Improvement of enzymatic saccharification yield in Arabidopsis thaliana by ectopic expression of the rice SUB1A-1 transcription factor

Plant material

Arabidopsis thaliana Col-0 accession was used as the wild-type (WT). Transgenic genotypes were described previously (Peña-Castro et al., 2011). Briefly, SUB1A-1 or SUB1C-1 cDNAs from Oryza sativa cv M202(SUB1) were expressed under Cauliflower Mosaic Virus 35S promoter with a N-terminal immunogenic FLAG-tag in Col-0. Two independent single-copy T4 generation transgenics were used for each transgene: OxSUB1A-L5 and -L12 and OxSUB1C-L6 and -L10.

Plant growth conditions

Arabidopsis seeds were surface sterilized (70% v/v EtOH for 5 min followed by 6% v/v hypochlorite for 2 min and six 1-min rinse steps with ddH₂O) and germinated in half-strength Murashige and Skoog agar medium (MS, salts 0.215% w/v, 1% w/v sucrose, 1% w/v agar, pH 5.7) in vertical plates. Seedlings were transferred when 7-day-old to substrate (Sunshine Mix #3 plus 1:4 volume perlite:substrate, autoclaved for 2 h and mixed with 2% w/w slow liberation fertilizer NPK 12:12:17) and watered every 2 days. Germination and growth was under long-day conditions (16 h light/8 h dark, 150 µE m⁻² s⁻¹, 60% humidity) in a growth chamber (Conviron CMP6010). ZT0 (Zeitgeber Time) was the start of the light cycle (day). Genotypes were grown side-by-side in a randomized manner to minimize experimental noise.

Reducing sugars, cell wall digestibility and starch content

All rosette leaves were harvested at the time described in each experiment, frozen in liquid nitrogen and stored at −80 °C. For all experiments, leaves were ground to a fine powder in liquid nitrogen with mortar and pestle, weighted and further stored or processed. An experimental strategy was designed to quantify the three main components of saccharification yield, namely free reducing sugars, cell wall digestibility and starch content (Fig. 1).

To measure free reducing sugars, 100–120 mg FW of powdered leaves were incubated with ddH₂0 for 5 min and centrifuged (13,000 rpm for 5 min) to remove debris. The supernatant (100 µl) was mixed 1:1 with DNS reagent (1% w/v 3,5-dinitrosalicylate, 30% w/v sodium potassium tartrate, 1.6% w/v NaOH) and incubated in a boiling water bath for 5 min, then diluted with 1 ml ddH₂0 and absorbance was determined at 545 nm in a spectrophotometer (Miller, 1959). A glucose standard curve (0.1 to 5 mg Glucose/ml, R = 0.985) was analyzed and used as reference.

Cellulose digestibility and starch content were enzymatically assayed as previously described (Chuck et al., 2011). To test cellulose saccharification yield, commercial cellulase enzyme complex Accellerase 1500 (Genencor, Cedar Rapids, Iowa, USA) composed of proprietary exoglucanase, endoglucanase, hemicellulase and beta-glucosidase was used. Powdered leaves were weighted in 15 ml capped plastic tubes (100–125 mg FW) and 200 µl of 80% ethanol were added, and the sample vortexed. Next, 3 ml of acetate buffer plus 0.74% w/v CaCl₂ (pH 5.0) with 1.7% v/v Accellerase 1500 were added, mixed by inversion and incubated at 50 °C for 24 h with rotation (11 rpm) in an oven. Saccharification was stable from 12–36 h as determined in a preliminary kinetics assay (Fig. S1). Reactions were stopped by incubation at 70 °C for 15 min with rotation in an oven. To measure starch content, samples were treated as described above and further hydrolyzed using the manufacturer’s instructions for the Total Starch K-TSTA kit (Megazyme, Bray, Ireland), which includes a thermostable α-amylase digestion in boiling water for 12 min with vigorous stirring every 4 min, and an amyloglucosidase digestion in a 50 °C water bath for 30 min.

Glucose from cellulose and starch was quantified by glucose oxidase/peroxidase at 510 nm in a spectrophotometer as indicated in the commercial kit Total Starch K-TSTA kit. In parallel, Accellerase buffer (blank), carboximethylcellulose and soluble starch (efficiency probes) were processed. The blank was subtracted from calculations and only experiments with efficiency >93% based on the two probes were employed.

Iodine staining

Starch was visualized in rosettes by iodine staining as previously described (Bahaji et al., 2011; Ovecka et al., 2012) with the following modifications. Whole plants were harvested and immediately infiltrated under vacuum with 3.7% v/v formaldehyde in 0.1 M potassium phosphate buffer (pH 6.6) for ∼10 min. Plants were incubated with hot 80% ethanol for 30 min under constant agitation, stained with iodine solution (KI 2% w/v, I₂ 1% w/v) for 30 min in the dark and rinsed until the blue precipitate of starch was distinguishable from the yellowish background.

Hardness tests

Fracture properties of leaves were measured using a texture analyzer (Brookfield CT325k; Brookfield Engineering Laboratories, Inc., Middleboro, Massachusetts, USA). The three largest rosette leaves from 23-day-old plants were stacked and placed in a fixture base and perforated in the middle of the left blade (avoiding the central vein) with a puncture test probe for fine films (TA-FSF). Resistance was expressed as the force (Newton) applied to break through the tissue.

Starch granule isolation and scanning electron microscopy

Rosette tissue pulverized in liquid nitrogen (2.5 g) was hydrated in 40 ml of water, sonicated for 10 min (100% power, 20% amplitude, 50% intensity, Hielscher Ultrasonic Processor UP200ST) and centrifuged for 5 min at 4,750 × g. The pellet was washed twice with 50 ml of water, resuspended and filtered through a 100 µm and then a 20 µm membrane. The filtrate was centrifuged again at 4,750 × g and the pellet washed with 20 ml of 100% ethanol. Granules were covered with a gold coat and observed in a scanning electron microscope (Helios NanoLab™ 600; FEI, Hillsboro, Oregon, USA).

Quantitative reverse transcription polymerase chain reaction (qRT-PCR)

Total RNA was extracted from Arabidopsis complete seedlings and qRT-PCR was performed as previously described (Peña-Castro et al., 2011). Primers for TUBULIN2 (TUB2, At1g65480) were previously reported (Wenkel et al., 2006). Primers for EXPANSIN2 (AtEXP2, At5g05290, 5′-TTACACAGCCAAGGCTATGGGCTA-3′ and 5′-GCCAATCATGAGGCACAACATCGT-3′) and GLUCAN-1,3,-BETA-GLUCOSIDASE (BGL2, AT3G57260, 5′-TCCTTCTTCAACCACACAGCTGGAC-3′ and 5′-CCAACGTTGATGTACCGGAATCTGA-3′) were obtained from the AtRTPrimer database (Han & Kim, 2006). Primers for GLUCAN WATER-DIKNASE 1 (GWD1, At1g10760) and STARCH EXCESS 4 phosphoglucan phosphatase (SEX4, At3g52180) were previously reported (Weise et al., 2012). Primers for SQUAMOSA PROMOTER BINDING PROTEIN-LIKE 3 to 5 (SPL3-5, At2g33810 At1g53160 At3g15270), SUPPRESSOR OF OVEREXPRESSION OF CONSTANS 1 (SOC1, At2g45660) and FRUITFUL (FUL, At5g60910) were previously reported (Wahl et al., 2013).

Improved production of fermentable sugars in OxSUB1A plants

We analyzed two independent Arabidopsis over-expressing transgenic lines for each SUB1 gene, OxSUB1A-L5 and OxSUB1C-L6 are weak over-expressing lines while OxSUB1A-L12 and OxSUB1C-L10 are strong over-expressing lines (Peña-Castro et al., 2011). As an experimental starting point for analysis of rosettes, we selected Col-0 (Wild-Type, WT) bolting time (22-day-old) since OxSUB1A lines have a late flowering-genotype (Peña-Castro et al., 2011). We hypothesized that flowering inhibition allows OxSUB1A plants to accumulate more carbohydrates at ZT16 (end of day).

To determine starch content and cell wall digestibility in rosette tissue of OxSUB1A and OxSUB1C transgenics, we followed a protocol reported to evaluate saccharification efficiency in switchgrass (Panicum virgatum L.), where plant tissue is digested in two sequential steps (Chuck et al., 2011). In the first reaction, the plant material is used as substrate in a cellulolytic enzymatic cocktail (Accellerase 1500) to breakdown cellulose and hemicellulose into glucose and measure cell wall digestibility. In the second reaction, the tissue is digested with α-amylase and amyloglucosidase to quantify starch. For both digestions, saccharification is expressed as released glucose (Fig. 1).

The carbon pool that yielded most glucose in WT Arabidopsis was starch (8.5 mg of glucose g⁻¹ FW), and then free reducing sugars (0.4 mg of glucose equivalents g⁻¹ FW) followed by cell walls (0.1 mg of glucose g⁻¹ FW).

When only free reducing sugars were determined (no enzymatic treatment), OxSUB1A-L5 had 37% more than WT whereas OxSUB1A-L12 did not show a significant difference (Table 1). If only cellulolytic treatment was applied, an improvement in cell wall saccharification was detected: OxSUB1A-L5 and OxSUB1A-L12 generated 16% and 23% higher yields than WT, respectively (Table 1 and Fig. S1). OxSUB1A-L5 and OxSUB1A-L12 rosette tissue generated 88% and 36% more glucose from starch than WT, respectively (Table 1). These results indicate that ectopic expression of SUB1A-1 allows plants to conserve carbohydrates, mainly starch, under non-stress conditions.

It has been reported similar saccharification improvement in switchgrass that over-express miRNA156, a strong inhibitor of the progression to flowering (Chuck et al., 2011). Weak miRNA156 over-expressing lines of switchgrass had better saccharification yield from starch than strongly expressing lines, probably because their growth was less impaired. In this work we also observed that the weak overexpressing line OxSUB1A-L5 had a better saccharification yield than the strong overexpressing line OxSUB1A-L12. It has been recently shown that Arabidopsis ERFs activate strong feedback loops through the enzyme plant cysteine oxidase 1 and 2 (PCO1, PCO2; Weits et al., 2014) and the inhibitory protein hypoxia response attenuator 1 (HRA1; Giuntoli et al., 2014) that when constitutively expressed, lead to suboptimal growth under normal and submergence stress conditions. Therefore, moderate and weak expression of SUB1A-1 is necessary to balance these mechanisms.

These data is consistent with previous research where rice UBI:SUB1A-1 plants had a higher free sugar concentration when measured in aerial tissue (Fukao, Yeung & Bailey-Serres, 2012). However, these rice transgenics do not show a constitutive starch accumulation but the development is severely delayed. The effects of SUB1A-1 on starch accumulation in rice and Arabidopsis may be due to different carbon allocation strategies among monocots and dicots (monocots use stems as storage organ), wild and cultivated plants, environmental cues and development stages (Bennett, Roberts & Wagstaff, 2012; Streb & Zeeman, 2012; Slewinski, 2012; Wang et al., 2013). For example, sex1-1 (gwd) mutants in Arabidopsis accumulate starch and have severe developmental defects since they cannot efficiently match growth and anabolism (Weise et al., 2012; Paparelli et al., 2013), whereas development of rice gwd mutants is normal and only impacts grain yield even when they accumulate up to 400% more starch than WT (Hirose et al., 2013). Overexpression of miRNA156 promoted starch accumulation in switchgrass but not in Arabidopsis, maize or tobacco (Chuck et al., 2011).

When the weak over-expressing line OxSUB1C-L6 was analyzed, it did not show a significant saccharification yield improvement in starch, cell wall or free reducing sugars (Table 1). The strong over-expressing line OxSUB1C-L10 showed decreased saccharification yield for starch and cell wall (−38% and −17%, respectively; Table 1) and lower free reducing sugars levels (−17%; Table 1). These data support the hypothesis that SUB1A-1 and SUB1C-1 control opposing biochemical mechanisms, despite belonging to the same ERF-VII gene family (Fukao et al., 2006; Fukao, Yeung & Bailey-Serres, 2012).

To visualize starch accumulation, we used iodine staining of 14-day-old plants of all transgenic lines and WT. ZT24 was selected as the testing point to increase contrast and observe if accumulation was distinct at the end of the night. The staining showed that both OxSUB1A lines leaves retained more starch in leaves. By contrast, the OxSUB1C lines retained less starch at ZT24 than WT (Fig. 2). Together these biochemical and histological data indicate that maintenance of significantly higher leaf starch is the main contributor to the improved saccharification yield of OxSUB1A plants. Differences in cell wall saccharification and free-sugar content are also distinct from WT but are less determining factors.

Diurnal and developmental starch accumulation patterns of OxSUB1A plants

Leaf starch accumulation has a diurnal pattern with a peak at the end of day and consumption during the night (Bahaji et al., 2013; Ortiz-Marchena et al., 2014). To quantify if starch content could be maintained during the diurnal oscillations as suggested by iodine staining (Fig. 2), we collected 21-day-old OxSUB1A and WT plants at the start and middle of both day and night. WT plants accumulated starch in an expected pattern for transitory starch (Ortiz-Marchena et al., 2014) with a peak at the end of day (Fig. 3). OxSUB1A lines had the same normal accumulation pattern but conserved more starch than WT at all points presented. As previously observed at the end of the day, the weakly over-expressing OxSUB1A-L5 significantly doubled starch content relative to WT whereas the strongly overexpressing OxSUB1A-L12 had only 13%–30% more (Fig. 3).

Plant development and starch accumulation are genetically coordinated, especially during the vegetative phase change marked by the floral transition (Chuck et al., 2011; Yang et al., 2013; Ortiz-Marchena et al., 2014). To investigate the developmental stages where SUB1A-1 can influence starch conservation improvement, we collected leaves at ZT16 at two WT pre-flowering points (adult vegetative, 18 and 21-day-old) and two WT post-flowering points (reproductive stage, 27 and 31-day-old). In WT plants, starch increased as plants reached bolting time and decreased and stabilized after flowering time when cauline leaves begin to contribute to photosynthetic carbon gain (Early et al., 2009). Interestingly, starch content was higher in both pre-flowering OxSUB1A-L5 and -L12 lines (278% ± 23 S.E. and 189% ± 9 S.E.). This difference decreased until all plants had the same starch content after flowering (Fig. 4A). Iodine starch staining at ZT24 of 14, 21 and 28-day-old rosette leaves matched the pattern of improved starch content (Figs. 4B–4D).

This evaluation of diurnal and developmental kinetics further supports the conclusion that starch accumulation is responsible for the improved saccharification yield of OxSUB1A plants. The data also indicate that SUB1A-1 is responsible for the starch conservation trait of the LOQS and that this phenotype is regulated in a developmental manner. Two factors involved in this developmental process are likely the flowering transcription factor CONSTANS (CO) and the florigen gene FLOWERING LOCUS T (FT); both transcripts are significantly down-regulated in OxSUB1A rice and Arabidopsis plants, leading to a late transition to reproductive stage even under an inductive flowering photoperiod (Peña-Castro et al., 2011).

Until recently, an involvement of CO/FT in starch metabolism was not evident because ft and co mutants accumulate similar levels of starch as WT when grown under continuous light; however, mutants of GIGANTEA (GI), an upstream circadian regulator of CO, are strong starch hyperaccumulators (up to 300% of WT levels) (Eimert et al., 1995). Recently, the role of photoperiod in starch accumulation during the floral transition was studied and demonstrated that CO controls starch granule structure via differential diurnal DNA-binding patterns and developmental and diurnal regulation of GRANULE BOUND STARCH SYNTHASE (GBSS; Ortiz-Marchena et al., 2014). Through these events, CO promotes accumulation of starch granules with a higher amylose:amylopectin ratio that can be readily digested proposed to enable a carbohydrate burst that create an optimum metabolic state for flowering. With these results, we hypothesize that down-regulation of CO/FT by SUB1A-1 allows OxSUB1A transgenics to conserve starch that would be otherwise employed for developing inflorescence structures.

The mechanism of starch content improvement mediated by SUB1A-1

Late flowering has been related to improved starch saccharification by mechanisms other than those directly regulated by CO. For example, in switchgrass engineered to over-express miRNA156, young nodes accumulated more starch than WT mature nodes (Chuck et al., 2011). However, miRNA156 is a repressor of vegetative-reproductive transition through a CO parallel pathway that was recently shown to be connected to T6P (Wahl et al., 2013; Yang et al., 2013), a repressor of starch catabolism through KIN10 signaling (Baena-González et al., 2007; Delatte et al., 2011). Defects in enzymatic starch mobilization also lead to late flowering, starch accumulation and size defects (Streb & Zeeman, 2012; Paparelli et al., 2013).

To obtain insight into mechanisms that are different or parallel to CO regulation of starch accumulation in OxSUB1A transgenics, we measured polyphasic fluorescence rise (OJIP kinetics). This method has been used to detect photosynthetically-improved plants with increased carbohydrate accumulation (Gururani et al., 2012). However, no significant differences were detected between the five genotypes analyzed in this study (Table S1), indicating that neither OxSUB1A nor OxSUB1C transgenics posses photosystem efficiency that differs from WT.

Although starch granule architecture and biosynthesis is not a well-understood process (Fettke et al., 2011; Streb & Zeeman, 2012), altered shape and size have been reported in mutants of starch breakdown enzymes GWD1 (also known as SEX1) and SEX4 (Zeeman et al., 2002) and starch synthase 2 and 3 (SS2, SS3; Zhang et al., 2008). We isolated and examined starch granules architecture by scanning electron microscopy and found that starch from OxSUB1A 21-day-old rosettes had the same size and characteristic ellipsoid-like shape of those of WT leaves of the same age (Fig. S2).

In our previously reported microarray studies at ZT8 of OxSUB1A and OxSUB1C seedlings (Peña-Castro et al., 2011), we did not detect a significant change in accumulation of mRNAs related to starch biosynthesis or catabolism. However, since these genes have a circadian oscillation, mostly peaking after midday (Smith et al., 2004), we searched our datasets for statistical outliers associated with this biological process that were up- or down-regulated and evaluated them in RNA from seedlings samples collected at ZT16 (end of day). We tested transcripts encoding starch degrading enzymes GWD1 and SEX4 between WT and OxSUB1A or OxSUB1C but found not significant difference in expression (Table S2).

Recently, it was demonstrated that in parallel to CO, transcription factors of the SQUAMOSA PROMOTER BINDING PROTEIN gene family (SPL3-5) connect carbohydrate metabolism to the vegetative phase change and also lead to late flowering phenotypes (Wahl et al., 2013). In switchgrass, down-regulation of SPL3-5 by miRNA156 promotes late flowering and improvement of saccharification yield by both amylolytic and cellulolytic treatments, without modulation of CO/FT ortholog transcripts (Chuck et al., 2011), supporting the idea that CO and SPL/miRNA156 are parallel pathways in leaves that impact flowering time (Wahl et al., 2013). To test if delayed vegetative phase change in OxSUB1A is related to SPL3-5, we measured transcripts of SPL3-5 and downstream genes SOC1 and FUL in 7-d-old seedlings at ZT16. These transcripts were also statistical outliers down-regulated in our microarrays. The expression of all these transcripts was similar to that of WT plants suggesting independent activity from SPL/miRNA156.

In addition to CO and SPL/miRNA156, post-translational regulation of starch synthesis enzymes by reactive oxygen species (Lepisto et al., 2013) and T6P signaling through the stress integrating kinase SnRK1 regulate starch levels (Baena-González et al., 2007; Mattos Martins et al., 2013). T6P is of particular interest for further research since microarray studies of submergence stress response in different plants indicate there is a dynamic change in the transcripts of trehalose-6-phosphate synthase and trehalose phosphate phosphatase (Jung et al., 2010; Lee et al., 2011; Narsai & Whelan, 2013; Van Veen et al., 2013; Tamang et al., 2014).

Hardness of OxSUB1A leaves

In earlier transcriptome analysis (Peña-Castro et al., 2011) we found that SUB1A-1 promoted in Arabidopsis the up-regulation of 17 genes associated with modification of the cell wall and/or biotic stress response, including endotransglycosylase (XTR3, XTR6), expansin (AtEXP2) and glucan-1,3,-beta-glucosidase (BGL2; Table S3). This latter gene was the most up-regulated transcript relative to WT in 7-d-old seedlings. In addition to their biological importance, cell wall associated proteins are also of technological interest for the development of bioethanol fuel. They consist of enzymes and proteins that can change the mechanical properties of cell wall polymers (cellulose, hemicellulose, lignin and callose) improving cell wall digestibility and saccharification yields (Arantes & Saddler, 2010).

To evaluate if the expression of cell wall associated genes in our transgenics was correlated with a phenotype with modified mechanical properties, we employed a texture analyzer to measure fracture properties of leaves in 23-day-old rosette leaves. Both OxSUB1A-L5 and OxSUB1A-L12 leaves offered significantly less resistance to fracture than WT (67% ± 18 S.D. and 70% ± 11% S.D., respectively). OxSUB1C lines were not statistically different from WT (Fig. 5A). To confirm expression of BGL2 and AtEXP2 in OxSUB1A and OxSUB1C, RNA from 7-day-old seedlings at ZT8 were tested by qRT-PCR. OxSUB1A-L5 and OxSUB1A-L12 expressed more BGL2 (Fig. 5B) and AtEXP2 transcripts (Fig. 5C). WT and OxSUB1C accumulated similar BGL2 mRNA levels, whereas OxSUB1C-L6 had 3-fold more AtEXP2 than WT; however this was not replicated in OxSUB1C-L10 (Fig. 5C).

BGL2 belongs to a multigene family of hydrolytic enzymes involved in fungal pathogen defense and developmental processes related to callose, a transitory β-1,3-glucan relevant for cell wall maturation (Doxey et al., 2007; Park et al., 2014). The rice response during submergence stress includes the expression of genes associated to pathogen stress, and the presence of SUB1A-1 improves this induction (Jung et al., 2010). In Arabidopsis, SUB1A-1 also promoted the constitutive expression of these genes (Peña-Castro et al., 2011). The biotic stress component of the submergence stress response primes plants to resist the pathogens that may increase their access to plant tissue during submergence (Hsu et al., 2013).

When rice plants are submerged, plants encoding SUB1A-1 induce EXPANSIN transcripts early in the stress and restrict them in later stages to conserve energy (Fukao et al., 2006). Expansins are cell wall morphogenic proteins that allow non-enzymatic loosening of cellulose and make it more accessible for enzymes during cell expansion (Arantes & Saddler, 2010). Our expression analysis indicates that EXPANSIN induction is conserved in OxSUB1A transgenics in non-stress growth conditions (Fig. 5C). AtEXP2 is a GA-responsive EXPANSIN normally active during seed germination (Yan et al., 2014).

Together, these data provide evidence that the expression of cell wall and biotic response associated genes mobilized by SUB1A-1 is correlated to a phenotype with decreased mechanical strength and improved cellulose digestibility.

Growth penalty in OxSUB1A plants

Strong constitutive starch conservation in plants is frequently accompanied with a growth penalty derived from their inability to efficiently use this energy reserve (Chuck et al., 2011; Weise et al., 2012; Paparelli et al., 2013; Pfister et al., 2014). We did not observe in our young vegetative OxSUB1A transgenics such penalty (Fig. S3A); however, we detected reduced size (Figs. 4C and 4D) and dry weight stagnation (Fig. S3B) in both pre-flowering (40% loss) and reproductive plants (65% loss). Weak cell walls also risk the plant to suffer pathogen attacks or suboptimal biomechanics (Nigorikawa et al., 2012; Petersen et al., 2012).

These negative features would compromise the development of industrial applications based on plants with improved saccharification traits. Peña-Castro et al. (2011) reported that SUB1A-1 heterozygous individuals have a larger rosette size than Col-0, normal flowering time and fertility; these characteristics are lost when SUB1A-1 is homozygous indicating that control of gene expression dosage is important to achieve optimal results. A proposed solution to these drawbacks is the use of inducible promoters (Weise et al., 2012) or tissue-specific promoters (Petersen et al., 2012) that allow fine-tuning the expression of saccharification traits.