Increased bioethanol production from commercial tobacco cultivars overexpressing thioredoxin f grown under field conditions
- First Online:
- Cite this article as:
- Farran, I., Fernandez-San Millan, A., Ancin, M. et al. Mol Breeding (2014) 34: 457. doi:10.1007/s11032-014-0047-x
- 353 Views
Bioethanol is mainly produced from food crops such as sugar cane and maize, and this has been held partly responsible for the rise of food commodity prices. Tobacco, integrated in biorefinery facilities for the extraction of different compounds, could become an alternative feedstock for biofuel production. When grown for energy production, using high plant densities and several mowings during the growing season, tobacco can produce large amounts of inexpensive green biomass. We have bred two commercial tobacco cultivars (Virginia Gold and Havana 503B) to increase the carbohydrate content by the overexpression of thioredoxin f in the chloroplast. Marker-free transplastomic plants were recovered and their agronomic performance under field conditions was evaluated. These plants were phenotypically equivalent to their wild types yet showed increased starch (up to 280 %) and soluble sugar (up to 74 %) contents in leaves relative to their control plants. Fermentable sugars released from the stalk were also higher (up to 24 %) for transplastomic plants. After heat pretreatment, enzymatic hydrolysis and yeast fermentation of leaf and stalk hydrolysates, an average of 20–40 % more ethanol was obtained from transplastomic plants than their wild-type controls. We propose an integral exploitation of the entire tobacco plant managed as a forage crop (harvesting sugar and starch-rich leaves and lignocellulosic stalks) that could considerably cheapen the entire production process.
KeywordsTobaccoChloroplast transformationThioredoxinBioethanolField trialCarbohydrate
Transportation fuels depend basically on petroleum, a non-renewable fossil hydrocarbon. Oil consumption has grown continuously, the USA being the foremost end user followed by China with a rising economic growth rate. It has been predicted that petroleum resources will be drastically constrained in the mid-term, with a consequent steady price increase. In addition to the high cost of petroleum, the release of polluting gases and their implication in global climate change have fostered the use of renewable resources for the production of transportation fuels, such as ethanol. These alternative fuels could substantially reduce the worldwide dependence on petroleum.
Ethanol can be blended with gasoline in any ratio, thus obtaining fuels with different properties. Extensive international experience has already demonstrated that blends containing up to 10 % ethanol do not require further engine modifications (Rutz and Janssen 2007). In Brazil, all brands of gasoline engines may run on 20–25 % ethanol. Currently, flexible fuelled vehicles are manufactured with engines which can run on any type of blend ranging up to 85 % ethanol, while even completely dedicated ethanol vehicles have been commercialized with an efficient use of absolute ethanol (Rutz and Janssen 2007). Consequently, it is not surprising that the fuel ethanol industry has been growing extensively worldwide, with a global production of 82,600 million litres recorded in 2012, the USA and Brazil being the main producers, representing more than 85 % of the world production (http://ethanolrfa.org/pages/World-Fuel-Ethanol-Production).
First-generation bioethanol production is mainly based on food crops. Sugar (sugar cane and sugar beet) and starch (maize and wheat) crops are the most widely used for ethanol production. To state the point, grain ethanol refineries consumed up to almost 40 % of the total USA maize production in the 2010/2011 crop season (Du and McPhail 2012). Due to competition with the food supply, the biofuel boom has been held partly responsible for food commodity price increases (OECD 2008; Food and Agricultural Organization 2011; Kretschmer et al. 2012). Hence, a future increase in ethanol production will need to rely on exploiting other non-food/feed-associated biomass products such as lignocellulosic feedstock (Hahn-Hägerdal et al. 2006; Sarkar et al. 2012), food wastes (Uncu and Cekmecelioglu 2011) and alternative plant species which are well adapted to particular agroecosystems and possess a developed infrastructure for crop management and harvest processing. Tobacco (Nicotiana tabacum L.) could become one of these dedicated energy crops, given the highly plastic nature of the species, grown in over 125 countries and across 4 million hectares of land, a third of which are located in China alone. The tobacco plant is a high biomass producer, yielding up to 160 tonnes per hectare through multiple harvests (Wildman 1979). Production of tobacco in developed countries has lately declined due to the weak demand of both the internal and the export markets. The cultivated area in the last 10 years has been reduced by 31 % in the USA and 45 % in Europe (http://faostat3.fao.org/home/index.html). This tendency has sharply increased in the European Union due to the reduced incentives derived from the Common Agricultural Policy (Food and Agricultural Organization 2003). Part of this arable land area could be dedicated to tobacco biomass production and ethanol conversion. In fact, tobacco was commercially used as a feedstock for ethanol production in the 1980s in the USA (Floyd Agricultural Energy Corporation, Virginia) and has currently been proposed as the ideal crop for bio-based products including ethanol (NewAgriculture Inc., http://www.newagriculture.com/home.html). For cost-effective ethanol production, tobacco could be integrated into a biorefinery facility for the extraction of other compounds (e.g. proteins and xanthophylls).
Genetic engineering provides a tool for improving plant characteristics in order to derive raw biomass materials appropriate for use as biofuel feedstocks. Improvement includes increasing the overall biomass, enhancing polysaccharide content and inducing cell wall modifications so as to diminish the need for pretreatments (expressing deconstructing enzymes in plants or down-regulating lignin biosynthesis) (Vanholme et al. 2008; Verma et al. 2010). Recently, it has been shown that the overexpression of plastidial thioredoxin f (Trx f) from the chloroplast genome leads to enhanced starch accumulation in leaves of the cultivar Petite Havana, a rather low biomass producer (Sanz-Barrio et al. 2013). Thioredoxins are oxidoreductases that mediate the thiol-disulfide exchange of Cys residues. Many starch metabolizing enzymes have been reported to be redox-regulated (Kötting et al. 2010; Glaring et al. 2012). Trx f could therefore act as a reductant of the redox-regulated enzymes involved in carbohydrate metabolism and, as a consequence, allow an increase in the rate of carbohydrate biosynthesis (Sanz-Barrio et al. 2013). In the present work, we have overexpressed Trx f in the plastids of two commercial tobacco cultivars of high biomass production. Transplastomic plants grown under field conditions showed increased starch and soluble sugar contents in leaves. The integral exploitation of tobacco biomass (leaves and stalks) was studied. Pretreatment experiments, enzymatic hydrolysis and yeast fermentation showed a better performance in these transplastomic plants as feedstock for ethanol production compared to their wild-type relatives.
Materials and methods
Development of chloroplast transgenic plants overexpressing the trx f gene
Two commercial cultivars of N. tabacum (Virginia Gold and Havana 503B) were used for the plastid transformations. Tobacco leaves grown in vitro were transformed with the pL3-PrrnG10LTrxf chloroplast transformation vector (Sanz-Barrio et al. 2013), where the tobacco trx f gene (Sanz-Barrio et al. 2012) was expressed under the control of the rrn promoter fused to the leader sequence of the bacteriophage T7 gene 10. The aadA gene, conferring resistance to spectinomycin and streptomycin, was used as a selectable marker gene. The Biolistic PDS-1000/He device (Bio-Rad) was utilized for the integration of the transgenes into the plastid genome as previously described (Daniell 1997). Two rounds of selection and shoot development on RMOP medium containing 500 mg/L of spectinomycin were performed. Regenerated transformants were transferred to soil.
Southern and Western blot analysis
Total plant DNA (10 μg) was digested with BglII, separated in a 0.8 % (w/v) agarose gel and transferred to a nylon membrane. A 0.8-kb probe (P1), homologous to the flanking sequences, was used for hybridization (Farran et al. 2008). Probe labelling and hybridization were performed using the chemiluminescent Dig High Prime DNA labelling and detection kit (Roche, Indianapolis, IN, USA). After Southern blot analysis confirmation of the T0 generation, selected plants were grown in the greenhouse for seed production.
For Western blot analysis, proteins from transformed and untransformed leaf samples were extracted and blotted using a specific Trx f antibody as previously described (Sanz-Barrio et al. 2012).
Removal of the antibiotic resistance gene by the Cre/lox site-specific recombination system
The transit peptide of the tobacco RuBisCO small subunit was translationally fused to the Cre recombinase from the P1 bacteriophage. The chimeric gene was introduced into a pBin20 binary vector under the control of the CaMV 35S promoter and the nopaline synthase terminator. The Agrobacterium tumefaciens GV3101 strain was used for nuclear transformation of the transplastomic plants. Briefly, nuclear gene transformants were selected by kanamicin resistance on regeneration medium containing 50 mg/L kanamycin and 250 mg/L cefotaxime. Kanamycin-resistant shoots were subcultured to a rooting medium containing 50 mg/L kanamycin. Removal of the aadA gene in the regenerated plants was confirmed by Southern blot analysis (using P1 and aadA probes) and by the in vitro seed germination of the T2 generation in a medium containing 500 mg/L spectinomycin. Additionally, loss of the nuclear nptII gene in seed progeny (T3 generation) was confirmed by kanamycin sensitivity.
Transplastomic and wild-type Virginia Gold and Havana 503B tobacco plants were grown under standard field conditions in the Experimental Station of the INTIA (Sartaguda, Navarre, Spain). The notification for the release of genetically modified higher plants (B/ES/12/16) was authorized by the Ministerio de Agricultura, Alimentación y Medio Ambiente. A randomized complete block design with three replications was used. For each transplastomic and wild-type lines, 15 plants were cultured per block (plantation density of 18,500 plants/ha; 90 cm between rows and 60 cm between plants). Field transplantation was undertaken on the 23rd of May 2012. Topping (removal of the flower head and the first few top leaves to prevent the production of seeds) was performed 55 days after transplanting. Plant height and relative leaf chlorophyll content (measured with SPAD 502 chlorophyll meter, Minolta Optics Inc, Tokyo, Japan) were recorded 60 days after transplanting. Total weight per plant and specific leaf weight (SLW) were measured at the date of harvest (71 days after transplanting). For the calculation of the SLW, five disks per leaf were cut with a cork borer. Disks from all plant leaves were collected and dried at 80 °C for 2 days. Total dry weight refers to the total disk area and is expressed as mg/cm2.
Total leaves and stalks derived from two plants per block (yielding a total of six individual wild-type or transplastomic plants per cultivar) were harvested, pooled and stored at 4 °C until sample conditioning (less than a week).
Conditioning of leaf and stalk samples
The pooled leaf and stalk samples were dried separately in an air-dry oven at 45 °C to further stop any biological reactions and possible contaminations. The dried plant material was milled to a particle size of 1 mm using an ultra-centrifugal mill (Retsch ZM200, Haan, Germany) and stored in hermetic plastic bags at 45 °C until further experimental usage. Representative samples obtained from the pools were processed and analysed.
Leaf and stalk samples were dried at 105 ± 2 °C to a constant weight and compositional values refer to dry weight. Ash content was determined after air combustion for 1 h at 550 ± 10 °C in a muffle furnace. Starch content was quantified by an enzymatic method including α-amylase, protease and amyloglucosidase (Total Dietary Fibre AOAC 985.29, Megazyme, Bray, Ireland) and subsequently sugars were quantified by HPLC as described below.
For the quantification of soluble sugars, 0.5 g of dried samples were mixed with 100 mL water and incubated for 1 h at 70 °C. Samples were centrifuged at 6,000×g for 10 min and the supernatant stored at −20 °C until analysis. Monomeric sugars (soluble monosaccharides), cellobiose and total sugars (monosaccharides and oligosaccharides) were determined according to the procedure published by the National Renewable Energy Laboratory (NREL, Golden, CO, USA) (www.nrel.gov/biomass/pdfs/42632.pdf). Oligomeric sugars were subjected to acid hydrolysis by adding enough 72 % H2SO4 (w/w) to bring the acid concentration of an aliquot of 10 mL up to 4 %. Samples were stored at −20 °C until analysis.
Sugar analysis was carried out by HPLC. The content of monosaccharides (d-glucose, d-xylose-fructose-galactose, l-arabinose), disaccharides (cellobiose, maltose) and trisaccharides (maltotriose) were quantified by an HPLC system (Agilent Technologies, model 1200) equipped with a refractive index detector. The separation was performed with an ICSep ION 300 column (7.8 × 300 mm, Transgenomics, Glasgow, United Kingdom) at 72 °C, with the mobile phase consisting of 8.5 mM H2SO4 diluted in water.
The contents of cellulose, hemicellulose (xylans and arabinans) and lignin were analysed using a two-step acid hydrolysis method according to the procedure published by the NREL (www.nrel.gov/biomass/pdfs/42618.pdf). Dried samples (0.3 g) were treated with 3 mL of 72 % H2SO4 (w/w) and placed in a water bath at 30 °C for 1 h. Samples were diluted with 84 mL of water to give a H2SO4 concentration of 4 %. Samples were autoclaved at 121 °C for 1 h. After cooling, the samples were vacuum filtered through a 0.22-μm mesh and an aliquot was collected for the analysis of sugars, and the acid soluble lignin was measured at 320 nm in a spectrophotometer. Insoluble acid solids were washed with water, dried at 105 °C, weighed and finally ashed at 575 °C ± 25 °C for acid insoluble lignin content determination.
Protein content was determined by the Dumas method, which consists of high temperature combustion (about 900 °C) of the sample in the presence of oxygen, leading to the release of carbon dioxide, water and nitrogen. Nitrogen is separated from the rest of the components using a conductivity column. A conversion factor of 6.25 was used for protein content determination.
Pretreatment and enzymatic hydrolysis of leaf samples
Dried leaf samples were mixed with 0.1 M citrate buffer, pH 4.8, in a 10 % (w/w) proportion of solid to liquid, and pretreated at 110 °C for 1 h. The final weight of the sample was approximately 28 g.
Four different enzyme cocktails (Novozymes A/S, Bagsvaerd, Denmark) were used for the enzymatic hydrolysis. The concentration of enzymes is expressed as % (w/w dry sample): mixture 1 (α-amylase, 0.001; β-glucoamylase, 0.002; cellulases Cellic® Ctec 3, 10.0), mixture 2 (α-amylase, 0.001; β-glucoamylase, 0.002; cellulases Cellic Ctec 3, 10.0; carbohydrolase + protease, 1), mixture 3 (α-amylase, 0.001; β-glucoamylase, 0.002; cellulases Cellic Ctec 2, 10.0), and mixture 4 (cellulases Cellic Ctec 3, 10.0). Samples were incubated in an orbital shaker (180 rpm) for 72 h at 50 °C. With regard to mixtures 1, 2 and 3, samples were kept at 80 °C during the first hour. Cellulases were always added after the incubation at 80 °C. Soluble sugars were quantified by HPLC as described above.
The efficiency of the enzymatic hydrolysis was calculated as follows: [(glucose released after hydrolysis × hydrolysis volume/sample initial dry weight) × 100]/[% cellulose × 1.11 (conversion factor of cellulose to glucose) + % starch × 1.11 (conversion factor of starch to glucose) + % fermentable soluble sugars contained in the initial biomass].
Pretreatment and enzymatic hydrolysis of stalk samples
Stalk samples were mixed with water in pressurized stirred-tank reactors (Autoclave Engineers, model M010SS) with a solid:liquid proportion of 10 % (w/w dry sample) reaching a final weight of 50 g, and subjected to three different temperature regimes (170, 190, 210 °C) for 30 min. In addition to temperature, acid (1 % w/w of H2SO4 with respect to dry sample) and pressure (20 bar, supplied from 200-bar pressurized pure nitrogen bottle) treatments were also applied. Nine different combinations were assayed: temperature alone, temperature plus acid, and temperature along with acid plus pressure, for each of the three temperature regimes.
After pretreatment, samples were weighed to calculate the recoveries of the solid (insoluble) and liquid fractions. Prior to enzymatic hydrolysis, solid fractions had been dried at 45 °C to a constant weight. Soluble sugars and sugar degradation products (acetic acid and furfurals) of the liquid fractions were also determined. Acetic acid was quantified by HPLC as described above for the soluble sugars. Furfural and 5-hydroxymethylfurfural (HMF) were quantified by HPLC by way of a DAS index detection using a Zorbax column (250 × 4.6 mm, Agilent Technologies) at 50 °C, and a mobile phase consisting of water: acetonitrile in a proportion of 80:20.
Enzymatic hydrolysis was performed as described above for leaf samples, but using only enzyme mixture 4 [cellulases Cellic Ctec 3, 10 % (w/w dry sample)]. Soluble sugars were quantified by HPLC as described above. The efficiency of the enzymatic hydrolysis was calculated as follows: [(glucose released after hydrolysis × hydrolysis volume/sample initial dry weight) × 100 × pretreatment recovery factor]/[% cellulose × 1.11 (conversion factor of cellulose to glucose) of the initial biomass].
After enzymatic hydrolysis, plant samples were subjected to fermentation with Saccharomyces cerevisiae (strain Ethanol Red, Fermentis, Marcq-en-Baroeul, France) with an inoculum concentration of 1 × 107 cfu/mL of hydrolysed material, in final volumes of approximately 28 and 4 mL for leaves and stalks, respectively. The process was conducted at 37 °C with a constant orbital agitation (180 rpm) for 24 h. Ethanol production was determined by a gas chromatography–flame ionization detector using a DB Wax 15-m column (Agilent Technologies, model 7890A) and 1-propanol as the internal standard.
Generation and characterization of the transplastomic tobacco plants overexpressing Trx f
Plastid transformation efficiency was assessed in six high-biomass tobacco cultivars with gfp as the reporter gene. Taking into account the number of transformed buds per bombarded leaf, shoot development, root formation and leaf fluorescence (data not shown), Virginia Gold and Havana 503B cultivars were selected (hereafter Virginia and Havana, respectively). Leaves from plantlets of these two cultivars grown in vitro were transformed by biolistics. Single regenerated shoots selected after two rounds of spectinomycin application were rooted and transplanted. The transformation efficiency was similar to that of the well-established Petite Havana cultivar: 5.2 and 3.9 transformed shoots were obtained per bombarded leaf in Havana and Virginia cultivars, respectively. In the case of Petite Havana, usually 4–5 shoots are observed per bombarded leaf (Molina et al. 2004).
Southern blot analysis was performed to verify the site-specific integration and to confirm homoplasmy. Total plant DNA was digested with BglII. The flanking region probe (P1, Fig. 1a) identified a 4.5-kb fragment in wild-type plants but a 6.7-kb fragment in the plastid transgenic plants (Fig. 1b). The absence of 4.5-kb bands in the transformed lines indicated homoplasmy.
Overexpression of Trx f was confirmed by Western blotting with a specific Trx f antibody. Trx f was detected in both transplastomic cultivars (Virginia and Havana) (Fig. 1c). No signal was detected in the wild-type plants. A lack of signal was also observed in wild-type Petite Havana plants when 10 μg of total leaf protein were loaded per lane (Sanz-Barrio et al. 2013). The spot intensity of Virginia and Havana transplastomic plants was equivalent to that of the transplastomic Petite Havana cultivar (Fig. 1c), indicating a high Trx f expression level.
Development of marker-free transplastomic plants
Since the Cre-lox site-specific recombination system has already been efficiently used to obtain marker-free transplastomic plants (Corneille et al. 2001), transplastomic T1 plants were nuclearly transformed with the cre gene from the P1 phage via A. tumefaciens. The transit peptide of the tobacco RuBisCO small subunit (SSU) fused to the N-terminus of Cre recombinase (SSU-Cre) targeted Cre to the plastid (Supplementary Fig. S1a), allowing the excision of the lox-flanked aadA gene from the plastid genome. Specific excision of the aadA gene was confirmed by Southern blotting. The flanking sequence P1 probe identified the expected 5.4-kb fragment in the nuclear cre-transformed transplastomic plants (VIR-TCre and HAV-TCre) (Supplementary Fig. S1b), which was smaller than the 6.7-kb fragment identified in their related transplastomic plants (VIR-T and HAV-T). Confirmation of the aadA gene removal was obtained via the aadA probe. VIR-T and HAV-T showed the specific band, whereas VIR-TCre and HAV-TCre lacked the signal (Supplementary Fig. S1b). The nuclear nptII gene was subsequently removed by segregation in the T2 seed progeny of the VIR-TCre and HAV-TCre plants. In vitro germination in the presence of kanamycin allowed for the identification of the sensitive seedlings. These seedlings were recovered, cultured in kanamycin-free medium and transplanted to obtain seeds. T3 generation plants were again analysed to confirm the antibiotic sensitivity by in vitro germination assays (Supplementary Fig. S1c,d). As expected, VIR-TCre and HAV-TCre were sensitive to both spectinomycin and kanamycin, equivalent to the wild-type plants. An unrelated nuclear transgenic plant (NUC) was used as a control for kanamycin resistance.
Phenotypic and compositional characterization of Trx f-overexpressing plants grown under field conditions: increased leaf starch and soluble sugar levels
Phenotypic analysis of Trx f-overexpressing field grown tobacco plants (cultivars Virginia and Havana) relative to their wild types
Total height (cm)
140 ± 1.4
144 ± 1.2
143 ± 2.3
137 ± 3.2
Total weight/plant (g)
2,940 ± 380
2,750 ± 410
1,980 ± 350
1,880 ± 250
Specific leaf weight (SLW, mg/cm2)
4.24 ± 0.2
4.66 ± 0.2
4.85 ± 0.3 a
5.99 ± 0.2 b
36.3 ± 0.8
35.3 ± 0.7
39.1 ± 0.8
41.3 ± 0.7
Compositional characterization of leaf and stalk samples
Parameter (% w/dw)
5.01 ± 0.04
4.87 ± 0.11
4.10 ± 0.02
5.00 ± 0.16
6.66 ± 0.01 a
7.60 ± 0.13 b
7.20 ± 0.04 a
12.50 ± 0.03 b
1.88 ± 0.01 a
4.32 ± 0.05 b
0.94 ± 0.00 a
3.41 ± 0.06 b
5.27 ± 0.18 b
4.30 ± 0.02 a
4.83 ± 0.09
4.86 ± 0.22
2.29 ± 0.04
2.13 ± 0.00
2.57 ± 0.05
2.15 ± 0.11
0.49 ± 0.02
0.43 ± 0.03
0.49 ± 0.00
0.47 ± 0.02
3.37 ± 0.02 b
2.85 ± 0.02 a
3.26 ± 0.07
2.80 ± 0.03
30.20 ± 0.44
28.60 ± 0.19
26.90 ± 0.04 b
24.12 ± 0.25 a
19.03 ± 0.04 b
17.98 ± 0.07 a
20.13 ± 0.01 b
18.69 ± 0.04 a
Generally, Trx f overexpression in tobacco plastids did not seem to affect the stalk composition (Table 2). Cellulose was the most abundant fraction [~30 % weight/dry weight (w/dw)], followed by proteins (~18 %), hemicellulose (~15 %, mainly xylans plus a small proportion of arabinans) and lignin (~11 %). Starch was not detected and the content of soluble sugars was very low, approximately 3 % (w/dw). These values were similar to those that have previously been reported for tobacco stalks (Martin et al. 2008; Akpinar et al. 2010).
Increased ethanol production from leaves of Trx f-overexpressing plants
Leaf sample quantification of total and fermentable sugars released after pretreatment and enzymatic hydrolysis followed by ethanol production after fermentation
Total sugars (% w/dw)
16.6 ± 0.1 a
20.3 ± 0.0 b
17.9 ± 0.7 a
23.9 ± 0.0 b
Fermentable sugars (% w/dw)
15.9 ± 0.1 a
19.8 ± 0.0 b
17.1 ± 0.6 a
23.2 ± 0.0 b
Efficiency of enzymatic hydrolysis (%)
Fermentation efficiency (%)
Ethanol (% w/dw)
6.9 ± 0.2 a
8.6 ± 0.1 b
7.1 ± 0.2 a
10.2 ± 0.2 b
Preparation of hydrolysates from tobacco stalks for ethanolic fermentation
In preliminary experiments, we had observed that relatively mild hydrothermal pretreatments (up to 130 °C) were not sufficient for an efficient degradation of stalk cellulose. Therefore, we analysed the effect on the production of sugar-enriched hydrolysates of harsher pretreatments based on higher temperatures (170–210 °C) combined with dilute acid or dilute acid plus pressure. To monitor the severity of the pretreatments, some degradation products were quantified (Supplementary Table S2). The maximum levels recorded were 5.19, 0.96 and 0.24 g/L of acetic acid, furfural and HMF, respectively. It has been demonstrated, working with different feedstocks, that these concentrations do not decrease ethanol yields (Larsson et al. 1999; Palmqvist et al. 1999; Pienkos and Zhang 2009; Erdei et al. 2010).
Comparing the different pretreatments, no clear effects were detected with regard to total sugars (Supplementary Fig. S2a). In contrast, a positive effect of temperature on the production of fermentable sugars after enzymatic hydrolysis of the solid fraction was observed (Supplementary Fig. S2b), which was not enhanced by the combination of temperature with dilute acid or pressure. Increased heat treatments of 170–190 °C gave an increase in ethanol production after fermentation, which did not further increase at 210 °C (Supplementary Fig. S2c). However, when data on the liquid and solid fractions (ratios of recovery) were considered, no marked differences between the pretreatments were observed, especially as to final ethanol production (Supplementary Table S3), thus showing that the addition of acid and the use of a higher pressure were unnecessary. The lack of differences with regard to ethanol production between the pretreatments could be explained by a reduction in the solid fraction’s recovery with increasing temperatures. Differences between the total and fermentable sugar fractions within each sample were mainly due to hemicellulose hydrolysis, rendering xylans and arabinans that cannot be fermented by S. cerevisiae. The enzymatic hydrolysis efficiency of the cellulase mixture (cellulases Cellic Ctec 3) was far from complete (55–63 %, Supplementary Table S3). This is therefore an important aspect for further improvement. Taking these results together, the pretreatment at 190 °C was selected for further experiments as the enzymatic hydrolysis ratio was one of the highest. In addition, taking into consideration a scaled-up industrial process, in which there is no precise way to control temperature, 190 °C would seem to guarantee ethanol yields similar to those obtained in the present experiment.
Tobacco stalks are a relevant biomass source for ethanol production
Stalk sample quantification of total and fermentable sugars of the liquid and solid fractions released after pretreatment at 190 °C and enzymatic hydrolysis together with ethanol production after fermentation
Total sugars (% w/dw) liquid fraction
0.30 ± 0.01 a
0.56 ± 0.03 b
0.45 ± 0.01
0.51 ± 0.02
Fermentable sugars (% w/dw) liquid fraction
2.59 ± 0.02 a
5.82 ± 0.01 b
4.15 ± 0.07 a
4.66 ± 0.09 b
Recovery of solid fraction (%)
Total sugars (% w/dw) solid fractiona
20.9 ± 0.7 a
29.7 ± 0.4 b
23.4 ± 0.3
24.1 ± 0.7
Fermentable sugars (% w/dw) solid fractiona
17.9 ± 0.1 a
22.2 ± 0.3 b
18.3 ± 0.2
20.1 ± 0.6
Efficiency of enzymatic hydrolysis (%)
51.5 ± 0.3 a
62.8 ± 0.9 b
55.5 ± 0.7
60.1 ± 1.9
Fermentation efficiency (%)
Ethanol (% w/dw)
8.1 ± 0.0 a
10.0 ± 0.0 b
8.5 ± 0.2
9.2 ± 0.1
The relative leaf and stalk proportions of the transplastomic plants were calculated. The leaf biomass (based on fresh weight) was 60 and 68 % in the Virginia and Havana cultivars, respectively. These values are in accordance with a previous field experiment carried out in Cadreita (Navarre, Spain) in which 17 commercial cultivars were grown and evaluated under standard conditions, with a mean leaf biomass corresponding to 62 %. The proportions of the final ethanol yield derived from leaf and stalk biomass were calculated. For the Virginia cultivar, 47 % of ethanol was derived from stalks and 53 % from leaves. The figures for the Havana cultivar were 36 % (stalks) and 64 % (leaves). These results reflect the importance of stalks for biomass processing when tobacco is cultivated for ethanol production: around 35–50 % of the final ethanol yield was derived from stalks of plants grown under standard conditions.
Although bioethanol is mainly produced from sugar cane, maize or other grain cereals, alternative starchy and lignocellulosic species are under investigation. Cassava, alfalfa, poplar and miscanthus are just a few examples of biomass crops that have been proposed for ethanol production. Crop yields strongly depend on local agro-ecological conditions, which therefore influence the choice of the best feedstock for ethanol production. For instance, the yield of cassava in some tropical countries out-performs that of maize, favouring cassava cultivation in such environments (Sánchez and Cardona 2008). In addition, the genetic modification of herbaceous plants to increase carbohydrate content could facilitate their use as feedstocks for biofuel production. Hence, plants well-adapted to local conditions and with an improved carbohydrate content could foster the geographical expansion of bioethanol production.
Here, we propose that tobacco, a high biomass crop, with enriched carbohydrate content could develop into an alternative energy crop, particularly in some developed countries with dedicated infrastructure yet with a continuous reduction in cultivated area. Tobacco has been previously proposed as a production platform for biofuel by increasing the amount of lipids in leaves which would favour use for biodiesel production (Andrianov et al. 2010). The integration of tobacco biomass in biorefinery facilities for the extraction and purification of additional compounds could facilitate a more cost-effective ethanol production. Tobacco produces large amounts of protein (18–30 %, Table 2). These proteins are nutritionally complete with all of the human essential amino acids and could be used as a supply source for human diets or animal feed, inclusive of a higher protein efficiency rating than that of either milk or soy (Wildman 1979). A protein yield of up to 3 tonnes/ha has been estimated, of which approximately half is extractable and consists of food-grade protein (Long 1984). Nicotine content should not be a problem since tobacco biomass would be harvested very young, well before significant amounts of nicotine have been formed. Other valuable co-products could consist of solanesol (used in the synthesis of vitamins E and K) and xanthophylls (additive in poultry food).
We have obtained marker-free transplastomic plants of two commercial cultivars overexpressing Trx f. These plants were phenotypically indistinguishable from their wild-type relatives although they showed an increased content of leaf starch (up to 280 %) and soluble sugars (up to 74 %) relative to the control plants (Table 2). These results are in accordance with those previously obtained for cultivar Petite Havana, a low-biomass producer (Sanz-Barrio et al. 2013). Moreover, fermentable sugar levels released from the stalk after enzymatic hydrolysis were higher in the transplastomic than the control plants (Table 4). However, soluble sugar differences were not observed for the initial stalk compositional characterization (Table 2), probably due to the mild pretreatment applied which impeded the solubilization of stalk components. The observed differences after a harsher pretreatment at 190 °C could be explained by a less organized cell wall matrix structure of the transplastomic plants that would favour the enzymatic hydrolysis and release of sugars. In previous studies, a reduction in the cell wall recalcitrance was achieved by modifying the cellulose structure (Harris and DeBolt 2010), for instance by expression of a foreign gene such as a bacterial endoglucanase (Brunecky et al. 2011) or by using synthetic biology to decrease lignin content (Yang et al. 2013). Thioredoxins catalyse redox reactions and control a wide range of biological pathways. It is to date unknown whether Trx f affects the cell wall structure, yet there is some evidence connecting the redox status with cellulose synthesis. DTT treatment stimulated carbon flux to cell wall components, especially cellulose (Kolbe et al. 2006), and cellulose synthase has been reported to be redox-regulated (Kurek et al. 2002). Therefore, we hypothesize that Trx f overexpression might mediate the modification of the cell wall structure, easing hydrolytic enzyme access. Finally, after fermentation of the leaf and stalk hydrolysates, an average of 20–40 % more ethanol was obtained from the transplastomic plants than the wild-type controls (Tables 3, 4).
The efficiency of enzymatic hydrolysis of the pretreated leaves was absolute. However, the efficiency of enzymatic hydrolysis of the pretreated stalks with the cellulase mixture used (including cellulase, hemicellulase and β-glucosidase activities) was about 60 %. Due to the fact that the carbohydrate composition of tobacco stalks (including mannans, galactans and arabinans) is similar to that of hardwoods, an increase in pretreatment severity has been proposed to improve the enzymatic hydrolysis (Martin et al. 2002). In our experiments, temperatures of up to 210 °C combined with 1 % H2SO4 or 20 bar of pressure were applied. Acid or pressure did not improve the pretreatment efficiency (Supplementary Table S3). Steam pretreatment at 205 °C (Martin et al. 2002) and wet oxidation at 195 °C (Martin et al. 2008) have been previously used for tobacco stalks, with similar results to those obtained in our experiments. A broad range of other chemical pretreatments such as alkali, organosolvent and ionic liquids have been used with different lignocellulosic materials to increase cellulose digestibility and efficient lignin solubilization (Alvira et al. 2010; Zhu and Pan 2010), and could also be appropriate for tobacco biomass. Additional physicochemical pretreatments include liquid hot water (up to 240 °C) and ammonia fiber explosion (AFEX) (Alvira et al. 2010). In particular, AFEX could be well suited to agricultural feedstocks as it reduces lignin content, removes some hemicellulose components and decrystallizes cellulose. It is also a very promising method with regard to operating costs (Hu et al. 2008).
Improvement in enzymatic saccharification could also be achieved by increasing the enzyme loading (Sánchez and Cardona 2008) or by the addition of specific xylanases and β-glucosidases to the enzyme cocktail. For example, a cellulose conversion of over 90 % was achieved in wheat (Saha and Cotta 2006) and barley (Saha and Cotta 2010) straw using preparations which included the following three enzymes: cellulase, xylanase and β-glucosidase. A cost-effective way of performing the enzymatic hydrolysis would be the expression of cell-wall degrading enzymes in the same plant species which is used for biomass production. The chloroplast is an ideal subcellular compartment for this purpose, with very high levels of foreign protein accumulation. Exo- and endoglucanases, β-glucosidases (Gray et al. 2009, 2011; Ziegelhoffer et al. 2009; Verma et al. 2010; Petersen and Bock 2011), endo-β-mannanase (Agrawal et al. 2011), hemicellulases (Kolotilin et al. 2013) and additional enzymes (pectate lyase, cutinase, swollenin, xylanase, acetyl xylan esterase and lipase) (Verma et al. 2010) have been successfully expressed in transplastomic tobacco plants in their active forms. In some cases, plant crude-extract enzyme cocktails showed a higher activity level and yielded more glucose than commercial cocktails (Verma et al. 2010), demonstrating the great potential associated with plant-derived enzyme production.
In order to use tobacco efficiently as a feedstock for biofuel production, maximum biomass yields of green tissue should be generated. Growing tobacco for biomass or cigarette production requires very different crop management practices. As a biomass crop, tobacco should be managed more as a forage crop with a very high plant density, developing a kind of grass lawn. Plants would be harvested by mowing when they reached a height of 50 cm, leaving the stumps to resprout, thus forming a dense tangle of new shoots and leaves within days (van Beilen et al. 2007). Multiple harvests could be possible under optimal environmental conditions. These young shoots comprise the sugar-richest and highest protein stage of the plant’s growth cycle (van Beilen et al. 2007), which would make the process of ethanol production easier. It has been estimated that a cropping season of 6–8 months could generate up to 160 tonnes/ha of green tissue based on wet weight, which is equivalent to 16–32 tonnes of dry biomass (Wildman 1979). Tobacco farming as a biomass crop is expected to be cheap (sowing instead of planting, reduced weed management and mechanized harvesting by mowing) (van Beilen et al. 2007). Integral exploitation of the whole plant (sugar and starch-rich leaves and lignocellulosic stalks) would be performed in a single production process. Analogous to the tobacco leaf and stalk scenario, this idea of mixed feedstocks has already been investigated to facilitate the introduction of second-generation technology for bioethanol production by mixing wheat straw and meal (Erdei et al. 2010).
In conclusion, the overexpression of Trx f in tobacco chloroplasts led to high-starch and high-sugar phenotypes in field trials. These improved tobacco commercial cultivars, grown in very high plant densities for maximum biomass yield, could be an alternative production platform for biofuel, with an increase of up to 40 % in ethanol production over control plants. To our knowledge, this is the first field trial of genetically modified plants via chloroplast transformation for biofuel production. Furthermore, this technology also brings about the possibility of developing alternative crops potentially useful for improved biofuel generation.
The authors are grateful to Irantzu Alegría and co-workers of the Biomass Energy Department (Renewable Energy National Centre, Spain) for technical assistance with regard to biomass processing and for the critical reading of the manuscript; Lorea Villanueva for the collaboration undertaken with plant transformations; and Jaime Zabaleta (INTIA, Spain), Esther Molinero and Amaia Altuna for field experiment collaborations. This work was supported by Grant AGL2010-15107 from the Ministerio de Ciencia e Innovación (Spain).