Screening and directed evolution of transporters to improve xylodextrin utilization in the yeast Saccharomyces cerevisiae

The economic production of cellulosic biofuel requires efficient and full utilization of all abundant carbohydrates naturally released from plant biomass by enzyme cocktails. Recently, we reconstituted the Neurospora crassa xylodextrin transport and consumption system in Saccharomyces cerevisiae, enabling growth of yeast on xylodextrins aerobically. However, the consumption rate of xylodextrin requires improvement for industrial applications, including consumption in anaerobic conditions. As a first step in this improvement, we report analysis of orthologues of the N. crassa transporters CDT-1 and CDT-2. Transporter ST16 from Trichoderma virens enables faster aerobic growth of S. cerevisiae on xylodextrins compared to CDT-2. ST16 is a xylodextrin-specific transporter, and the xylobiose transport activity of ST16 is not inhibited by cellobiose. Other transporters identified in the screen also enable growth on xylodextrins including xylotriose. Taken together, these results indicate that multiple transporters might prove useful to improve xylodextrin utilization in S. cerevisiae. Efforts to use directed evolution to improve ST16 from a chromosomally-integrated copy were not successful, due to background growth of yeast on other carbon sources present in the selection medium. Future experiments will require increasing the baseline growth rate of the yeast population on xylodextrins, to ensure that the selective pressure exerted on xylodextrin transport can lead to isolation of improved xylodextrin transporters.


Introduction
Cellulosic biofuel production from plant biomass requires efficient use of all abundant carbohydrates in the plant cell wall [1,2]. To make the fermentation process economically feasible, engineered yeast should be able to consume a mixture of sugars naturally released by enzyme cocktails simultaneously, including hexose and pentose sugars [3][4][5]. Xylodextrins (XDs), such as xylobiose and xylotriose, are oligomers of β-1,4-linked xylose. They are derived from hemicellulose, one of the major forms of biomass in lignocellulose, and are hydrolyzed to xylose by β-xylosidase. The first example of ethanol production from xylan was reported in 2004, wherein xylanase and β-xylosidase were displayed on the cell surfaces of S. cerevisiae expressing a xylose consumption pathway [6]. However, the xylan degrading and xylose utilization abilities of this recombinant S. cerevisiae strain requires further optimization to be industrially useful. Recently, a XD utilization pathway from N. crassa, which requires a transporter CDT-2 along with two intracellular β-xylosidases GH43-2 and GH43-7, was identified and subsequently engineered into S. cerevisiae, enabling the yeast to grow aerobically on XD, or co-ferment XDs with xylose or other hexose sugars [7]. However, the consumption rate of XDs remains quite slow and thus needs to be improved for future industrial applications.
The transporter CDT-2 belongs to the Major Facilitator Superfamily (MFS), one of the largest and most ubiquitous secondary transporter subfamilies [8]. Its members exist in all species from bacteria to mammals, and they range in size from 400 to 600 amino acids, organized into 12 trans-membrane α-helices [9]. CDT-2 belongs to the hexose family of MFS transporters, which includes yeast hexose transporters, human glucose transporters, and xylose transporters [10][11][12][13].
For industrial applications, naturally occurring transporters are generally not optimal without further engineering. Protein engineering, including rational design and directed evolution approaches, has been widely used to improve the performance of a wide variety of enzymes and pathways. However, engineering membrane proteins such as MFS transporters has proven to be challenging [14][15][16]. Directed evolution is an important means for carrying out protein engineering [17], but requires developing a high-throughout screening system [18,19]. In 2014, Ryan et al. developed a new method for performing directed evolution experiments in yeast using the CRISPR-Cas9 system [20]. Integrating linear DNA into the genome by homologous recombination mediated CRISPR-Cas9 overcame high levels of copy number variation seen with plasmid-based expression, and allowed directed evolution of CDT-1 to isolate an improved cellodextrin transporter [20].
To improve xylodextrin utilization in S. cerevisiae, we sought to identify potential XD transporters by characterizing a library of CDT-1 and CDT-2 orthologues. We analyzed cellular localization, transport activity and aerobic growth profiles. Moreover, with one of the best performing transporters identified in the screen, we attempted to use CRISPR-Cas9 mediated directed evolution to improve XD consumption in yeast.

Results
To identify XD transporters that could be used to improve utilization of xylodextrins, we carried out a screening of codon-optimized CDT-1 and CDT-2 transporter orthologues, named STX (X spanning from 1 to 17) (S1 Table). Phylogenetic analysis suggested that orthologues of CDT-2 are widely distributed in the fungal kingdom, indicating that many fungi are able to consume xylodextrins derived from plant cell walls ( Figure 1). These orthologues are 28-67% identical in amino acid sequence to CDT-2.
We first constructed plasmids expressing ST1-ST17 fused to enhanced GFP at the C-terminus, using the strong TDH3 promoter and the CYC1 terminator, and expressed the transporters in S. cerevisiae strain D452-2. Using epi-fluorescent microscopy, we found that most transporters localize to the plasma membrane ( Figure 2). Transporters ST5, ST8, ST10 and ST14 are dispersed in the cytoplasm or vacuole (Figure 2), indicating these transporters are not well folded in S. cerevisiae, and they were not analyzed further.
Xylodextrins, including xylobiose and xylotriose, are β-1,4-linked xylose oligomers with different degree of polymerization. As a direct test of transport activity for cellobiose (G2), xylobiose (X2) and xylotriose (X3), we chose transporters that localized to the plasma membrane ( Figure 2) and conducted a yeast cell-based sugar uptake assay. Notably, we found that ST3, ST15 and ST16 are XD-specific transporters, while other transporters showing XD transport activity are able to transport cellobiose as well, although the overall activity varied ( Figure 3). The uptake activity of CDT-2 for both X2 and X3 was more than ten times that of ST9, ST11 and ST12 ( Figure 3). Thus, these transporters were not studied further. The sugar preferences for ST transporters are summarized and shown on the phylogenetic tree in Figure S1. which has led to many efforts to relieve this inhibition [21,22]. We had previously observed that cellobiose inhibits xylobiose transport by CDT-2 [7]. We therefore tested xylobiose transport activity in the presence of increasing molar concentrations of cellobiose, up to 10 times the concentration of X2. Interestingly, we found that the uptake of xylobiose by ST15 and ST16 was not inhibited by cellobiose (Figure 4), making them good candidates for co-fermentation with glucose or cellodextrin from lignocellulosic biomass under a number of pretreatment scenarios. However, for other transporters, there was less X2 consumed per OD when increasing molar concentration of G2 were present in the media, i.e. as observed for CDT-2 ( Figure 4).
To further correlate the sugar uptake activity with yeast growth, we engineered the xylose-consuming S. cerevisiae strain SR8 [23] using CRISPRm (S2 Figure), by insertion of GH43-2 and GH43-7 at the TRP1 and LYP1 loci, respectively. The engineered SR8 strain was named SR8A. Yeast strain SR8A cells with plasmids expressing different transporters were cultured in oMM media under aerobic conditions for 96 hours using xylodextrin as the only carbon source. ST2 and ST16 enabled S. cerevisiae to grow more rapidly on xylodextrin ( Figure 5). The area under the curve (AUC) of ST2 and ST16 were 1.16 and 1.26 times higher than CDT-2, respectively. We also tested the anaerobic growth profile of ST2 and ST16 transporter on XD plus sucrose, but there were no significant improvements compared with CDT-2 in anaerobic conditions. Taken together, three potentially useful xylodextrin transporters were identified from the screen, each with different advantages and disadvantages (S1 Table). Of these, ST16 is a XD-specific transporter, showing faster aerobic growth and its xylobiose transport activity is not inhibited by cellobiose. We therefore chose ST16, which is derived from Trichoderma virens, as the starting point for setting up a directed evolution experiment to improve XD consumption. Based on experiments published previously, in which XD consumption was rapidly turned on and off in anaerobic conditions [7], we hypothesized that XD transport is limiting in terms of consumption rate. Notably, when we inserted the transporter ST16 under control of PGK1 promoter and TPS3 terminator into the yeast genome at the LEU2 locus, there was a dramatic decrease in the growth rate compared with overexpression ST16 from a plasmid.
Furthermore, the strain with incorporated ST16 grew almost as slow as the control ( Figure 6A). GFP fluorescence emission spectroscopy verified that ST16 was expressed from the chromosome, but at lower levels than overexpression from a plasmid ( Figure 6B).
To further test whether transport is limiting in XD-utilizing strains, we compared the aerobic growth profile of the SR8 strain with a plasmid expressing CDT-2, GH43-2 and GH43-7 and the SR8A strain, which has GH43-2 and GH43-7 expressed from chromosomal copies, with a plasmid only expressing the CDT-2 transporter. We reasoned that incorporation of the genes for GH43-2 and GH43-7 into the chromosome would drop the protein expression level compared with overexpression using a plasmid, because of the much higher copy number of plasmids [20,24]. As expected for XD transport being the limiting factor for XD utilization, expressing GH43-2 and GH43-7 from the chromosome did not affect aerobic growth on xylodextrin (S3 Figure). These results further supported that the XD transporter is the rate-limiting step of the XD utilization pathway.
Next, we attempt to develop a CRISPR-Cas9 mediated directed evolution selection experiment [20] using growth in liquid media followed by screening based on colony size. We first used error prone PCR to generate a mutant library of ST16 of 2x10 3 variants and transformed the library into the SR8A strain. Pooled strains from YPAD+G418 plates were grown aerobically in liquid synthetic media containing XD for 2 days and plated on plates with synthetic media containing XD to enrich for functional ST16 mutants. Surprisingly, after 4 days of aerobic growth on plates, all of the colonies had similar sizes, and we were not able to identify any individual colony that demonstrated faster aerobic growth. Furthermore, we conducted a serial dilution of xylodextrin concentration in making the synthetic plates, and found that the major roadblock with the selection strategy is the background growth of the strains on contaminating xylose and amino acids (S4 Figure), resulting in difficulty identifying functional ST16 mutants. Thus, the yeast strain was able to grow on alternative carbon sources rather than XD, circumventing the selective pressure of XD-dependent growth.

Discussion
CDT-1 and CDT-2 are two important cellodextrin transporters used by N. crassa in cellulose degradation [25], of which only CDT-2 is also involved in hemicellulose utilization [7,26]. In previous results, the discovery of the xylodextrin consumption pathway consisting of transporter CDT-2 along with two β-xylosidases GH43-2 and GH43-7 provided new modes for utilization of xylose derived from hemicellulose [7].
However, the XD pathway as isolated from N. crassa requires significant improvement for its future use industrially. In our study, we confirmed that the XD transporter is the rate-limiting step of the XD utilization pathway ( Figure 6A and S3 Figure). Through characterization of a library of CDT-1 and CDT-2 transporter orthologues, we identified three potential transporters from other fungi that could be used as targets for exploring faster xylodextrin utilization (S2 Table). Of these, the transporter ST16 isolated from Trichoderma virens has advantages over CDT-2, including the fact that it is a XD-specific transporter with good membrane localization, and enables faster aerobic growth. Furthermore, the xylobiose transport activity of ST16 is not inhibited by the presence of cellobiose.
Given the superior properties of the ST16 transporter, we attempted to develop a CRISPR-Cas9 mediated directed evolution experiment, using ST16 as the starting point to screen mutants that could further improve xylodextrin utilization efficiency.
We used liquid medium containing XD for selection, and subsequent screening for larger colonies on XD-containing plates. Through mutagenesis and chromosomal integration, we screened 2x10 3 ST16 mutants. However, due to the background growth of these strains on xylose and amino acids, it was hard for us to select large colonies, resulting in difficulty in identifying more functional ST16 mutants. To overcome remaining bottlenecks of the current selection strategy, it will be necessary to remove the contaminating xylose from XD preparations, and minimize the amino acid concentrations in the media. In addition, further experiments need to be done for increasing the expression level of chromosomally integrated ST16, as we found that using the PGK1 promoter was not sufficient to drive growth above background levels on XD-containing plates. Although not ideal, it may be necessary to first carry out directed evolution using ST16 expression from a plasmid, to increase ST16 expression levels. Other strategies might also be explored for directed evolution of XD transporters, including starting with ST2 or ST15. Alternatively, it might prove useful to perform a directed evolution experiment on a cellodextrin transporter like CDT-2 to improve cellobiose uptake, then test whether isolated mutations will also facilitate faster XD consumption.
Although both ST2 and ST16 did not improve anaerobic utilization of XD compared with CDT-2, the presence of xylose or other simple sugars could be used to initiate the consumption of XD, according to our previous results [7]. These anaerobic fermentation experiments suggest that metabolic sensing might require additional tuning for optimal xylodextrin fermentation. Based on the rapid turn-off and turn-on of XD consumption in anaerobic conditions [7], we suspect that the XD transporters are internalized in the absence of hexose or xylose sugars [27]. Currently, how CDT-2 and ST16 are regulated in xylodextrin-only media remains unknown. Other systems-level experiments such as transcription and ribosome profiling could also be used to better understand the mechanism by which the yeast strain senses xylodextrins under anaerobic conditions. Finally, the cofactor imbalance problem of the XR/XDH pathway may lead to the accumulation of xylose in the culture supernatant [28], indicating that the metabolic sensing and xylose assimilation pathway might require additional tuning for optimal xylodextrin fermentation.

Strains and plasmids
The N. crassa and the codon-optimized versions of all ST transporters were cloned into the pRS316 plasmid (CEN URA3)-under the control of the S. cerevisiae TDH3 promoter and the CYC1 terminator-using the In-Fusion HD Cloning Kit (Clontech Laboratories, Inc., Mountain View, CA). The cleavage site for the HRV 3C protease followed by eGFP was fused to the C-terminus. cdt-1 and cdt-2 were PCR-amplified from cDNA synthesized from mRNA isolated from N. crassa (FGSC 2489) grown on minimal media plus Avicel (microcrystalline cellulose) as the sole carbon source [25].
Atum (formerly DNA 2.0) performed gene synthesis and codon-optimization for all ST transporters. S. cerevisiae strain D452-2 [29], was used for XD and cellobiose transport studies. Codon-optimized gh1-1 was cloned into pRS315 plasmid (CEN LEU) and was expressed under the control of the PGK1 promoter. Details of the codon optimization process for gh1-1 are described in [30].
S. cerevisiae strain SR8 [23] was used for strain engineering. We used CRISPR-Cas9 genome editing to insert GH43-2 and GH43-7 into the TRP1 and LYP1 loci, respectively, and the engineered SR8 strain was named SR8A. GH43-2 was under the control of TDH3 promoter and the ADH1 terminator, while GH43-7 was drove by CCW12 promoter and the CYC1 terminator.

Epi-fluorescence Microscopy
D452-2 cells expressing the transporters were grown in oMM-Ura plus 2% glucose media to mid-exponential phase at 30 °C. The cultures were centrifuged, spotted onto glass slides and examined on a Leica DM 5000B Epi-fluorescence microscope at 100x DIC (Leica, Germany). Transporters were visualized using the L5 filter cube; images were captured using the Leica DFC 490 camera and analyzed with the accompanying microscope software.

Yeast-cell based sugar uptake assay
Yeast strain D452-2 cells transformed with pRS316-CDT1, -CDT2 or pRS316-ST were grown to mid-exponential phase. Cells were harvested and washed twice with Transport Assay Buffer (5 mM MES, 100 mM NaCl, pH 6

HPLC analysis
HPLC analysis was performed on Dionex ICS-3000 HPLC using a CarboPac PA200 analytical column (150 x 3 mm) and a CarboPac PA200 guard column (3 x 30 mm) at room temperature. 25 µL of sample was injected and run at 0.4 mL/min with a mobile phase using 0.1 M NaOH. Acetate gradients were used to resolve xylodextrin samples. For xylobiose and xylotriose, the acetate gradients were as follows: 0 mM for 1 min, increasing in 8 min to 80 mM, then to 300 mM in 1 min, keeping at this concentration for 2 min, followed by equilibration back to 0 mM for 3 min.