Construction of a novel CBS biocatalyst by genetic engineering of C. thermocellum
Because the cellobiose inhibition to the cellulosome is considered one of the major problems that hinder the application of C. thermocellum as a CBS biocatalyst, and the introduction of beta-glucosidases (BGL) can greatly release the inhibition effect, we have previously constructed a C. thermocellum recombinant strain ∆pyrF::CaBglA to produce a fusion protein of the cellobiohydrolase Cel48S with a heterologous BGL [28]. Although ∆pyrF::CaBglA showed high cellulosic sugar productivity, the fusion protein was expressed at a significantly decreased level compared to that of the wild-type Cel48S in the parent strain [28]. Cel48S plays key roles in the cellulosome of C. thermocellum for cellulose degradation [29,30,31]. Hence, the low expression of Cel48S may influence the saccharification efficiency.
To avoid the reduced Cel48S expression, we determined to fuse the BGL with another cellobiohydrolase Cel9K in C. thermocellum [32] and constructed a new C. thermocellum strain ∆pyrF::KBm using previously developed seamless genome editing system [28]. The strain ∆pyrF::KBm would produce a fused protein of Cel9K with BGL containing three functional modules (GH9–CaBglAm–DocI) under the control of the endogenous Cel9K promoter. The fused sequence was confirmed by PCR and sequencing of the genomic DNA (Additional file 1).
The cellulosomal and extracellular proteins of ∆pyrF::KBm were prepared and analyzed by SDS-PAGE to confirm the expression of the fusion protein with a theoretical size of ~ 150 kDa. The samples from the parent strain ΔpyrF and the previously constructed strain ΔpyrF::CaBglAm were also analyzed in parallel. Compared to the parent strain ΔpyrF, an additional 150-kDa band was detected for ∆pyrF::KBm but the ~ 90-kDa band referring to the wild-type Cel9K protein was rarely observed. This indicated that ∆pyrF::KBm produced the fusion protein instead of the Cel9K protein (Fig. 1). The intensities of the bands referring to the primary scaffoldin ScaA, Cel48S, Cel9K and the fusion protein Cel9K–BGL were determined using the Quantity One software based on the peak intensity analysis, and the relative protein expression levels were determined by dividing their band intensities with that of the ScaA protein according to a previously described ScaA-based estimation method [33]. The Cel48S protein was with comparative expression level in ΔpyrF and ∆pyrF::KBm, indicating that the expression of Cel48S was not influenced in ∆pyrF::KBm. However, reduced expression of Cel9K–Bgl was detected in ∆pyrF::KBm compared to Cel9K in ΔpyrF. In consideration of various molecular weights (~ 100 kDa for Cel9K and ~ 150 kDa for Cel9K–BGL), the expression intensity of Cel9K–BGL was only about 40% of that of Cel9K in ΔpyrF (Fig. 1). Enzyme assay showed the cellulosomes of ∆pyrF::KBm and ΔpyrF::CaBglAm had the BGL activity of 13.1 ± 0.3 and 14.7 ± 0.2 U/mg, respectively, indicating ∆pyrF::KBm could express cellulosomal BGL at a comparative level of the previously constructed biocatalyst.
The biocatalyst ∆pyrF::KBm showed higher cellulose saccharification efficiency
The strain ∆pyrF::KBm was then used as a whole-cell biocatalyst for CBS using 100 g/L Avicel or 40 g/L sulfite-pretreated wheat straw substrate (SPS) as the sole carbon source with the inoculum size of 100% or 1%, respectively. SPS contains 65.31% of cellulose, 16.26% of hemicellulose and 9.92% of lignin. The parent strain ∆pyrF and the previously constructed biocatalyst ΔpyrF::CaBglAm were also used for saccharification under the same conditions.
With 100 g/L of Avicel as the cellulosic substrate, ∆pyrF::KBm produced 72.5 g/L reducing sugar in 18 days and the saccharification level was 65.9%, which is higher than that of ∆pyrF or ΔpyrF::CaBglAm (33% and 58.5%, respectively). The saccharification process of the C. thermocellum strains showed two phases, Phase I and Phase II, including the first 6 and later days, respectively (Fig. 2), with different sugar production rates. The sugar production curves in both Phase I and II fitted to a linear relationship with R2 values of 0.812–0.989. The slopes of the trend lines were then calculated to determine the sugar production rates in different phases. For all tested strains, the production rates in Phase I were generally higher than those in Phase II. The reduced saccharification efficiency indicated the reduced activity of the cellulosome, which might be caused by feedback inhibition or enzymatic instability. It has been reported that the cellulosomal activity was not inhibited by the produced acids and alcohols [34], and low amount of cellooligosaccharides and cellobiose also have slight influence on the cellulosomal activity [28, 35, 36]. Thus, the decreased cellulolytic activity of the cellulosome might result from the instability of the cellulosome after the long-time reaction. The production rate of ∆pyrF::KBm in Phase I was 7.87 g/L/day, which was 1.2- and 1.5-fold of that of ΔpyrF::CaBglAm and ΔpyrF, respectively. In Phase II, ∆pyrF::KBm and ΔpyrF::CaBglAm had similar sugar production rates of 1.9 and 2.1 g/L/day, respectively, but the parent strain ΔpyrF produced a low amount of reducing sugar in this phase. This result indicated that ∆pyrF::KBm had increased saccharification efficiency in terms of both saccharification level and sugar production rate compared to the parent strain and the previous generation biocatalyst.
In our previous study, the supplementation of free BGL protein slightly stimulated the saccharification efficiency of ∆pyrF::CaBglA which produces a fused protein of Cel48S and CaBglA [28]. 15 and 50 U/g cellulose of CaBglA were supplemented in the saccharification system of ∆pyrF::KBm with 100 g/L Avicel as the substrate (Additional file 2). The result showed that the saccharification level was greatly stimulated to over 75% by adding 50 U/g cellulose of the purified BGL protein. Because ∆pyrF::KBm produced a high amount of Cel48S protein but a low amount of the fusion protein Cel9K–BGL, this result indicated the importance of matching the expression level of BGL with that of Cel48S. Thus, we consider that the key to an effective CBS biocatalyst is the high and balanced expression levels and activities of Cel48S and BGL. We have previously tried to express a dockerin-bearing BGL under the control of cel48S promoter using a replicating plasmid in C. thermocellum DSM1313 to avoid the decreased expression of cellulosomal components, but detected low BGL activity and abundance in the cellulosome [28]. It might be caused by the improper promoters or plasmid backbone. The chromosomal integration of the BGL-encoding gene could also be tried in future to obtain high expression of the free BGL.
With 40 g/L SPS as the cellulosic substrate, ∆pyrF::KBm produced 30.75 g/L reducing sugar determined by 3,5-dinitrosalicylic acid (DNS) method. High-performance liquid chromatography (HPLC) analysis revealed that the produced reducing sugar contained 22.9 g/L glucose and 7.0 g/L xylose, and no cellooligosaccharides and cellobiose were detected. Because SPS contains 65.3% cellulose (0.71 g/g in glucose equivalent) and 16.3% hemicellulose (0.18 g/g in xylose equivalent), the saccharification levels of cellulose and hemicellulose were calculated as 80.6% and 97.2%, respectively. The higher saccharification level of hemicellulose than cellulose was reasonable because C. thermocellum cannot use xylose as the carbon source but can assimilate glucose, cellobiose and other cellooligosaccharides for growth [37]. In comparison, ΔpyrF::CaBglAm could produce over 30 g/L reducing sugar but required two more days and ΔpyrF only produced 22.6 g/L sugar after 10-day saccharification (Fig. 3). Thus, ∆pyrF::KBm showed the highest saccharification efficiency among tested strains. However, it still took a long time (8 days) for ∆pyrF::KBm including a 2-day lag phase which might be caused by the low inoculum size, and the degree of efficiency improvement was not as obvious as that in microcrystalline cellulose (MCC) saccharification. Furthermore, the GS-2 medium commonly used for C. thermocellum cultivation contains expensive ingredients that are feasible economically. Hence, it is necessary to optimize the process and the medium to enhance the solubilization efficiency and reduce the cultivation cost.
Optimization of saccharification medium
GS-2 medium was commonly used for the cultivation of C. thermocellum and other clostridial strains [38]. It contains cellobiose or cellulose as the carbon source, urea as the inorganic nitrogen source, cysteine hydrochloride as sulfur supply [39], yeast extract as a supply of both nitrogen and necessary trace elements, and other phosphates and salts. For saccharification, pretreated biomass is used instead of cellobiose or cellulose to reduce the cost of carbon source. However, the sulfur and nitrogen supplies are still at a high cost.
We investigated the growth curves of cells cultivated in various modified media to determine the influence on cell growth. The media were derived from GS-2 containing cellobiose as the carbon source and substitutive ingredients. The results showed that cells growing in media containing 2 g/L sodium sulfide instead of 1 g/L cysteine hydrochloride or medium with 8 g/L corn steep liquor instead of yeast extract had similar growth patterns with those cultivated in GS-2 medium as shown in Additional file 3, indicating that sodium sulfide could be used as an alternative sulfur supply and corn steep liquor was used to replace yeast extract. The cell biomass obtained from the modified GS-2 (mGS-2) medium containing 2 g/L sodium sulfide and 8 g/L corn steep liquor was about 1.2-fold of that from GS-2 medium (Fig. 4).
∆pyrF::KBm was then used for saccharification using 100 g/L Avicel as the substrate and GS-2 or mGS-2 as the cultivation medium. As shown in Fig. 2, the sugar production rate obtained using mGS-2 maintained stably in Phase I but was greatly enhanced in Phase II from 1.90 to 3.26 g/L/day. Over 70 g/L reducing sugar was produced in 14 days using mGS-2 medium, which was 4 days earlier than that using GS-2 medium (Fig. 2). Additionally, the nitrogen and sulfur supplies in mGS-2 were cheap substitutes that only cost 0.23 and 0.087 US cents per liter, respectively. In contrast, the cost of yeast extract and cysteine hydrochloride is about 2.8 and 3.2 US cents per liter for the GS-2 medium, respectively. Thus, the cost of mGS-2 is much lower than that of GS-2. This result suggested that mGS-2 was a cost-effective medium for C. thermocellum cultivation to obtain more cell biomass for higher saccharification efficiency. Although ∆pyrF::KBm can grow in both GS-2 and mGS-2 media without extra addition of uracil, stimulated cell growth was observed by adding uracil. This indicated that the complementation of the pyrF gene in the biocatalysts would be necessary before industrial application to further enhance the cell growth without increasing the medium cost. The mGS-2 medium was used for further saccharification analysis in this study.
Optimization of biocatalyst cultivation and inoculation for SPS saccharification
1% inoculum size may cause a lag phase of the saccharification process (Fig. 3). To optimize the inoculum size of SPS saccharification, the ∆pyrF::KBm cells grown on Avicel until mid-log phase were used as the inoculum with a size of 1–300%, and stimulated saccharification efficiency was observed along with the increased inoculum size (Fig. 5a). With 5% inoculation, no apparent lag phase was observed and 30 g/L of reducing sugar was produced in 7 days instead of 8 days. With 10% inoculation, the saccharification process was further shortened by another 1 day to obtain a saccharification level of about 90%. Interestingly, no further enhancement of the saccharification process was observed when the inoculum size was increased from 10 to 100%. But the saccharification level could reach 88% in 4 days with 300% inoculation. In light of the operational feasibility and cost, 5 or 10% of inoculum size was considered to be conducive to eliminating the lag phase and promoting the saccharification process.
For the industrial purpose, the biocatalyst cultivation should not be performed using cellobiose or Avicel as the sole carbon source because of the high cost. C. thermocellum can also grow on glucose, a cheaper carbon source and one of the main products of cellulosic substrate saccharification, but with a long adaption phase [37]. SPS can be used as the carbon source for inoculum growth but the cells may attach to the substrate residues and cannot be easily separated for inoculation. Thus, to reduce cultivation cost under the premise of not reducing saccharification efficiency, glucose and SPS were tested as the carbon source for inoculum cultivation. Cells grown on cellobiose were inoculated in media with a mixture of glucose and SPS with a ratio of 5:0, 1:4, 1:1 or 0:5 as the carbon source and subcultured for two times for adaption and were cultivated till the mid-log phase. Avicel was used as the positive control. As shown in Fig. 5b, similar saccharification patterns were detected with cells grown on glucose or Avicel as the inoculum, and when SPS was used to grow the cells, with or without glucose and independent of the supplementation ratios, the saccharification process was stimulated and shortened by 1 day to reach the saccharification level of ~ 85%. This result suggested that SPS could be used as the sole carbon source for the cultivation of the whole-cell biocatalyst without adding high-price substrates.
Influence of hemicellulases on SPS saccharification process
Although C. thermocellum produces several cellulosomal hemicellulases, it cannot grow on hemicellulose-derived sugars. It is presumed that the main role the hemicellulases play is to expose the cellulose fibers and make them more accessible to hydrolysis [40], and the hemicellulase activity of the cellulosome may not be sufficient in terms of the saccharification of pretreated biomass substrate. Thus, we added commercial hemicellulases into the saccharification system to enhance the hemicellulase activity. The result showed that the saccharification process was greatly shortened from 8 days to 5 days with the addition of 150 U/g hemicellulase cocktail, mainly due to the elimination of the 2-day lag phase (Fig. 6). This result demonstrated that high hemicellulose degradation activity was essential for increasing the initial hydrolysis rate. Thus, cellulosomal hemicellulases of the current biocatalyst could be further enhanced by overexpressing endogenous enzymes or introducing heterologous enzymes.
Influence of substrate load on SPS saccharification process
As mentioned above, over 30 g/L reducing sugar could be produced from 40 g/L SPS substrate using the newly developed biocatalyst ∆pyrF::KBm, resulting in a saccharification level of around 90% (Fig. 3). To determine whether the saccharification efficiency would be influenced by substrate load, up to 80 g/L SPS was used for saccharification through batch and fed-batch substrate supplementation (Fig. 7). In the first 2 days, less than 5 g/L reducing sugars was produced with the initial SPS substrate load of either 20, 40 or 80 g/L. After the 2-day lag phase, the sugar production rates varied along with the substrate load significantly.
With 40 g/L initial substrate load, the production rate was around 5.5 g/L/day, but the value reduced to 4.0 and 1.8 g/L/day with 80 g/L and 20 g/L initial substrate load, respectively (Fig. 7). The results indicated that higher (80 g/L) or lower (20 g/L) initial substrate load was not beneficial for saccharification under the experimental conditions because low substrate load might cause insufficient carbon source for cellulosome production and, therefore, resulted in the decreased hydrolysis efficiency, and the horizontal shaking mode used in this study might result in inefficient mass transfer in the system when high substrate load was used, and further led to a declined sugar production. Thus, 40 g/L was the optimal initial substrate load to enhance saccharification efficiency under the experimental conditions.
At the 6th and 12th day, 20 g/L SPS was supplemented in the reactions with 40 g/L or 20 g/L initial substrate subsequently for fed-batch saccharification, termed “4 + 2 + 2” and “2 + 2 + 2”, respectively (Fig. 7). To reach a total substrate load of 80 g/L, 40 g/L SPS was supplemented at the 6th day in the system with 40 g/L initial substrate, termed “4 + 4” fed-batch process. For “4 + 4” and “8” processes which had the same total substrate load, decreased sugar production rates of 2.262 and 1.687 g/L/day were detected after 6 or 8 days’ saccharification, respectively. Because the initial sugar production rate of “4 + 4” process was higher than the 80 g/L SPS saccharification process, a higher amount of reducing sugars was produced in the fed-batch process (Fig. 7). This indicated that fed-batch, instead of batch saccharification, should be used to obtain a high amount of reducing sugars. For “4 + 2 + 2”, the sugar production rate was reduced to about 2.3 g/L/day and 1.3 g/L/day after the first and second substrate feeding, respectively. In contrast, the sugar production rate of the “2 + 2 + 2” process slightly increased from 1.8 g/L/day to 2.2 g/L/day and 2.8 g/L/day after the first and second substrate feeding (Fig. 7). However, the sugar production rate of the “2 + 2 + 2” process generally maintained at a low level, and it took 18 days to produce about 39.5 g/L reducing sugars. In comparison, a similar amount of reducing sugars was produced in “4 + 2 + 2” process at the 12th day before the second feeding. It is noteworthy that “4 + 2 + 2” and “4 + 4” showed similar saccharification patterns. This suggested that the substrate load affected saccharification efficiency at the beginning but had a slight influence in the later process.
Consolidated bio-saccharification of SPS under optimal conditions
Consolidated bio-saccharification of 40 g/L SPS was performed under optimal conditions (∆pyrF::KBm as the biocatalyst, inoculum size of 5%, mGS-2 medium, 150 U/g hemicellulase) or the regular conditions without modification (∆pyrF::CaBglA, inoculum size of 1%, GS-2 medium) in both 100-mL anaerobic bottles (Fig. 8a) and a 10-L anaerobic fermenter (Fig. 8b). Additionally, we observed enhanced saccharification process at 60 °C compared to 55 °C, which was reasonable because the optimal growth temperature of C. thermocellum is 60 °C [41]. Under optimal conditions, over 30 g/L reducing sugar was produced in 5 days and the saccharification process was shortened by 50% compared to that under regular conditions (Fig. 8). The saccharification level was 89.3% and 0.795 g/g reducing sugar was produced from SPS. Hence, the saccharification efficiency was significantly stimulated using the newly constructed biocatalyst under the improved conditions. In this way, the consolidated bio-saccharification strategy could be applied for highly efficient lignocellulose solubilization.