Design of chimeric expression elements that confer high-level gene activity in chromoplasts


For correspondence (e-mail


Non-green plastids, such as chromoplasts, generally have much lower activity of gene expression than chloroplasts in photosynthetically active tissues. Suppression of plastid genes in non-green tissues occurs through a complex interplay of transcriptional and translational control, with the contribution of regulation of transcript abundance versus translational activity being highly variable between genes. Here, we have investigated whether the low expression of the plastid genome in chromoplasts results from inherent limitations in gene expression capacity, or can be overcome by designing appropriate combinations of promoters and translation initiation signals in the 5′ untranslated region (5′–UTR). We constructed chimeric expression elements that combine promoters and 5′–UTRs from plastid genes, which are suppressed during chloroplast-to-chromoplast conversion in Solanum lycopersicum (tomato) fruit ripening, either just at the translational level or just at the level of mRNA accumulation. These chimeric expression elements were introduced into the tomato plastid genome by stable chloroplast transformation. We report the identification of promoter-UTR combinations that confer high-level gene expression in chromoplasts of ripe tomato fruits, resulting in the accumulation of reporter protein GFP to up to 1% of total cellular protein. Our work demonstrates that non-green plastids are capable of expressing genes to high levels. Moreover, the chimeric cis-elements for chromoplasts developed here are widely applicable in basic and applied research using transplastomic methods.


Gene expression from the plastid (chloroplast) genome is regulated through a complex interplay of regulatory processes acting at the transcriptional, translational and post-translational levels. The transcriptional apparatus of plastids comprises three RNA polymerases, one of which is homologous with the multisubunit RNA polymerase of bacteria, whereas the other two resemble single subunit polymerases of bacteriophages (Hedtke et al., 2002; Liere and Börner, 2007). Chloroplast primary transcripts are subject to an intricate diversity of RNA maturation processes, including terminal endo- and exoribonucleolytic trimming, intercistronic cleavage of polycistronic precursors into monocistronic or oligocistronic units, group–I and group–II intron splicing and RNA editing (Hayes et al., 1996; Bock, 2000; Stern et al., 2010; Barkan, 2011). Protein biosynthesis occurs on 70S ribosomes that, in addition to the conserved bacterial-type RNA and protein subunits, also contain a few plastid-specific components that may have been newly acquired during evolution (Tiller et al., 2012). The (nucleus-encoded) trans-acting factors regulating chloroplast transcription, RNA processing, transcript stability and translation often act in a highly mRNA-specific fashion, suggesting that a large fraction of nuclear genes is dedicated to the control of plastid gene expression (Kim and Mullet, 1995; Cheng et al., 1997; Boudreau et al., 2000; Barneche et al., 2006; Zghidi et al., 2007; Hammani et al., 2009; Boulouis et al., 2011; Fujii and Small, 2011; Prikryl et al., 2011).

In vitro transcription and translation systems using chloroplast extracts have contributed greatly to the characterization of cis-acting elements (promoters and 5′ untranslated regions, 5′–UTRs) involved in the expression of chloroplast genes (e.g. Gruissem and Zurawski, 1985a,b; Sun et al., 1989; Chen et al., 1990; Wada et al., 1994; Hirose and Sugiura, 1996, 2004; Hirose et al., 1998). More recently, the availability of a workable chloroplast transformation system for Nicotiana tabacum (tobacco) plants has also facilitated the investigation of chloroplast gene expression in vivo. Reporter gene fusions have identified promoter elements and translation signals in the 5′–UTR of plastid mRNAs that are involved in the transcriptional and translational regulation of chloroplast gene expression (e.g. Staub and Maliga, 1994; Allison and Maliga, 1995; Sriraman et al., 1998; Eibl et al., 1999; Herz et al., 2005).

Whereas gene expression in chloroplasts has been intensely studied both in vitro and in vivo, much less is known about gene expression in non-green plastid types. Recent genome-wide analyses of RNA accumulation and translational activity in chromoplasts of Solanum lycopersicum (tomato) fruits (Kahlau and Bock, 2008) and amyloplasts of Solanum tuberosum (potato) tubers (Valkov et al., 2009) has revealed that, in non-green tissues and organs, a strong general downregulation of the expression of the plastid genome occurs. Because of the limited availability of plastid transformation systems for higher plants other than tobacco, and the much more laborious and time-consuming procedures involved in generating transplastomic plants in the few other species currently amenable to plastid genome manipulation (Maliga, 2004; Bock, 2007; Koop et al., 2007; Maliga and Bock, 2011), no systematic in vivo analysis of gene expression in non-green plastids has been possible so far. Only very recently, two studies have begun to characterize expression elements in the colorless plastids of tobacco roots and potato tubers (Valkov et al., 2011; Zhang et al., 2012). Testing of reporter gene fusions revealed that the expression elements tested so far conferred only low levels of reporter protein accumulation, with 0.02% of the total soluble protein (TSP) being the maximum in tubers (Valkov et al., 2011), although tissue culture-grown (chlorophyll-containing) microtubers can accumulate substantially higher levels of recombinant protein (Segretin et al., 2012). This is consistent with the very low expression levels of the plastid genome in amyloplasts of potatoes (Valkov et al., 2009).

To date, no characterization of cis-elements determining plastid gene expression in any other non-green plastid type has been undertaken. A recent plastid genome-wide survey of transcription and translation during fruit development and chloroplast-to-chromoplast conversion in tomato revealed striking gene-specific differences in the regulation of gene expression (Kahlau and Bock, 2008). Whereas many genes were found to be strongly downregulated during chromoplast development at both the transcriptional and the translational levels (e.g. psbC, psbD, psbN, psbZ, atpF and atpH), a few genes are predominantly downregulated at only one level (e.g. psbA and ycf2 at the translational level), or are not significantly downregulated at all (Kahlau and Bock, 2008). The latter group comprises only a small number of weakly expressed genes, such as accD, the only plastid gene involved in lipid metabolism (encoding a subunit of the acetyl-CoA carboxylase), and rpoB, encoding a subunit of the bacterial-type plastid RNA polymerase (Igloi and Kössel, 1992; Hess and Börner, 1999). The model derived from these findings is that, in chromoplasts: (i) all genes that are highly expressed in chloroplasts (including all photosynthesis-related genes) are strongly downregulated; and (ii) all other genes are either downregulated as well or continue to be expressed at a low level. This model is consistent with most plastid-encoded proteins being undetectable in the chromoplast proteomes of tomatoes (Barsan et al., 2010) and fruits of Capsicum spp. (bell pepper; Siddique et al., 2006). The strongly downregulated expression of the plastid genome, together with the very low abundance of plastid-encoded proteins in chromoplasts raises interesting questions about possible general limitations of the gene expression capacity of chromoplasts, and has cast doubt on the possibility to express plastid (trans)genes to high levels in chromoplast-containing tissues, such as carotenoid-accumulating fruits, flowers and tap roots.

To gain insights into the gene expression capacity of chromoplasts, we decided to take a transplastomic approach that aimed at determining the maximum expression level of plastid genes in tomato fruits. To this end, the reporter gene gfp (encoding the green fluorescent protein, GFP) was combined with different 5′ expression signals comprising a promoter (conferring transcriptional activity) and a 5′–UTR known to harbor the cis-acting elements that control translational activity (Staub and Maliga, 1994; Zerges et al., 1997; Hirose et al., 1998; Eibl et al., 1999; Hirose and Suguira, 2004; Scharff et al., 2011). We report here the construction and in vivo testing of hybrid expression elements combining a promoter that is highly active in chromoplasts with a 5′–UTR that confers strong translational activity. The best-performing promoter-UTR combinations triggered GFP accumulation to up to 1% of the total protein of the ripe tomato fruit, which is close to the GFP accumulation levels achievable in chloroplasts of green leaves. Our data demonstrate that there is no strong inherent limitation to high-level gene expression in chromoplasts, and show that chimeric expression elements can trigger significantly higher gene activity in chromoplasts than native elements in the plastid genome.


Design of chimeric promoter-UTR combinations for expression in chromoplasts

Based on previously obtained transcript profiling and polysome profiling data sets for tomato leaves and fruits of different ripening stages (Kahlau and Bock, 2008), we sought to design plastid expression elements that confer high activity in ripe tomato fruits, and define the upper limit of plastid gene activity in chromoplasts. To this end, four different strategies were pursued. First, the promoter and 5′–UTR from accD, the only gene for which expression increases during chloroplast-to-chromoplast conversion in tomato fruit ripening (Kahlau and Bock, 2008), were tested. Second, the promoter from the plastid rRNA operon (residing upstream of the 16S rRNA gene, rrn16) was combined with the 5′–UTR from accD known to confer active translation in chromoplasts. The rrn16 promoter is one of the strongest promoters in chloroplasts (with rRNA levels accounting for more than 90% of the total RNA) and, although downregulated in non-photosynthetic plastid types, still retains significant activity in non-green tissues (Zhang et al., 2012). Third, the promoter of the psbA gene was combined with translation initiation signals active in chromoplasts. This promoter is unique in that it is one of the strongest promoters in chloroplasts (driving the expression of the D1 protein of photosystem II that is subject to constant replacement synthesis; Herz et al., 2005), but yet its activity is not downregulated during tomato fruit development. Instead, the downregulation of psbA expression occurs exclusively at the translational level (Kahlau and Bock, 2008). To combine the psbA promoter with translation signals active in chromoplasts, the 5′–UTR from rpoB was chosen, a gene that had been found to be unaffected by the translational downregulation of plastid gene expression during tomato fruit ripening (Kahlau and Bock, 2008). Fourth, the two strong promoters (from the rrn16 and psbA genes) were fused to the strongest known synthetic translation initiation signals that are derived from the Shine–Dalgarno sequence of gene 10 of bacteriophage T7 (T7g10; Ye et al., 2001; Kuroda and Maliga, 2001; Oey et al., 2009a,b). Thus, initially the following expression elements were constructed: (i) the accD promoter with the accD 5′–UTR (subsequently referred to as accD::accD; Table 1); (ii) the rrn16 promoter combined with the accD 5′–UTR (rrn16::accD); (iii) the rrn16 promoter combined with the T7g10 5′–UTR (rrn16::T7g10); (iv) the psbA promoter combined with the psbA 5′–UTR (psbA::psbA); (v) the psbA promoter combined with the rpoB 5′–UTR (psbA::rpoB); and (vi) the psbA promoter combined with the T7g10 5′–UTR (psbA::T7g10).

Table 1. Foreign protein accumulation in leaves and fruits (of three different ripening stages: green, turning and red) of transplastomic tomato plants expressing gfp from different chimeric expression cassettes
ConstructPromoter5′–UTR3′–UTRTransplastomic lineGFP accumulation (% TP)
  1. –, tissue not available; n.d., not detectable; TP, total cellular protein.

pRC18 accD accD rbcL Sl-pRC18 (accD::accD)n.d.
pRC20 rrn16 accD rbcL Sl-pRC20 (rrn16::accD)n.d.n.d.n.d.n.d.
pRC21 psbA psbA rbcL Sl-pRC21 (psbA::psbA)
pRC22 psbA T7g10 rbcL Sl-pRC22 (psbA::T7g10)2111
pRC23 psbA rpoB rbcL Sl-pRC23 (psbA::rpoB)0.04
pRC24 rrn16 T7g10 rbcL Sl-pRC24 (rrn16::T7g10)1.510.750.75
pRC30 accD accD* rbcL Sl-pRC30 (accD::accD*)n.d.
pRC31 psbA accD* rbcL Sl-pRC31 (psbA::accD*)
pRC32 rrn16 accD* rbcL Sl-pRC32 (rrn16::accD*)

The expression elements were fused to the reporter gene gfp (Figure 1) to be able to visually detect gene activity via GFP fluorescence. In addition, the availability of sensitive anti-GFP antibodies facilitates the easy quantification of reporter gene expression in transplastomic plants (Apel et al., 2010; Drechsel and Bock, 2010). All chimeric GFP expression cassettes were finally incorporated into a plastid transformation vector harboring the aadA selectable marker gene for chloroplast transformation (Figure 1a,b; Svab and Maliga, 1993; Ruf et al., 2001).

Figure 1.

Introduction of gfp constructs driven by chimeric promoter-UTR elements for chromoplast expression into the plastid genome of tomato plants. (a) Physical map of the targeting region in the tomato plastid genome. Genes above the line are transcribed from the left to the right; genes below the line are transcribed in the opposite direction. The transgenes are targeted to the intergenic region between trnfM and trnG. The location of the RFLP probe is indicated with a black bar. (b) Map of plastid transformation vectors of the pRC series. The vectors carry two transgene cassettes: a gfp cassette driven by the chimeric expression elements (comprising promoter and 5′–UTR) designed for high-level expression in chromoplasts and the aadA selectable marker conferring spectinomycin resistance. The location of the gfp probe used for northern blotting is shown with a black bar. The expected sizes of DNA fragments in restriction fragment length polymorphism (RFLP) analyses with the restriction enzyme BglII are indicated in (a) and (b). (c) RFLP analysis of transplastomic tobacco lines. Total plant DNA was digested with BglII and hybridized to a radiolabelled probe detecting the region of the plastid genome that flanks the transgene insertion site. The absence of the 3.5–kb hybridization signal diagnostic of the wild-type genome indicates the homoplasmy of all transplastomic lines. The additional weakly hybridizing band in the Sl-pRC23 (psbA::rpoB) line arises from flip-flop recombination between the rpoB 5′–UTR driving the gfp transgene and the endogenous copy of the 5′–UTR upstream of the rpoB gene in the plastid genome (Rogalski et al., 2006, 2008a,b). The band is absent from all other transplastomic lines, because the 5′–UTRs from all other genes used in this study are much shorter than the rpoB 5′–UTR. For plasmid names and combinations of expression elements, see Table 1. Sl–wt, wild type.

Transplastomic tomato plants were generated using the biolistic protocol and selection for spectinomycin resistance (Ruf et al., 2001; Wurbs et al., 2007). Primary transplastomic lines were subjected to two additional cycles of regeneration under spectinomycin selection to eliminate residual copies of the wild-type plastid genome and isolate homoplasmic tissue. Integration of the transgene cassettes into the plastid genome by homologous recombination and homoplasmy of the transplastomic tomato lines were confirmed by DNA gel blot analyses (Figure 1c).

Phenotypic analysis of transplastomic tomato plants

To assess the phenotypes of the tomato lines expressing GFP from the various promoter-UTR combinations, homoplasmic transplastomic plants were transferred to soil and grown to maturity under standard glasshouse conditions (Figure 2). Surprisingly, the expression of some of the GFP cassettes caused mutant phenotypes. The Sl-pRC18 (accD::accD) transplastomic lines expressing GFP from the accD::accD cassette were most severely affected, with pale plants incapable of sustained autotrophic growth after transfer from tissue culture to soil (Figure 2). Likewise, the Sl-pRC23 lines expressing GFP from psbA::rpoB, although growing autotrophically, showed severe pigment deficiency and strongly reduced growth. The Sl-pRC24 (rrn16::T7G10) lines displayed a mild pigment-deficient phenotype, but otherwise grew well and showed normal fruit set and development. All other transplastomic lines, Sl-pRC20 (rrn16::accD), Sl-pRC21 (psbA::psbA) and Sl-pRC22 (psbA::T7g10), grew normally and had a wild type-like phenotype.

Figure 2.

Phenotypes of transplastomic tomato lines. Pictures were taken 8–10 weeks after the transfer of plants to soil, with the exception of the Sl-pRC18 (accD::accD) line, which had to be photographed after 4 weeks (because its strong mutant phenotype did not allow for longer survival times under autotrophic growth conditions). Note that the Sl-pRC30 (accD::accD*) and Sl-pRC23 (psbA::rpoB) plants are also strongly retarded in growth and display a pigment-deficient phenotype. No fruits could be obtained from these plants. Sl-pRC31 (psbA::accD*) plants and, to a lesser extent, Sl-pRC24 (rrn16::T7g10) plants showed light-green leaves and reduced fruit production compared with the wild type-like lines Sl-pRC20 (rrn16::accD), Sl-pRC21 (psbA::psbA), Sl-pRC22 (psbA::T7g10) and Sl-pRC32 (rrn16::accD*). Scale bars: 10 cm.

To exclude unwanted recombinations (Rogalski et al., 2006, 2008a,b; Gray et al., 2009) or other mutations in the plastid genome as a possible source of the mutant phenotypes, all GFP expression cassettes were PCR amplified from extracted plant DNA and resequenced. No mutations or recombination events were identified, confirming that the mutant phenotypes are directly caused by transgene expression. We, therefore, suspect that the mutant phenotypes resulting from the expression of some of our GFP cassettes result from competition between duplicated expression elements for trans-acting factors, such as mRNA-specific transcription factors, RNA processing factors acting at the 5′–UTR or translation factors. In this scenario, the endogenous copy of the promoter and/or 5′–UTR and the copy driving the gfp transgene compete for a limiting protein factor, resulting in depletion of this factor from the endogenous plastid gene. Previous work has shown that this trans-factor competition can lead to mutant phenotypes in transplastomic plants (Kuroda and Maliga, 2002).

Because of their severe mutant phenotypes and their inability to produce fruits and seeds, the Sl-pRC18 (accD::accD) and Sl-pRC23 (psbA::rpoB) transplastomic lines were not further analyzed. For all other transplastomic lines, fruits were obtained, harvested at different stages of ripening and included in all subsequent molecular analyses.

Identification of a functional accD 5′–UTR

All transplastomic lines were subjected to an initial test for GFP fluorescence using material from tissue culture. In these analyses, we noted that none of the constructs containing the accD 5′–UTR showed any detectable fluorescence in transplastomic callus tissue or regenerating leaves. This was surprising and prompted us to reinvestigate the sequence of the accD 5′–UTR. The length of the accD 5′–UTR used in our constructs and in previous studies (Hajdukiewicz et al., 1997; Hirata et al., 2004) was based on the accD start codon annotation derived from the tobacco sequence (Shinozaki et al., 1986). When comparing the sequences from tobacco and tomato (Kahlau et al., 2006) with accD sequences outside the Solanaceous family, we noticed that the assumed accD start codon was not conserved in all species (Figure 3a,b). Although an ATG was present in most species, it did not always give rise to a contiguous reading frame. Interestingly, we identified a downstream ATG that was much better conserved (Figure 3a,b). This raised the possibility that the previously assumed accD start codon does not represent the correct translation initiation site used in vivo. The immediate consequence from this would be that the accD 5′–UTR used in our expression cassettes was 51 nucleotides too short, possibly accounting for the lack of GFP expression in our transplastomic lines.

Figure 3.

Identification of a functional accD 5′–UTR. (a) Alignment of the nucleotide sequences surrounding the two possible ATG start codons (boxed) of accD from tomato (Sl), tobacco (Nt) and Arabidopsis thaliana (At). Asterisks indicate nucleotides that are identical in all three sequences. (b) Alignment of the N–terminal amino acid sequences of the AccD proteins from tomato, tobacco and Arabidopsis. Asterisks denote residues identical in all three species; colons indicate conserved substitutions. The two possible initiator methionines are boxed. Note that, in Arabidopsis, the first ATG does not give rise to a contiguous accD reading frame. (c) Analysis of gfp mRNA accumulation in leaves and fruits of transplastomic tomato plants. The gfp probe detects three major bands. The ~1.1–kb band represents the mature monocistronic gfp mRNA. The two larger bands represent stable read-through transcripts that have been observed before in plastid transformation experiments using pKP9-derived vectors (Zhou et al., 2008; Apel and Bock, 2009; Oey et al., 2009a,b). Samples of 3 μg of total RNA were loaded in all lanes. To confirm equal loading, the 18S rRNA-containing region of the ethidium bromide-stained gel is shown below the blot.

To test this possibility experimentally, we generated three plastid transformation vectors in which the GFP expression cassettes harbored the longer version of the accD 5′–UTR (subsequently referred to as accD*). The accD* 5′–UTR was combined with the accD promoter (accD::accD* in vector pRC30), the psbA promoter (psbA::accD* in vector pRC31) and the rrn16 promoter (rrn16::accD* in vector pRC32; Table 1). Transplastomic tomato lines were produced with these three constructs, purified to homoplasmy (Figure 1c) and subjected to an initial test for GFP fluorescence, as well as a preliminary analysis of gfp mRNA and protein accumulation (Figures 3c and S1). Readily detectable GFP fluorescence, as well as higher mRNA accumulation and strong accumulation of the green fluorescent protein in transplastomic lines generated with constructs harboring the accD* 5′–UTR (contrasting the lack of detectable GFP accumulation in lines harboring the accD 5′–UTR; Figure S1), strongly suggested that only the second ATG is used as the translation initiation codon in vivo. We therefore conclude that the 5′–UTR of the accD mRNA is 51 nucleotides longer than previously assumed.

mRNA accumulation levels conferred by chimeric promoter-UTR combinations in transplastomic tomato plants

Among the three additionally produced sets of transplastomic lines (expressing GFP from cassettes containing the accD* 5′–UTR), the Sl-pRC30 (accD::accD*) plants displayed a strong mutant phenotype and were unable to produce fruits, whereas the Sl-pRC31 (psbA::accD*) plants showed a mild pigment-deficient phenotype and the Sl-pRC32 (rrn16::accD*) plants were indistinguishable from the wild type (Figure 2). These phenotypes were consistent with those observed before in transplastomic lines expressing GFP from the short version of the accD 5′–UTR, in that the Sl-pRC30 (accD::accD*) plants were similarly pale as the Sl-pRC18 (accD::accD) plants (although the Sl-pRC18 plants grew even slower than the Sl-pRC30 plants), whereas both the Sl-pRC32 (rrn16::accD*) and Sl-pRC20 (rrn16::accD) lines had a wild type-like phenotype (Figure 2).

To determine the expression levels conferred by the various GFP cassettes in our transplastomic tomato lines, we first examined gfp mRNA accumulation by performing a series of northern blot experiments (Figures 4 and S2). For all lines that developed fruits, mRNA accumulation levels were comparatively analyzed in leaves and fruits of three different ripening stages: (i) green fruits mainly containing chloroplasts; (ii) turning fruits, representing the transition from chloroplasts to chromoplasts; and (iii) ripe red fruits containing chromoplasts. The constructs harboring the synthetic translation signals derived from gene 10 of bacteriophage T7 (T7g10) were found to accumulate the gfp mRNA to the highest levels in leaves. Surprisingly, mRNA abundance appeared to be more dependent on the 5′–UTR than on the promoter. In combination with the T7g10 5′–UTR, both the psbA promoter and the rrn16 promoter gave rise to similarly high mRNA accumulation in leaves, whereas both promoters conferred much lower transcript levels when combined with other 5′–UTRs (from accD or psbA; Figure 4). This may indicate that the translation rate has a significant impact on RNA accumulation, in that highly active translation (as conferred by the T7g10 5′–UTR; Figures 3c and 4; Ye et al., 2001; Kuroda and Maliga, 2001; Oey et al., 2009a) protects the mRNA from ribonucleolytic degradation. This is reminiscent of the situation in bacteria, where transcript coverage with ribosomes can strongly reduce the rate of mRNA decay (Braun et al., 1998; Deana and Belasco, 2005).

Figure 4.

Comparison of gfp mRNA accumulation levels in transplastomic tomato lines. For each line, leaves and three different fruit ripening stages (green, turning and red) were investigated. Samples of 3 μg of total cellular RNA were electrophoretically separated, blotted and hybridized to a gfp-specific probe. Sizes of marker bands are indicated in kb. For major transcript species, see Figure 3c. RNA samples from line Sl-pRC22 (psbA::T7g10) were used as a reference to adjust exposure time and facilitate comparison of gfp mRNA accumulation between blots. To confirm equal loading, the 18S rRNA-containing region of the ethidium bromide-stained gel prior to blotting is shown below each blot.

The maximum mRNA accumulation in fruits was significantly lower than in leaves of the lines expressing GFP from the T7g10 5′–UTR, but in the same range as in leaves of the other transplastomic lines. During the fruit-ripening process, the most pronounced changes were observed in lines with the accD* 5′–UTR. In these plants (Sl-pRC32 and Sl-pRC31), gfp mRNA levels increased from green to red fruits (Figure 4), consistent with increased expression of the endogenous accD gene during fruit ripening in tomato (Kahlau and Bock, 2008). Unexpectedly, this increase in transcript levels appears to be caused by the 5′–UTR rather than the promoter (Figures 3c and 4), again pointing to a possible role of the translation rate in determining transcript abundance. Alternatively, transcript-stabilizing proteins that bind to the 5′–UTR could have an impact on mRNA accumulation.

Reporter protein accumulation in transplastomic tomato plants

We next examined reporter gene expression in transplastomic tomato plants at the protein level. To this end, we compared GFP accumulation in leaves and fruits representing different ripening stages. To facilitate quantitative assessment, we compared the protein accumulation in different tissues of each individual transplastomic line (Figure 5), and also performed a side-by-side comparison of the same tissue from different transplastomic lines (Figure 6).

Figure 5.

Protein accumulation in transplastomic tomato plants expressing gfp from different expression cassettes. (a) GFP accumulation in leaves and fruits of Sl-pRC22 plants expressing gfp from psbA::T7g10. (b) GFP accumulation in leaves and fruits of Sl-pRC24 plants expressing gfp from rrn16::T7g10. (c) GFP accumulation in leaves and fruits of Sl-pRC32 plants expressing gfp from rrn16::accD*. (d) GFP accumulation in leaves and fruits of Sl-pRC21 plants expressing gfp from psbA::psbA. (e) GFP accumulation in leaves and fruits of Sl-pRC31 plants expressing gfp from psbA::accD*. Note that the total quantity of protein loaded differs between the different transplastomic lines. (The loading was adjusted to be comparable with the same dilution series of recombinant GFP, rGFP). Dilutions of total protein samples from leaves and fruits were loaded to be able to quantify GFP accumulation by comparison with the dilution series of rGFP.

Figure 6.

Comparison of GFP accumulation levels in leaves and fruits of transplastomic tomato plants. The loaded quantities of total protein from the individual transplastomic lines are indicated. For semiquantitative assessment of foreign protein accumulation, a dilution series of recombinant GFP (rGFP) was also included. Protein loading for the various transplastomic lines was adjusted to be comparable with the rGFP standard. (a) Western blot analysis to compare GFP accumulation in leaves. (b) Western blot analyses comparing GFP accumulation levels in the transplastomic tomato lines over three stages of fruit ripening. Line Sl-pRC22 (psbA::T7g10) was included in both blots as a reference to adjust exposure time and facilitate comparison of protein accumulation levels between blots.

Green fluorescent protein (GFP) accumulation was determined using an anti-GFP antibody and a dilution series of purified recombinant GFP as standard for quantification. When the different tissue types were compared within each plant, the three 5′–UTRs analyzed exhibited different patterns of GFP accumulation. Consistent with the increasing levels of accD gene expression during fruit ripening (Kahlau and Bock, 2008), GFP expression in fruits of Sl-pRC32 plants (expressing gfp from rrn16::accD*) and Sl-pRC31 plants (expressing gfp from psbA::accD*) was higher than in leaves and, moreover, increased during fruit ripening (Figure 5c,e; Table 1). In contrast, GFP accumulation in Sl-pRC21 (psbA::psbA) plants was lower in fruits than in leaves, and decreased during fruit ripening, consistent with the strong translational downregulation of psbA gene expression during chloroplast-to-chromoplast conversion (Kahlau and Bock, 2008; Figure 5d). The synthetic translation signals of the T7g10 5′–UTR showed an intermediate behavior, in that GFP expression in Sl-pRC22 plants (expressing gfp from psbA::T7g10) and Sl-pRC24 plants (expressing gfp from rrn16::T7g10) was very similar in all tissues investigated (Figure 5a,b).

When the different transplastomic lines were compared, analysis of leaves revealed strong differences in GFP accumulation (Figure 6a). In agreement with the very high RNA accumulation levels, GFP levels in leaves of Sl-pRC22 and Sl-pRC24 (expressing gfp from the T7g10 5′–UTR) plants were at least 10-fold higher than in all other lines. This finding is also well in line with the T7g10 5′–UTR representing the strongest translation initiation signal identified to date in leaf chloroplasts (Kuroda and Maliga, 2001; Ye et al., 2001; Oey et al., 2009a). The two transplastomic lines expressing gfp from the T7g10 5′–UTR also showed the highest protein accumulation levels in fruits (Figure 6b; Table 1). In ripe red fruits, the difference compared with the best-performing construct among the remaining lines (Sl-pRC32 expressing gfp from rrn16::accD*) was only approximately fourfold, and thus less pronounced than in green leaves. This is mainly caused by the increase in GFP accumulation during fruit ripening conferred by the 5′–UTR from accD (Figure 5c). The highest protein accumulation in chromoplasts of ripe tomatoes was 1% of the total cellular protein. This is only 50% lower than the highest level obtained in leaf chloroplasts (Table 1) and, to our knowledge, represents the highest expression level achieved in non-green plastids to date.

Finally, we wanted to confirm that the successful GFP expression in chromoplasts also provides a fluorescence label for fruit plastids, thereby facilitating the in vivo observation of the peculiar morphological features of chromoplasts (Pyke and Howells, 2002; Waters et al., 2004). GFP fluorescence was readily detectable in both chloroplasts of green fruits and chromoplasts of ripe red tomatoes by confocal laser-scanning microscopy (Figure 7). This is consistent with the protein accumulation data (Figures 5 and 6; Table 1), and confirms the potential of fruit plastids to accumulate significant levels of plastid genome-expressed recombinant proteins.

Figure 7.

Detection of GFP accumulation in leaves and fruits of transplastomic tomato plants by confocal laser-scanning microscopy. Cross sections of leaves, and green and red fruits were analyzed. Three transplastomic lines, Sl-pRC24 (rrn16::T7g10), Sl-pRC32 (rrn16::accD*) and Sl-pRC31 (psbA::accD*) are compared with the wild type (Sl–wt). As expected, the red chlorophyll fluorescence is observable in leaves and green fruits, but not in ripe red fruits. Note that GFP fluorescence is also detectable in leaves and fruits of the weakly expressing transplastomic line Sl-pRC31 (psbA::accD*). Scale bars: 10 μm.


Very little is known about the regulation of plastid gene expression in non-green plastid types. Until recently, in vivo analysis of gene expression in chromoplasts was hampered by the lack of a workable plastid transformation technology for a plant species with chromoplast-containing organs. With the development of a transformation protocol for tomato plastids (Ruf et al., 2001) and recent improvements in transformation efficiency (Wurbs et al., 2007; Apel and Bock, 2009; Ruf and Bock, 2012), it has become possible to produce transplastomic tomato lines on a somewhat larger scale, although the method remains significantly more laborious and time consuming than plastid transformation in the routinely used tobacco system (Maliga, 2004; Maliga and Bock, 2011). In this work, we have used plastid transformation in tomato plants to test combinations of promoters and 5′–UTRs for their potential to confer active gene expression in chromoplasts. We identified chimeric expression elements that trigger high-level protein accumulation in chromoplasts of ripe tomato fruits. The highest expression levels obtained for the reporter protein GFP reached 1% of the total cellular protein of the fruit, which is comparable with the maximum expression level of GFP obtained in this study in tomato leaves and in earlier work in tobacco leaves (from 2% of total cellular protein to 5.5% of total soluble protein; Table 1; Reed et al., 2001; Newell et al., 2003). This demonstrates that, although the expression of the plastid genome is strongly downregulated in non-green tissues (Kahlau and Bock, 2008; Valkov et al., 2009), there is no inherent limitation to high-level gene expression in chromoplasts. The combination of a strong promoter with a strong 5′–UTR, not prone to severe developmental downregulation during fruit development and/or chloroplast-to-chromoplast conversion, represents a viable strategy to maximize plastid gene activity in chromoplasts. The best-performing expression elements for tomato fruits identified here combine the strong psbA promoter (taken from a plastid gene that, during fruit development, is exclusively downregulated at the translational level) with the synthetic T7g10-derived 5′–UTR that harbors the strongest known ribosome-binding site (Shine–Dalgarno sequence) for translation initiation, and is unlikely to require specific translation factors. This could also be the reason why, despite the high GFP expression levels in both leaves and fruits, transplastomic tomato plants generated with this construct did not develop any mutant phenotype (in contrast to some other transplastomic lines expressing GFP to much lower levels; Figure 2).

Competition for gene-specific transcription factors and/or mRNA-specific translation factors is the likely cause of the mutant phenotypes seen in some of our transplastomic plants (Kuroda and Maliga, 2002). These phenotypes provide interesting information about specific trans-acting factors involved in the expression of the genes from which the cis-elements were taken. For example, the severe mutant phenotype resulting from the combination of the psbA promoter with the rpoB 5′–UTR, as opposed to the entirely normal phenotype of plants expressing the combination of the psbA promoter with the T7g10 or psbA 5′–UTR, suggests the existence of a translation factor specific to the rpoB 5′–UTR that is depleted by expression of the chimeric gfp mRNA under the control growth of a second copy of the rpoB 5′–UTR. Similarly, the severe pale phenotypes of all transplastomic lines containing the accD promoter in the gfp expression cassette may indicate that a gene-specific transcription factor becomes limiting upon the ectopic presence of an additional copy of the accD promoter in the plastid genome.

Previous attempts to optimize plastid transgene expression levels in non-green plastids of potato tubers have met with limited success. Testing expression elements that had been found reasonably active in transcript and polysome profiling studies of amyloplasts (Valkov et al., 2009) resulted in maximum expression levels of only 0.02% of the total soluble protein (Valkov et al., 2011), which is almost two orders of magnitude lower than the maximum expression level achieved here in chromoplasts of tomato fruits. Whether or not there are fundamental differences between gene expression in chromoplasts and amyloplasts or, alternatively, more successful combinations of promoters and 5′–UTRs for high-level gene expression in amyloplasts remains to be discovered, and is currently unclear. Testing our best-performing combinations in potato plastid transformation experiments would be a first step towards addressing this question.

It is also important to note that protein accumulation levels in plastids can depend strongly on the stability of a given protein (Birch-Machin et al., 2004; Zhou et al., 2008; Elghabi et al., 2011). In this work, a reporter protein was chosen (GFP) that accumulates to medium levels in leaf chloroplasts (i.e. in the low percentage range; Reed et al., 2001; Newell et al., 2003). Proteins that are significantly more stable in plastids, such as Bt toxins (De Cosa et al., 2001), the reporter protein β–glucuronidase (GUS; Herz et al., 2005; Segretin et al., 2012) and some phage-derived bacteriolytic proteins (Oey et al., 2009a; Bock and Warzecha, 2010) are known to accumulate in transgenic chloroplasts to much higher levels than GFP, in extreme cases reaching up to more than 70% of the total soluble protein (Oey et al., 2009a). Thus, the 1% of GFP protein obtained in ripe tomatoes in this study is most likely to be surpassed by expressing more stable recombinant proteins.

Our work described here has also led to the identification of expression elements that are more active in chromoplasts than in chloroplasts. This most remarkable property is conferred by the 5′–UTR from the accD gene (more precisely, its revised accD* version identified here), which had been shown previously to confer active translation in ripe tomato fruits (Kahlau and Bock, 2008). The accD* 5′–UTR doubles protein accumulation in chromoplast- compared with chloroplast-containing leaves and fruits (e.g. in Sl-pRC32 plants expressing gfp from rrn16::accD*; Table 1), providing a useful tool for all transplastomic applications, in which higher expression in fruits than in leaves in desirable.

In view of the rapidly growing interest in the exploitation of the transplastomic technology in biotechnology, and especially for the high-level production of recombinant proteins (Maliga, 2004; Bock, 2007; Daniell et al., 2009; Bock and Warzecha, 2010), promoter and UTR elements that trigger high transgene expression levels in non-green edible tissues, such as chromoplast-containing fruits and tap roots, are of great value. Therefore, in addition to providing new insights into the control of plastid gene expression in chromoplasts, the chimeric cis-elements described here are also widely applicable as tools for future transgenic studies in both basic research and plant biotechnology.

Experimental Procedures

Plant material and growth conditions

Tomato plants (Solanum lycopersicum cv. IPA–6) for plastid transformation experiments were raised under aseptic conditions on agar-solidified MS medium (Murashige and Skoog, 1962), containing 20 g L−1 sucrose. Homoplasmic transplastomic lines were rooted and propagated on a similar medium that additionally contained spectinomycin (500 mg L−1). To obtain seeds, plants were transferred to soil and grown to maturity under standard glasshouse conditions. Plants with pale-green to yellow phenotype were grown under low light conditions. Fruits were harvested at different ripening stages (green, turning and red), as defined by the California Tomato Commission (

Vector construction

The plastid transformation vectors constructed in this study are based on the previously described plasmid pKP9 (Zhou et al., 2008). The coding region of gfp was obtained by PCR amplification with primers P1 and P2 (for descriptions and sequences of primers, see Table S1) using a plasmid clone (Kato et al., 2002) as the template. The primer P1 introduces an NsiI restriction site spanning the first and second codon of gfp. The modified gfp reporter gene was cloned into pHK20 (Kuroda and Maliga, 2001) as an NdeI/XbaI fragment, generating vector pRC1. Chimeric expression elements for gfp, which consist of different combinations of promoters and 5′–UTRs, were generated using PCR-based strategies. The accD promoter combined with the accD 5′–UTR (accD::accD) was amplified by PCR with primers P15 and P16 using cloned tomato plastid DNA (Kahlau et al., 2006) as the template. The accD promoter combined with the full-length accD 5′–UTR (accD::accD*) identified in this study was produced by the same strategy, with primer P21 replacing P16. The psbA promoter combined with the accD* 5′–UTR (psbA::accD*) was obtained by PCR with primers P11 and P21 using cloned tomato plastid DNA as template. The combination of the rrn16 promoter with the accD 5′–UTR (rrn16::accD) was obtained by two rounds of PCRs. First, the rrn16 promoter was amplified with primers P3 and P4 using cloned tomato plastid DNA as the template, and the accD 5′–UTR was amplified with primers P5 and P16 using cloned tomato plastid DNA as the template. Subsequently, a second PCR was performed with the primers P3 and P16 using the products from the first two reactions as templates. Promoter-UTR fusion was facilitated by the overlapping ends present in the PCR products from the first round of reactions. The rrn16 promoter was combined with the accD* 5′–UTR (rrn16:accD*) employing a similar two-round PCR strategy, but using primer P21 instead of P16 (Table S1). The psbA promoter with its 5′–UTR was amplified with primers P13 and P14 using cloned tomato plastid DNA as the template. The psbA promoter was combined with the T7g10 5′–UTR (psbA::T7g10) by PCR amplification with primers P9 and P10, using pHK20 as the template. The combination of psbA promoter and rpoB 5′–UTR (psbA::rpoB) was produced by amplification with primers P12 and P8, using cloned tomato plastid DNA as the template. Finally, the combination of the rrn16 promoter with the T7g10 5′–UTR (rrn16::T7g10) was obtained by two rounds of PCR. First, the rrn16 promoter was amplified with primers P3 and P17 using cloned tomato plastid DNA as the template, and the T7g10 5′–UTR was amplified with primers P18 and P10 using pHK20 as the template. In a second round of PCR (using primers P3 and P10), the two amplification products from the first round of reactions served as templates, and promoter-UTR fusion was achieved via the overlapping ends of the two fragments. All expression cassettes generated by PCR were cloned into vector pRC1 as SacI/NsiI fragments (restriction sites were added with the sequences of the PCR primers; Table S1). Finally, the cassettes were cloned into plastid transformation vector pKP9 as a SacI/HindIII fragment (Zhou et al., 2008), and the correctness of all cloning steps was verified by resequencing.

Plastid transformation and selection of homoplasmic transplastomic tomato lines

Tomato plastid transformation was carried out using the biolistic protocol (Svab and Maliga, 1993; Ruf et al., 2001; Ruf and Bock, 2012). Young leaves from tomato plants grown under aseptic conditions were bombarded with plasmid DNA-coated gold particles using a PDS100He biolistic gun equipped with a Hepta adaptor (Bio-Rad, Primary spectinomycin-resistant lines were selected on a modified MS-based regeneration medium (Wurbs et al., 2007) containing 500 mg L−1 spectinomycin. For each construct, several independent resistant lines were subjected to two additional rounds of regeneration on spectinomycin-containing medium to obtain homoplasmic tissue. To confirm homoplasmy by inheritance assays, seeds were surface sterilized and germinated on medium containing 100 mg L−1 spectinomycin (Bock, 2001).

Isolation of nucleic acids and gel blot analyses

Total DNA from tomato plants was extracted from fresh leaf samples by a cetyltrimethylammonium bromide (CTAB)-based method (Doyle and Doyle, 1990). Total cellular RNA was isolated using the peqGOLD TriFast reagent (Peqlab, For Southern blot analysis, samples of 5 μg of total cellular DNA were digested with the restriction enzyme BglII, separated by gel electrophoresis in 1.0% agarose gels and transferred onto Hybond nylon membranes (GE Healthcare, by capillary blotting. A PCR product covering part of the psaB coding region (Wurbs et al., 2007) was used as a hybridization probe to verify plastid transformation and assess homoplasmy. For northern blot analysis, total cellular RNA samples from different tissues were electrophoresed in formaldehyde-containing 1.5% agarose gels and blotted onto Hybond nylon membranes. To detect gfp transcripts in transplastomic lines, a hybridization probe was generated by PCR amplification of the gfp coding region with specific primers (forward primer, 5′–GGAGAAGAACTTTTCACTGG–3′; reverse primer, 5′–GCCATCGCCAATTGGAGTAT–3′), producing a 560–bp fragment. Probes were labeled with α[32P]dCTP using the Multiprime DNA labeling system (GE Healthcare). Hybridizations were performed at 65°C using standard protocols and signals were analyzed using a Typhoon Trio+ variable mode imager (GE Healthcare).

Protein extraction and immunoblot analyses

Total protein from plant material was isolated using a phenol-based extraction method (Cahoon et al., 1992). The protein pellets obtained were dissolved in 1% SDS and the protein concentration was determined using the BCA Protein Assay (Pierce, using known concentrations of bovine serum albumin (BSA) as standard. Protein samples were separated by electrophoresis in 12% SDS-containing polyacrylamide gels, and the proteins were directly visualized by Coomassie blue staining or blotted onto polyvinylidene difluoride (PVDF) membranes (GE Healthcare). Membranes were treated with blocking buffer (20 mm Tris–HCl, pH 7.6, 137 mm NaCl, 0.5% BSA) for 1 h and then incubated with a mouse polyclonal anti-GFP antibody (JL–8; Clontech, diluted 1:20 000. Detection was performed with the ECL Plus system (GE Healthcare) and an anti-mouse secondary antibody (Sigma-Aldrich, For quantitative analysis, purified recombinant GFP (Roche Applied Science, was loaded in an appropriate dilution series followed by visual comparison of signal intensities.

Confocal laser-scanning microscopy

A confocal laser-scanning microscope (TCS SP5; Leica, was used for observation of GFP fluorescence signals in plastids of tomato leaves and fruits. The instrument was equipped with an argon laser and the following objectives: 63 × Plan Apo water and 20 × Plan Apo dry. For the detection of GFP fluorescence, the excitation wavelength was 488 nm and the barrier filter BP 530 (band pass, 515–545 nm) was used. The chlorophyll fluorescence was analyzed using an excitation wavelength of 568 nm and the barrier filter BP 590 (long pass, >590 nm).


We thank Steffi Seeger for help with plant transformation, Dr Eugenia Maximova (all MPI-MP) for help with microscopy and the MPI-MP Green Team for plant cultivation and care. RC was the recipient of a doctoral fellowship from the Deutsche Akademische Austauschdienst (DAAD, Germany) and the Comisión Nacional de Investigación Científica y Tecnológica, (CONICYT, Chile). KAH was supported by an Alexander von Humboldt Foundation Research Fellowship. This work was financed by a grant from the European Union to RB (EU-FP7 METAPRO 244348).