Batch Enzymatic hydrolysis
The effect of enzyme loading on batch cellulose hydrolysis was studied at the optimal condition of temperature 50 °C and pH 4.5 [26–28]. It is evident from Fig. 1 that the initial hydrolysis rate and the final percentage conversion remains similar for enzyme loading beyond 15 FPU/gbiomass. Therefore, this value (15 FPU/gbiomass) was taken as the saturation loading for batch cellulose hydrolysis. The effect of pH was studied at the saturation enzyme loading and optimal temperature (50 °C) at two conditions, pH 4.5 (optimal pH for enzymatic hydrolysis) and 5.5 (fermentation pH). There was no significant difference observed in terms of initial rate and final percentage conversion between these two pH conditions (Fig. 2a). Thus, we considered pH 5.5 as the optimal pH for SSF process, since it would be more favorable for fermentation and would not affect the enzymatic hydrolysis. The enzymatic hydrolysis was also found to be more efficient at 50 °C than at 37 °C (Fig. 2b, Additional file 1: Fig. S1). Since the optimal growth temperature of most of the lactic acid bacteria is around 37 °C, a SSF process would require the use of a thermo-tolerant bacterial strain to make the process more efficient.
Product Inhibition In Batch Enzymatic Hydrolysis
Product inhibition, in the form of cellobiose and glucose, is a major limitation in a batch enzymatic hydrolysis of cellulose [29, 30]. Lignocellulosic biomass (pretreated rice straw) spiked with different concentrations of glucose (0–60 g/L) or cellobiose (0–5 g/L) was subjected to batch enzymatic hydrolysis to understand the effect of product inhibition. Glucose concentration at a minimal level (≤ 5 g/L) was found to have no significant effect on the initial hydrolysis rate (during 0–2 h), cellobiose accumulation (Fig. 3a) and final percentage conversion (Fig. 3b). Glucose level beyond 20 g/L significantly inhibited the initial rate as well as the final percentage conversion. Higher initial glucose concentration (20–60 g/L) was also found to result in accumulation of cellobiose, the intermediate in conversion of cellulose to glucose (Fig. 3a). Moreover, cellobiose accumulation, in turn, inhibits the overall conversion of cellulose (Additional file 1: Fig. S2).
Therefore, removal of glucose during the hydrolysis process can result in reducing glucose and cellobiose inhibition, due to improved cellobiose utilization. This will increase the hydrolysis rates and overall conversion of cellulose to glucose. In an independent set of experiments, it was observed that removal of glucose (by intermittent replacement of supernatant with buffer) could enhance the conversion when compared to a batch hydrolysis where the glucose was not removed (Additional file 1: Fig. S3). Furthermore, the effect of enzyme loading has significant differences only at the early stage of the hydrolysis process. As can be seen in Fig. 1, increasing the enzyme loading from 5 FPU/gbiomass to 15 FPU/gbiomass can double the hydrolysis rate within the initial 2 hours, while the difference is relatively much less after 12 hours of hydrolysis. Towards the late stage of hydrolysis (24–72 h), the rate of hydrolysis slows down which could be due to combined effect of product accumulation (inhibition), decreased availability of substrate, non-productive binding of enzymes to lignin present in the acid-pretreated biomass and partial enzyme deactivation [31, 32]. By removing the glucose during the hydrolysis, the inhibition can be reduced and the rate of hydrolysis can be improved at a lower enzyme loading (Additional file 1: Fig S3). A similar result can be brought about by employing a SSF process, wherein the glucose formed by hydrolysis of cellulose is simultaneously converted to the fermented product. However, an optimal process condition needs to be found for the saccharification as well as fermentation.
To validate our assumption that SSF would improve the conversion efficiency because of concomitant glucose consumption by the microbe, batch SSF experiments were conducted at 37 °C and pH 5.5 and compared with separate hydrolysis and fermentation (SHF) experiments. Separate batch fermentation experiments were conducted at 37 °C with 30 g/L pure glucose to compare the growth parameters of the organism in batch fermentations with lignocellulosic hydrolysate.
Simultaneous saccharification and fermentation by L. bulgaricusWT
A separate fermentation process for converting the cellulosic hydrolysate to D-lactate at 37 °C and pH 5.5 gave a yield of 0.86 (gDLA/gglucose) (Table 1). The yield of cellulose to glucose by enzymatic hydrolysis at similar conditions of 37 °C and pH 5.5 using 10% solid loading (w/v) was 48.22% (Table 2). Thus the estimated conversion of cellulose to D-LA that we could have achieved in a SHF process with similar condition would have been around 41%, which is much lower than the yield we achieved in SSF process (49%) at 37 °C (Table 2). Also, we found that there is no significant difference in growth parameters when L. bulgaricus WT was grown on lignocellulosic hydrolysate and pure glucose (Table 1). This indicates L. bulgaricus WT is a suitable candidate for producing D-LA by valorization of LCBs through a biochemical route.
Table 1
Comparison of specific growth rate, specific glucose uptake rate and D-LA yield between L. bulgaricus WT and ET45
Strain | Condition | µ (h− 1) | qs (g.g− 1.h− 1) | YD−LA/glucose (g.g− 1) |
L. bulgaricus WT | 30 g/L glucose, 37 °C, pH 5.5 | 0.30 | 3.31 | 0.86 |
L. bulgaricus ET45 | 30 g/L glucose, 45 °C, pH 5.5 | 0.29 | 3.18 | 0.87 |
L. bulgaricus WT | Hydrolysate (30 g/L glucose equivalent), 37 °C, pH 5.5 | 0.33 | 3.41 | 0.87 |
L. bulgaricus ET45 | Hydrolysate (30 g/L glucose equivalent), 45 °C, pH 5.5 | 0.32 | 3.75 | 0.86 |
However, the 37 °C temperature for the SSF process was sub-optimal for enzymatic hydrolysis, while the fermentation could not be carried out above 40 °C due to the lack of growth of L. bulgaricus WT strain beyond this temperature. To further enhance the conversion of cellulose to D-LA, SSF needs to be carried out closer to the optimal temperature for enzymatic hydrolysis. This necessitated the generation of a thermo-tolerant strain of L. bulgaricus WT by ALE as described in Materials and Methods section.
Batch fermentation experiments with thermo-tolerantL. bulgaricus
Using ALE strategy, a stable thermo-tolerant L. bulgaricus ET45 was developed from the parental strain (WT). L. bulgaricus ET45, unlike the parental strain, is capable of growing at an elevated temperature of 45 °C (no growth was observed above 45.3 °C). Batch reactor studies using glucose (30 g/L) as substrate suggest that the specific growth rate (µ), specific glucose uptake rate (qs) of ET45 strain are similar to that of WT strain (Table 1). Moreover, there was no significant change observed in D-LA yield (0.87 g/g) of L. bulgaricus ET45 at 45 °C as compared to WT strain at 37 °C (Table 1).
To further ascertain its growth parameters in hydrolysate, we carried out batch SHF experiment with hydrolysate containing 30 g/L glucose. In batch SHF at 45 °C, we achieved a specific growth rate and D-LA yield with the thermo-tolerant L. bulgaricus ET45 which was similar to that achieved with pure glucose substrate at 45 °C as well as to the parent strain grown in hydrolysate at 37 °C (Table 1). Subsequently, SSF experiments were conducted with L. bulgaricus ET45 at 45 °C, and compared with the hydrolysis rate and conversion of cellulose to glucose obtained in SHF experiments.
Table 2
Comparison of percentage cellulose to glucose conversion among enzyme hydrolysis (EH) and SSF experiments at different conditions
Process | Total Solid loading (% w/v) | D-LA (g/L) | Cellulose to D-LA conversion (%) | Cellulose to glucose conversion (%) |
EH 15FPU/50 °C/ pH 4.5 | 10 | | 57.31# | 66.64 |
EH 15FPU/45 °C/ pH 5.5 | 10 | | 52.13# | 60.62 |
EH 5FPU/45 °C pH 5.5 | 10 | | 43.07# | 50.08 |
EH 5FPU/37 °C/ pH 5.5 | 10 | | 41.47# | 48.22 |
Batch SSF (L. bulgaricus WT) 5FPU /37 °C /pH 5.5 | 5 | 12.54 | 49.11 | 57.11* |
Batch SSF (L. bulgaricus ET45) 15FPU / 45 °C /pH 5.5 | 10 | 32.37 | 63.4 | 73.72* |
Batch SSF (L. bulgaricus ET45) 5FPU /45 °C /pH 5.5 | 10 | 30.6 | 59.93 | 69.68* |
Batch SSF (L. bulgaricus ET45) 5FPU /45 °C /pH 5.5 | 5 | 15.02 | 58.83 | 68.41* |
Batch SSF (L. bulgaricus ET45) 3FPU /45 °C /pH 5.5 | 5 | 11.82 | 46.3 | 53.84* |
Pulse-fed SSF (L. bulgaricus ET45) 5FPU /45 °C /pH 5.5 | 35 | 108.58 | 60.76 | 70.65* |
#Estimated cellulose to D-LA conversion = Cellulose to glucose conversion*YD−LA/Glucose *Estimated cellulose to glucose conversion = Cellulose to D-LA conversion/YD−LA/Glucose YD−LA/Glucose =0.86, calculated from independent SHF experiments at 45 °C and pH 5.5 (Table 1) |
Simultaneous saccharification and production of D-LA byL. bulgaricusET45
The estimated cellulose to glucose conversion (73.7%) in SSF, with 10% (w/v) solid loading and 15 FPU/gbiomass at 45 °C, pH 5.5 (Fig. 4a), is substantially higher than that achieved in similar conditions with batch enzyme hydrolysis (60.6%) and also higher than the batch enzymatic hydrolysis at the optimal conditions of 50 °C, pH 4.5 and 15 FPU/gbiomass (Table 2).
The higher conversion achieved in SSF is obviously due to the concomitant microbial consumption of glucose which reduces the inhibition on enzymatic hydrolysis. In earlier study, we have already proved that glucose removal during enzymatic cellulose hydrolysis improves the conversion efficiency (Additional file 1: Fig. S3). This allows for the possibility of reducing the enzyme usage in SSF, since the optimal value of 15 FPU/gbiomass for batch enzyme hydrolysis was achieved without removing glucose inhibition. Therefore, we carried out the SSF process (at 45 °C and pH 5.5) at a reduced enzyme loading of 5 FPU/gbiomass (Fig. 4b). As can be seen from Table 2, the estimated glucose conversion and the D-LA conversion from cellulose is only marginally (~ 5%) less with an enzyme loading of 5 FPU/gbiomass as compared with that obtained at 15 FPU/gbiomass.
To determine the possibility of further reduction in enzyme usage, we carried out the 45 °C SSF experiment with 3 FPU/gbiomass (Additional file 1: Fig. S4b). The estimated glucose conversion was ~ 54% and D-LA conversion was ~ 46%, which is much lower than that obtained under the same conditions with SSF at 5 FPU/gbiomass (Table 2). Therefore, we carried out further SSF experiments with L. bulgaricus ET45 at 5 FPU/gbiomass, which is much more economical and has only marginally lower conversion as obtained at 15 FPU/gbiomass (Fig. 4).
In the batch SSF process, the D-LA accumulation with 10% (w/v) biomass loading is only ~ 30 g/L. The difficulty of mixing of pretreated biomass in reactor precludes the biomass concentrations higher than 10% solid loading. Therefore, we investigated a pulse-fed SSF process, involving intermittent feeding of pretreated biomass and enzyme addition of 5 FPU/gbiomass, to enhance D- LA accumulation.
Pulse-fed Simultaneous Saccharification And Fermentation For D-la Production
To avoid the difficulty associated with mixing at initial solid loading of 10%, the pulse-fed SSF was started as batch SSF with 5% (w/v) initial solid loading. In the initial 24 hours, we could observe accumulation of glucose (Fig. 5) during the lag phase of bacterial growth. After 24 hours, the glucose level started to fall with concomitant increase in D-LA accumulation due to the bacterial fermentation. Pulse-feeding of biomass and enzyme, at 5% (w/v) with 5 FPU/gbiomass, was started at 36th hour after the glucose dropped to a low level and continued till 96th hour. Transient accumulation of glucose was observed after each pulse, due to higher initial rate of hydrolysis than microbial glucose uptake rate. The increased accumulation of glucose in the later stages of pulse-feed can be attributed to lesser glucose uptake rate by the bacteria than the rate of hydrolysis.
The D-LA titer achieved in this pulse-fed SSF was ~ 108 g/L, among the highest reported with SSF processes using pre-treated lignocellulosics as the carbon source (Additional file 1: Table S1). The estimated conversion of cellulose to glucose during SSF was 70.65%, consistent with our results from batch SSFs.