Multiple Arabidopsis genes primed for recruitment into C4 photosynthesis

Authors


(fax +44 (0)1223 333953; e-mail julian.hibberd@plantsci.cam.ac.uk).

Summary

C4 photosynthesis occurs in the most productive crops and vegetation on the planet, and has become widespread because it allows increased rates of photosynthesis compared with the ancestral C3 pathway. Leaves of C4 plants typically possess complicated alterations to photosynthesis, such that its reactions are compartmented between mesophyll and bundle sheath cells. Despite its complexity, the C4 pathway has arisen independently in 62 separate lineages of land plants, and so represents one of the most striking examples of convergent evolution known. We demonstrate that elements in untranslated regions (UTRs) of multiple genes important for C4 photosynthesis contribute to the metabolic compartmentalization characteristic of a C4 leaf. Either the 5′ or the 3′ UTR is sufficient for cell specificity, indicating that functional redundancy underlies this key aspect of C4 gene expression. Furthermore, we show that orthologous PPDK and CA genes from the C3 plant Arabidopsis thaliana are primed for recruitment into the C4 pathway. Elements sufficient for M-cell specificity in C4 leaves are also present in both the 5′ and 3′ UTRs of these C3A. thaliana genes. These data indicate functional latency within the UTRs of genes from C3 species that have been recruited into the C4 pathway. The repeated recruitment of pre-existing cis-elements in C3 genes may have facilitated the evolution of C4 photosynthesis. These data also highlight the importance of alterations in trans in producing a functional C4 leaf, and so provide insight into both the evolution and molecular basis of this important type of photosynthesis.

Introduction

Having evolved in at least 62 separate lineages of land plants, C4 photosynthesis is considered one of the most remarkable examples of convergent evolution (Sage et al., 2011). Current estimates are that land plants first started to use C4 photosynthesis around 30–32 million years ago, and that it evolved in response to high temperatures and reductions in atmospheric CO2 content (Christin et al., 2008; Edwards et al., 2010; Vicentini et al., 2008). The C4 pathway allows CO2 to be concentrated around the central photosynthesis enzyme Ribulose-1,5-Bisphosphate Carboxylase Oxygenase (RuBisCO). Because RuBisCO does not completely distinguish between CO2 and O2 (Bowes et al., 1971), the increased supply of CO2 to RuBisCO in C4 plants reduces the oxygenation reaction, and in turn limits the wasteful reactions of photorespiration (Hatch, 1987). In all cases, C4 plants use a biochemical pump to increase the amount of CO2 in compartments within which RuBisCO is limited. Although photosynthesis reactions are compartmented within a single cell in some C4 lineages (Reiskind et al., 1989; Voznesenskaya et al., 2001), most C4 species partition photosynthetic reactions between two distinct cell types that are arranged in concentric circles around veins, generating so-called Kranz anatomy (Brown et al., 2005; Hatch, 1987; Langdale and Nelson, 1991). This separation of metabolic reactions between bundle sheath (BS) and mesophyll (M) cells mean that C4 acids are produced in M cells and then diffuse to BS cells where the Calvin–Benson cycle operates (Hatch, 1987; Leegood, 2002), and is achieved by restricting the accumulation of proteins to either BS or M cells. For example, carbonic anhydrase (CA), phosphoenolpyruvate carboxylase (PEPC), NADP-malate dehydrogenase (MDH), pyruvate,orthophosphate dikinase (PPDK) and the PPDK regulatory protein (RP) typically accumulate in M cells in C4 plants, whereas a C4 acid decarboxylase, phosphoribulokinase (PRK), ribose-5-phosphate isomerase (RPI) and RuBisCO are restricted to BS cells (Kanai and Edwards, 1999; Ku et al., 1976; Sinha and Kellogg, 1996).

Compartmentalization of photosynthesis proteins between M and BS cells is generated by both transcriptional and post-transcriptional mechanisms (Hibberd and Covshoff, 2010). Mechanisms underlying the M-cell specificity of PEPC and PPDK are probably the best studied. For example, in maize (Zea mays), accumulation of PEPC in M cells is related to specific changes in sequences within the promoter, DNA methylation, and also histone modifications. M-cell-specific accumulation of ZmPEPC transcripts in maize is correlated with demethylation of a PvuII site 3.1 kb upstream of the PEPC gene during greening (Langdale et al., 1991), but this is unlikely to be the only mechanism responsible for M-cell-specific accumulation of PEPC because when regions of only 1.7 or 0.6 kb upstream of ZmPEPC are fused to the uidA gene encoding β-glucoronidase, GUS staining was strong in developing and mature M cells (Kausch et al., 2001; Taniguchi et al., 2000). Sequences within this 0.6 kb region of ZmPEPC1 bind unidentified proteins termed PEPIb and PEPIc (Taniguchi et al., 2000). High trimethylation of histone H3K4 tails was detected within ZmPEPC1 in M cells but not BS cells (Danker et al., 2008), although the functional significance of this is not certain. In C4Flaveria, a cis-element that is necessary and sufficient for M-cell-specific expression of the GUS reporter has been identified (Akyildiz et al., 2007; Gowik et al., 2004). This element is contained within a 41 bp region known as mesophyll-enhancing module 1 (MEM1) from C4 species of Flaveria. MEM1 is located in a distal region of the promoter (−1566 to −2141), and contains a CACT tetranucleotide and a G→A substitution. Placing MEM1 into the C3Ppc promoter from Flaveria pringlei represses accumulation of the GUS reporter in BS cells rather than increasing it in M cells (Akyildiz et al., 2007). Analysing the amount of Ppc transcripts in Flaveria bidentis on the basis of GUS accumulation driven by the FbPpc promoter indicates that control is predominantly via transcription (Stockhaus et al., 1997). Although a mesophyll-enhancing module has been identified in the promoter to the FbCA3 gene from F. bidentis, its function has not been confirmed (Tanz et al., 2009).

Pyruvate,orthophosphate dikinase also accumulates preferentially in M cells, and a region between −301 and −296 from the translational start site is important for this. Gel-retardation assays showed that this region binds a protein termed PPD1 (Matsuoka and Numazawa, 1991). When the region between −370 and −76 in the maize promoter is fused to the rice (Oryza sativa) promoter, high expression is observed in M protoplasts (Matsuoka and Numazawa, 1991). In stable transformants of F. bidentis, use of the region between −1212 and +279 from the transcriptional start site fused to the uidA gene results in high levels of GUS in M cells, but lower amounts in BS cells (Rosche et al., 1998). In summary, these data are consistent with transcriptional regulation of both the PEPC and PPDK genes being important for M-cell specificity in maize and Flaveria.

It is known that genes encoding NAD-dependent malic enzyme (NAD-ME) and NADP-dependent malic enzyme (NADP-ME) from C3 species contain cis-elements sufficient for BS specificity in C4 leaves, and thus cell specificity in the C4 leaf can be mediated by changes in trans (Brown et al., 2011). The location of cis-elements within the coding region of these ME genes and the requirement for transcription implies post-transcriptional regulation (Brown et al., 2011). To determine the importance of regulation of C4 genes by elements within their transcripts, and the extent to which these elements are present in C3 orthologs, we used the Cleome genus, which is closely related to Arabidopsis thaliana and contains C3 and C4 plants (Brown et al., 2005; Marshall et al., 2007; Voznesenskaya et al., 2007). The phylogenetic proximity to A. thaliana facilitates annotation of likely gene function, and also provides relatively simple resources for functional analysis of orthologous genes in a C3 species. To investigate mechanisms underlying M-cell specificity in Cleome gynandra, we isolated cDNAs and genomic sequences encoding PPDK and CA4, and investigated regions of each gene that were sufficient for accumulation in M cells. By comparison with regulation of the orthologous PPDK and β-CA4 genes from C3A. thaliana, we were also able to investigate the extent to which mechanisms have evolved de novo in the C4PPDK and CA4 genes from C. gynandra.

Results

Cloning PPDK from C. gynandra

Compartmentation of photosynthesis proteins between M and BS cells in C4 species can be generated by both transcriptional and post-transcriptional mechanisms, and in most cases alterations in cis-elements have been linked to this compartmentation (Akyildiz et al., 2007; Hibberd and Covshoff, 2010; Westhoff and Gowik, 2004). To investigate mechanisms underlying the accumulation of proteins in M cells, we used a combination of degenerate PCR, 5′ and 3′ RACE, 454 sequencing and genome walking to isolate the gene encoding PPDK from C. gynandra, the most closely related C4 plant to the C3 model Arabidopsis thaliana (Brown et al., 2005; Marshall et al., 2007). This allowed us to assemble the sequence of the PPDK gene from C. gynandra (Figure 1a). To confirm that the CgPPDK gene is highly expressed, targeted to chloroplasts and therefore probably important for C4 photosynthesis, we used two approaches. First we created a fusion between the promoter sequence (including the 5′ UTR) isolated from C. gynandra and the uidA gene encoding β-glucuronidase (GUS) and investigated whether it contained enhancer elements compared with the orthologous promoter from the closely related C3 species A. thaliana (Figure S1A,B). This promoter drives expression of transcripts that encode the longer form of the PPDK protein targeted to chloroplasts (Parsley and Hibberd, 2006; Rosche and Westhoff, 1995; Sheen, 1991). Second we created a translational fusion between the CgPPDK coding sequence and GFP under the control of the CaMV 35S promoter, and used laser scanning confocal microscopy to determine the subcellular localization of the fusion protein. Staining stable transformants of A. thaliana indicated that use of the A. thaliana PPDK promoter resulted in accumulation of GUS in older senescing leaves (Figure 1b) (Taylor et al., 2010), but use of the promoter from C. gynandra extended accumulation of GUS into mature leaves (Figure 1c). Quantitative assays of GUS activity using methylumbelliferyl-d-glucuronide (MUG) indicated that use of the C. gynandra PPDK promoter increased GUS activity by at least 20-fold in mature leaves (Figure 1d). Laser scanning confocal microscopy showed that GFP localized to chloroplasts after microprojectile bombardment was used to deliver the CgPPDK::GFP fusion into leaves (Figure 1e–g), in contrast with the cytosolic localization of GFP when under the control of the CaMV 35S promoter alone (Figure S2). The increased accumulation of GUS driven by the PPDK promoter from C. gynandra compared with the promoter from A. thaliana and localization of the protein to chloroplasts are consistent with this CgPPDK gene being important in the C4 pathway.

Figure 1.

 The PPDK promoter from C4C. gynandra contains enhancer elements that are recognized in C3A. thaliana.
(a) Structure of the C4PPDK gene from C. gynandra. Exons are numbered and indicated by boxes, with black representing those that encode the mature protein, gray representing the chloroplast transit peptide, and white indicating 5′ or 3′ untranslated regions. The translational start and stop codons are annotated. The promoter directing expression of the chloroplastic protein is upstream of the first exon, while intron 2 contains a promoter that generates transcripts encoding the cytosolic PPDK.
(b, c) GUS staining of stable transformants of A. thaliana containing either the endogenous A. thaliana (b) or C. gynandra (c) chloroplastic PPDK promoter fused to the uidA gene. Scale bars are 0.5 cm.
(d) Quantitative analysis of GUS activity shows a 20-fold enhancement when the C. gynandra promoter was used. Boxes indicates upper and lower quartiles, median values are represented by horizontal lines within the boxes, and minimum and maximum MUG activities for each construct are indicated by black dots and whiskers.
(e–g) Confocal laser scanning microscopy after microprojectile bombardment of Arabidopsis thaliana leaves shows that the PPDK::GFP fusion protein localizes to chloroplasts. (e) Chlorophyll channel, (f) GFP channel, and (g) chlorophyll and GFP images merged. Scale bars = 10 μm.

Untranslated regions are sufficient for accumulation of PPDK in M cells

Hybridization in situ and quantitative RT-PCR after laser microdissection of M and BS cells showed that CgPPDK transcripts accumulate preferentially in M cells of C. gynandra (Figure 2a,b). To investigate the regions of the CgPPDK gene that generate M-cell specificity, we used microprojectile bombardment to deliver various parts of the CgPPDK gene fused to the uidA reporter into leaves of C. gynandra, and counted the number of M and BS cells accumulating GUS. When uidA was transcriptionally fused to the constitutive CaMV 35S promoter (Figure S1C), M and BS cells accumulated GUS (Figure 2c,d). Under our assay conditions, this control construct labeled more BS than M cells (Figure 2e). Despite being under the control of the CaMV 35S promoter, when the CgPPDK cDNA including 5′ and 3′ UTRs (Figure S1D) was incorporated into this construct, GUS accumulated preferentially in M cells (Figure 2f and Table S1). This indicates that neither the endogenous promoter nor the introns of CgPPDK are necessary for preferential accumulation in M cells of C. gynandra, and that elements contained within the transcript are sufficient for M-cell specificity.

Figure 2.

 PPDK accumulates in M cells of C. gynandra due to elements in both the 5′ and 3′ UTRs.
(a) In situ hybridization shows that transcripts encoding PPDK localize to M cells of C. gynandra leaves.
(b) Quantitative RT-PCR after extraction of RNA from either mesophyll (M) or bundle sheath (BS) cells shows preferential accumulation of PPDK transcripts in M cells.
(c–e) The constitutive CaMV 35S promoter drives expression of the uidA reporter in mesophyll (c) and bundle sheath (d) cells, and more BS cells than M cells showed staining under our assay conditions (e).
(f) Elements within either the 5′ or 3′ UTR of PPDK from C. gynandra are sufficient to generate M-cell specificity in leaves of C.  gynandra.
(g) Elements within either the 5′ or 3′ UTR of PPDK from A. thaliana are sufficient to generate M-cell specificity in leaves of C. gynandra.
Cells accumulating GUS were counted. Three or more biological replicates were performed. Error bars represent standard errors. P values were determined by a one-tailed t test.

To investigate regions of the CgPPDK transcript that are sufficient for preferential accumulation in M cells, we produced a series of truncations fused to uidA (Figure S1E–G). The results showed that the CgPPDK coding region was not necessary for preferential accumulation in M cells, and that either the 5′ UTR or 3′ UTR alone was sufficient (Figure 2f and Table S1). Compared with each UTR alone, the presence of both 5′ and 3′ UTRs slightly increased the proportion of M cells accumulating GUS.

The 5′ and 3′ UTRs of PPDK from A. thaliana generate M-cell specificity in C. gynandra

Comparison of the UTRs from the PPDK genes of A. thaliana and C. gynandra showed regions of sequence conservation between the two species (Figure S3). The 5′ and 3′ UTRs from A. thaliana PPDK were sufficient for preferential accumulation of GUS in M cells of C. gynandra (Figure 2g and Figure S1H–J), and, as with the orthologous regions from C. gynandra, each UTR alone was sufficient for this specificity. These data indicate that the PPDK gene from A. thaliana possesses cis-elements that are sufficient for M-cell specificity when placed into a closely related C4 leaf. We next investigated how the UTRs from CgPPDK behave in A. thaliana. This analysis showed that, in A. thaliana, neither the 5′ UTR nor the 3′ UTR of CgPPDK, or both together, led to a large alteration in the relatively constitutive accumulation of GUS driven by the CaMV 35S promoter (Figure 3a–d). However, quantitative assays of GUS activity using MUG showed that both UTRs from CgPPDK and the 5′ UTR alone slightly increased the activity of GUS in leaves of A. thaliana compared with the CaMV 35S:uidA controls (Figure 3e).

Figure 3.

 Untranslated regions of PPDK from Cleome gynandra show relatively constitutive accumulation of GUS in stable transformants of A. thaliana.
(a) The CaMV 35S::uidA fusion leads to accumulation of GUS in most leaves and cells of A. thaliana, particularly the vascular system.
(b–d) Including both the 5′ and 3′ UTRs of CgPPDK (b), the 5′ UTR alone (c) or the 3′ UTR alone (d) in the CaMV35S::uidA fusion had little impact on the spatial accumulation of GUS in A. thaliana.
(e) Quantitative assays show that both UTRs and the 5′ UTR increase expression compared with the CaMV 35S control.
Scale bars = 1 cm.

UTRs from C. gynandra or A. thaliana CA4 genes are sufficient for accumulation in M cells

Laser capture microscopy and quantitative RT-PCR designed against contigs derived from 454 sequencing (Bräutigam et al., 2011) detected strong and preferential accumulation of CgCA4 transcripts, encoding a carbonic anhydrase, and CgBASS2 transcripts, encoding a pyruvate transporter (Furumoto et al., 2011), in M cells (Figure S4). We used RACE and PCR on genomic DNA to identify full-length transcripts as well as the gene structure for each of these genes. In both cases, coding sequences were very similar to orthologs in A. thaliana, and the number of introns was conserved (Figure 4a). A fusion between GFP and the coding sequence of CgCA4 led to accumulation of GFP in close proximity to the plasma membrane (Figure 4b–d), and a fusion between CgBASS2 and GFP labeled the outside of chloroplasts (Figure 4e–g). In addition to the high level of their transcripts as detected by 454 sequencing (Bräutigam et al., 2011), the localization of CA4 and BASS2 to the plasma membrane and chloroplast envelopes, respectively, is consistent with their localization in A. thaliana and their roles in C4 photosynthesis.

Figure 4.

 CA accumulates in M cells of C. gynandra due to elements in either the 5′ or 3′ UTR.
(a) Gene structures of CA4 and BASS2 genes from C. gynandra and their A. thaliana orthologs. Exons are numbered and indicated by black boxes, with arrowheads representing transcriptional start sites. The translational start and stop codons are indicated.
(b–g) Confocal laser scanning microscopy after microprojectile bombardment of Arabidopsis thaliana leaves shows that CA4::GFP (b–d) localizes close to the plasma membrane and BASS2::GFP (e–g) localizes to the chloroplast envelope. (b, e) Chlorophyll channel, (c, f) GFP channel, and (d, g) chlorophyll and GFP images merged. Scale bars = 10 μm.
(h) Elements within either the 5′ or 3′ UTR of CA4 from C. gynandra are sufficient to generate M-cell specificity in leaves of C. gynandra.
(i) Elements within either the 5′ or 3′ UTR of CA4 from C. gynandra and equivalent regions from A. thaliana CA4 are sufficient to generate M-cell specificity in leaves of C. gynandra. Cells accumulating GUS were counted. Three or more biological replicates were performed. Error bars represent standard errors. P values were determined by a one-tailed t test.

Microprojectile bombardment showed that elements present in CgCA4 transcripts were sufficient for M-cell specificity, but elements in CgBASS2 were not (Figure 4h, Table S1 and Figure S1K–P). The coding region of CgCA4 was not necessary, but the 5′ and 3′ UTRs were sufficient for M accumulation (Figure 4h and Table S1). As with PPDK, either the 5′ or the 3′ UTR was sufficient for accumulation of GUS in M cells (Figure 4i). Taken together, these data indicate that elements in either the 5′ or the 3′ UTR of CgPPDK and CgCA4 are sufficient for preferential accumulation of PPDK in M cells of C. gynandra.

Alignments of the CA4 UTR sequences from C. gynandra and A. thaliana showed some similarities in their sequences (Figure S5). When introduced into C. gynandra leaves by microprojectile bombardment, the 5′ and 3′ UTRs from A. thaliana CA4 (Figure S1Q,R) were sufficient for preferential accumulation in the M cells (Figure 4i and Table S1), and, as with the orthologous regions from C. gynandra, each UTR alone was sufficient to confer such specificity (Figure 4i and Table S1). These data indicate that the CA4 gene from A. thaliana possesses cis-elements that are sufficient for M-cell specificity when placed into the C4 environment.

Discussion

Functional latency within multiple C3 genes co-opted into C4 photosynthesis

As elements in the UTRs of PPDK and CA4 from C3A. thaliana are sufficient for preferential accumulation in M cells of C. gynandra, alterations to cis-elements are not required to generate M-specific function in the C4 pathway. Combined with work on NAD-ME1 and 2 (Brown et al., 2011), these data indicate that four genes from the last common ancestor of A. thaliana and C. gynandra have been recruited into cell-specific accumulation in C. gynandra without changes to cis-elements. The fact that genes can be recruited to M- and BS-specific function in C4 leaves without alterations to cis-elements may have facilitated evolution of the pathway. As the CaMV 35S promoter drives expression in both M and BS cells of C4 leaves (Bansal et al., 1992; Brown et al., 2011; Patel et al., 2006), the reduction in the proportion of BS cells containing GUS when PPDK or CA4 UTRs are present implies that post-transcriptional regulation is probably important. However, we cannot rule out the possibility that transcriptional mechanisms contribute to cell specificity. The fact that the BS specificity of NAD-ME proteins important for C4 photosynthesis in C. gynandra is mediated by elements that are present in their coding regions, together with the data on CgPPDK and CgCA4, suggests that transcript sequence is important for M-cell and BS-cell specificity in this species.

Genes recruited into M- or BS-specific function in the C4 pathway are present in C3 plants, but prior to recruitment into C4 photosynthesis, tend to be expressed at relatively constitutive and low levels (Aubry et al., 2011; Brown et al., 2010; Taylor et al., 2010). When we expressed the uidA gene under the control of the CgPPDK promoter in A. thaliana, accumulation of GUS was increased in mature leaves compared with use of the endogenous AtPPDK promoter. This indicates that the C4PPDK promoter has acquired strong enhancer elements that extend expression from senescent leaves in the C3 species A. thaliana (Taylor et al., 2010) into mature photosynthetic leaves of C. gynandra. Furthermore, our data indicate that these enhancer elements are recognized by factors in A. thaliana, and thus trans factors responsible for enhanced expression in the C4 plant C. gynandra are already present in A. thaliana. The increased accumulation of GUS when fused to the 5′ UTR of CgPPDK alone indicates that at least part of this increased expression may be due to elements within the 5′ UTR. Our data are also consistent with the PPDK gene recruited into the C4 pathway from C. gynandra possessing the same dual promoter system that produces either cytosolic of chloroplastic isoforms of PPDK as reported for other species (Parsley and Hibberd, 2006; Rosche and Westhoff, 1995; Sheen, 1991).

Mechanisms underlying cell specificity in C4 leaves

Because a large number of genes in multiple plant lineages are co-opted into the pathway, it is not surprising that a number of mechanisms, ranging from transcriptional through post-transcriptional to translational, are known to control accumulation of proteins in either M or BS cells (Berry et al., 1987; Hibberd and Covshoff, 2010; Sheen, 1999). However, although post-transcriptional regulation (Patel et al., 2004, 2006) and translational elongation (Berry et al., 1986) have been implicated in the accumulation of protein in BS cells, there have been no reports that these mechanisms result in accumulation of proteins in M cells of C4 leaves. The regulation of CgPPDK and CgCA4 is therefore more similar to that of RbcS in the BS cells of F. bidentis (Patel et al., 2004, 2006) and NAD-ME in C. gynandra (Brown et al., 2011) than to that of PPDK or other M-cell-specific proteins studied to date (Akyildiz et al., 2007; Gowik et al., 2004; Kausch et al., 2001; Matsuoka and Numazawa, 1991).

The 5′ or 3′ UTR of CgPPDK or equivalent regions of CgCA4 are sufficient to generate M-cell specificity in C. gynandra. The most parsimonious explanation is that all four UTRs share a characteristic that is recognized by a common trans factor to generate M-cell specificity. In eukaryotes, high AU content of UTRs is important to induce instability and degradation of mRNAs (Chen and Shyu, 1995). However, we detected no clear difference in the AU content of the CgCA4 and CgPPDK UTRs compared with those from CgBASS2, which were not sufficient for M-cell specificity. This implies that AU content per se is not sufficient for M-cell specificity in C. gynandra. A polypyrimidine tract after the 5′ cap (Levy et al., 1991) and the downstream element (DST) motif (Newman et al., 1993) are also known to regulate transcript stability; however, neither of these elements were present in the UTRs of the PPDK and CA4 genes that we studied. In the C4 species F. bidentis, the 5′ and 3′ UTRs of transcripts encoding the small subunit of RuBisCO can specify accumulation of the GFP reporter in BS cells, and it was proposed that RNA-binding proteins mediate turnover of transcripts via AU-rich elements identified in these UTRs (Patel et al., 2006). This is also a possible mechanism in accumulation of CgPPDK and CgCA4 transcripts in M cells of C. gynandra, although a clear UUAUU motif (Patel et al., 2006) is not present. Irrespective of the exact mechanism, our data show that M-cell specificity in C4 leaves is under the control of elements in UTRs, and importantly that this has occurred for multiple genes. Our data also show that elements in the UTRs of PPDK and CA4 from C3A. thaliana are sufficient for accumulation in M cells of C. gynandra, and demonstrate that alterations to cis-elements are not required for M-cell-specific function in the C4 pathway. The results also suggest that post-transcriptional regulation is particularly important in maintaining the C4 pathway in leaves of this species.

Experimental Procedures

Generation of constructs

cDNA sequences were obtained after performing degenerate PCR followed by 5′ and 3 ‘RACE using a FirstChoice RLM-RACE kit (Ambion, http://www.ambion.com/). Coding regions were amplified from both cDNA and genomic DNA. Promoter regions were isolated by genome walking using a BD GenomeWalker Universal Kit (BD Biosciences, http://www.bdbiosciences.com/). Reporter gene constructs containing the gfp gene or the uidA gene were generated by cloning the UTRs and coding sequence (CDS) into a modified pENTR/D-TOPO vector (Invitrogen, http://www.invitrogen.com/) containing a gfp::uidA::nosT cassette (Brown et al., 2011). The 3′ UTRs were inserted between uidA and nosT. The 5′ UTRs were cloned together with the CgPPDK promoter or fused to the CaMV 35S promoter by PCR and inserted in front of the cassette. Spliced coding sequences amplified from cDNA were inserted between the 5′ UTRs and the gfp gene. The assembled constructs were used for microprojectile bombardment (Hibberd et al., 1998), and recombined with LR clonase (Invitrogen) into the Gateway destination vector pGWB1 (Nakagawa et al., 2007), transferred into Agrobacterium tumefaciens GV3101, and then into Arabidopsis thaliana.

Plant transformation

Transient expression of the constructs in C. gynandra was achieved by microprojectile bombardment as described by Brown et al. (2011). C. gynandra seeds from intact pods were germinated on moist filter paper in darkness at 30°C for 24 h, transferred onto MS medium with 1% w/v sucrose and 0.8% w/v agar (pH 5.8), and grown for 13 days at 22°C, 200 μmol m−2 sec−1 photon flux density with a 16 h photoperiod. Gold particles (1.0 μm diameter; Bio-Rad, http://www.bio-rad.com/) were coated with 0.5 pmol plasmid DNA of pENTRY/D-TOPO-based constructs. Two-week-old seedlings were positioned abaxial side up on filter paper moistened with half-strength MS medium, 6 cm below the stopping screen (Bio-Rad), and bombarded three times (at 1100 psi) using a Bio-Rad PDS-1000/He particle delivery system. After bombardment, seedlings were held vertically with the base of their stems in half-strength MS medium, and incubated for 48 h at 22°C, 200 μmol m−2 sec−1 photon flux density with a 16 h photoperiod, prior to GUS staining or laser scanning confocal microscopy.

Transgenic A. thaliana Col-0 plants were generated by floral dipping (Clough and Bent, 1998). Primary transformants were identified by selection on 50 μg ml−1 kanamycin for 7 days (Harrison et al., 2006), and grown for a further 3 weeks on a 3:1 mixture of compost and vermiculite at 20°C, 200 μmol m−2 sec−1 photon flux density with a 16 h photoperiod, prior to GUS staining.

RNA hybridization and GUS staining

RNA in situ hybridization was performed using digoxigenin-labeled RNA probes on 7 μm transverse sections of C. gynandra leaf primordia as previously described (Brown et al., 2011). Riboprobe template DNA of the PPDK coding sequence was amplified by PCR using primers with T3 and T7 polymerase binding sites, allowing generation of both antisense and sense probes.

For analysis of GUS accumulation in transgenic plants, tissue was placed in 0.1 m Na2HPO4 (Fisher Scientific, http://www.fishersci.com/), 0.5 mm K ferricyanide (Fisher Scientific), 0.5 mm K ferrocyanide (Sigma-Aldrich, http://www.sigmaaldrich.com/), 10 mm EDTA, 0.06% v/v Triton X-100 (Sigma-Aldrich) and 2 mm X-GlcA (5-bromo-4-chloro-3-indolyl beta-d-glucuronide) (Sigma-Aldrich), vacuum infiltrated (2 × 30 sec for A. thaliana, 3 × 3 min for C. gynandra), and incubated at 37°C for 20 h. Tissue was then fixed in 3:1 ethanol/acetic acid for 30 min at room temperature, and chlorophyll was cleared using 70% v/v ethanol at 37°C for 24 h and then 5% w/v sodium hydroxide at 37°C for 2 h. M and BS cells containing GUS were identified and counted using phase-contrast microscopy. In each experiment, at least 50 cells were counted, and at least three independent repetitions of the bombardment were performed for each construct. MUG activity assays were performed on 3-week-old seedlings. Samples were frozen in liquid nitrogen and then ground in 200 μl 50 mm NaH2PO4, 0.07%β-mercaptoethanol, 0.1% Triton X-100 prior to centrifugation at 13 000 g for 5 min. A dilution series was run for each sample to ensure linearity of the assay. The assay was stopped using 200 mm Na2CO3 after various time intervals in order to ensure linearity.

Confocal laser scanning microscopy

The subcellular localization of translational fusions of coding sequences of PPDK and CA4 to GFP was visualized by laser scanning confocal microscopy. Sections of transformed leaves were placed on microscope slides adaxial side up, a drop of water was added, and a cover slip was placed on top. The cells were viewed using a Leica TCS SP5 confocal microscope (http://www.leica.com/) with an excitation wavelength of 488 nm and a 63 × water-immersion objective. Images were collected using separate channels for GFP (500–545 nm), chlorophyll fluorescence (630–720 nm) and transmission microscopy, and these channels were overlaid using LAS AF software (Leica).

Quantitative PCR and laser capture microscopy

For quantitative RT-PCR, C. gynandra was grown a 3:1 compost/vermiculite mixture, at 22°C, 350 μmol m−2 sec−1 photon flux density, 65% relative humidity and a 16 h photoperiod. Mature leaf samples were collected from 4-week-old plants, cut into 2 mm strips, fixed in 3:1 ethanol/acetic acid for 16 h, and treated for 1 h each in 75, 85, 100, 100 and 100% v/v ethanol, followed by 1 h each in 75:25, 50:50, 25:75, 0:100, 0:100 and 0:100% v/v ethanol/Histoclear (Thermo Scientific, http://www.thermoscientific.com), all at 4°C (Kerk et al., 2003). Paraplast X-tra (Sigma-Aldrich) was added to the final Histoclear wash, and the samples were incubated at 58°C for 5 days, washing the sample with molten Paraplast X-tra every 12 h. The sections were positioned upright in Paraplast X-tra in Peel-A-Way S22 moulds (Agar Scientific, http://www.agarscientific.com/), allowed to cool and stored at 4°C. The sample blocks were sectioned using a rotary microtome (Jung) and Shandon MX35+ Premier microtome blades (Thermo Scientific) to 7 μm thickness. The sections were floated in sterile water on Probe-On Plus microscope slides (Fisher Scientific) to straighten them, the water was drained off, and the slides were dried at 42°C for 16 h. The slides were stored at 4°C and used within 4 weeks after sectioning. Prior to laser capture microdissection, slides were washed in Histoclear for 1 h and then air-dried for 15 min.

Laser capture microdissection was performed using a Veritas microdissection instrument model 704 and CapSure HS LCM caps (both MDS Analytical Technologies, http://www.moleculardevices.com/) according to the manufacturer’s instructions. Approximately 500 cells were captured per cap. Immediately after LCM, RNA was extracted directly from the caps using a PicoPure RNA extraction kit (MDS Analytical Technologies), and synthesized into cDNA using SuperScript II and random primers (Promega, http://www.promega.com/). Quantitative RT-PCR was performed exactly as described by Bräutigam et al. (2011) using a DNA Engine thermal cycler, a Chromo4 real-time detector (Bio-Rad, http://www.bio-rad.com) and SYBR Green JumpStart Taq Ready Mix (Sigma-Aldrich). The PCR program comprised initial denaturation at 94°C for 2 min, followed by 40 cycles of 94°C for 20 sec, 60°C for 30 sec, 72°C for 30 sec and 75°C for 5 sec. CT values were generated for three technical replicates for three independent biological replicates per cell type. Standard errors were calculated from inline image values of each combination of biological replicates.

Acknowledgements

We thank the Frank Smart fund for a studentship to K.K., the Biotechnology and Biology Sciences Research Council for funding, and Samart Wanchana (C4 Rice Centre, International Rice Research Institute) for advice.

Ancillary