Self-Assembled Amyloid-Like Oligomeric-Cohesin Scaffoldin for Augmented Protein Display on the Saccharomyces cerevisiae Cell Surface

Abstract

In this study, a molecular self-assembly strategy to develop a novel protein scaffold for amplifying the extent and variety of proteins displayed on the surface of Saccharomyces cerevisiae is presented. The cellulosomal scaffolding protein cohesin and its upstream hydrophilic domain (HD) were genetically fused with the yeast Ure2p N-terminal fibrillogenic domain consisting of residues 1 to 80 (Ure2p^1-80). The resulting Ure2p^1-80-HD-cohesin fusion protein was successfully expressed in Escherichia coli to produce self-assembled supramolecular nanofibrils that serve as a novel protein scaffold displaying multiple copies of functional cohesin domains. The amyloid-like property of the nanofibrils was confirmed via thioflavin T staining and atomic force microscopy. These cohesin nanofibrils attached themselves, via a green fluorescent protein (GFP)-dockerin fusion protein, to the cell surface of S. cerevisiae engineered to display a GFP-nanobody. The excess cohesin units on the nanofibrils provide ample sites for binding to dockerin fusion proteins, as exemplified using an mCherry-dockerin fusion protein as well as the Clostridium cellulolyticum CelA endoglucanase. More than a 24-fold increase in mCherry fluorescence and an 8-fold increase in CelA activity were noted when the cohesin nanofibril scaffold-mediated yeast display was used, compared to using yeast display with GFP-cohesin that contains only a single copy of cohesin. Self-assembled supramolecular cohesin nanofibrils created by fusion with the yeast Ure2p fibrillogenic domain provide a versatile protein scaffold that expands the utility of yeast cell surface display.

INTRODUCTION

Yeast surface display provides a promising platform for protein engineering (9). The ability to anchor and display functional proteins on the yeast cell surface is also being exploited for developing novel whole-cell biocatalysts (11). Although whole-cell biocatalysis using yeast surface display is generally considered an established technology, challenges remain regarding the efficient display of protein complexes and the further increase in the amount of protein that can be displayed on the yeast surface without destabilizing the cell. To this end, unique properties of protein building blocks found in natural higher-order protein complexes may be exploited to create novel synthetic protein chimeras that self-assemble into nano-scaffolds on the yeast surface to display increased levels and varieties of proteins. One such protein building block is the cohesin (Coh) domain, which is found with multiple copies in cellulosomes.

The cellulosome is an extracellular multienzyme complex found on the cell surface of anaerobic bacteria such as Clostridium cellulolyticum, facilitating highly efficient hydrolysis of amorphous and crystalline cellulose. It is composed of numerous cellulolytic enzymes and other functional units assembled, via high-affinity interactions between the cohesin and dockerin domains (22), on a protein scaffold (scaffoldin) that contains multiple copies of cohesin. Cellulases and related enzymes localized on the scaffoldin are thought to function synergistically in hydrolyzing crystalline cellulose. Several types of free designer cellulosomes (not bound to cell surfaces) have been proposed that mimic the natural cellulosomes (8, 18). For display on the yeast surface, only relatively short and simple minicellulosomes that imitate truncated forms of the native Clostridium cellulosomes have been reported (17, 25–27).

The native Clostridium scaffoldin contains multiple cohesin units joined covalently by disordered linker regions. Here, we report an alternative molecular design for synthetic scaffoldin. It is based on a fusion protein that contains a cohesin domain fused to a cellulosomal hydrophilic domain and a yeast fibrillogenic domain. We hypothesized that the yeast fibrillogenic domain could effectively promote self-assembly of the fusion proteins into novel amyloid-like scaffoldins displaying multiple copies of the cohesin modules. Such a self-assembled oligomeric-cohesin scaffoldin can potentially display a much higher number of cohesin domains than the native bacterial scaffoldin and can be made from relatively small fusion protein monomers. Together with the high-affinity cohesin-dockerin interaction and the use of dockerin fusion proteins, the supramolecular-cohesin scaffoldin could serve as a promising protein scaffold for constructing synthetic protein complexes with novel catalytic or binding functionalities.

The main objective of this study was to develop a novel protein scaffold to enable display of increased levels and varieties of proteins on the yeast surface and, thus, expand the utility of yeast cell surface display. Here, we discuss the synthesis and performance of a supramolecular cohesin scaffold designed to fulfill this objective.

MATERIALS AND METHODS

Strains, plasmids, and media.

Escherichia coli strain DH5α was used for genetic manipulations. E. coli BL21(DE3)TrxB was used for recombinant protein production. Saccharomyces cerevisiae strain EBY100 (3) was used for the surface display of green fluorescent protein (GFP)-nanobody, a GFP-binding camelid-derived single-domain antibody fragment (16). All E. coli strains were grown in LB medium supplemented with 100 μg/ml ampicillin and 50 μg/ml kanamycin (for BL21) or 100 μg/ml ampicillin alone (for DH5α). EBY100 cells containing no plasmid were grown in YPD (yeast extract, peptone, dextrose) medium. EBY100 cells harboring the surface-display plasmid were precultured in SD-CAA (synthetic dextrose medium with Casamino Acids) medium (3) and induced using SG-CAA medium (replacing dextrose in SD-CAA with galactose).

Plasmid construction and transformation.

To display GFP-nanobody on the yeast surface, its coding sequence was amplified with forward primer NGF (5′-CGCGCGCTAGCCAGGTTCAACTGGTGGAAAG-3′) and reverse primer XGR (5′-CGCCTCGAGTTAGTGATGGTGATG-3′) using plasmid pOPINE2-GFP-NT (16) as the template. The PCR product was then digested with NheI and XhoI and ligated into the NheI/XhoI-digested surface display vector pCT302 (3) to form pCTNG. Plasmid pETGC, encoding GFP and the His₆-tagged first cohesin domain from C. cellulolyticum, was obtained by two-step cloning. First, the GFP coding sequence was obtained by PCR from pH469 (provided by H. Edske) with forward primer GFN (5′-CGCGCGCTAGCATGTCTAAAGGTGAAGAATTATTC-3′) and reverse primer GRK (5′-CGCGGTACCGGATCCTTTGTACAATTCATCCATAC-3′); it was digested with NheI and KpnI and ligated into pET21a to form pETG. Subsequently, cohesin sequence was PCR amplified from the genomic DNA of C. cellulolyticum H10 (supplied by Z. He) with forward primer CKF (5′-CGCGGTACCGTTCTTCCAAAAGATATTCCAGGCGATTCTTTGAAAGT-3′) and reverse primer CXR (5′-GTGCTCGAGAACTGCAACTTTAAGTTCTTTAGTTGGTTGGGTTCCAGGGTCGATTGTAAC-3′), digested with KpnI and XhoI, and ligated into pETG to result in pETGC. Plasmids pETGD and pETMD, carrying a His₆-tagged dockerin from C. cellulolyticum endoglucanase CelA fused to GFP and an mCherry fragment, respectively, were obtained by the following procedure. First, the His₆-tagged dockerin coding sequence was PCR amplified from pJFAc, which carries a His₆-tagged CelA (obtained from H. P. Fierobe), with forward primer DAB (5′-CGCGGATCCGTAATTGTATATGGAGATTATAAC-3′) and reverse primer DRS (5′-CGCGAGCTCTTAGTGGTGGTGGTGGTGGT-3′); the fragment was digested with BamHI and SacI and ligated into pETG to form pETGD. Subsequently, an mCherry-harboring fragment was obtained by PCR with forward primer MNF (5′-CGCGGGCTAGCGTGAGCAAGGGCGAGGAGGA-3′) and reverse primer MBR (5′-CGCGGATCCCTTGTACAGCTCGTCCATGC-3′), using pRTL2-mCherry (provided by S. B. Gelvin) as the template. The amplified fragment was digested with NheI and BamHI and cloned into pETGD to form pETMD. Plasmid pUHC carries a fusion protein designated UHC that encompasses N-terminal residues 1 to 80 of the yeast Ure2p (Ure2p^1-80) fused with the hydrophilic domain (HD) and the His₆-tagged cohesin domain (Coh). The plasmid was constructed by two-step cloning. The Ure2p^1-80 was amplified from pH469 with forward primer NF (5′-CGCGCTAGCATGATGAATAACAACGGCAAC-3′) and reverse primer UR (5′-GTTGAGCTCGATGTTGTTCTAAAGTATTCTTGATATTATTCTCG-3′), digested with NheI and SacI, and integrated into pET21a to form pETU. Subsequently, the gene encoding the cellulosomal scaffoldin protein cohesin (Coh) and its upstream HD was PCR amplified from the genomic DNA of C. cellulolyticum H10 with forward primer HDFN (5′-CGCGAGCTCATTCTAGAGGATCCGACGGTGGAAACCCACCTCC-3′) and reverse primer HCRS (5′-GCGGTCGACTACTGCTACTTTAAGTTCCTTTG-3′), digested with SacI and SalI, and ligated into pETU to form pUHC.

Protein expression and purification.

Recombinant proteins used in this study include the following: AGN (yeast Aga2p fused with GFP-nanobody), GC (GFP fused with the cohesin from C. cellulolyticum), GD (GFP fused with dockerin from CelA), MD (mCherry fused with dockerin from CelA), UHC (Ure2p^1-80 fused with HD and cohesin), and CelA. E. coli strains were grown at 37°C to an optical density at 600 nm (OD₆₀₀) of 0.5 (MD, GD, GC, and UHC) or 1.5 (CelA) in 250 ml of LB medium supplemented with appropriate antibiotics. For the dockerin-tagged proteins (MD, GD, and CelA), 1.2% glycerol and 1 mM CaCl₂ were also added. The cultures were then cooled to 22°C, and isopropyl-β-d-thiogalactopyranoside (IPTG) was added to a final concentration of 1 mM (MD, GD, GC, and UHC) or 400 μM (CelA). After 16 h, the cells were centrifugally harvested (5,000 × g for 5 min) at 4°C, resuspended in TBS-Ca buffer (25 mM Tris-HCl, pH 8, 137 mM NaCl, 10 mM CaCl₂) containing a protease inhibitor cocktail, and lysed with an XL2000 ultrasonic cell disruptor. His₆-tagged proteins were purified by immobilized metal affinity chromatography (IMAC) using a Ni-loaded HiTrap Chelating HP column (GE Healthcare) connected to a Biologic Duo Flow chromatography system (Bio-Rad). The concentration of purified proteins was estimated by Bradford assay.

Amyloid formation.

Purified UHC fusion protein (1 mg/ml) in TBS buffer was incubated at 22°C with shaking. Large visible aggregates that appeared within an hour were removed by centrifugation at 18,000 × g for 10 min. The UHC protein concentration in the supernatant was then adjusted to 0.8 mg/ml with 0.5% bovine serum albumin (BSA) added to prevent disordered aggregation. The kinetics of UHC fibril formation was analyzed using thioflavin T (ThT) binding fluorescence. At regular time intervals (1 h), 12-μl aliquots of fibril solution were removed and mixed with 12 μl of ThT stock solution (100 μM) and 276 μl of TBS buffer. The mixture was vortexed for 30 s before fluorescence measurements. Fluorescence emission of ThT-UHC fibril complexes was measured with an excitation wavelength of 450 nm to monitor the formation of the fibrils, using a Hitachi F-2500 fluorescence spectrophotometer. TBS without fibrils was used as a control. All measurements were carried out in triplicates.

AFM.

A 10-μl drop of the UHC fibril solution was spotted on freshly cleaved mica, exposed to air for 10 min, then washed with 200-μl aliquots of Milli-Q ultrapure water three times, and finally dried with a stream of nitrogen for 4 min. Topographical images were obtained in air with the contact mode using an Innova AFM (Bruker, Santa Barbara, CA). A silicon nitride cantilever with a spring constant of 0.12 N/m and a nominal tip radius of 2 nm was used. Each sample was characterized at multiple locations with various scan areas to ensure the detection of representative structures. Analysis of the atomic force microscopy (AFM) images was carried out using Nanoscope software.

Display of GFP-nanobody on the yeast cell surface.

Plasmid pCTNG was used to transform EBY100 cells by electroporation with a Gene Pulser (Bio-Rad). To induce display of GFP-nanobody, cells harboring pCTNG were precultured in SD-CAA medium to an OD₆₀₀ of 2 to 5 at 30°C. The preculture was then centrifuged, and the cell pellet was resuspended in 50 ml of SG-CAA medium to an OD₆₀₀ of 1 and incubated for 48 h at 22°C with shaking.

Protein assembly on the yeast cell surface.

To assemble the dockerin-tagged proteins on the yeast surface, purified GFP-dockerin (GD) or GFP-cohesin (GC) proteins were first incubated with the yeasts that display the GFP-nanobody (termed Yn yeast) for 15 min with shaking at 4°C in the TBS-Ca buffer to obtain Yn-GD or Yn-GC yeasts, respectively, and then washed to remove unbound GFP-dockerin or GFP-cohesin. The Yn-GD yeasts were subsequently incubated with excess UHC amyloid fibrils for 30 min at 4°C and washed to remove unbound fibrils. Both types of yeast cells were then incubated with mCherry-dockerin or CelA for 30 min at 4°C and harvested by centrifugation (4,000 × g for 5 min) at 4°C. Pelleted yeast cells were washed five times to remove free proteins and resuspended in TBS-Ca buffer for subsequent experiments.

Confocal laser scanning microscopy and flow cytometry.

Upon initial examination using epifluorescence microscopy on an Olympus BX60 microscope, yeast cells displaying GFP and mCherry fusion proteins were examined using confocal laser scanning microscopy to further characterize localization of the fusion proteins on the cell surfaces. Cells were examined with an Olympus Fluoview FV-1000 confocal laser scanning system mounted on an Olympus IX-81 inverted microscope. GFP and mCherry were excited at 488 nm and 543 nm, respectively, and fluorescence emission was collected through BA505-525 nm and BA560IF filters, respectively. Observations were made with a UPLSAPO 60× oil immersion lens (numerical aperture [NA], 1.35) and a 4.0× zoom factor. Z-stacks of 1,024 by 1,024 pixels were obtained by taking optical slices at 0.46-μm intervals. Yeast cells with different protein assembly display treatments were also analyzed on a flow cytometer (Beckman Coulter Altra, Fulerton, CA) using a 488-nm laser for GFP or a 543-nm laser for mCherry to compare the fluorescence intensities of the cell populations. Fifty thousand cells were analyzed for each sample, and data were analyzed using the CellQuest Pro software.

Enzyme assays.

Low-viscosity carboxymethyl cellulose (CMC; Sigma-Aldrich) was used as a substrate to measure the activity of CelA immobilized on the yeast cell surface with different protein assembly treatments. Enzyme activity was assayed in the presence of a 0.8% (wt/vol) concentration of CMC at 37°C in 25 mM TBS-Ca buffer (pH 7). Samples of 100 μl were removed periodically from the reaction mixture and immediately mixed with 100 μl of DNS reagents (10 g/liter dinitrosalicylic acid, 10 g/liter sodium hydroxide, 2 g/liter phenol, 0.5 g/liter sodium sulfite). After the sample was incubated at 95°C for 5 min, 50 μl of 40% KNa tartrate was added to fix the color. After centrifugation at 18,000 × g for 1 min, a 200-μl aliquot of the resultant supernatant was placed in a 96-well plate, and the absorbance at 580 nm was measured using a Tecan microplate reader. Product yields were determined from a standard curve prepared using glucose as a standard.

RESULTS

Self-assembly of UHC fusion protein monomers into amyloid-like nanofibrils.

The UHC fusion protein is highly expressed in E. coli BL21(DE3)TrxB at an estimated 50 mg/liter of culture, based on anti-His tag Western blot analysis of the cell extract. Upon staining with ThT, E. coli cells expressing UHC displayed a strong green fluorescence, whereas control cells (E. coli expressing CelA) showed no such fluorescence (Fig. 1). ThT binds specifically to amyloid-like fibrils and emits green fluorescence at an excitation wavelength of 450 nm, which is commonly used to diagnose amyloid fibrils (15). The observation of intense green fluorescence in UHC-expressing cells, but not in control cells, indicates that the UHC fusion proteins self-assembled into amyloid-like fibers inside the E. coli host.

Fig 1.

UHC-expressing E. coli BL21 cells showed green fluorescence when incubated with 10 μM ThT. No fluorescence was noted when ThT was incubated with E. coli ells expressing CelA, which were used as controls.

After purification, the UHC fibril formation kinetics was monitored with ThT binding fluorescence at 22°C. Upon incubation with ThT, the UHC fibrils displayed a fluorescence spectrum characteristic of amyloid-like materials (Fig. 2a). UHC fibrillation in solution occurred very rapidly, as indicated in the time course of peak fibril ThT fluorescence (Fig. 2b). UHC nanofibrils formed in the presence of BSA were found to be fluffy and fibrous, whereas large clumps appeared in the absence of BSA (Fig. 2c versus d).

Fig 2.

Characterization of UHC fibrils. (a) ThT fluorescence emission spectrum of UHC fibrils incubated with ThT and excited at 450 nm. (b) Kinetics of UHC fibril formation monitored using ThT binding. (c and d) Influence of BSA on the formation of UHC fibrils. UHC (20 μM) was incubated in TBS buffer without (c) or with (d) 0.5% BSA at 22°C with shaking and then stained with ThT before being viewed with a fluorescence microscope. Scale bar, 10 μm. (e) Formation of amyloid-like fibrils by the UHC proteins, as revealed by AFM (scan area, 5 by 5 μm²). AU, arbitrary units.

Morphology of the UHC nanofibrils was further analyzed using AFM. The UHC fibril solution was adsorbed onto a mica surface and scanned with a nanometer-sized probe on a flexible microcantilever. The AFM images (Fig. 2e) revealed the detailed structure of the UHC nanofibrils, which is highly similar to that of native Ure2p fibrils (5, 14). The diameter of the UHC fibrils based on the height measurement of the fibril cross-section using AFM was approximately 5 nm. In comparison, the diameter of wild-type Ure2p fibrils was estimated by Bousset et al. (5) to be 11 nm based on AFM measurement. The wild-type Ure2p monomer contains a C-terminal glutathione S-transferase (GST)-like domain of 33 kDa, while the HD-Coh moiety of the UHC fusion protein is about 27 kDa. In the present study, the UHC fibrils were prepared in a buffer containing BSA to minimize disordered aggregation (Fig. 2). As such, the UHC fibrils were embedded within a BSA monolayer coated on the hydrophilic mica as a substrate for AFM measurement. After adsorption to the mica surface, the BSA monolayer has a thickness of about 4 nm (29). AFM measures only relative height. Therefore, the 5 nm we measured may be the height difference between fibers and the BSA monolayer on mica. This will give a total thickness of the fibers to be about 9 nm. As indicated in the AFM image, most UHC fibrils are single, and only a few are bundled or paired. These fibrils, however, do intertwine with each other and form a large fibril network with fluffy structures, as seen in Fig. 2d. The self-assembled amyloid-like supramolecular UHC scaffoldin was confirmed to display multiple units of functional cohesin domains that are capable of binding dockerin fusion proteins. This was achieved initially by labeling the UHC nanofibrils using dockerin-tagged fluorescent proteins. Intense fluorescence was found to associate with the UHC fibrils after they were labeled with GFP-dockerin or mCherry-dockerin, while no fluorescence was noted on the UHC fibrils when they were labeled using GFP and mCherry (data not shown). Additional proof of active cohesin in the UHC fibrils comes from yeast display results, which are discussed below.

Formation of functional UHC scaffoldin on the yeast surface.

To create yeasts that display the GFP-nanobody (i.e., the Yn yeast), EBY100 harboring the chromosomal Aga1 gene and episomal plasmid pCTNG was induced with galactose to express both the Aga1p and Aga2p::GFP-nanobody (AGN) fusion protein under the control of the GAL1 promoter. The AGN fusion protein thus expressed binds Aga1p through a pair of disulfide bonds and is secreted and anchored on the yeast cell surface via Aga1p. To anchor the UHC scaffoldin on the yeast cell surface, Yn yeasts were preincubated with GFP-dockerin so that multiple dockerin domains were anchored on the yeast surface via the strong interaction between GFP and GFP-nanobody. The highly specific cohesin-dockerin interaction is then utilized for efficient docking of the UHC scaffoldin on the GFP-dockerin-tethered Yn yeast cells. Subsequently dockerin-tagged mCherry was used to label the surface-displayed UHC scaffoldin complex (Fig. 3a). For comparison, Yn yeast pretreated with a GFP-cohesin fusion protein was also labeled with mCherry-dockerin (Fig. 3b). Given that the cohesin domains in the free UHC scaffoldin were found to be accessible to dockerin fusion proteins as discussed earlier, it was anticipated that an elevated level of mCherry-dockerin would be immobilized on the yeast surface displaying the UHC scaffoldin, which contains many copies of the cohesin domains. Here, we confirmed that this is indeed the case, as revealed first by epifluorescence microscopy and subsequently by using confocal laser scanning microcopy, whole-cell fluorescence measurement, and flow cytometry. As seen in Fig. 4, when examined using an epifluorescence microscope equipped with a GFP or red fluorescent protein (RFP) filter cube, Yn-GD-UHC-MD (Fig. 3a) and Yn-GC-MD (Fig. 3b) cells displayed comparable GFP fluorescence intensities while the latter exhibited much stronger mCherry fluorescence. Here, Yn-MD, Yn-GD-MD, and Yn-UHC-MD cells were used as controls to confirm the absence of nonspecific binding between MD and Yn or GD. UHC also showed no nonspecific interaction with Yn (i.e., GD is required to link UHC to Yn).

Fig 3.

Proposed schematics of the protein complexes assembled on the cell surface of Yn-GD-UHC-MD yeast (a) and Yn-GC-MD yeast (b). I, AGN (yeast Aga2p fused with GFP nanobody); II, GD (GFP-dockerin); III, UHC (Ure2p^1-80 fused with HD and cohesin); IV, MD (mCherry-dockerin); V, GC (GFP-cohesin).

Fig 4.

Probing protein complex formation on the surface of yeasts displaying GFP-nanobody (Yn) via epifluorescence microscopy. Yn+MD, Yn incubated with MD; Yn-GD+MD, Yn sequentially incubated with GD and MD; Yn+UHC+MD, Yn sequentially incubated with UHC and MD; Yn-GC-MD, Yn sequentially incubated with GC and MD; Yn-GD-UHC-MD, Yn sequentially incubated with GD, UHC fibrils, and MD; Yn-GD-UHC-MD/GD, Yn sequentially incubated with GD, UHC fibrils, and a mixture of MD and GD.

To further characterize the UHC scaffoldin-mediated protein complex display on yeast surface, Yn-GC-MD and Yn-GD-UHC-MD cells (Fig. 3) were examined using confocal laser scanning microscopy to obtain higher-resolution optical images of the cells that allow a clearer view of the spatial distribution of the fluorescent reporter proteins on the yeast surface. As shown in Fig. 5 with composite images (composite of green and red fluorescent images) of Yn-GC-MD and Yn-GD-UHC-MD cells, the former exhibited a thin and uniform yellow protein layer around the cell surface (Fig. 5, left), which indicated colocalization of GFP and mCherry, which is expected of Yn-GC-MD. The Yn-GD-UHC-MD cells, on the other hand, had a thin layer of GFP (indicated by a thin and uniform layer of green fluorescence around the cells when they were excited using a 488-nm laser [data not shown]) covered by a thick layer of mCherry fluorescence with varied thickness around the cell surface, indicating that a large amount of MD was anchored on the cell surface via the oligomeric-cohesin UHC scaffoldin (Fig. 5, right).

Fig 5.

Confocal laser scanning microscopy composite images of Yn-GC-MD and Yn-GD-UHC-MD cells.

In order to quantify and compare the extent of mCherry-dockerin and GFP-dockerin display on Yn-GD-UHC-MD (Fig. 3a) versus Yn-GC-MD (Fig. 3b) yeast cells, whole-cell GFP and mCherry fluorescence levels of the respective cultures were determined using a fluorescence spectrophotometer. The result showed that the GFP fluorescence intensities of these cells were similar (differed by less than 2%), indicating that the cells adsorbed similar quantities of GFP fusion protein, which is expected. On the other hand, the mCherry fluorescence intensity of the Yn-GD-UHC-MD yeast was significantly higher than that of the Yn-GC-MD yeast (differed by 24.8-fold).

While the whole-cell fluorescence data indicate the overall fluorescence of the entire cell population, they do not provide information on the variation of specific fluorescence properties among the cells within the population tested. To this end, Yn-GD-UHC-MD and Yn-GC-MD cells were subjected to flow cytometry analysis using a 543-nm laser (to excite mCherry). As seen in Fig. 6, the Yn-GD-UHC-MD cell population contains a significantly higher portion and number of cells displaying intense mCherry fluorescence than the Yn-GC-MD population. A large number of the Yn-GD-UHC-MD cells exhibited elevated side scatter signal intensity with high mCherry fluorescence, likely resulting from display of large UHC scaffoldin-MD complexes on the yeast surface, as revealed in the confocal images (Fig. 5). Side scatter signals typically provide information about internal granularity and cell surface roughness (23). The flow cytometry analysis therefore further confirmed the presence of large UHC scaffoldin-MD complexes on the surface of the Yn-GD-UHC-MD cells, which should increase the roughness of the yeast cell surface. Yn-GD-UHC-MD and Yn-GC-MD cells were also subjected to flow cytometry analysis using a 488-nm laser (to excite GFP). In contrast to the mCherry fluorescence, GFP fluorescence distributions for Yn-GD-UHC-MD and Yn-GC-MD cells were nearly identical (data not shown). This is as expected since both cell types ought to accommodate similar levels of GFP on their cell surfaces. Taken together, these data demonstrate that the self-assembled UHC scaffoldin serves as an effective protein scaffold to significantly amplify the amount of proteins that can be immobilized and displayed on the yeast surface.

Fig 6.

Flow cytometry analysis of yeast cells with different molecular assembly display treatments. Note that significantly higher numbers of cells display an elevated level of mCherry fluorescence when UHC fibrils are used (i.e., Yn-GD-UHC-MD).

Given that the UHC scaffoldin contains many copies of cohesin domains, it should be possible to bind different types of dockerin fusion proteins concurrently. This was verified by incubating both GD and MD with the Yn-GD-UHC cells to create the Yn-GD-UHC-MD/GD cells. From the epifluorescence images presented in Fig. 4, it is evident that the GFP fluorescence from these cells is different from that of the Yn-GC-MD or Yn-GD-UHC-MD cells. For the latter, a single layer of GFP is expected to anchor on the cell surface since one GD binds one GFP-nanobody, and only a single layer of GFP-nanobody is expected on the yeast surface. For the Yn-GD-UHC-MD/GD cells, on the other hand, the additional GD proteins are expected to adhere to the fibrous UHC scaffoldin, and thus the green fluorescence on the cell surface no longer appears as a smooth layer. The fact that the fibrous structures displayed on the Yn-GD-UHC-MD/GD cells emitted both GFP and mCherry fluorescence (Fig. 4) indicates that GD and MD successfully codisplayed on the cell surfaces by binding to the cohesin domains on the UHC scaffoldin. The result has important implications for improving or introducing novel biocatalytic activities by assembling on the yeast surface novel multienzyme complexes or coimmobilizing cofactor-dependent enzymes and cofactor-regenerating enzymes on the same protein scaffold to benefit from the proximity of the enzymes (6, 24).

Augmented enzyme display on yeast surfaces using the UHC scaffoldin.

To investigate the utility of the UHC-engineered yeasts, experiments were conducted to determine whether the unique ability of these yeasts to display increased amounts of enzymes can lead to augmented overall enzymatic activity, especially with macromolecular substrates. Here, the endoglucanase CelA from C. cellulolyticum that carries a native dockerin domain at its C terminus is used as a model enzyme. To evaluate the functionality of the UHC scaffoldin for augmented enzyme display, recombinant CelA expressed in E. coli and purified with IMAC was incubated with Yn-GC or Yn-GD-UHC cells to create Yn-GC-CelA or Yn-GD-UHC-CelA cells, respectively. Hydrolysis of CMC (average molecular mass of 90 kDa) using these two types of yeast cells was compared. Based on the initial rate of reducing sugar release from CMC, the specific hydrolytic activity of Yn-GD-UHC-CelA was estimated to be about 8.5-fold higher than that of Yn-GC-CelA (Fig. 7).

Fig 7.

Time course of CMC hydrolysis using yeast cells displaying CelA via different assembly methods (Fig. 3).

DISCUSSION

In this study, we have established a method for amplifying the extent and variety of proteins displayed on the surface of Saccharomyces cerevisiae and demonstrated its utility. A key component of this method is a self-assembled supramolecular scaffoldin that displays a large number of functional cohesin domains. Protein complexes built from self-assembly of identical subunits (monomers) are interesting molecular structures that can be exploited for displaying multiple protein/peptide cargos. Mitsuzawa et al. (19) fused a Clostridium thermocellum cohesin module to a subunit of group II chaperonins (also known as Rosettasomes) from Sulfolobus shibatae and showed that the resulting fusion protein self-assembled into 18-unit protein complexes in the presence of ATP and Mg²⁺. Similarly, Heyman et al. (12, 20) fused a type I cohesin to stable protein 1 (SP1) from Populus tremula, and the resulting Coh-SP1 fusion protein was able to self-assemble into a ring-shaped protein complex containing 12 units of Coh-SP1 per complex. In this study, we fused a cohesin domain to the yeast Ure2p fibrillogenic domain, via the HD module as a structural linker, and demonstrated self-assembly of the UHC fusion protein monomers into an amyloid-like oligomeric-cohesin scaffoldin. This approach enables presentation of a much larger number of cohesin domains per scaffoldin compared with previous approaches. It also offers a molecular architecture (mainly fibrous) different from those in Rosettasomes or SP1 complexes (both are ring shaped). The UHC monomers that make up the supramolecular UHC-based scaffoldin are relatively small and hence are encoded by a short open reading frame (ORF). UHC can be expressed as soluble protein with high yield. In native Clostridium scaffoldins, multiple cohesin modules are linked covalently and encoded on a single and very long ORF. For instance, the CipC scaffoldin coding sequence is more than 4.6 kb. Expression of such a long ORF that contains many repeated sequences in heterologous hosts might be challenging especially if high protein yield is necessary. To this end, self-assembly from small protein subunits into an ordered supramolecular complex offers an attractive alternative in creating oligomeric or polymeric cohesin scaffolds.

We found that inclusion of the C. cellulolyticum cellulosomal hydrophilic domain (HD) in the UHC fusion protein facilitates formation of the amyloid nanofibrils. When the HD module was omitted from the fusion protein, i.e., with direct fusion between Ure2p^1-80 and cohesin, though amyloid fibrils still formed, the process was much slower, and a larger proportion of disordered aggregates was found (data not shown). It is known that the property of the C-terminal protein appended to the Ure2p fibrillogenic domain can have a considerable effect on fibril formation (1, 2, 4). It is also important to incorporate an appropriate linker between the fibrillogenic domain and the appended cargo protein to avoid steric hindrance that may hamper proper packaging of the cross-beta sheets to form an ordered fibril structure with a properly oriented C-terminal cargo protein (10). The HD domain (first X2 module) is found directly linked upstream of the first cohesin module in the C. cellulolyticum CipC scaffoldin. Since cohesin is a relatively hydrophobic protein, the presence of the HD domains in CipC is thought to increase the solubility of the scaffoldin protein (21). In our UHC scaffoldin, the HD domain may function as a structural linker that improves the presentation of the cohesin module to render the domains more exposed and accessible, and it may also help the folding of the fusion protein. We also noticed a considerable improvement in forming smaller and more uniform fibrils by including BSA in the UHC fusion protein solution. This likely resulted from reducing any nonspecific protein interaction in solution.

Recently, the usefulness of synthetic protein scaffolds that spatially recruit synergistic pathway enzymes was demonstrated by putting three mevalonate biosynthetic enzymes on a single synthetic complex to significantly improve mevalonate biosynthesis (6). Other well-known examples that involve multiple synergistic enzymes on the same protein scaffold are the cellulosomes. Tsai et al. successfully displayed on S. cerevisiae a synthetic trifunctional mini-CipC₁-like scaffoldin (22) that contains three divergent cohesin domains (25, 26). Upon binding to three different types of cellulolytic enzymes on the surface-displayed scaffold, Tsai et al. were able to show ethanol fermentation from these engineered yeasts using cellulosic substrates. Similarly, Lilly and collaborators engineered a chimeric scaffoldin protein Scaf3p from C. cellulolyticum that contains two different cohesin domains and targeted the Scaf3p to display on the S. cerevisiae cell surface (17). Wen and coworkers constructed yeast strains displaying a series of uni-, bi-, and trifunctional minicellulosomal scaffoldins composed of a cellulose binding domain and cohesins (27). Instead of using cohesins from different species, Ito and colleagues made use of the cohesin from C. cellulovorans and the Z domain of protein A derived from Staphylococcus aureus to synthesize a scaffoldin protein and displayed it on S. cerevisiae cell surfaces (13). In these previous studies, no more than three cohesin modules were incorporated in the miniscaffoldin that was displayed on the yeast surface. The UHC-based supramolecular scaffoldin reported in the present study incorporates a much higher number of cohesin modules capable of binding a large amount of dockerin-bearing proteins, as shown using dockerin-tagged fluorescent proteins and the CelA enzyme.

The unique design of the supramolecular cohesin scaffoldin is potentially useful for presenting multiple types of enzymes on the same scaffold to synergistically catalyze complex reactions. With properly designed scaffolded multienzyme systems, one may accelerate conversion by taking advantage of the enhanced local intermediate concentrations resulting from the spatial control over enzyme organization (28). Finally, while this study demonstrated a new scaffoldin concept in yeast, such a technique should be equally applicable for improving protein display on other hosts as novel whole-cell biocatalysts, or even on nonbiological nanoparticles, to assemble nanofactories that comprise multiple functional modules (7).