3.1 Compositional analysis of waste sandwich bread
The compositions of the WB used in this study are shown in Table 1. This accords with Pietrzak and Kawa-Rygielska, 2014 who reported that WB contained 68.9% starch. Sanchez et al 2014 and Dewettinck et al 2008 also reported that bread contains between 55.3–83.3% glucose equivalents. The moisture content of the mixed bread, made up of a combination of 22 different types of wholemeal and white bread, was 35% ± 13.5. After oven drying at 45°C for 24hr, the moisture content was reduced to 8.5% ± 0.8 (w/w) to prevent spoilage during storage. The total carbohydrate content of the mixed WB (Table 1) was 70.58% starch with a small amount of hemicellulose (2.26%) derived mainly from the wholemeal variety. The parent strain of TM333 can ferment hemicellulose-derived sugars (such as xylose and arabinose) and associated short chain oligosaccharides but does not encode a secreted xylanase or arabinanase (Lisowska 2016).
Table 1
Carbohydrate composition (% w/w) of WB used in SSF
Type of bread | Starch | Xylan | Arabinan |
Wholemeal bread | 67.8 | 1.2 | 0.5 |
White bread | 75.0 | 0.3 | 0.01 |
Mixed bread | 70.6 | 0.8 | 1.5 |
3.2 Alpha-amylase and neopullulanase production by TM333
The crude α-amylase secreted by TM333 into the culture supernatant after 17h of fermentation had a maximum activity of 428.5 U/ml at pH 7 and 85°C but retained some activity at pH 5.5 and 35°C (yeast fermentation condition), and also at the P. thermoglucosidasius TM333 optimal fermentation conditions of pH7 and 60°C (Table 2). Malhotra et al (2000) reported that maximum Ca+-independent α-amylase production by thermophilic B. thermooleovorans NP54, using a 2% inoculum was between 12-16hr, similar to the 17h optimum fermentation time reported here.
Table 2
Alpha amylase activities of the crude enzyme from P. thermoglucosidasius TM333 under various assay conditions similar to SSF conditions used in this study.
pH | Temperature ⁰C | Activity *U/ml | Condition used |
7.0 | 60 | 195.5 | P. thermoglucosidasius fermentation |
7.0 | 85 | 428.5 | Liquefaction |
7.0 | 100 | 8.1 | Gelatinisation |
5.5 | 35 | 29.2 | Yeast fermentation |
Endoamylases are known to produce various chain lengths of linear and branched oligosaccharides from starch by acting on α-1-4, or α-1-6 glucose linkages (Gupta et al., 2003). In addition to α-amylase, low levels of neopullulanase activity (1.35 to 1.79 U/ml) were also measured in the culture supernatant produced either from soluble starch or WB at 18hr to 24hr (Fig. 1). Neopullulanase and pullulanase are known to increase the hydrolysis of starch by α-amylase, especially via the removal of α 1–6 branching or cleavage of dextrin to form panose, but they are expensive enzymes that are produced in low titers and are rarely added to starch hydrolysis for economic reasons. The use of pullulanases can reduce the requirement for glucoamylase by ~ 50 % and can reduce the total reaction time of industrial starch conversion. A higher neopullulanase titre was produced on WB (1.79 U/ml) than on soluble starch (1.35 U/ml), suggesting that this might be a cheaper substrate for both α-amylase and neopullulanase production compared with purified soluble starch. The presence of protein, various vitamins and minerals in WB compared with purified soluble starch might have resulted in better enzyme production from WB. A DHSS (1981) report on the nutritional aspects of bread and flour established that UK breads contain between 8 to 9% protein, several essential amino acids, fats and fatty acids, minerals, vitamins and trace elements which might enhance microbial growth and enzyme production.
3.3 WB processing for SSF 1 to 5
Figure 2 shows the sugar composition of the WB in media without pretreatment, after gelatinization and after both gelatinization and liquefaction in SPY media; the same treatments done in YP media gave virtually identical results. The untreated WB suspended in media shows the presence of soluble malto-oligosaccharides and maltose with trace amounts of glucose (glucose < 0.5g/l, maltose 3.6 to 3.9g/l, malto-oligosaccharides 8.7-9.2g/l) due to the presence of partially hydrolyzed starch. The low levels of glucose and maltose after 1hr of liquefaction suggest that TM333 α-amylase is not a maltogenic α-amylase. Van Zyl et al (2012), showed that α-amylases hydrolyze the internal α-1,4-bonds of starch amylose and amylopectin randomly, leading to the production of maltodextrins with a length of 10 to 20 glucose residues as major products, as well as smaller amounts of maltose and free glucose. These dextrins are known to be water soluble (Gupta et al 2003; Van Der Maarel et al 2002). In a control experiment (results not shown), WB without autoclave treatment was processed in the same way, in the presence of 0.2% (w/v) sodium azide to reduce microbial growth. The sugar profiles were similar to those in Fig. 2 indicating that the autoclave treatment did not contribute to starch hydrolysis in the WB.
3.4 SSF1: SSF of gelatinized and liquefied WB with amyloglucosidase + pullulanase.
Fermentation results for SSF 1 with amyloglucosidase and pullulanase included in the process are shown in Fig. 3. As described, the bread samples had been pre-gelatinized and liquefied with α-amylase from TM333 to yield mainly maltodextrins (Fig. 2). While AMG is important for the generation of glucose from straight chain malto-oligosaccharides, pullulanase attacks the 1,6 linkages, debranching the dextrins and also prevents the reverse reaction of glucose condensation to maltose or isomaltose, a process known to be catalysed by amyloglucosidase (Roy and Gupta 2004; Presecki et al 2013). The fermentation results show that complete fermentation of the WB was possible with yeast or TM333 at yields of > 98% of the maximum theoretical from all WB carbohydrates. Thus, the malto-oligosaccharides in the pretreated bread (Fig. 2) were being rapidly converted to maltose or glucose by the amyloglucosidase and/or pullulanase in combination with the remaining crude TM333 α-amylase under both sets of culture conditions.
3.5 SSF 2: SSF of gelatinized and liquefied WB with amyloglucosidase added during fermentation .
In SSF 2 the debranching pullulanase was omitted from the SSF, but otherwise the conditions were the same as SSF 1. Fermentation results (Fig. 4) show that this SSF generated similarly high yields (> 98% of theoretical) for both yeast and TM333, indicating that the combination of TM333 α-amylase and AMG alone was sufficient to hydrolyse all of the malto-oligosaccharides to maltose or glucose, even at the sub-optimal temperature of 35oC and pH5.5, with a fermentation profile similar to that in SSF 1. The α 1,6 debranching activities of AMG appeared to be sufficient to obviate the need for pullulanase, although it is not clear whether the low neopullulanase activity in the crude amylase might have been beneficial.
3.6 SSF 3: SSF of gelatinized WB with amyloglucosidase added during fermentation
SSF 3 differed from 1 and 2 by not going through a liquefaction process after the initial gelatinization. The compositions of the SSF3 media prior to fermentation (Fig 1 and 2) suggested that much of the initial α-amylase activity occurred during gelatinization, mainly producing malto-oligosaccharides with a small amount of maltose and less than 0.5% of glucose. Fermentation results (Fig 5 and Table 2) showed that both TM333 and yeast fully fermented this media with yields of >98% of theoretical. This shows that, with the quantities of α-amylase added, there was no need for liquefaction and pre-saccharification before fermentation when the crude enzyme of TM333 was added for both yeast and TM333 fermentations. However, TM333 fermentation was slower than seen in SSF 1 and 2 and there was a transient increase in glucose accumulation in the media, suggesting that the amyloglucosidase activity exceeded the capacity of cells to grow on the resulting glucose.
3.7 SSF 4; SSF of untreated WB with crude α-amylase and amyloglucosidase added during fermentation.
Given that some maltose and free glucose was present in the autoclaved breadcrumbs, this should allow initial growth of both organisms while supplemented enzymes hydrolyse the oligosaccharides and any residual unhydrolysed starch. Therefore, in SSF 4 untreated WB without gelatinization or liquefaction was used as the substrate and α-amylase and amyloglucosidase were added to the fermentation media (Fig. 6). Both yeast and TM333 were still able to fully ferment the WB in 24-48h with > 98% of the maximum theoretical yield with comparable results to SSF 1 to 3. The fermentation profile for TM333 was similar to that of SSF 3, with a spike of glucose at 24h.
The ability to degrade the residual non-hydrolysed starch within 24h shows that the synergistic activity of the α-amylase and amyloglucosidase, even under the unfavorable conditions of 35oC and pH5.5 was high, and that the autoclaved, but otherwise particulate starch was accessible to the TM333 amylase. It is, therefore, surprising that TM333 grew marginally more slowly under these conditions than in SSF 2. Given that there was glucose accumulating after 24h, this suggests that intermediates in the breakdown of the polymeric starch were actually reducing the rate of growth, possibly by directly or indirectly (via regulation) affecting the glucose transporter. Given that Parageobacillus spp are known to be capable of transporting complex oligosaccharides this could be an example of evolutionary adaptation leading to impairment of growth under an artificial abundance of monosaccharide. Nevertheless, the cost savings in removing a pre-hydrolysis step and using a one-pot SSF should outweigh any marginal reduction in productivity. Furthermore, this effect might be resolved by optimising the balance of enzyme activity and nutrient uptake rate of the cells. The only comparative work on SSF of WB was by Pietrzak and Kawa-Rygielska, 2014 and 2015b, whereby a commercial enzyme cocktail (α-amylase and amyloglucosidase plus a protease) were used in combination with several types of pretreatment including heat, microwave, ultrasound (Table 3), but yields were still lower than those from our current work with only crude α-amylase.
3.8 SSF 5: SSF of untreated WB with TM333 α-amylase alone.
As the TM333 α-amylase was clearly capable of degradation of the polymeric starch from WB, potentially down to maltose and branched limit dextrins the effect of a simple SSF on the autoclaved WB was examined in SSF 5. The lack of the saccharifying and 1,6-debranching amyloglucosidase was expected to reduce the amount of carbohydrate available for the yeast fermentation, but production of a neopullulanase by TM333 suggested that it should be able to import the product panose for further degradation. Given the results from SSF 4, the yeast fermentation in the presence of α-amylase alone was surprisingly poor (Fig. 7A), with less than 30% of the ethanol production seen in SSF 1 and 2. While this is more than can be accounted for by the maltose and glucose present in the untreated bread, suggesting that some maltose or glucose had been generated by the α-amylase, this is clearly fairly limited, with the saccharifying AMG being critical for good yeast SSF on WB. This is consistent with reports that maltose and glucose are minor products from the action of α-amylase on starch, (Favaro et al 2013; Gibreel et al 2009). However, Fig. 7B shows that WB was fully fermented by TM333 in SSF with a native (actually a mixture or native and recombinant from the closely related G. stearothermophilus) α-amylase, with a very similar profile to that which was observed in SSF 4, suggesting that AMG had minimal contribution to the latter. The yield of ethanol using TM333 in this SSF was 94–96% of theoretical (14.24 g/l ethanol) based on the total sugars in WB, and seemingly even higher after 72-92h fermentation (Table 3). With S. cerevisiae, however, the SSF with α-amylase alone gave a maximum ethanol yield of 26.8% of theoretical (3.74g/l) based on total carbohydrate (Fig. 7A), with no increase at longer fermentation times. (Table 3). The profile of the SSF 5 fermentation with TM333 was very similar to that of SSF 4 with a transient increase in glucose concentration after 24h despite the lower ethanol production rate. The similarity suggests this was not the result of AMG activity but potentially the result of transglycosylation activity involving malto-oligosaccharides or neopullulanase activity, again suggesting that oligosaccharides were being metabolised in preference to glucose. Fermentation of other oligosaccharides by TM242, the parent strain of TM333 has been reported, including mannobiose and mannotriose from Palm kernel cake mannan (Raita et al., 2016), and also xylo-oligosaccharides and cellobiose (Cripps et al., 2009; Lisowska, 2016) with the help of its oligosaccharide active transport systems. Any regulatory effects of oligosaccharide utilization over monosaccharide uptake can potentially be modified through mutation.
3.9 Comparison with previous studies.
As far as we are aware, the only previous reports of WB SSF is the work of Pietrazak and Kawa-Ragieska et al (2014 and 2015b) using SSF after various pretreatments and supplementing with commercial α-amylase, protease and amyloglucosidase. In some cases, cellulases, pentosanases and β-glucanases were also added to improve yields (Table 3). When adjusted to a common scale, these studies produced yields of 333-425g ethanol per kg of WB (Pietrzak and Kawa-Rygielska (2015b), Pietrzak and Kawa-Rygielska (2014)), with the highest yield only obtained by employing pre-liquefaction, in comparison with 437–447 g/kg produced by simple SSF with TM333 at 48hr and 72h respectively in the current study.
Table 3
Comparison with previous reports of bread to ethanol process including SSF and SHF
Bread type | Treatment/Fermentation conditions | Bread concentration % w/w | **Ethanol yield per kg of bread @ 48hr | Enzymes used | Reference |
Waste wheat-rye bread | SSF (slurried with enzymes at 35°C prior to fermentation) | 15 | 333.31 (*354.36) | α-amylase and glucoamylase, plus protease | Pietrzak and Kawa-Rygielska, 2014 |
Waste wheat-rye bread | SSF (slurried at 45°C for 20min prior to fermentation) | 15 | 373.15 (*398.4) | α-amylase, β-glucanase, pentosanase, cellulase, plus protease | Pietrzak and Kawa-Rygielska, 2014 |
Waste wheat-rye bread | SSF of liquefied WB | 30 | 416–425 | α-amylase, amyloglucosidase plus protease | Pietrzak and Kawa-Rygielska, 2015b |
Waste wheat-rye bread | SSF post microwave pretreatment | 15 | 375.5 (*384.6) | α-amylase and glucoamylase, plus protease | Pietrzak and Kawa-Rygielska, 2014 |
Waste wheat-rye bread | SSF post Ultrasonic pretreatment | 15 | 365.1 (*366.8) | α-amylase and glucoamylase, plus protease | Pietrzak and Kawa-Rygielska, 2014 |
Waste wheat-rye bread | Separate hydrolysis and fermentation (SHF) standard process | 15 | 378.7 (*386.0) | STARGEN 002 (α-amylase and glucoamylase) plus protease | Pietrzak and Kawa-Rygielska, 2014 |
mixture of wheat and buckwheat flours | SHF standard process | 16 | 303.2-408.1 (67hr) | α-amylase plus a glucoamylase | Acanski et al 2014 |
Sandwich WB | Yeast SSF of gelatinized and/liquefied WB | 3 | 456-478.4 (*466-474.3) | α-amylase plus a glucoamylase | This work |
Sandwich WB | Yeast SSF of untreated WB | 3 | 448.9 (*459.9) | α-amylase plus a glucoamylase | This work |
Sandwich WB | Thermophilic bacteria SSF of untreated WB | 3 | 437.1 (*446.8) | P. thermoglucosidasius TM333 crude α-amylase only | This work |
*Yields at 72hr fermentation; **Yields are based on Eq. 1; g ethanol per g of WB as in Section 2.5. All referenced fermentations were with Saccharomyces cerevisiae. |
The work of Pietrazak and Kawa-Ragieska et al (2014, 2015) showed that initial high temperature gelatinization was beneficial for obtaining good ethanol yields, but this was less obvious in the current study. Two major starch transformations take place during the bread baking and storage process: firstly, crystalline starch is converted largely to its amorphous form during baking, while secondly during storage for > 2 days, the amylopectin units become more crystalline through retrogradation (Hug-Item et al 2001). These two reactions are known to present a starchy material that normally requires high temperature gelatinization to be completely accessible to α-amylase. Additionally, amylose interacts with lipids to form complexes that are difficult for α-amylases to access and requires a liquefaction/melting temperature of around 105–116˚C (Eliason, 1994;; Hug-Item et al 2001; Gudmundsson and Eliason 1992) to improve accessibility, while amylose crystals are much more thermostable and melt at 150˚C (Eberstein et al 1980). Though fermentation was slightly faster in our experiments when gelatinization and liquefaction pre-treatment was incorporated, the yields without any treatment were comparable and not significantly different at P ≤ 0.05.
The SSF yields of ethanol obtained using the TM333 α-amylase, whether using P thermoglucosidasius or yeast (in combination with AMG) were higher in the current study than in previous reports (Table 3). Although this study used a lower bread concentration, Torabi et al (2020) showed that over a range of 9–16% (w/v), the concentration of WB did not affect the sugar and subsequent ethanol fermentation yields. Experimentally, it is difficult to demonstrate a P thermoglucosidasius SSF using higher concentrations of substrate without setting up a process of continuous ethanol removal, because ethanol toxicity starts to be observed above 2% (v/v). So, to confirm that the observations made using TM333 α-amylase hold true at higher loadings, SSF3 with yeast was repeated at WB loadings up to 25% (w/w).
Table 4
Concentration of solubilized carbohydrates after gelatinization pre-treatment at various WB concentrations
Concentration of waste bread % g/g | glucose g/l | maltose g/l | glycerol g/l | Malto-oligos g/l |
2.5 | 0.34 ± 0.04 | 2.99 ± 0.28 | 0.07 ± 0.01 | 18.09 ± 0.46 |
5 | 0.81 ± .04 | 4.77 ± 0.26 | 0.18 ± 0.01 | 37.10 ± 1.5 |
10 | 1.23 ± 0.09 | 7.97 ± 1.2 | 0.29 ± 0.01 | 73.97 ± 1.61 |
15 | 1.49 ± 0.22 | 10.64 ± 0.6 | 0.39 ± 0.03 | 112.23 ± 1.72 |
20 | 0.15 ± 0.05 | 10.14 ± 0.66 | 0.42 ± 0.01 | 156.23 ± 1.29 |
25 | 0.96 ± 0.31 | 10.82 ± 1.0 | 0.57 ± 0.03 | 184.25 ± 3.0 |
Ethanol and glycerol yields after 24h increased linearly over the waste bread concentrations tested (Fig. 8). The linearity (R2 = 0.989 for ethanol and 0.999 for glycerol) of the profiles indicates that there was no observed loss of enzyme performance or yeast fermentation inhibition as the waste bread concentrations increased from 3-27.5% during gelatinisation and from 2.5–25% during SSF ethanol production. The only notable effect of high substrate concentrations was that the concentration of glucose and maltose measured after gelatinization, reached a plateau at 15% (w/w) WB (Table 4). Glycerol production during yeast fermentation (Fig. 8) is a reaction to stress and acts as osmotic stabilizer for the yeast cells against high ethanol and acid concentrations produced during fermentation (Sakwa et al. 2018), but it diverts sugar away from ethanol production. The glycerol levels are consistent with those reported by Ebrahimi et al. 2008 from 20% waste bread (43g/kg of waste bread).
With 25% (w/w) WB as substrate SSF3 produced approximately 11% (w/v) ethanol, equivalent to 440g/kg WB, higher than virtually all previous reports and similar to that produced after liquefaction of the WB using α-amylase, amyloglucosidase and a protease (Pietrzak and Kawa-Rygielska 2015b). This is also close to the yields obtained with both yeast and TM333 at lower substrate loadings. A possible explanation for these very high yields is that the crude α-amylase preparation from TM333 also contained a low activity of neopullulanase. Neopullulanase is a starch trimming enzyme with α-1-4 activity and some α-1-6 activity enabling conversion of highly branched amylopectin to the 3-sugar compound panose (Hii et al 2012; Hussein et al 2015) for direct transport into P. thermoglucosidasius or further hydrolysis by AMG for uptake into yeast. Indeed, Takata et al have shown that neopullulanase can catalyse hydrolysis of α-(1–4)-glucosidic linkage, hydrolysis of α-(1–6)-glucosidic linkage, transglycosylation to form α-(1–4)-glucosidic linkage, and transglycosylation to form α-(1–6)-glucosidic linkages (Takata et al 1992) and may be able to covert liquefied starch to malto-oligosaccharides without the presence of an α-amylase or amyloglucosidase.
It is notable that the disruption of the neopullulanase gene, susA, in Bacteroides thetaiotaomicron was found to reduce its rate of growth on starch by about 30% (D’elia and Salyers 1996). It is therefore possible that there was incomplete starch hydrolysis in most of the previous works shown in Table 3 due to the lack of a starch debranching enzyme. Interestingly, Novozymes (2017) have recently launched an enzyme mixture (Spirizyme T) that contains both amyloglucosidase and pullulanase, with reported increase in both sugar and ethanol yields from starch. In a review on the use of pullulanases as debranching enzymes, Nisha and Satyanarayana (2016) stated that the addition of pullulanase in starch hydrolysis would allow a reduction of glucoamylase usage by approximately 50% and reduce the total reaction time of an industrial starch conversion process. The possibility of using thermostable amylopullulanases in an economic one-step starch liquefaction and saccharification process, which replaces amylolytic enzymes thus resulting in the overall reduction in the cost of sugar production was also mentioned (Nisha and Satyanarayana 2016).