In order to obtain steam-exploded liquors with high inhibitor concentrations, samples of Phragmites australis (common reed), Cynara cardunculus (cardoon), and Saccharum officinarum (sugarcane) bagasse pre-treated by applying different conditions (residence time and temperature) resulting in increasing severity factors (LogR₀) were investigated in this study (Table 1). LogR₀-values (Overend and Chornet, 1987; Espirito Santo et al., 2020) were obtained according to Equation [1]:

where t is the residence time (sec/min) and T is the temperature (°C).

Briefly, pre-treatment liquors of cardoon and common reed were obtained by steam explosion. For every LogR₀-value, six consecutive explosions were executed using 500 g of dry biomass for each explosion. Liquors were then separated from the WIS fraction using a stainless-steel filter with a cutoff of 1 mm (Cotana et al., 2015a,b; Cagnin et al., 2018).

Sugarcane pre-hydrolysate was obtained in a steam explosion plant composed of a 19 L reactor, a collection tank, and a 40-bar electrical boiler. Milled sugarcane bagasse samples, dried in a drying chamber to a final moisture content of 10% (w/w), were loaded into the reactor and treated for 10 min at 200°C. The pre-hydrolysate was then removed using a locally manufactured dead-end press. All the pre-hydrolysates were refrigerated until use. The inhibitor and sugar contents are reported in Table 1.

All chemicals, media components, and supplements were of analytical grade standard.

The phenotypes and sources of the S. cerevisiae strains used in this work are summarized in Table 2. Yeast strains pre-cultures were grown in YPD medium (g/L: yeast extract, 10; peptone, 20; glucose, 20) at 30°C on a rotary shaker set at 130 rpm unless otherwise stated.

Seven natural yeast strains (Fm17, Fm89, Fm90, Fm96, M2n, MEL2, and YI30) were screened for their industrial fitness using Ethanol Red as benchmark industrial yeast. Inhibitor tolerance in the presence of four synthetic inhibitor mixtures and eight inhibitor-rich lignocellulosic pre-hydrolysates was assessed.

Fermentation performances of S. cerevisiae Fm17 and Ethanol Red were evaluated in YNB medium formulated with cardoon (Cc3) or common reed (Pa3) pre-hydrolysates supplemented with up to 20 g/L of glucose. Moreover, the two pre-treatment liquors were supplemented with YNB and 40 or 92 g/L of glucose, respectively, to simulate the highest glucose concentration obtained by enzymatic saccharification of each steam-exploded WIS (Cotana et al., 2015a, b; Cavalaglio et al., 2016). After pH was adjusted to 5.0 using 5 M NaOH, broths were sterilized using a 0.22 μm sterile filter.

Pre-cultures of yeast cells grown to stationary phase in YNB medium containing 20 g/L of glucose were used to inoculate 50 mL medium to an initial OD₆₀₀ of 1.0 in 55 mL glass serum bottles. The small-scale fermentations were carried out in triplicate under oxygen-limited conditions. Bottles were sealed with rubber stoppers, incubated at 30°C, and mixed on a magnetic stirrer. Growth was measured as OD₆₀₀ and samples, taken through a capped syringe needle pierced through the bottle stopper, were stored at −20°C. Collected samples were filtered through a 0.22 μm pore filter and diluted prior to HPLC (high-performance liquid chromatography) analysis performed as described in the section “Analytical Methods, Calculations, and Statistical Analysis.”

Samples of lignocellulosic pre-hydrolysates and liquid fractions during small scale fermentations were analyzed for ethanol, glycerol, arabinose, galactose, glucose, xylose, mannose, sucrose, maltose, cellobiose, acetic acid, formic acid, levulinic acid, furfural, and HMF. Liquid chromatography analysis was performed using a Shimadzu Nexera HPLC system, equipped with a RID-10A refractive index detector (Shimadzu, Kyoto, Japan). The chromatographic separations were performed using a Phenomenex Rezex ROA-Organic Acid H⁺ (8%) column (300 mm × 7.8 mm). The column temperature was set at 65°C, and the analysis was performed at a flow rate of 0.6 mL/min using isocratic elution, with 0.01 M H₂SO₄ as a mobile phase (Favaro et al., 2010). Analytes were identified by comparing their retention times, and the concentrations were calculated using calibration curves of the corresponding external standard.

The ethanol yield (Y_E/G) from glucose was calculated as the highest amount of ethanol produced per gram of consumed glucose (g/g). The volumetric productivity (Q) was based on grams of the highest ethanol produced per liter of culture medium per hour (g/L h^–1). Q_max was calculated as the highest volumetric productivity along the fermentations.

Statistical analyses were obtained using the Graphpad Prism 5 package (Graphpad Software, Inc., San Diego, CA, United States). Mean values, standard deviations, and descriptive statistics were calculated. Fermentations performances were analyzed by ANOVA (ANalysis Of Variance) using Duncan test post hoc means differentiation.

Seven S. cerevisiae strains, namely, Fm17, Fm89, Fm90, Fm96, M2n, MEL2, and YI30, were previously described for their potential in various bioethanol applications (Favaro et al., 2013a, 2014; Jansen et al., 2018). In this study, these strains were further characterized with the final aim of assessing their promise to be used for lignocellulosic ethanol production. As such, S. cerevisiae Ethanol Red was specifically chosen as reference industrial yeast (Walker and Walker, 2018; Favaro et al., 2019b).

Since developing industrial yeast with high fermentative capacity from different pre-treated feedstocks, rather than a preferred substrate, is one of the ultimate goals, pre-hydrolysates from cardoon (Cc3) and common reed (Pa3) were used as a substrate to simulate the industrial environment as closely as possible. Firstly, the pre-treatment liquors were supplemented up to 20 g/L glucose and used in small-scale fermentations to compare the fermenting abilities of S. cerevisiae Fm17 to those of the reference industrial strain Ethanol Red (Figures 1A–D and Table 6). The acidity of the medium was adjusted to pH 5.0 with 5 M NaOH. A fermentation medium formulated without Cc3 or Pa3 was used as control (Figures 1E,F and Table 6).

Once exposed to the pre-hydrolysate Cc3 (C. cardunculus), glucose consumption was completed within the first 20 h by both strains (Figures 1A,B). The industrial yeast Ethanol Red produced higher biomass levels with the final OD₆₀₀ approaching 4.0. On the contrary, the novel yeast strain Fm17 achieved higher ethanol production (9.34 g/L) which was 1.13-fold that of Ethanol Red (Figures 1A,B and Table 6). As such, ethanol yield and maximum productivity (Q_max) values were greater (Table 6). In particular, the selected S. cerevisiae Fm17 exhibited an ethanol yield of 0.46 g/g of consumed glucose, corresponding to almost 90% of the theoretical (0.51 g/g), whereas the industrial benchmark stopped only at 0.40 g/g, which corresponds to 79% of the maximum yield.

The strains produced similar fermenting patterns also in the presence of the pre-treatment liquor Pa3 from P. australis (Figures 1C,D and Table 6). Biomass yield, detected as OD₆₀₀-values, was higher in the case of S. cerevisiae Ethanol Red (Figures 1C,D); meanwhile, ethanol performances were better for the selected strain Fm17 (Figures 1C,D and Table 6), which produced up to 9.04 instead of 8.01 g/L. The resulting ethanol yields were 0.45 and 0.40 g/g of consumed glucose, corresponding to 88 and 78% of the theoretical for S. cerevisiae Fm17 and Ethanol Red, respectively (Table 6).

In the control medium, YNB with 20 g/L of glucose (Figures 1E,F), S. cerevisiae Fm17, and Ethanol Red readily consumed all the glucose available and OD₆₀₀ levels were higher than those detected in the presence of both Cc3 and Pa3. Furthermore, Ethanol Red reached OD₆₀₀ levels greater than those of Fm17 with values of almost 5.0 (Figures 1E,F). Noteworthy, ethanol levels produced by both strains were lower than those detected in the pre-hydrolysates. S. cerevisiae Fm17 and Ethanol Red yielded 8.17 and 8.21 g/L of ethanol, corresponding to 80 and 81% of the theoretical, respectively (Figures 1C,D). Since pre-hydrolysates have a complex chemical composition, presence of additional carbon sources in the medium containing Cc3 and Pa3 is possible, resulting in greater ethanol productions. Furthermore, a higher amount of ethanol produced in the pre-hydrolysates can also be ascribed to the presence of furfural and HMF. Although these chemical compounds exhibit a negative impact on yeast metabolism, their reduction to less toxic compounds can act as a redox sink, thus preventing redox imbalances and increasing final ethanol yield (Wahlbom and Hahn-Hägerdal, 2002; Ask et al., 2013; Favaro et al., 2013a). Furfural and HMF were completely metabolized by the strains (Figures 1A–D). Lower glycerol production observed in Cc3 and Pa3 when compared to the control broth further supports this hypothesis, since glycerol production as redox sink is less favored than furan conversion (Palmqvist et al., 1999; Martín and Jönsson, 2003).

As reported in Table 6, in the presence of both pre-treatment liquors, the volumetric productivities of the yeast strains were generally greater than those recorded in the reference medium (without inhibitor supplementation). This could be due to the presence of weak acids, which can boost the fermentation rate at concentrations below 100 mM (Palmqvist and Hahn-Hägerdal, 2000; Favaro et al., 2013a; Jönsson et al., 2013; Jönsson and Martín, 2016).

To further assess the fermenting abilities of both strains in industrially relevant conditions, their performances were evaluated in YNB medium formulated with Cc3 (Figure 2) or Pa3 (Figure 3) pre-hydrolysates supplemented up to 40 and 92 g/L of glucose, respectively, to mimic the highest glucose concentration obtained by enzymatic saccharification of each steam-exploded WIS (Cotana et al., 2015a, b; Cavalaglio et al., 2016). pH-value was adjusted to 5.0 with 5 M NaOH. Fermentation medium formulated without Cc3 or Pa3 was used as control (Figures 2C,D, 3C,D).

In the presence of the pre-hydrolysate Cc3 (C. cardunculus) and 40 g/L glucose, the strains utilized all glucose available by 20 h of fermentation (Figures 2A,B). Ethanol Red produced higher biomass than Fm17 in both broths (with and without Cc3): final OD₆₀₀ was 4.9 in the control medium and 4.8 in presence of Cc3, amounting to 16% and 20% higher than Fm17, respectively (Figure 2). On the contrary, Fm17 displayed better volumetric productivities and ethanol production (Figure 2 and Table 6). Fm17 and Ethanol Red produced 18.29 g/L and 17.25 g/L of ethanol in the medium formulated with Cc3, respectively, corresponding to 88% and 82% of the theoretical yield (Figures 2A,B and Table 6). Volumetric productivities of Fm17 was significantly higher than those of the industrial yeast, with Q_max values of 0.90 instead of 0.85 (g/L h^–1) for S. cerevisiae Fm17 and Ethanol Red, respectively (Table 6).

In terms of furans reduction, furfural and HMF were completely metabolized by both strains (Figure 2). In the control medium supplemented with 40 g/L glucose, S. cerevisiae Fm17 and Ethanol Red produced lower ethanol levels: 16.92 and 16.95 g/L of ethanol, respectively, corresponding to 82% of the theoretical (Figures 2C,D and Table 6). This finding is consistent with the ethanol performances described earlier (Figure 1), further supporting the hypothesis that the occurrence of additional carbon sources and/or redox sinks in the pre-hydrolysate may have enhanced ethanol production by both strains.

Once exposed to the pre-hydrolysate Pa3 (P. australis) with 92 g/L glucose, the strains confirmed their ability to withstand high inhibitor concentrations (Figures 3A,B). However, glucose consumption of both strains took longer than in the reference medium (Figures 3C,D). Nevertheless, ethanol production was high with S. cerevisiae Fm17 having again the most promise (39.09 g/L) corresponding to about 88% of the theoretical. The volumetric productivity values of both strains were comparable (Table 6).

As already described and discussed above, ethanol levels in Pa3 were again higher than those detected in the reference broth (YNB supplemented with 92 g/L). S. cerevisiae Fm17 and Ethanol Red quickly converted glucose to comparable amounts of alcohol (35.56 and 36.34 g/L, respectively), with an ethanol yield of about 80% of the theoretical (Figures 3C,D and Table 6).

Taken together, the results of small-scale fermentations in the presence of two pre-hydrolysates and increasing concentrations of glucose showed that S. cerevisiae Fm17 outcompeted ethanol performances of the reference strain Ethanol Red currently used in industrial bioethanol production (Figures 1–3 and Table 6). Interestingly, the lower glucose concentrations, the greater ethanol yields and productivities (p ≤ 0.05) were achieved by the superior yeast S. cerevisiae Fm17. Glycerol levels were comparable in both strain fermentations; meanwhile, biomass yields were always higher in the case of the industrial benchmark yeast (Figures 1–3). These findings could be explained considering the ecological origin of Fm17, which has been isolated from grape marcs, an extreme environment with a limited amount of glucose (Favaro et al., 2013a). On the contrary, the industrial strain Ethanol Red has been specifically selected for high alcohol yield and tolerance especially during very high gravity fermentation, typical of the corn ethanol industry where at least 200 g/L glucose is available (Walker and Walker, 2018). As such, the novel S. cerevisiae Fm17 seems to be able to withstand better the inhibitors at lower glucose concentrations, reducing biomass yield in favor of ethanol production. These observations are in agreement with the fact that, under anaerobic conditions, yeast cells use alcoholic fermentation of sugars as the sole pathway to obtain energy in the form of ATP for cellular maintenance and, if sufficient ATP is available, for growth. When ATP is utilized for growth, yeast biomass and associated glycerol are produced at the expense of sugars that are not converted to alcohol (Gombert and van Maris, 2015). Furthermore, considering that under SSF or Consolidated BioProcessing settings, glucose levels, released by commercial or, respectively, recombinant enzymes produced by the engineered yeast, do not usually accumulate because of the quick yeast utilization (Hasunuma and Kondo, 2012; Cripwell et al., 2019, 2020), Fm17 should be considered as a very promising lignocellulosic ethanol strain.

In conclusion, this paper was successful in employing undetoxified steam-exploded lignocellulosic residues for second-generation ethanol production. The liquors were used in a close simulation of industrial conditions, considering that as the key point for strain selection.

A cluster of yeast strains demonstrated inhibitor tolerance higher than those of S. cerevisiae Ethanol Red, the most used microorganism for lignocellulosic ethanol. This finding is of great value considering that to obtain large additional profits, first-generation ethanol plants strive for an increase of even 1% in ethanol yield. Techno-economical evaluations are in progress to determine the weight of using S. cerevisiae Fm17 in the overall process efficiency. Moreover, further studies are on-going to confirm its promising industrial fitness both at higher scale (i.e., bioreactor) and in SSF settings in the presence of WIS collected after steam-explosion of selected lignocellulosic materials.

This study also implies that there are interesting opportunities to isolate or engineer natural yeast variants with performances better than those currently exploited in well-known industrial yeast strains. Moreover, the phenotypic differences between the screened yeast in terms of inhibitor-tolerance indicated that the choice of strain is critical when contemplating the design of a process involving fermentation of lignocellulosic pre-treated materials at industrial scale.

The most burgeoning strain, capable of growing well in undiluted liquors, was also able to ferment in more than one pre-treated feedstocks. This confirmed its great versatility to multiple pre-treated materials, which is one of the requirements for an efficient second-generation ethanol yeast.