• Open Access

Enhanced polyhydroxybutyrate production in transgenic sugarcane

Authors


(Tel 1 940 369 5072; fax 1 940 369 8656; email sbrumbley@unt.edu)

Summary

Polyhydroxybutyrate (PHB) is a bacterial polyester that has properties similar to some petrochemically produced plastics. Plant-based production has the potential to make this biorenewable plastic highly competitive with petrochemical-based plastics. We previously reported that transgenic sugarcane produced PHB at levels as high as 1.8% leaf dry weight without penalty to biomass accumulation, suggesting scope for improving PHB production in this species. In this study, we used different plant and viral promoters, in combination with multigene or single-gene constructs to increase PHB levels. Promoters tested included the maize and rice polyubiquitin promoters, the maize chlorophyll A/B-binding protein promoter and a Cavendish banana streak badnavirus promoter. At the seedling stage, the highest levels of polymer were produced in sugarcane plants when the Cavendish banana streak badnavirus promoter was used. However, in all cases, this promoter underwent silencing as the plants matured. The rice Ubi promoter enabled the production of PHB at levels similar to the maize Ubi promoter. The maize chlorophyll A/B-binding protein promoter enabled the production of PHB to levels as high as 4.8% of the leaf dry weight, which is approximately 2.5 times higher than previously reported levels in sugarcane. This is the first time that this promoter has been tested in sugarcane. The highest PHB-producing lines showed phenotypic differences to the wild-type parent, including reduced biomass and slight chlorosis.

Introduction

Sugarcane (Saccharum sp. interspecific commercial hybrids) is a high biomass crop well suited to play a critical role in the developing bio-based economy (Brumbley et al., 2007, 2008) exemplified by the large-scale transformation of the Brazilian sugarcane industry into a sugars, biofuels, bioproducts and bioenergy industry (Velho and Velho, 2006). Sugarcane is readily amenable to crop improvement through genetic engineering and efficient protocols exist for tissue culture and the delivery of transgenes (Arencibia et al., 1995, 1998; Bower et al., 1996; Brumbley et al., 2008). Several novel bioproducts have already been successfully engineered into sugarcane, including p-hydroxybenzoate, a major component of liquid crystalline polymers (McQualter et al., 2005), sorbitol (Chong et al., 2007), isomaltulose (Wu and Birch, 2007), polyhydroxybutyrate (PHB), a biodegradable plastic from bacteria (Petrasovits et al., 2007) and various polyhydroxyalkanoate (PHA) copolymers (Anderson et al., 2011; Tilbrook et al., 2011), demonstrating the potential of sugarcane as a biofactory for the production of industrial chemicals.

Polyhydroxyalkanoates are a family of biodegradable plastics produced naturally in bacteria as a carbon storage compound (Poirier, 2002). Bioplastics are already being marketed as alternatives to petroleum-based polymers. The family of PHA bioplastics is of particular interest because of the diverse range of physical properties found in these naturally produced or engineered polyesters (Steinbüchel and Valentin, 1995). To date, most of the emphasis on PHA production has been by bacterial fermentation (Dias et al., 2006; Lee et al., 1999). There is increasing interest in producing PHAs in plants because of the potential to reduce production costs, and pathways for the production of PHAs have been introduced into a number of crop species (Mooney, 2009; Snell and Peoples, 2009). Production of PHAs in high biomass crops is thought to be the most likely mechanism to achieve this result, and the short-chain PHA, PHB, has successfully been produced in crops such as maize (Poirier and Gruys, 2001), sugarcane (Petrasovits et al., 2007), switchgrass (Somleva et al., 2008) and poplar (Dalton et al., 2011) but not yet at production levels considered commercially feasible. Therefore, additional research is required to understand the constraints to higher PHB production in plants. Some constraints to PHB production may include the level of recombinant protein expression; expression of the transgenes in the most appropriate cell types at the appropriate stages of plant growth and development and problems associated with substrate availability.

One strategy to increase PHB production is to increase the amount of recombinant protein in the plant. A detailed analysis of the spatio-temporal accumulation profile of PHB in transgenic sugarcane lines expressing the PHB biosynthetic pathway found good correlations between polymer content and transcript abundance (Purnell et al., 2007). PHB accumulated in leaf tips at levels of up to 1.88% DW without an apparent agronomic penalty (Petrasovits et al., 2007). These findings suggested that there was potential for significant improvement in PHB yield by increasing expression of the three genes of the PHB biosynthetic pathway. In switchgrass (Somleva et al., 2008), some plants accumulating more than 2% DW PHB in senescing leaves were observed to have slight chlorosis and growth reduction. A similar phenomenon was noted in transgenic poplar trees producing PHB levels higher than 1% (Dalton et al., 2011). It is expected that a similar threshold will be reached in transgenic sugarcane where phenotypic effects will become apparent.

Several strategies have been developed to increase the levels of recombinant protein expression in transgenic plants. These include the use of tissue-specific or constitutive promoters, optimization of the initiation codon position and context, control of gene expression by 5′ and 3′ untranslated regions and optimization of transgene coding regions (Sharma and Sharma, 2009). In the present study, we focussed on the effect of stronger promoters to drive higher levels of transgene expression. Besides the maize ubiquitin promoter (ubi1), other promoters with demonstrated efficacy in sugarcane include the Cavendish banana streak badnavirus promoter (Cv) (Schenk et al., 2001) and the rice ubiquitin 2 (rubi2) promoter (Wang and Oard, 2003; Wang et al., 2000). Of these, the Cv promoter has been shown to be the strongest in sugarcane with a performance comparable to that of the Cauliflower Mosaic Virus 35S promoter in dicots (Schenk et al., 2001). In other C4 crops engineered for the production of PHB (Poirier, 2002; Somleva et al., 2008), the maize chlorophyll A/B-binding protein (cab-m5) promoter has been used to control transgene expression. Up to 3.72% and 5.73% DW PHB were observed in portions of leaf tissue in switchgrass (Somleva et al., 2008) and maize (Poirier and Gruys, 2001), respectively. To our knowledge, the cab-m5 promoter has not been tested in sugarcane prior to this study. The production of PHB in sugarcane brings an added level of complexity to sugarcane engineering as it requires the concerted expression of three biosynthetic genes phaA, phaB and phaC (Brumbley et al., 2002; Petrasovits et al., 2007; Purnell et al., 2007). The successful production of PHB in sugarcane was the first reported engineering of a multigene pathway into this crop. It was achieved (Petrasovits et al., 2007) by cotransformation of sugarcane with three separate vectors containing each of the three PHB biosynthesis genes, respectively, under the control of the maize polyubiquitin (ubi1) promoter (Christensen and Quail, 1996). Of the total number of transgenic sugarcane lines generated, only 20% produced PHB at levels quantifiable by high-performance liquid chromatography (HPLC) and only 4% at levels over 1% dry weight in mature leaf tips, indicating that ‘high-producing’ lines occur at low frequencies (Petrasovits et al., 2007). This meant that large numbers of plants needed to be generated in order to obtain plants with PHB levels higher than previously achieved and that these populations needed to be screened for high-producing lines at the pre-glasshouse stage.

In both maize and switchgrass studies, the PHB pathway was introduced on a single, multigene transformation vector. Multigene transformation vectors have not previously been tested in sugarcane. It is logical to assume that a multigene pathway contained on a single construct would have a higher probability of all three genes integrating into an optimal locus within the plant genome than if the pathway was introduced as individual genes in single-gene vectors. In fact, Bohmert et al. (2000) demonstrated fourfold higher levels of PHB in Arabidopsis when the three PHB biosynthetic pathway genes were introduced on a single construct. Using a multigene construct may reduce the need to screen large numbers of sugarcane plants to identify high PHB producers.

In this report, we describe the use of alternative promoters to drive high levels of transgene expression and the use of multigene transformation vectors in efforts to increase PHB production in sugarcane to levels comparable or higher to those previously achieved in switchgrass. A high-throughput screening technique is described for identifying PHB producers at the tissue culture stage just prior to transfer to the glasshouse, enabling significant savings in time, space and labour. Additionally, we explore the effects of high PHB production on biomass production in transgenic sugarcane under glasshouse conditions.

Results

A pre-glasshouse screening system

A selection system was developed based on Nile Blue A staining of tissue-cultured plantlets. Previously identified PHB-positive sugarcane plants (Petrasovits et al., 2007) were used as explant material to produce tissue culture plantlets. Leaf sections were collected from these tissue culture plantlets as small as 2 cm in height and stained with Nile Blue A. Polymer granules were readily detectable by compound fluorescent microscopy at 200× magnification (Figure 1).

Figure 1.

 Nile Blue staining was found to be a sensitive and reliable tool for detecting polyhydroxybutyrate (PHB)-positive plants in tissue culture. Results are tabulated for tissue culture plantlets of Q117 wild type (WT) and lines B2-12, B3-5 and TA4 that produced 0.03%, 0.9% and >2% DW PHB, respectively, in leaves at maturity. Typical staining results of ca. 0.2-mm leaf sections are shown in the panel above.

To evaluate the system, a population of 1484 newly generated Cv-promoter plantlets was examined and approximately 20% of these plantlets were found to contain discernable fluorescent granules. A subset of 300 plantlets (both positive and negative for PHB production) was selected for further analysis. These plants were transferred to the glasshouse and grown for 3 months before analysis of PHB content by HPLC of the oldest green leaf from the base of the plant (Table 1).

Table 1.   Comparison of Nile Blue staining and HPLC results in 300 new lines
CategoryNumber of plantsPositives identified by Nile Blue AHPLCRange (% DW PHB)
  1. HPLC, high-performance liquid chromatography; PHB, polyhydroxybutyrate.

  2. Positives by Nile Blue A refer to the number of plants where PHB granules were present in leaf section at the pre-glasshouse stage; positives by HPLC represent the number of plants within each subcategory that contained polymer levels quantifiable at 3 months in the glasshouse; range represents PHB contents determined by HPLC in tissue from the oldest green leaf blade in these plants. The following criteria were used to classify PHB producers: negative, <2 granules; weak positive, <10 granules; positive, >20 granules; and strong positive, >100 granules per field of view.

Negatives134010.031
Weak positives474730.015–0.022
Positives7373350.018–0.098
Strong positives4646450.05–3.52

High-performance liquid chromatography analysis revealed that only one of 134 plants identified as negative for PHB production using Nile Blue screening contained PHB. This false negative, however, contained only a trace amount of PHB (0.031% DW PHB; Table 1). Of the subset of tissue culture plants identified as positive, 45 of 46 (97.8%) of the strongly positive, 35 of 73 (47.9%) of the positive and 3 of 47(6%) of the weakly positive plants contained polymer quantifiable by HPLC 3 months after transfer to the glasshouse. Owing to the high success rate of identifying the high producers using this technique, all subsequent putative transformants were pre-screened in tissue culture by Nile Blue A staining and plants that were negative or weakly positive were discarded.

Use of strong promoters in transgene expression cassettes

Transgenic plants were mass-produced by biolistic transformation of sugarcane callus fortnightly over a 4-month period. The constructs employed, the total number of regenerated and screened plantlets and the number of PHB producers identified by Nile Blue A staining are summarized in Table 2. For each set of constructs, the number of positives was similar, ranging from 21.5% (pCvA, B, C) to 27.9% (pMBXS159). Strong positives occurred at frequencies of 4.7% (pCvA, B, C) to 9.4% (pUbiA, B, C).

Table 2.   Results of biolistic transformation of sugarcane embryogenic callus
ConstructBiolistic transformation eventsTotal number of transgenic plants+++
  1. +, Average number of fluorescing granules seen per field of view was >20. Three fields were viewed per line tested; ++, Average number of fluorescing granules seen per field of view was >100. Three fields were viewed per line tested.

  2. The number of transformations reflects individual bombardments carried out in six replicates each time.

  3. UbiA, B and C: the maize polyubiquitin promoter driving each of the polyhydroxybutyrate (PHB) biosynthetic pathway genes, phaA, phaB and phaC, each on an individual sugarcane transformation vector. CvA, B and C are the same as UbiABC, and only the Ubi promoter was replaced with the Cavendish banana streak virus (Cv) promoter. pMBXS155 is a megaconstruct with all three PHB biosynthetic genes on a single construct each with the maize cab-m5 promoter driving expression. pMBXS159 is also a megaconstruct but with the rice polyubiquitin promoter (rUbi) driving expression. pMBXS165, pMBXS166 and pMBXS167 are individual constructs with phbA, phbB and phaC, respectively, each driven by the same rUbi promoter as pMBX159 (refer to Materials and Methods).

pUbiA, B and C81029241 (23.4%)97 (9.4%)
pCvA, B and C72807604 (21.5%)132 (4.7%)
pMBXS15571886520 (27.6%)130 (6.9%)
pMBXS15971760492 (27.9%)109 (6.2%)
pMBXS165,166 and 16771409382 (27.1%)95 (6.7%)
Total3688912239 (25.2%)563 (6.3%)

A total of 563 plantlets identified as strongly positive by Nile Blue A staining were transferred into the glasshouse and screened by HPLC after 3 months. The oldest green leaf from the base of the plant was collected. The majority of these plants contained PHB at levels below 0.1% DW. Over 60 lines were found to contain polymer levels similar to or higher (Figure 2a) than the best transgenic sugarcane line TA4 previously reported (Petrasovits et al., 2007). At this stage, the highest levels of PHB were observed in plants generated with the Cv promoter with an average of 1.71 ± 0.23% DW. The top two individual lines produced as much as 3.50% and 2.95% of leaf DW, respectively. Plants generated with the cab-m5 promoter had an almost twofold higher average PHB content (0.57 ± 0.07% DW) than plants containing the three PHB biosynthesis genes under the control of the maize ubi1 promoter (0.33 ± 0.03% DW). Plants transformed with the rice rubi2 promoter-driven phaA, B and C cassettes (pMBXS159 or pMBXS165, pMBXS166 and pMBXS167, respectively) did not produce PHB at levels higher than those observed for the maize ubi1 promoter. In these lines, averages reached 0.34 ± 0.05% DW for the multigene construct pMBXS159 and 0.14 ± 0.02% DW in lines simultaneously transformed with the three individual expression cassettes (Figure 2b).

Figure 2.

 (a) Polyhydroxybutyrate (PHB) content was determined by high-performance liquid chromatography in leaf blades of plants three (light shaded bars) and 6 months (dark shaded bars) after transfer to soil. Lines were generated with individual expression cassettes driven by the maize ubiquitin (UbiABC), the rice ubiquitin (pMBXS165, 166 and 167, represented here as R-Ubi S) or the banana streak virus promoter (CvABC). In lines transformed with pMBXS159 (R-Ubi M) and pMBXS155 (cab-m5), the expression cassettes were located on a single large construct with each gene under the control of the rubi2 and cab-m5 promoters, respectively. (b) Box plot representation of the PHB content of transgenic sugarcane lines grouped by the expression construct(s) used. In each plot, the horizontal lines represent the 10th, 25th, 50th (median), 75th and 90th percentiles. Dots denote outliers.

Six months after transfer to the glasshouse, the oldest green leaf from each plant was sampled and reanalysed by HPLC (Figure 2a). An increase in PHB production was observed in all lines except those containing the pha genes under the control of the Cv promoter that contained an average of 0.03 ± 0.01% DW of PHB. The cab-m5 lines performed best on average at 1.41 ± 0.16% DW followed by the maize ubi1 lines at 0.79 ± 0.12% DW. The rubi2 lines reached PHB levels of 0.62 ± 0.07% and 0.45 ± 0.14% DW for multigene and single-gene constructs, respectively (Figure 2b).

To investigate the apparent instability of PHB production in the Cv lines further, explants from six lines were reintroduced into tissue culture. Two months after transfer to the glasshouse, polymer production was re-established at detectable levels in over 50% of regenerants as shown by HPLC (Table 3). Polymer levels in the regenerated plantlets did not reach the levels observed in the primary transformants and did not appear to maintain rank order. Polymer levels subsequently declined and were undetectable by HPLC in four of the six Cv lines tested and were negligible in the other two lines after 6 months of greenhouse growth (Figure 3). Neither PhaA nor PhaB was detected in protein extracts from leaves collected at the 6-month time point (data not shown).

Table 3.   Recovery of the PHB production in Cv promoter-driven plants obtained after passage through tissue culture and grown in the glasshouse for 3 months
Parent (% DW PHB)Sugarcane lines recoveredPercentage of positive plantsMax % DW PHB
  1. PHB, polyhydroxybutyrate.

  2. The number in brackets indicates the highest amount of polymer observed for the line when originally screened as first-generation transgenic material sampled 3 months after transferring from tissue culture to glasshouse. Polymer content was determined for the oldest green leaf.

2–55 (2.05)6500.20
2–50 (1.12)13540.14
2–44 (3.52)16560.33
2–43 (1.59)10600.77
2–42 (1.07)26650.69
2–39 (2.95)15600.65
Figure 3.

 Polyhydroxybutyrate (PHB) content in previously silenced Cv lines after re-initiation and regeneration from tissue culture. PHB production was initially restored in 3-month-old plants but at 6 months was no longer detected in leaves in four of the six lines.

After the second sampling, low PHB-producing plants were culled and the best lines for each promoter replanted in replicates. At this time, an analysis of PHB accumulation in leaves of different ages was undertaken. We collected four different leaf types based on their position and/or morphology: tops included leaf material above the leaf with the first visible dewlap (FVD), young leaves that had green tips and a dry-to-fresh weight ratio of 0.2–0.3 (data not shown), green leaves that were fully mature with senescing tips and old leaves that were intact but dry. Additionally, dry leaf sheaths (LSs) from the base of stalks were collected. Samples from at least two lines for ubi1, rubi2 and cab-m5 promoter constructs were analysed. The results are shown in Figure 4. With all three promoters, the PHB levels in the leaf tissue were highest in mature leaves and lowest in the youngest, which is consistent with constitutive expression of these genes throughout the life of the plant (Figure 4). The pattern of PHB accumulation was different in the LS, the part of the leaf at the base that wraps around the stalk. For the ubi1 promoter lines, PHB levels were highest in this tissue; for rubi2, PHB levels were about the same as in young leaves but less than in old or green leaves; and for cab-m5, PHB levels in the LS were the lowest of the five tissue types sampled.

Figure 4.

 Polymer distribution in the best two plants transformed with different promoter constructs. Leaves were grouped on the basis of their physiological appearance and polyhydroxybutyrate (PHB) content determined for each leaf within the group (n > 6). The error bars represent the standard error. LS, leaf sheath.

Polyhydroxybutyrate accumulation in mesophyll cells

Sections were taken from mature leaves and fixed for Nile Blue A staining and transmission electron microscopy (TEM). Samples taken from the same leaves were also analysed for PHB content. In transverse leaf sections, fluorescent granular inclusions were observed in transgenic plants but not in the wild type (Figure 5a–c). In plant material that contained polymer levels of less than ca. 1% dry weight, these PHB granules were detectable in all cell types with the exception of mesophyll cells (Figure 5b). In lines that had higher polymer levels, granular inclusions were visible in all cell types (Figure 5c). The mesophyll cells contained fewer polymer granules than the other leaf cells. Electron microscopic investigation confirmed the presence of electron-lucent granules in mesophyll plastids in high-producing lines (Figure 5e, f). However, by far the highest concentration of polymer was found to accumulate in chloroplasts of the bundle sheath cells (Figure 5g).

Figure 5.

 Leaf transverse sections stained with Nile Blue A (Panel a—Wild type, Panel b—approximately 1% DW polyhydroxybutyrate (PHB), Panel c—greater than 2% DW PHB) viewed by fluorescence microscopy and by transmission electron microscopy (Panels d–f). At the highest levels produced (Panel g), PHB is seen in plastids of all cell types with the highest concentrations in the bundle sheath cells. Note the presence of polymer granules in mesophyll plastids in panels c, f and g. Size bars represent 25 μm (Panel a–c, g) and 1 μm (Panel d–f). VB, vascular bundle; BS, bundle sheath; M, mesophyll; L, lipid; St, starch.

Glasshouse trial

Examples of the best PHB-producing sugarcane lines for each promoter were grown to maturity in a replicated, randomized block design, in a PC2 glasshouse. At the end of the trial, the plants were harvested and height and weight data were collected. The results are shown in Figure 6. Analysis of variance (ANOVA) was used to compare each transgenic line to the wild-type control parent for differences in height and weight. No significant difference in height or weight was observed between the wild-type sugarcane line and the best maize ubi1 (TA4) or rice ubi2 promoter (4A1, 4F1) lines. Of the three cab-m5 promoter lines included in the trial, 8C8 and 7B4 were significantly shorter than the wild-type parent (P < 0.05), while only 7B4 weighed significantly less (P < 0.05) than the wild-type parent sugarcane line.

Figure 6.

 Total fresh weight and height of the aerial biomass of replicated sugarcane plants in a randomized 9-month glasshouse trial. The asterisk indicates significant differences (P < 0.05) between wild-type and transgenic lines. WT, wild type; GFP, green fluorescent protein expressing; polyhydroxybutyrate (PHB), negative control line.

In a subsequent trial, plants were replicated and grouped so as to reduce competition, as shading appeared to affect the performance of high PHB producers. The highest polymer levels were found in cab-m5 promoter lines, whereas the maize and rice ubi promoter lines performed similarly (Table 4). In the highest PHB-producing lines, chlorosis of the leaves was observed compared with the wild type (Figure 7). Analysis of total leaf protein showed similar levels of PhaA in the maize and rice ubi lines and higher levels in the cab-m5 lines. The levels of PhaB appeared to reflect polymer content more closely than those of PhaA (Figure 8). PhaC was not assayed for as we did not have a sufficiently specific antiserum.

Table 4.   Mean, range and standard error of polyhydroxybutyrate content as a percentage of whole-leaf dry weight in best-performing ubi1, rubi2, and cab-m5 promoter-driven transgenic sugarcane lines
Line IDPromoterSize (n)Mean (+/− SE)Range
TA4ubi130.45 (0.02)0.42–0.50
1C3ubi130.59 (0.06)0.52–0.70
4A1rubi240.29 (0.07)0.11–0.44
4F1rubi231.21 (0.24)0.75–1.56
5C5rubi230.56 (0.03)0.51–0.61
7B4Cab-m552.98 (0.39)1.87–4.08
7C3Cab-m573.11 (0.31)2.45–4.86
8C8Cab-m581.30 (0.11)0.72–1.66
Figure 7.

 Chlorotic appearance of leaves from lines 7C3 (left) and 7B4 (right) compared with a wild-type leaf (back).

Figure 8.

 Western blot of total protein extract probed with antisera to PhaA, PhaB and the Rubisco large subunit (RbcL) to show equal loading of lanes. Polyhydroxybutyrate (PHB) content by leaf dry weight of a subsample of the leaf material is given by the number underneath the blot images.

Discussion

There is insufficient leaf material available for analysis of PHB levels in sugarcane leaves by HPLC at the tissue culture stage necessitating the transfer of plantlets to PC2 glasshouse facilities for growth. However, PHB granules can be visualized by histological staining requiring small sample sizes. The implementation of a pre-glasshouse screening system allowed a large number of putative transformants to be screened quickly and efficiently at an early stage of the transformation procedure with only minimal amounts of sample (∼1-cm leaf blade) providing substantial savings in time and glasshouse space. The pre-glasshouse screening confirmed our previous observations that >80% of the putative sugarcane transformants produce either no or very low levels of PHB. Nile Blue A staining of leaf segments visualized by fluorescence microscopy proved to be a highly sensitive and reliable method for screening plantlets as small as 2 cm in height in a semiquantitative manner (Figure 1, Table 1). However, we were unable to detect polymer granules by Nile Blue staining in callus cultures of the lines described in this study. A possible explanation for this is that the proplastids present in callus cells (Shang et al., 2009) are not readily capable of PHB production.

Pre-glasshouse screening by the Nile Blue method was validated in a subpopulation of the same 300 plants 3 months after transfer to the glasshouse: 98% of the putative strong positives, 48% of the positives and 6% of the weak positives contained PHB levels quantifiable by HPLC (Table 1). A developmentally regulated decrease in transgene expression may have contributed to the discrepancy between Nile Blue and HPLC results, particularly in the less strongly PHB-positive categories (Table 1, further discussed below). Alternatively, Nile Blue A and its derivative Nile Red are lysochrome dyes that stain lipids nonspecifically (Greenspan and Fowler, 1985; Ostle and Holt, 1982), and therefore, the presence of intracellular and intercellular lipids has the potential to interfere with the detection of PHB granules, particularly in weakly positive samples (see Figure 1). Indeed, Peters et al. (2007) observed false positives when putative PHB-positive bacterial colonies were grown in the presence of Nile Red. To date, we have found only four examples of false negatives out of 8891 plants tested. In these cases, HPLC analysis showed that these plants produced PHB, although Nile Blue A staining failed to detect the presence of the polymer (data not shown). In all cases, PHB levels were low (<0.05% DW) so these plants would not have been carried forward beyond the initial HPLC screening 3 months after transfer to soil.

Four different promoters were compared in this study, namely the ubi1, rubi2, Cv and cab-m5 promoters. This is the first report where the cab-m5 promoter has been tested in sugarcane. The percentages of PHB-positive plants out of the total number of plants produced, as determined by Nile Blue A staining, were similar for all promoters tested: the lowest values were obtained for the Cv constructs (21.5% PHB positive) and the highest for the rubi2 promoter multigene construct pMBXS159 (27.9% PHB positive; Table 2). These values are in good agreement with our previous finding that 20% of putative transformants produced detectable quantities of PHB. Similarly, the percentage of strong positives (∼6% of plants screened) was also consistent with our previous observations in which 4% of transgenic sugarcane plants contained polymer levels in excess of 1% DW in the tips of the oldest green leaves 9 months after transfer to soil (Petrasovits et al., 2007). The threshold distinguishing strong positives (>100 granules per field of view) for transfer to the glasshouse and plants to be discarded appeared to be consistent with a minimal plant polymer content of between 0.05 and 0.1% DW at 3 months in the glasshouse (Tables 1 and 2).

Three months after transfer of tissue culture–derived plants to the glasshouse, the plants with the highest polymer content were lines generated with the Cv promoter with averages approximately three times higher than those observed for the Ubi lines (Figure 2b). This finding agreed with published information on the Cv promoter (Schenk et al., 2001) where green fluorescent protein (GFP) expression in transgenic sugarcane was threefold higher in sugarcane plants transformed with the gfp gene driven by the Cv promoter compared to the same gene under the control of the maize ubi1 promoter. However, at the 6-month stage, the PHB content of all Cv lines had drastically decreased (Figure 2b). This apparent developmental silencing was unexpected as previous work indicated stable expression from the Cv promoter in sugarcane (Schenk et al., 2001), although the authors described a reduction in gfp expression concomitant with the transfer of plantlets from tissue culture to the glasshouse, suggesting a developmentally regulated decrease in the promoter activity. Of the four promoters used in this study, the Cv promoter was the only promoter of viral origin. Additionally, the constructs did not contain an intron downstream of the promoter, while the ubi1, rubi2 and cab-m5 promoters did. Wilmink et al.(1995) showed that the presence of an intron downstream of the promoter increased gus expression in most monocots. Developmentally silenced gene expression has previously been reported in sugarcane when transgenes are introduced under the control of native and heterologous promoters (Mudge et al., 2009). The activity of some of these promoters was restored through tissue culture with the application of demethylating agents such as azacytidine (Hansom et al., 1999). We were able to restore some Cv promoter activity by passing lines back through tissue culture without the addition of demethylating agents as evident by restoration of the PHB phenotype that requires the presence of the three biosynthetic enzymes. However, regenerated plants did not maintain the same rank order or achieve the same polymer levels as the original lines. Furthermore, we did not detect the presence of PhaA or PhaB in these plants at 6 months. This may be attributable to temporal differences in the onset of silencing, and indeed, not all regenerated lines lost PHB production after 6 months (Table 3 and Figure 3). We concluded that while the Cv promoter is not a good candidate for driving stable transgene expression in mature sugarcane plants, our results suggest that it may be useful for expressing selectable markers during the tissue culture phase, particularly where the expression of the marker genes is not desired in mature plants.

The rubi2 transformants, whether generated from simultaneous bombardment with individual expression cassettes or with the single multigene construct, did not perform better than the maize ubi1 transformants. Interestingly, when the PHB biosynthetic genes were introduced on the single multigene construct, the resulting plantlets initially produced on average 2.5 times more polymer than when individual expression cassettes were cobombarded into callus (averages 0.34 ± 0.05 and 0.14 ± 0.02, respectively; cf. Figure 2). This was expected because the use of a multigene construct theoretically facilitates the integration of all transgenes into the same locus, thus eliminating locus-dependent variation in expression. This apparent advantage was not as obvious in 6-month-old plants with average polymer contents at 0.45 ± 0.14 for individual and 0.62 ± 0.07 for the multigene (rubi2 promoter) constructs. Moreover, the best-performing line of these, 4F1 (see Table 4), was obtained from cobombardment with individual expression cassettes. The reason for this observation is unclear but may be due to differences in the number and orientation of the inserts and other effects that are independent of the choice of construct. Our data suggest that there is no clear advantage of using the rubi2 promoter over using the ubi1 promoter for sugarcane transformation when strong expression of transgenes in mature plant tissue is required.

The cab-m5 promoter was the best-performing promoter, within the set of promoters that we analysed for leaf-based, plastid-targeted production of PHB in sugarcane with observed levels of polymer production reaching 2.74% DW in leaf tissue samples of primary transgenic plants (Figure 2a) and 4.86% DW (average for this line was 3.11 ± 0.31% DW, Table 4) in an individual leaf blade of line 7C3 after plants were replicated. This is more than twice the polymer level reported previously for transgenic sugarcane lines under the control of the ubi1 promoter at a comparable age (Purnell et al., 2007). Similar levels were achieved in switchgrass (Somleva et al., 2008) using the cab-m5 promoter to drive the expression of the PHB biosynthesis genes. While this is a significant improvement in PHB production, further work is needed to reach the >7.5% DW production required for the commercialization of PHB production in sugarcane (Bohlmann, 2006; Somleva et al., 2008).

Polyhydroxybutyrate content increased with time in mature green leaves in nearly all the lines generated in this study with the exception of lines containing expression cassettes controlled by the Cv promoter (Figures 2 and 4). A vertical gradient of PHB distribution existed in all lines with polymer concentrations being highest in the mature leaves and lowest in the youngest leaves (Figure 4). This effect was most pronounced in the cab-m5 lines and is consistent with our expectations as PHB is not metabolized by sugarcane (Purnell et al., 2007). There appeared to be promoter-specific differences in PHB accumulation in LSs: in the ubi1 and rubi2 lines, polymer abundance in this tissue was comparable to that of mature leaves, whereas in the cab-m5 lines, LSs contained very little polymer (Figure 4). In the youngest, most rapidly growing tissues, polymer content was similar regardless of which promoter was employed. This suggested that polymer production in these tissues is metabolically limited. This observation is being investigated further.

Interestingly, PHB accumulation in the different cell types in sugarcane appeared to be hierarchical in that, regardless of line or promoter constructs, leaf tissue producing polymer at up to 1% dry weight contained granular inclusions predominantly in bundle sheath and epidermal cells with fewer inclusions observable in bulliform cells and none in mesophyll. In the leaves with higher PHB content, polymer inclusions were found in all cell types but noticeably fewer granules were observed in mesophyll cells (Figure 5) as previously reported (Petrasovits et al., 2007; Somleva et al., 2008). While this demonstrates that sugarcane mesophyll plastids are capable of polymer production, the reasons for the relative inefficiency when compared with other cell types are unclear. It is possible that endogenous processes such as fatty acid biosynthesis limit the availability of substrates for PHB production.

Polyhydroxybutyrate production in sugarcane does appear to affect biomass yield to some extent in lines producing over 2% total leaf DW PHB (Table 4). Under replicated conditions in the glasshouse, the best PHB lines were on average shorter and weighed less than lines producing less than 2% or negative controls but only one of the lines significantly so (Figure 6). Under these conditions, it is difficult to determine to what extent biomass production was directly affected by PHB production as other factors including shading can have a dramatic effect on biomass yield in sugarcane (Pinto et al., 2005). A true understanding of biomass reduction will need to be determined in field trials where the high-producing sugarcane lines can be grown in blocks without the risk of shading effects inherent in the glasshouse trials.

The highest producing lines in the trial exhibited chlorosis of the leaves (Figure 7).This is consistent with observations in Arabidopsis (Bohmert et al., 2000) and poplar (Dalton et al., 2011). While this could potentially explain the observed growth penalty as a function of reduced photosynthetic capability, the underlying cause for this is not clear and is being investigated further.

Other factors may also influence yield and biomass accumulation in PHB-producing sugarcane. For example, transplastomic tobacco has been shown to produce considerably more PHB (Bohmert-Tatarev et al., 2011) or p-hydroxybenzoic acid (Viitanen et al., 2004) than the nuclear-transformed, plastid-targeted counterparts (Siebert et al., 1996; Yamaguchi et al., 1999) with significantly fewer side effects. Therefore, a clear understanding of how PHB is produced in sugarcane will need to be obtained, in particular to determine the underlying causes of biomass reduction and the apparent PHB production inefficiency of mesophyll plastids. We are currently establishing a systems biology approach using some of the more powerful ‘omics’ tools now available.

Analysis of total leaf protein from glasshouse-grown plants showed that expression from the cab-m5 promoter yielded substantially higher levels of both PhaA and PhaB than from either ubi promoter (Figure 8). Because the corresponding plants produced higher levels of PHB, this was expected. We were, however, unable to reliably detect PhaC immunologically. Based on the expression of PhaA and PhaB, it is reasonable to conclude that the cab-m5 promoter drives the strongest expression of all three transgenes in sugarcane.

Conclusions

By using these strategies, we have significantly increased PHB production in sugarcane. The most substantial gains in mature plants were achieved using the cab-m5 promoter to drive transgene expression. However, additional approaches are still needed to reach commercial levels. Significant gains might be achieved if polymer levels in mesophyll cells can be increased to match levels in bundle sheath cells as mesophyll cells occupy around 2.4 times more leaf volume in sugarcane leaves than bundle sheath cells (Hattersley, 1984). Improving yields in mesophyll cells could significantly boost overall PHB production. We are currently investigating this problem and exploring additional strategies to achieve commercially relevant PHB levels in sugarcane.

Materials and methods

Tissue culture and transformation

All in vitro sugarcane culture was performed as described (Bower et al., 1996) with modifications. The pH of the medium was adjusted to 6.1 and pUKN (Joyce et al., 1998) replaced pEmuKn previously used in sugarcane (Petrasovits et al., 2007) for transformant selection. Transgenic callus was grown on EM3 or MS medium supplemented with 50 μg/L of Geneticin (Invitrogen, Thornton, NSW, Australia). Embryogenic callus used for the transformations was generated from the commercial sugarcane variety Q117. Biolistic transformation of callus was performed using the PDS 1000 gene delivery system (Biorad, Hercules, CA) in accordance with the manufacturer’s instructions. All bombardments were performed using 1100 psi rupture discs with 10 μg of plasmid DNA comprised of equimolar ratios of up to five plasmids coated onto 1.0-μm gold particles as described (Bower et al., 1996).

Construction of gene cassettes

Expression cassettes for sugarcane nuclear transformation driven by the maize Ubi promoter were constructed as previously described (Petrasovits et al., 2007). The Cv promoter was excised from pCvGFPT (Schenk et al., 2001) with XbaI and inserted into the respective sugarcane plastid targeting vectors cut with the same enzyme, replacing the Ubi promoter. Digestion with SmaI and EcoRI confirmed the presence of the correct size Cv promoter fragment in the correct orientation. Plasmids pMBXS155, a multigene construct in which the PHB genes are driven by the maize chlorophyll A/B-binding protein (cab-m5) promoter, and pMBXS159, a multigene construct in which the PHB genes are driven by the rice ubiquitin promoter (rubi2), have been previously described (Somleva et al., 2008). Plasmids pMBXS165, pMBXS166 and pMBXS167 contain phaA, phaB and phaC under the control of rubi2, respectively, and were supplied by Metabolix Inc. (Cambridge, MA).

Polymer detection

Polyhydroxybutyrate was detected in situ in sugarcane leaf tissue by Nile Blue staining (Kourtz et al., 2005). Stained leaf segments were examined at 200× magnification using a compound fluorescence microscope equipped with a dp70 imaging system (Olympus BX50; Olympus Australia, Windsor, Qld, Australia) and the following filter set: exciter HQ545/30, beam splitter d590/20m, emitter q570lp (Chroma Technology, Brattleboro, VT). Polymer presence was confirmed by HPLC as previously described (Petrasovits et al., 2007). Whole-leaf blades from the oldest green leaf were collected and analysed for the initial HPLC analysis. For the spatial distribution of polymer content as described in Figure 4 and polymer content in Table 4, whole leaves were used. Statistical analyses were performed as previously described. TEM was performed as previously described (Petrasovits et al., 2007).

Glasshouse trial

Single stalk cuttings with a one bud (one-eye sett) were planted in 4–6-L pots in December 2008, and mature plants were harvested in September 2009. The trial was planted in a randomized plot and watered automatically twice daily. Osmocote fertilizer (Scotts P/L, Baulkham Hills, Victoria, Australia) was applied as recommended by the manufacturer. At harvest, tiller heights and aerial biomass were determined on site. To determine PHB content, leaf material was dried at 70 °C for 1 week and measured using previously described HPLC procedures (Petrasovits et al., 2007).

Statistical analysis was performed using SigmaPlot for Windows Version 11.0 (Systat Software, Inc., Chicago, IL, USA). Data obtained on plant height and weight were statistically analysed using a one-way analysis of variance (ANOVA) test. Multiple comparisons versus the control group were conducted when statistically significant differences (P = <0.05) in the mean values among the treatment groups were detected.

Western blot analysis

Protein extractions and Western blotting were performed as previously described (Petrasovits et al., 2007). The RbcL antiserum was purchased (Agrisera A/B, Vännäs, Sweden) and used at a 1 : 1000 dilution as recommended by the supplier.

Acknowledgements

The expert technical assistance of Ms A. Su and S. Swapnil is gratefully acknowledged. Dr L. K. Gebbie is gratefully acknowledged for critically reviewing this manuscript. We express our thanks to Mr P. Abeydeera, Mr. M. Wang and Dr. J.O. Krömer for performing the HPLC separations. BSES Limited and the Cooperative Research Centre for Sugar Industry Innovation through Biotechnology are gratefully acknowledged for providing laboratory, office, tissue culture and glasshouse facilities to support this work and financial support, respectively.

Ancillary