alone was targeted to the scaffold (Figure 5). To verify if enzyme location also affected the overall resulting activity of the complex, the location of each enzyme partner was reversed. UidA activity decreased when LacZ-dock1 was incorporated prior to UidA-dock2, and this activity was only partially regained when the order of assembly was reversed (Figure 5C). LacZ activity was not affected by the order in which LacZ-dock1 and UidA-dock2 were bound into such complexes (Figure 5D). In addition, when UidA-dock1 was targeted to the coh1C3 cohesin (Figure 5A), the order in which LacZ was targeted to coh2S1 also had less of an effect on resulting UidA activity compared to when UidA-dock2 was targeted to coh2S1 (Figure 5C). From these results, it appears that when a fusion enzyme is targeted to the outermost position on the scaffold, distal to the cell surface, its binding to the scaffold may be less affected by enzyme partners, compared to when it is targeted to the innermost position, proximal to the cell surface.