Sugarcane (Saccharum hybrids) was evaluated as a production platform for p-hydroxybenzoic acid using two different bacterial proteins (a chloroplast-targeted version of Escherichia coli chorismate pyruvate-lyase and 4-hydroxycinnamoyl-CoA hydratase/lyase from Pseudomonas fluorescens) that both provide a one-enzyme pathway from a naturally occurring plant intermediate. The substrates for these enzymes are chorismate (a shikimate pathway intermediate that is synthesized in plastids) and 4-hydroxycinnamoyl-CoA (a cytosolic phenylpropanoid intermediate). Although both proteins have previously been shown to elevate p-hydroxybenzoic acid levels in plants, they have never been evaluated concurrently in the same laboratory. Nor are there any reports on their efficacy in stem tissue. After surveying two large populations of transgenic plants, it was concluded that the hydratase/lyase is the superior catalyst for leaf and stem tissue, and further studies focused on this pathway. p-Hydroxybenzoic acid was quantitatively converted to glucose conjugates by endogenous uridine diphosphate (UDP)-glucosyltransferases and presumably stored in the vacuole. The largest amounts detected in leaf and stem tissue were 7.3% and 1.5% dry weight (DW), respectively, yet there were no discernible phenotypic abnormalities. However, as a result of diverting carbon away from the phenylpropanoid pathway, there was a severe reduction in leaf chlorogenic acid, subtle changes in lignin composition, as revealed by phloroglucinol staining, and an apparent compensatory up-regulation of phenylalanine ammonia-lyase. Although product accumulation in the leaves at the highest level of gene expression obtained in the present study was clearly substrate-limited, additional experiments are necessary before this conclusion can be extended to the stalk.
Liquid crystal polymers (LCPs) have exceptional qualities compared with conventional resins. They are creep and electrically resistant, flame retardant, and perform well at elevated temperatures (Tullo, 1999). The global market for LCPs is approximately 10 000 metric tons per year (Tullo, 1999), with applications in electrical and optical connectors, integrated circuit boards, vehicle ignition components, and mobile phone components. LCPs are thermotropic polyesters and the major monomer used in the manufacture of these copolymers is the aromatic hydroxyacid, p-hydroxybenzoic acid (pHBA). The production of pHBA is currently achieved synthetically using the Kolbe–Schmitt process (Erickson, 1982), but this high-temperature, high-pressure carboxylation reaction is relatively expensive on an industrial scale. Green plants offer an attractive alternative for producing large quantities of pHBA in an environmentally sustainable manner that is less reliant on non-renewable resources.
All plants normally produce pHBA, albeit usually in small quantities. In Lithospermum erythrorhizon, pHBA is an intermediate in the synthesis of the commercially important red naphthoquinone pigment, shikonin. Radioisotope studies have suggested that pHBA is derived from the CoA ester of p-hydroxycinnamic acid through a β-oxidation-like mechanism (Löscher and Heide, 1994), although earlier studies with L. erythrorhizon (Yazaki et al., 1991) and carrot (Daucus carota) cell cultures (Schnitzler et al., 1992) suggested a cleavage mechanism that occurs via 4-hydroxybenzaldehyde. The accumulation of large quantities of pHBA has been accomplished in tobacco (Nicotiana tabacum) plants and tobacco cell cultures using a chloroplast-targeted version of Escherichia coli chorismate pyruvate-lyase (CPL) (Siebert et al., 1996). This enzyme catalyses the synthesis of pHBA from chorismate, an intermediate of the shikimate pathway (Figure 1), which in plants is largely, if not entirely, localized in plastids (Herrmann and Weaver, 1999). The tobacco-generated pHBA was quantitatively converted to two glucose conjugates, a phenolic glucoside and a glucose ester, that both contained a 1-O-β-d linkage. The pHBA glucose conjugates accumulated to 0.52% dry weight (DW), which is equivalent to 0.24%‘free’ pHBA after accounting for the attached glucose molecule. Remarkably, no visible phenotypic changes were detected (Siebert et al., 1996).
CPL has also been used for pHBA overproduction in E. coli (Barker and Frost, 2001), although this approach is plagued by difficulties that are easily circumvented in planta. High levels of pHBA are toxic to E. coli, thus restricting fermentation productivity. End-product inhibition is also a major problem for CPL-mediated microbial production, as the enzyme's Ki for pHBA (∼2 µm) is even lower than its Km for chorismate (Holden et al., 2002). These problems are minimized in plants because pHBA is glucosylated and subsequently stored in membrane-bound vacuoles (Yazaki et al., 1995; Bartholomew et al., 2002). Glucosylation of pHBA is required for vacuolar uptake.
Another bacterial protein that has been exploited for pHBA production in plants is 4-hydroxycinnamoyl-CoA hydratase/lyase (HCHL). This enzyme, originally isolated from Pseudomonas fluorescens (Gasson et al., 1998), hydrates and then catalyses a retro-aldol cleavage of 4-hydroxycinnamoyl-CoA thioesters to generate the corresponding 4-hydroxybenzaldehydes (Gasson et al., 1998; Narbad and Gasson, 1998; Mitra et al., 1999). 4-Hydroxycinnamoyl-CoA thioesters are intermediates of the phenylpropanoid pathway, which is one of the most active metabolic pathways in plants owing to its involvement in lignin and flavonoid biosynthesis. Constitutive expression of HCHL in tobacco resulted in leaf levels of pHBA glucose conjugates equivalent to 0.29% fresh weight, which corresponds to ∼1.3%‘free’ pHBA on a DW basis (Mayer et al., 2001). Apparently, 4-coumaroyl-CoA is the predominant substrate for HCHL in tobacco cytosol, and the vast majority of the resulting 4-hydroxybenzaldehyde is oxidized to pHBA by endogenous plant enzymes (Figure 1). Unfortunately, the HCHL-expressing tobacco plants suffered a severe depletion of phenylpropanoids that resulted in numerous phenotypic anomalies.
Significant achievements in plant genetics and breeding have made the in planta production of compounds traditionally synthesized from petrochemicals both technically feasible and economically attractive for major cash crops, such as maize and sorghum. Sugarcane is another plant suited to the task. It is a fast-growing C4 monocot that generates huge amounts of biomass (Alexander, 1973), which is a prerequisite for the economical production of bulk chemicals. The storage parenchyma of sugarcane is located above-ground, so that the product is easily harvested. Although sugarcane can flower, it is vegetatively propagated on a commercial scale (Bull, 2000), which minimizes the risk of transgene dispersal and guarantees germplasm stability. Finally, efficient sugarcane transformation protocols make it easy to simultaneously introduce multiple genetic traits at high frequency (Franks and Birch, 1991; Elliott et al., 2001), and the lead-time has been considerably shortened with the development of rapid tissue regeneration techniques (Lakshmanan et al., 2001).
As already indicated, CPL and HCHL have both previously been shown to increase pHBA levels in plants. However, these experiments were performed in different laboratories, and so a direct comparison of the two enzymes is not available. Moreover, pHBA accumulation in stem tissue was not examined in the earlier studies, and this is the part of the sugarcane plant that is normally harvested in the existing infrastructure. These important issues are addressed in the present study in carefully controlled experiments in which both proteins were expressed behind the same promoter. Finally, to the best of our knowledge, this is the first time that CPL or HCHL has been used to overproduce pHBA in a monocot.
Characterization of CPL-expressing and HCHL-expressing sugarcane plants
A chloroplast-targeted version of E. colicpl, situated between the maize ubi-1 promoter and nos terminator of the expression construct pU3z-mcs-nos, was cobombarded with a plasmid containing a selectable marker (pUKN) into embryogenic sugarcane callus to yield the UC series of plants. The UH lines were generated in the same manner using an analogous expression construct that contained the open reading frame (ORF) of the P. fluorescenshchl gene. To serve as controls, four non-transgenic lines (TC1–TC4) were regenerated from the same callus material omitting the transformation and selection steps. The regenerated plants were analysed by high-performance liquid chromatography (HPLC) for pHBA accumulation in leaf tissue after 4 weeks of growth in a greenhouse (Figure 2). The low number of plants tested was influenced by the fact that sugarcane is a very large plant and glasshouse space was limited. Moreover, only plants that had higher levels of pHBA than the control plants were ultimately included in the analysis (46% and 56% of the UC and UH lines, respectively).
Not surprisingly, none of the plants accumulated significant amounts of free pHBA. Similar to CPL-expressing tobacco plants (Siebert et al., 1996) and HCHL-expressing tobacco plants (Mayer et al., 2001), virtually all of the pHBA was converted to glucose conjugates: a phenolic glucoside and a glucose ester. Both compounds contained a single glucose molecule that was attached to the hydroxyl or carboxyl group of pHBA by a 1-O-β-d linkage. The phenolic glucoside was the predominant species in all plants examined, and accounted for > 90% of the total pHBA. Leaf levels of total pHBA glucose conjugates for 49 independent primary transformants expressing CPL are shown in Figure 2A. The mean value for the entire population was 0.41% ± 0.04% DW, which is almost 30-fold higher than the average value for the control plants (0.014% ± 0.01% DW). More importantly, the leaf content of the best UC line was 1.5% DW, which is equivalent to 0.69% DW free pHBA after correcting for the attached glucose molecule. To put this in perspective, the highest leaf levels of pHBA reported for tobacco plants expressing a different chloroplast-targeted version of CPL were almost three times lower (Siebert et al., 1996). The UH lines accumulated even higher levels of pHBA glucose conjugates (Figure 2B). The mean value was 0.70% ± 0.07% DW, and the highest level observed at this stage of development was 2.6% DW, which is nearly identical to the best value reported for fully mature tobacco plants expressing exactly the same enzyme (Mayer et al., 2001).
After 16 weeks of additional growth, a subset of the primary transformants was selected for further evaluation (Figure 3). Included in this analysis were the two best CPL lines and five HCHL plants that encompassed a broad range of pHBA levels when the plants were 4 weeks old. It was anticipated that product accumulation would continue throughout development, and that the 20-week-old plants would have higher levels of pHBA. However, the results were not very dramatic, nor was an increase in pHBA universally observed, when the results were expressed on a DW basis (Figure 3A). If the data from the UH lines are expressed on a fresh weight basis (Figure 3B), it becomes evident that each unit of leaf biomass harvested at week 20 contained more pHBA than at week 4. Therefore, it is clear that the leaf levels of pHBA increased as the plants continued to grow. The discrepancy between Figure 3A and 3B can be explained by the higher proportion of water in the biomass of the younger plants: the average dry weight to wet weight ratio for the 20-week-old plants was 0.23, compared with 0.15 for the 4-week-old plants. Interestingly, UC65 did not exhibit the same trend, but this could simply be due to the small number of plants tested.
The same 20-week-old primary transformants were large enough to screen for stalk levels of pHBA without damaging the plants. At this stage of development, the oldest stem tissue is semi-mature and new tillers start to emerge. As the sugarcane stalk is the only part of the plant that is normally harvested, stalk levels of pHBA are the most important gauge for technical success. The source of stem tissue for this analysis was the third internode from the bottom of the plant, and leaf samples were obtained from the third leaf from the top. Generally speaking, product accumulation in leaf tissue was greater than that in stem tissue, but the difference was much more pronounced for the CPL-expressing plants. For example, the average stalk to leaf ratio for the five UH lines was 0.324 ± 0.031, whereas the corresponding values for UC63 and UC65 were 0.135 and 0.133, respectively. More importantly, the highest stem level of pHBA glucose conjugates for the UH lines was 0.52% DW, compared with only 0.13% DW for the best CPL plant. Similar to the situation in leaves, almost all of the pHBA in the stalk was converted to the phenolic glucoside (> 80%). Taken together, the above results suggest that HCHL is a better catalyst than CPL for pHBA production in sugarcane, and subsequent studies focused on the UH lines.
To gain a better understanding of pHBA accumulation in different parts of the plant, several of the 20-week-old primary transformants were examined in greater detail. Leaf and stem tissue was sampled along the length of the primary shoot, and pHBA glucose conjugate levels were determined by HPLC. The first leaf at the top of the plant with a fully visible dewlap was designated ‘L1’ and consecutive leaves down the stalk were numbered in ascending order. Stem samples were numbered in a similar fashion, such that ‘S5’ corresponds to the internode immediately above ‘L5’, and so on. A representative experiment with UH1 is shown in Figure 4A. Product accumulation in leaves was relatively uniform along the length of the plant, except for the youngest leaf examined. A similar trend was observed in the mature part of the stalk, but the maximum level of pHBA glucose conjugates was about threefold lower than in the leaves, as already described. There was also a much larger difference between young tissue and old tissue in the stalk.
Further insight was obtained from dissection experiments similar to that shown in Figure 4B, which was also performed with 20-week-old UH1. Three different parts of the stalk were examined: rind, pith, and vascular bundles. The rind had the highest concentration of pHBA glucose conjugates (2.3% DW), which was almost equivalent to the values obtained with leaf lamina and leaf midrib. The pith and vascular bundles had three- to fourfold lower levels. Although a constitutive promoter was used throughout the present work, these observations provide useful benchmarks for future studies, and underscore the need to identify stronger promoters that more effectively target pHBA production to the stalk.
HCHL enzyme activity in second-generation UH lines
The preceding experiments provide no insight on how to achieve higher levels of pHBA in HCHL-expressing sugarcane plants and what the upper boundaries are. Specifically, it would be useful to know what the current limitation is – substrate availability or enzyme activity – in different parts of the plant. To address this issue, UH52, UH51, UH68, UH79, and a non-transformed control plant (TC1) were examined in greater detail as second-generation plants (Figure 5). These lines were chosen because they all originated from the same batch of callus and encompassed a broad range of pHBA levels as primary transformants. Although the number of plants in this analysis is rather small, it should be emphasized that the same trends were observed with second-generation UH lines that were derived from different batches of callus.
Propagative cuttings of stem tissue (setts), each containing an internode and associated bud, were regenerated in a greenhouse to yield multiple clones of each line. When the regenerated plants were 20 weeks old, mature leaf and stem tissue was analysed for pHBA glucose conjugates. Cell-free extracts were prepared from the same samples and HCHL enzyme activity was measured with pHCA-CoA as a substrate (Figure 5). Although there was a clear correlation between pHBA glucose conjugates and HCHL enzyme activity in leaf tissue, the relationship was far from linear and appeared to approach saturation with respect to the enzyme. Indeed, the data suggest that the current limitation for pHBA production in leaves is substrate availability, not HCHL enzyme activity. This notion is also supported by other experiments described below.
Unfortunately, the parallel experiments with stem tissue were not very illuminating. With the small number of data points and ambiguous shape of the curve, it was not clear whether pHBA production in the stalk was substrate-limited at the highest level of HCHL enzyme activity achieved. The phenylpropanoid pathway in stem tissue is a high-flux pathway largely devoted to lignin biosynthesis, and lignin accounts for up to 20% of the total DW of the stalk. Consequently, carbon flux through the phenylpropanoid pathway should not be a problem in the stalk unless substrate channelling sets an upper limit on the amount of 4-coumaroyl-CoA that is available to HCHL. The highest HCHL specific activities measured in leaf and stem tissue were very similar when normalized to total protein, but the pHBA content in leaves was about 2.5-fold higher. Although this might suggest that HCHL is substrate-limited in the stalk, this is not a valid conclusion, as the protein content of leaves is at least 10-fold higher than that of the culm on a fresh weight basis. Clearly, more experiments with a larger population of plants encompassing a broader range of HCHL enzyme activities are necessary to determine what currently limits stalk production of pHBA.
Effect of HCHL on other soluble phenolics, lignin, and phenylalanine ammonia-lyase (PAL) activity
As anticipated from previous experiments with HCHL-expressing tobacco plants (Mayer et al., 2001), the second-generation UH lines also accumulated vanillic acid glucose conjugates (Table 1). The HCHL substrate from which these compounds are ultimately derived is feruloyl-CoA, a phenylpropanoid intermediate downstream from 4-coumaroyl-CoA. Leaves accumulated more vanillic acid than the culm, but both tissues had much higher levels of pHBA (Table 1 and Figure 5). Although there was a distinct correlation between pHBA and vanillic acid accumulation, the results strongly suggested that vanillic acid production became substrate-limited at lower levels of HCHL enzyme activity.
Table 1. Accumulation of soluble phenolics in leaf and culm tissue of second-generation UH lines at 20 weeks. TC1 is a non-transformed control plant
Values shown are the mean ± standard error (SE) of replicate plants from each line (n = 4–5).
CA, chlorogenic acid, × 10−3% DW; VA, vanillic acid, × 10−2% DW; HCHL, 4-hydroxycinnamoyl-CoA hydratase/lyase enzyme activity with 4-coumaroyl-CoA as a substrate, × 10−2 pkat/µg protein.
94.2 ± 1.2
10.1 ± 5.6
57.6 ± 4.9
1.9 ± 0.2
7.2 ± 5.3
0.3 ± 0.03
1.4 ± 0.3
11.6 ± 2.8
12.1 ± 1.7
2.9 ± 0.1
3.7 ± 0.05
3.0 ± 0.1
3.2 ± 0.3
13.1 ± 1.6
3.8 ± 0.03
5.7 ± 2.9
3.2 ± 0.2
5.4 ± 1.2
16.4 ± 3.7
6.6 ± 0.6
8.3 ± 1.9
3.5 ± 0.4
Chlorogenic acid (5-caffeoylquinic acid) was another soluble phenylpropanoid derivative affected by HCHL expression. As shown in Table 1, the amount of this compound in leaves was reciprocally related to HCHL enzyme activity. Indeed, leaf chlorogenic acid was completely depleted in the two plants that exhibited the highest HCHL specific activity (UH68 and UH79). The drastic reduction of chlorogenic acid with high-level expression of HCHL provides further support for the notion that carbon flux through the phenylpropanoid pathway sets an upper limit on pHBA production in leaves. In contrast, with the exception of UH51, chlorogenic acid in stem tissue was only moderately affected by HCHL expression, and there was no apparent correlation between enzyme activity and the amount of this compound.
It was also of interest to determine whether there were any gross changes in lignin composition in the UH lines, as 4-coumaroyl-CoA is an important branch point intermediate in monolignol biosynthesis. Stem tissue from second-generation UH68 and from a non-transformed control plant (TC1) was used for the representative experiment shown in Figure 6. Multiple hand-cut sections from the fourth and fifth stem internodes were stained with phloroglucinol and subjected to bright-field microscopy. The most noticeable difference in staining pattern was in the lignified sclerenchyma layer of the vascular bundle, which stained dark cherry-red in the control plant. The extent and intensity of staining in this region were typically reduced in UH68, although the differences were not extreme. In contrast with HCHL-expressing tobacco plants (Mayer et al., 2001), there was no evidence of deformed vessels in sugarcane.
Phenylalanine ammonia-lyase (PAL) is the first enzyme in the phenylpropanoid pathway and exerts a major regulatory influence on carbon flux to downstream intermediates, including 4-coumaroyl-CoA. As HCHL-expressing tobacco plants had higher steady-state levels of PAL mRNA (Mayer et al., 2001), presumably reflecting elevated levels of PAL enzyme activity, it was important to determine whether this phenomenon also occurred in sugarcane (Table 2). Indeed, there was a five- to 10-fold increase in PAL enzyme activity in leaf extracts prepared from UH68 compared with the non-transformed control plant. A similar trend was observed in the culm, but the results were not as dramatic.
Table 2. In vitro enzyme activity (× 10−3 pkat/µg protein) for 4-hydroxycinnamoyl-CoA hydratase/lyase (HCHL) and phenylalanine ammonia-lyase (PAL) in UH68 and TC1
PAL activities are tabulated as the mean ± standard error (SE) of independent assays conducted on two separate clones from each line. HCHL activities were collated as described in Figure 5.
31 ± 3
38 ± 0.3
0.08 ± 0.005
0.85 ± 0.38
176 ± 50
363 ± 82
Of all the primary transformants monitored, UH98 consistently had the highest levels of pHBA glucose conjugates (Figure 2B). When UH98 was 20 weeks old, product accumulation in the mature leaf and stem tissue was 6.1% DW and 1.5% DW, respectively, yet the plant appeared to be perfectly healthy (Figure 7). Unfortunately, UH98 was generated from a later batch of callus and was not available when the experiments with the second-generation plants were performed.
The DW and specific protein content for leaf and stem were also measured in the four HCHL lines and the control line described previously (results not shown). A small, but statistically significant, decline in leaf DW and an elevated stem-specific protein level were observed with higher levels of transgene expression, but there was no discernible effect on stem DW or leaf-specific protein content. The explanation for these observations remains to be elucidated and warrants further investigation.
CPL and HCHL were used to evaluate sugarcane as a production platform for pHBA. Neither of these bacterial enzymes have counterparts in plants. pHBA is the major monomer in LCPs and a precursor for the chemical synthesis of parabens, which is widely used as a preservative in food and cosmetics. Although CPL and HCHL both converted a naturally occurring plant compound to pHBA, superior results were obtained with the latter when a constitutive promoter was used to drive transgene expression. The HCHL substrate that gives rise to pHBA in plants is 4-coumaroyl-CoA, a key intermediate in the phenylpropanoid pathway, which is synthesized in the cytosol. The substrate for CPL is chorismate, an important branch point intermediate in the shikimate pathway, which is only formed in plastids. Given the additional complexity of targeting CPL to plastids, it is very possible that the full potential of this enzyme was not realized in the present study, and that using a better chloroplast transit peptide might result in higher levels of product accumulation.
Similar to the results obtained with tobacco, virtually all of the sugarcane-produced pHBA was converted to a glucose ester and a phenolic glucoside. The latter compound was the predominant species in all tissues examined. In tobacco, both glucose conjugates are formed by distinct uridine diphosphate (UDP)-glucosyltransferases, and the preferential attachment of glucose to the aromatic hydroxyl group was attributed to the phenolic glucoside-forming enzyme having a lower Km for pHBA (Li et al., 1997). Leaf levels of pHBA glucose conjugates in HCHL-expressing sugarcane plants were as high as 7.3% DW (35-week-old UH98), which exceeds the best value reported for constitutive HCHL expression in tobacco by almost a factor of three (Mayer et al., 2001). Moreover, the transgenic tobacco plants that accumulated the highest levels of pHBA also showed a severe depletion of phenylpropanoids that resulted in a number of phenotypic abnormalities, including leaf chlorosis, altered lignin content, stunted growth, and male sterility. Although none of these side-effects were observed in our UH lines, a similar scenario would probably occur with higher levels of HCHL gene expression, especially if a constitutive promoter was used.
The expected HCHL cleavage product with 4-coumaroyl-CoA as a substrate is 4-hydroxybenzaldehyde, but none of the UH lines accumulated this compound or its glucose conjugate. This result is consistent with the observations reported for tobacco (Mayer et al., 2001) and Datura stramonium hairy root cultures (Mitra et al., 2002). Although the endogenous enzymes that convert 4-hydroxybenzaldehyde to pHBA have not been identified in any of these systems, it is reasonable to assume that oxidation occurs prior to glucosylation, at least in the case of the pHBA glucose ester. HCHL-expressing tobacco plants also accumulated glucose conjugates of 4-hydroxybenzyl alcohol, although they were not as abundant as pHBA derivatives (Mayer et al., 2001). These compounds, which were probably formed through the sequential reduction and glucosylation of 4-hydroxybenzaldehyde, may also be minor products in sugarcane, but no attempt was made to identify them.
At least two other intermediates in the phenylpropanoid pathway can serve as substrates for P. fluorescens HCHL, as revealed by studies with the purified enzyme (Mitra et al., 1999). The cleavage product with feruloyl-CoA as a substrate is vanillin (Figure 8) which, in plants, should give rise to vanillic acid glucose conjugates by analogy with 4-hydroxybenzaldehyde. Indeed, our UH lines accumulated these compounds, albeit at much lower levels than pHBA (Table 1). The other naturally occurring HCHL substrate in plants is caffeoyl-CoA, which should yield glucose conjugates of protocatechuic acid through the intermediacy of 4-hydroxy-3-methoxybenzaldehyde. However, HCHL-expressing sugarcane plants did not accumulate these compounds, similar to previous results reported for tobacco (Mayer et al., 2001).
Clearly, the kinetic properties of the purified enzyme (Mitra et al., 1999) do not entirely account for the product profile in HCHL-expressing plants. The apparent Km values for 4-coumaroyl-CoA, caffeoyl-CoA, and feruloyl-CoA are 5.3 µm, 1.6 µm, and 2.4 µm, respectively, and their catalytic efficiencies (Vmax/Km) differ by less than a factor of two. The fact that 4-coumaroyl-CoA is the predominant substrate for HCHL in plants may be at least partially explained by its favourable upstream location in the phenylpropanoid pathway, as caffeoyl-CoA and feruloyl-CoA are both believed to largely be derived from this compound (Whetten et al., 1998). If this is true, it is not too difficult to imagine that HCHL expression in plants could lower the steady-state level of both of these compounds. In support of this notion, leaf levels of chlorogenic acid were below the level of detection in UH lines that had the largest amounts of pHBA (Table 1), and the principal precursor for this normally abundant secondary metabolite is caffeoyl-CoA (Stöckigt and Zenk, 1974). On the other hand, the complete absence of protocatechuic acid derivatives in HCHL-expressing plants suggests that other factors may also be operating, such as substrate channelling, which has been proposed for certain enzymes in the phenylpropanoid pathway (Rasmussen and Dixon, 1999; Winkel-Shirley, 1999).
The first step in phenylpropanoid biosynthesis is catalysed by PAL, which also plays a key role in regulating carbon flux into the pathway (Bate et al., 1994). This enzyme is stimulated by numerous factors, including stress elicitors (Ni et al., 1996; Rivero et al., 2001; Rizhsky et al., 2002). PAL enzyme activity in HCHL-expressing sugarcane plants was significantly elevated, particularly in leaf tissue. Although PAL specific activities were not measured in the study with tobacco (Mayer et al., 2001), hyperexpression of HCHL was accompanied by a large increase in PAL transcripts. It therefore appears that, in both plant species, up-regulation of PAL is a compensatory mechanism that occurs in response to HCHL-mediated phenylpropanoid depletion. Although this probably takes place at the level of feedback control by downstream intermediates, it may also reflect to some degree a generalized response to stress. Regardless of the explanation, it is clear that the elevated PAL activity in tobacco plants expressing HCHL could not keep pace with the massive re-routing of 4-coumaroyl-CoA to pHBA glucose conjugates. Although the only evidence for phenylpropanoid depletion in sugarcane was a drastic reduction of leaf chlorogenic acid, other soluble phenylpropanoid derivatives were probably affected by HCHL expression. Reduced levels of chlorogenic acid in tobacco are thought to result in an increased disease susceptibility (Maher et al., 1994).
The curvilinear relationship between the pHBA content and HCHL enzyme activity (Figure 5) suggests that product accumulation in leaves is substrate-limited at the highest level of gene expression achieved in the present study. Consequently, using a stronger constitutive promoter to boost HCHL expression in sugarcane is not a viable strategy for commercial success, especially when the stalk is the compartment of interest. This approach would probably result in very sick plants, similar to the situation in tobacco. It might also have a detrimental effect on biomass yield and indirectly set an upper threshold on stalk levels of pHBA even under conditions in which substrate availability would not normally be limiting.
The large amounts of lignin that are typically deposited in stem tissue suggest that the stalk has a high-flux phenylpropanoid pathway, and that this part of the plant should provide an ideal environment for HCHL. However, what currently limits pHBA production in the stalk is still an open question that needs to be addressed in a more rigorous manner, as the results from our study are inconclusive. The best way to address this question is to use a strong stem-specific promoter and carry out experiments similar to those shown in Figure 5. Targeting HCHL to the stalk would minimize phenylpropanoid depletion in other parts of the plant that are more sensitive. As most tissue-specific promoters are associated with some degree of ‘leakiness’, it might also be prudent to overexpress PAL to provide additional protection.
Plants typically respond to biotic and abiotic stresses by producing ethylene, which initiates deleterious events such as senescence and chlorosis (Stearns and Glick, 2003). Although we did not measure ethylene in the transgenic sugarcane, the high levels of pHBA demonstrated in this study may stress the sugarcane plant. A closer examination of this aspect would be prudent in future work.
In summary, this study has provided some insight into the strategies that must be considered for a commercially viable pathway for pHBA production in sugarcane. As this crop is traditionally harvested for its stalk, it is desirable to target HCHL to this compartment to exploit existing infrastructure. However, if it is not possible to dissociate the positive and negative attributes of HCHL using a strong stem-specific promoter, further studies with CPL are warranted. There are many possible reasons why this protein was not as effective a catalyst as HCHL for stalk production of pHBA. Very little is known about protein import into non-photosynthetic plastids, and the chloroplast-targeting sequence used in this study may have been suboptimal. Because the sugar industry currently harvests just the stalk once a year, either burning off the leaves or cutting off the tops and leaving the leaf material behind as a trash blanket, and given the high-flux phenylpropanoid pathway in stem tissue, we have focused most of our effort to date on stalk production of pHBA. However, an economic appraisal of the situation might reveal that production in leaves is also a viable strategy. Three alternatives can be envisioned: (i) the entire plant is harvested and pHBA, sucrose, molasses and energy are products; (ii) the leaves and stalk are harvested separately for pHBA and sucrose, respectively; or (iii) the tops are harvested multiple times throughout the long growing season to obtain even more pHBA. This would have significant advantages for the sugarcane mills and farm machinery which currently operate for very short periods each year (20–24 weeks).
Construction of cTP-CPL
The monocot chloroplast-targeting sequence that was fused to the N-terminus of E. coli CPL was generated by polymerase chain reaction (PCR). The maize rbcS gene (GenBank accession number Y00322) that codes for the ribulose-1,5-bisphosphate carboxylase/oxygenase (rubisco) small subunit precursor was the target for amplification. Primer 1 (5′-CTA CTC ATA ACC ATG GCG CCC ACC GTG-3′) hybridized to nucleotides (nt) 489–505 and introduced an NcoI site at the transit peptide start codon. Primer 2 (5′-CAT CTT ACT CAT ATG CCG CAC CTG CAT GCA CCG GAT CCT TCC G-3′) hybridized to nt 616–639 and introduced an NdeI site five amino acid residues downstream from the chloroplast cleavage site. The PCR product was cut with NcoI and NdeI and inserted into pET24a-tTP-CPL, after the latter had been cleaved with the same enzymes. pET24a-tTP-CPL contains the gene for a chimeric protein that consists of the tomato rubisco small subunit transit peptide plus the first four amino acid residues of the ‘mature’ rubisco small subunit, fused to the N-terminus of E. coli CPL. The plasmid DNA was cut with NcoI and NdeI to remove the tomato chloroplast-targeting sequence, and the latter was replaced with the maize chloroplast-targeting sequence. The ligation mixture was introduced into E. coli DH10B, and growth was selected on Loria Bertwi (LB) medium plus kanamycin (50 µg/mL). A representative plasmid (pET24a-cTP-CPL) that contained no PCR errors was selected for further manipulation. The chloroplast cleavage product of cTP-CPL is a CPL variant with five extra N-terminal residues (i.e. MQVRH-CPL) that do not effect enzyme activity.
Generation of cpl and hchl expression constructs for sugarcane
The antibiotic selection plasmid pUKN (Joyce et al., 1998) contains the ubi-1 promoter, the neomycin phosphotransferase gene and the nos terminator. The plasmid pU3z-mcs-nos was used for cTP-CPL and HCHL expression in sugarcane. This plasmid is a modification of pAHC20 (Christensen and Quail, 1996) and contains the maize ubi-1 promoter and nos terminator. Both genes were inserted in the SpeI and KpnI sites of the multicloning region that immediately follows the maize ubi-1 intron. The gene coding for cTP-CPL was amplified from pET24a-cTP-CPL using primers 3 and 4. Primer 3 (5′-CTA CTC ATT TAC TAG TCA CCA TGG CGC CCA CCG TGA TG-3′) hybridized to the first 18 nt of the ORF of cTP-CPL and introduced a SpeI site upstream from the start codon. Primer 3 also contained a consensus monocot ribosomal binding site, CACC (Joshi et al., 1997), which is situated between the SpeI site and the initiator Met codon. Primer 4 (5′-CAT CTT ACT GGT ACC TTT AGT ACA ACG GTG ACG CC-3′) hybridized to the other end of the insert and introduced a KpnI site just after the cTP-CPL stop codon. The PCR product was cut with SpeI and KpnI, and ligated into similarly digested pU3z-mcs-nos. The ligation reaction mixture was used to transform E. coli DH10B and growth was selected on LB medium containing ampicillin (100 µg/mL). A representative plasmid (pU3z-mcs-nos-cTP-CPL) was sequenced to confirm the absence of PCR errors.
Primers 5 and 6 were used to amplify the P. fluorescens strain AN103 hchl gene (GenBank accession number Y13067) from plasmid pFI1039 (Gasson et al., 1998). Primer 5 (5′-CTA CTC ATT TAC TAG TCA CCA TGA GCA CAT ACG AAG GTC G-3′) hybridized to the first 20 nt of the hchl ORF and introduced a unique SpeI site upstream from the start codon. Primer 5 also contained a consensus monocot ribosomal binding site (CACC) that is located between the SpeI site and the initiator Met codon. Primer 6 (5′-CAT CTT ACT GGT ACC TTC AGC GTT TAT ACG CTT GCA-3′) hybridized to the other end of the insert and introduced a KpnI site just after the HCHL stop codon. The PCR product was cut with SpeI and KpnI, and ligated into similarly digested pU3z-mcs-nos. The ligation reaction mixture was used to transform E. coli DH10B and growth was selected on LB medium containing ampicillin (100 µg/mL). A representative plasmid (pU3z-mcs-nos-HCHL) was sequenced to confirm the absence of PCR errors.
Embryogenic callus from sugarcane cultivar Q117 was prepared essentially as described previously (Franks and Birch, 1991), and grown in the dark at 27 °C on Murashige–Skoog (MS) medium (Murashige and Skoog, 1962) supplemented with 3 mg/L of 2,4-dichlorophenoxy acetic acid (2,4-D). The calli were cotransformed with the antibiotic selection plasmid pUKN by microprojectile bombardment (Bower et al., 1996). Following bombardment and a 2-week recovery period in the dark, transformants were placed on MS-2,4-D selection medium supplemented with 60 mg/L geneticin. Individual callus clumps were maintained separately throughout the selection process. Antibiotic-resistant calli were transferred to MS medium supplemented with geneticin 6 weeks later and were incubated in the light to regenerate plantlets. At least four plantlets per callus clump were transferred to pots in a glasshouse certified for the physical containment of transgenic plants for further analysis.
Plant material and growth conditions
Sugarcane cultivar Q117 and all transgenic lines derived therefrom were grown in a soil substitute comprising a 1 : 1 mixture of perlite (Chillagoe Perlite, Mareeba, Qld, Australia) and vermiculite (Chillagoe Perlite, Mareeba, Qld, Australia). The soil was supplemented at 3-monthly intervals with Osmocote slow-release nitrogen fertilizer granules (Scotts Australia Pty Ltd, Baulkam Hills, NSW, Australia). Plants were grown in a glasshouse and planted individually in pots.
Stalk and leaf midrib lignin was visualized by histochemical staining with phloroglucinol. Hand-cut sections of tissue were soaked in an ethanol solution containing 1% (w/v) phloroglucinol for 2 min, and were then immersed in concentrated HCl until the red colour developed (∼1 min). The stained sections were rinsed in distilled water, mounted in distilled water, and immediately examined using bright-field microscopy (400× magnification).
Measurement of HCHL and PAL enzyme activities in leaf and stalk extracts
Leaf samples were collected from node 4 and stem tissue from internodes 5 and 6. For leaf assays, 1 g of lamina tissue was ground to a fine powder in liquid nitrogen, and resuspended in 3 mL of extraction buffer [100 mm Tris-HCl/pH 8.5, 10 mm ethylenediaminetetraacetic acid (EDTA), 300 mm NaCl, 5% v/v glycerol, 0.01% v/v Tween 20, 2 mm dithiothreitol (DTT)]. For stem assays, 4 g of tissue was removed from each segment and 8 mL of extraction buffer was used. Each internode was extracted independently. The samples were then homogenized at 0 °C using a Polytron® (IKA-Werke GMBK & Co. KG, Staufen, Germany); 5% w/w polyvinylpolypyrrolidone was added during the grinding process. Debris was removed by centrifugation at 4 °C, and the supernatant was concentrated using Ultra-4, 10 000 NMWL centrifugal filters (Millipore, Bedford, MA, USA). The cell-free extract protein concentration was determined by the Bradford method (Bradford, 1976) using bovine serum albumin as the standard, and all enzyme assays were conducted with freshly prepared extracts.
The 25-µl HCHL reactions were conducted in triplicate at 30 °C, and contained 100 mm Tris-HCl (pH 8.5), 800 µm 4-coumaroyl-CoA, and 100–400 µg of extract protein. Reactions were stopped with 25 µL of 12% (v/v) acetic acid/methanol after a 5–30-min incubation period. HCHL enzyme activity was monitored by following the formation of 4-hydroxybenzaldehyde, which was detected by HPLC using a Novapak C18, 60A, 4 µm, 3.9 mm × 150 mm column (Waters, Milford, MA, USA), which was maintained at 32 °C. Samples were clarified by centrifugation and 20 µL was injected. Solvent A was 1.5% phosphoric acid and solvent B was 1.5% phosphoric acid, 50% methanol. The flow rate was 1 mL/min and the following gradient was used: 0–10 min, 0% B; 19 min, 3% B; 20 min, 10% B; 40 min, 100% B; 44 min, 100% B; 45 min, 0% B. Adsorption at 284 nm was monitored and the area under the peak corresponding to 4-hydroxybenzaldehyde was quantified using an authentic standard (Sigma-Aldrich Co. St Louis, MI).
The 50 µL PAL assays contained 50 mm Tris-HCl (pH 8.5), 10 mm l-phenylalanine, and 20 µL of extract protein at 30 °C. They were conducted in a similar manner to the HCHL assays, except that stem and leaf reactions were stopped after 30 and 60 min, respectively, using 50 µL of the acidic methanol solution described above. The product measured was trans-cinnamic acid, which was detected by HPLC using the conditions described below for soluble phenolics.
Measurement of accumulated soluble phenolics by HPLC
Soluble phenolics (including pHBA and vanillic acid glucose conjugates and chlorogenic acid) were extracted from leaf or stem tissue (100–200 mg) with 1 mL of 50% (v/v) methanol. The tissue was disrupted with a bead beater (Bio101/Savant, Fastprep FP120, Holbrook, NY, USA), and the resulting homogenate was agitated on an orbital shaker (200 r.p.m.) for 1 h at 37 °C. Debris was removed by centrifugation and a 550-µL aliquot of the supernatant was transferred to a fresh microfuge tube and dried under vacuum. The dry residue was dissolved in 200 µL of dH2O and the sample was analysed by HPLC as described below. Alternatively, the dry residue was resuspended in 200 µL of 1 m HCl (with vigorous vortex mixing) and the tube was incubated for 2 h at 95 °C. After the addition of 200 µL of 1.2 m NaOH, the sample was analysed by HPLC as described below. This procedure completely hydrolyses pHBA and vanillic glucose conjugates and results in the quantitative recovery of free pHBA and vanillic acid.
Soluble phenolics were detected by HPLC at 32 °C using the Novapak C18 column described above. Samples were filtered through 0.2-µm syringe filters and 20 µL of filtrate was analysed. The flow rate and mobile phases were the same as described above, but the following gradient conditions were employed: 0 min, 0% B; 80 min, 80% B; 81 min, 100% B; 85 min, 100% B; 86 min, 0% B. To effectively separate pHBA and vanillic acid, the gradient was modified as follows: 0 min, 10% B; 20 min, 50% B; 21 min, 100% B; 22 min, 100% B; 25 min, 10% B. The identified peaks were quantified using authentic standards obtained from Sigma-Aldrich Co. (St Louis, MI), with the exception of the pHBA phenolic glucoside and glucose ester, which were synthesized and characterized at DuPont.
We are grateful to Nial Masel for technical assistance with the HPLC analyses and Karen Bacot for help with the DNA constructs. We also thank Dr Peter Allsopp for critically reviewing the manuscript. NJW receives support from the Biotechnology and Biological Sciences Research Council (BBSRC) through the Core Strategic Grant to the Institute of Food Research (IFR).