Introduction

The angiosperm genus Cuscuta (Cuscutaceae) comprises a range of species that have all adapted to a parasitic lifestyle. Although knowledge of their biology is still rudimentary and limited to only a few species, it has nevertheless become apparent that there is a wide range of adaptations to a parasitic lifestyle at the ultrastructural and physiological levels that has not been reported for any other parasitic plant genus.

Like holoparasitic plant species from other families (e.g. Orobanchaceae), many Cuscuta species (e.g. C. grandiflora, C. odorata) are completely devoid of chlorophylls and have lost their capacity for photosynthesis (van der Kooij et al. 2000). However, a number of Cuscuta species (e.g. C. pentagona, C. reflexa) were found to possess chloroplasts with intact photosystems and small amounts of chlorophyll (Panda and Choudhury 1992; Hibberd et al. 1998; Sherman et al. 1999; van der Kooij et al. 2000). Although photosynthesis in these species does not seem to exceed the compensation point (van der Kooij et al. 2000), light-dependent CO2 fixation was nevertheless shown to occur in C. reflexa (Machado and Zetsche 1990; Hibberd et al. 1998). Among the species that possess a potential for light-driven electron transport, some species are characterized by unusually high amounts of carotenoids (e.g. C. gronovii, C. campestris; van der Kooij et al. 2000). Compared to the 'green' species C. reflexa, these yellow 'intermediate' species have a lower PSII efficiency (van der Kooij et al. 2000). The physiological differences are reflected by obvious variations in the existence and quantity of thylakoid membranes as well as in the plastid ultrastructure in general (van der Kooij et al. 2000).

With the release of the selective pressure imposed upon the photosynthetic apparatus, the plastid genome of parasitic plants in general has undergone significant reductions when compared to green autotrophic land plants. Deletions in the plastid genome of several Cuscuta species have been reported. Haberhausen and Zetsche (1994) showed that a functional loss of all ndh genes from the plastome of C. reflexa has occurred, while van der Kooij et al. (2000) demonstrated that the plastid-encoded gene for the large subunit of Rubisco (rbcL) is missing in C. odorata but not in the other five Cuscuta species investigated. Recently, even the loss of the rpo genes coding for subunits of a plastid-encoded RNA polymerase (PEP) from the plastomes of C. subinclusa, C. gronovii and C. odorata was reported (Krause et al. 2003). PEP is the major RNA polymerase in chloroplasts of higher plants and is mainly responsible for the transcription of genes for the photosynthetic apparatus (Mullet 1993; Allison et al. 1996; Maliga 1998). Despite the loss of PEP, expression of these vestigial plastomes could be demonstrated by Northern and Western analysis (van der Kooij et al. 2000), providing evidence for the existence of an imported nuclear-encoded RNA polymerase (NEP). A single-subunit RNA polymerase with a plastid-targeting signal that is homologous to phage-type RNA polymerases was identified in the nuclear genome of Arabidopsis thaliana (Hedtke et al. 1997) and other higher plants (Chang et al. 1999; Ikeda and Gray 1999; Hedtke et al. 2002). Although direct biochemical evidence in Cuscuta is still missing, indirect evidence from plastid promoter analyses makes it highly likely that this phage-type RNA polymerase is responsible for complete transcription of the reduced plastid DNA in at least three Cuscuta species (C. gronovii, C. subinclusa and C. odorata) (Krause et al. 2003).

In contrast to these three species, the 'green' species C. reflexa has retained the four rpo genes (rpoA, B, C1 and C2) coding for PEP, which is in line with the assumption that PEP is needed for the expression of photosynthesis-related genes. A loss of the rpo gene products from plastids of higher plants, which can be artificially induced either by deletion of the corresponding plastid genes (e.g. Allison et al. 1996; DeSantis-Maciossek et al. 1999) or by plastid ribosome deficiency (Falk et al. 1993; Hess et al. 1993) has in all cases resulted in the inability of the plants to accumulate chlorophyll and to photosynthesize. Surprisingly, among the three Cuscuta species that have lost the rpo genes, two (C. gronovii, C. subinclusa) are still capable of producing small amounts of chlorophyll and performing photosynthesis, albeit at low levels (van der Kooij et al. 2000).

Here, a comparative analysis of plastid DNA and RNA from the green C. reflexa, the intermediate C. gronovii and the white C. odorata was performed in order to gain more information on plastid gene expression in these three different holoparasitic species.

Materials and methods

Plant material

Cuscuta species (C. reflexa, C. gronovii and C. odorata) were kept in a greenhouse at the University of Kiel on Pelargonium zonale as host plants as described before (van der Kooij et al. 2000, Krause et al. 2003).

Isolation of plastids

Plastids were isolated from C. reflexa and C. gronovii essentially as described by Krause et al. (2000). Briefly, shoot material (30 g) was homogenized in 250 ml buffer A [0.33 M sorbitol, 25 mM HEPES, 25 mM MES, 4 mM Na-ascorbate, 1.2 mM MnCl2, 0.8 mM MgCl2, 4 mM EDTA, 1 mM KH2PO4, 4 mM DTT, 0.2% (w/v) BSA, 0.1% (w/v) PVP-10, pH 6.8]. The homogenate was filtered through Miracloth (Calbiochem, La Jolla, CA, USA), and centrifuged at 3,000 g for 60 s. The pellet was resuspended in a small volume of buffer B [0.33 M sorbitol, 50 mM HEPES–KOH, pH 8.0] and was fractionated on 20–80% (v/v) Percoll gradients. Intact chloroplasts were washed and resuspended in buffer B. All steps were performed at 4 °C or on ice. For run-on transcription assays the number of plastids was determined microscopically, and the plastid suspension was adjusted to 1×106–2×106 plastids per μl.

Amyloplasts were isolated from C. odorata according to Wischmann et al. (1999) with some modifications. Shoot material (30 g) was homogenized in buffer A [1 mM MgCl2, 1 mM KCl, 1 mM EDTA, 5 mM DTT, 0.025% (w/v) BSA, 1% (w/v) PVP-40, 1 M sorbitol, 10% (v/v) ethylene glycol, 50 mM HEPES–KOH, pH 8]. The suspension was filtered through nylon gauze (40 μm) and fractionated in gradients [1% (w/v) agarose, 2% (v/v) Nycodenz in buffer A] by centrifugation for 30 min, 30 g and 4 °C. Amyloplasts collected on the agarose cushion were carefully removed and used for extraction of nucleic acids.

Preparation and labelling of plastid DNA and RNA

Plastid DNA was extracted from isolated plastids (see above) as described by Wulff et al. (2002). Prior to labelling with α-[32P]dCTP using a random primed labelling kit (Roche), the DNA was digested with BamHI. RNA was isolated using standard procedures (van der Kooij et al 2000; Krause et al. 2003). The transcripts were labelled at their 3′ ends using a T4 RNA ligase and 3′-5′-α-[32P]bisphosphate–CTP (pCp) according to the enzyme manufacturer's instructions.

Run-on transcription assays

Run-on transcription assays with 2×107 plastids were carried out in a 100-μl volume in the presence of heparin as previously described (Klein and Mullet 1990; Krupinska 1992; Krause et al. 1998). Incorporation of α-[32P]UTP into elongating transcripts was determined as described by Hallick et al. (1976) with aliquots spotted onto DE81 filters (Whatman, Maidstone, UK). Labelled transcripts were extracted from the plastids as previously described (Krause et al. 2000), resuspended in 50 μl of 50% (v/v) deionised formamide and denatured by heating for 5 min at 65 °C.

Hybridization with DNA fragments representing in sum the entire tobacco plastid chromosome

Recombinant pBR322 plasmids containing the BamHI fragments Ba1, Ba2, Ba5, B7, B13, B18, B19, B20, B22, B25, B27, B28 and B29 of tobacco plastid DNA (Sugiura et al. 1986) were digested with BamHI and EcoRI. A segment of plastid DNA not represented by these fragments was amplified by PCR using a primer pair corresponding to nucleotide positions 113071–113055 and 109663–109680 of the tobacco plastome (Wakasugi et al. 1998). The 3.4-kb PCR product was purified and directly digested with BamHI and EcoRI. All restriction fragments were fractionated on 1.2% (w/v) agarose gels and subsequently transferred onto a nylon membrane (Hybond N+; Amersham, Braunschweig, Germany) by capillary blotting using 0.4 M NaOH as transfer buffer. Hybridizations of 32P-labelled plastid DNA, RNA or run-on transcripts with immobilized tobacco plastid DNA fragments were performed according to the membrane manufacturer's protocol. Autoradiography was carried out at −80 °C using 'Hyperfilm MP' X-ray film (Amersham). The same filter was subsequently used for all hybridizations. After each hybridization, bound DNA or RNA was stripped off by boiling in 0.5% (w/v) SDS as described by the manufacturer.

Hybridization with single-stranded gene-specific RNA probes

Single-stranded gene-specific RNA probes were generated by in vitro transcription of cloned plastid DNA fragments from tobacco (Krause et al. 2000) and dotted onto nylon filters (Zeta Probe GT; BioRad Laboratories, Germany) in a series of dilutions (100, 400 and 1,600 fmol, respectively) as described previously (Krause et al. 2000). Hybridization conditions for heterologous probes (labelled steady-state RNA or run-on transcripts) were as recommended by the membrane manufacturer.

Results

Estimation of plastome sizes and coding capacities

Since no information was available on the degree of conservation of the plastid genomes of C. odorata and C. gronovii, and knowledge of the C. reflexa plastome was also fragmentary, we first performed a survey of coding capacities and approximate sizes of the three plastid genomes. For this purpose, Southern hybridization of labelled Cuscuta plastid DNAs with a set of cloned BamHI fragments of tobacco (Sugiura et al. 1986), and a 3.4-kb PCR fragment representing in sum the entire plastid genome was employed. Since each BamHI clone encompasses a large area of the plastid DNA, a restriction digest with BamHI and EcoRI was performed in order to receive smaller subfragments that are specific for individual operons or even genes. A small piece of DNA not represented by these fragments was amplified by PCR and treated likewise (see Materials and methods). The subfragments were separated by agarose gel electrophoresis and stained with ethidium bromide as shown in Fig. 1. The large single-copy region (LSC) of the circular plastid chromosome encompasses part of clone B28 and extends into clone B13. The small single-copy region (SSC) is represented by clone Ba2 and the PCR product, while clones B18 and Ba5 are part of the two inverted repeats (IRA, IRB). The junction between IRA and LSC is located in clone B13, and that between IRB and LSC in clone B28 (Fig. 1).

Fig. 1.
figure 1

Ethidium bromide-stained gel with DNA-fragments representing the entire tobacco (Nicotiana tabacum) plastid genome. Thirteen recombinant plasmids carrying BamHI fragments that represent 98% of the tobacco plastid chromosome (Sugiura et al. 1986) were digested with BamHI and EcoRI, and fractionated in a 1.2% agarose gel. A 3.4-kb segment of the plastid DNA not represented by these fragments was amplified by PCR (see Materials and methods) and treated similarly. The positions of the individual BamHI clones relative to the four major regions of the plastid genome (LSC, IRA, SSC, IRB) are indicated above the gel. The positions of size standards are shown on the left

Full size image

The separated fragments were transferred onto a nylon filter and hybridized with radiolabelled plastid DNA derived from the different Cuscuta species. While the hybridization pattern for C. odorata shows that very few fragments give hybridization signals, a substantially larger portion of the fragments hybridize with the plastid DNA of C. gronovii. In comparison, the majority of the fragments hybridize with labelled plastid DNA of C. reflexa (Fig. 2a).

Fig. 2a, b.
figure 2

Comparison of coding capacities and transcript compositions in plastids of three Cuscuta species. Filter-bound DNA fragments representing the entire plastid genome of tobacco (see Fig. 1) were subsequently hybridized with 32P-labelled plastid DNA (a) or RNA (b) from C. odorata, C. gronovii and C. reflexa. The filter was exposed to X-ray film. Positions and sizes (in kbp) of the molecular weight standards are indicated on the left of each autoradiogram

Full size image

The areas retained on the reduced plastid genome of C. odorata encompass the rrn operon and several tRNAs represented by clones B18 and Ba5, the trnEYD operon on the 344-bp subfragment of B20 and the clpP gene on the 2,125-bp fragment of clone Ba1. The 3,100-bp subfragment of clone B28 from tobacco, which shows a strong hybridization signal with labelled plastid DNA from C. odorata, contains part of the matK gene, exon II of trnK, the complete coding sequences of psbA and trnH as well as exon II of rpl2.

While many deletions in the plastid DNA are shared by C. gronovii and C. odorata (see, for example, the 1-kb subfragment of clone B28 consisting of the rpoB coding sequence and subfragments of clone B7 encompassing the rest of the rpoB/C1/C2 operon), some differences in the hybridization patterns between the two species (e.g. in clones Ba1, B13, B29, B25) are present (Fig. 2a). Among the genes that are represented by those fragments that hybridize only with C. gronovii plastid DNA are several genes required for photosynthesis (psaA, psaB, psbC, psbD on large subfragments of B20 and rbcL on the 1.2-kb fragment of B27 and B29), the atpB gene represented by the 896-bp fragment of clones B27 and B29, but also ribosomal protein genes (rpl22, rps19, rps3). Though the trnEYD operon on the smallest subfragment of clone B20 gave only a very weak signal, its presence was unambiguously demonstrated by hybridization with labelled plastid RNA (see Fig. 2b).

Out of the 13 subfragments of clone Ba2 only the 952-bp subfragment generated a hybridization signal with labelled pt DNA of C. reflexa (Fig. 2a). This fragment covers part of the ycf1 reading frame. The other subfragments mainly contain the ndh genes as well as two ribosomal protein genes (rpl32, rps15) located in the small single-copy region of the tobacco plastid genome (Sugiura et al. 1986; Wakasugi et al. 1998).

Composition of steady-state RNAs in plastids of the three Cuscuta species

The same tobacco plastid DNA filter that was used for Southern analysis of Cuscuta DNAs was subsequently used for hybridization with 32P-labelled RNA derived from plastids of the three Cuscuta species (see Materials and methods). A comparison of the three corresponding autoradiograms in Fig. 2b shows that in all cases the basic hybridization pattern is very similar. Although a few subfragments in the lanes 'B27', 'B29' and 'PCR' show hybridization signals with RNA from C. reflexa but not from the other two species (Fig. 2b, bottom), the overall similarity that exists between the RNA patterns of the three species is in striking contrast to the differences in DNA patterns (compare Fig. 2a with Fig. 2b). While the comparison of the two autoradiograms from C. odorata reveals that basically every fragment that hybridizes with C. odorata plastid DNA (Fig. 2a) also hybridizes to labelled RNA of this parasite (Fig. 2a, b, top), in the case of C. reflexa a considerable portion of the plastid genome is not represented at the RNA level (compare Fig. 2a and b, bottom). While, for example, subfragments of the clones B7, B22 and B13 hybridize with C. reflexa plastid DNA, no hybridization signals were obtained with labelled plastid RNA as probe. Fewer differences than in case of C. reflexa can be seen between the hybridization patterns obtained with DNA and RNA of C. gronovii, respectively (e.g. clones B27, B29, B13).

In all three species, the strongest hybridization signals obtained with labelled RNA can be observed with the ribosomal RNAs encoded by clones B18 and Ba5 (Fig. 2b). Another band that strongly hybridizes with labelled RNA from all three species is the 344-bp subfragment of clone B20, representing the trnEYD operon. Differences in the relative hybridization intensities of individual fragments within each filter are most likely caused by different homologies of the corresponding sequences between tobacco and the three Cuscuta species.

Transcription in plastids from C. reflexa

The apparent differences between the hybridization patterns obtained for C. reflexa using plastid DNA on one hand and steady-state RNA on the other hand, raised the question of whether this discrepancy is based on selective transcription of the plastid genome or on differential transcript stability. To investigate transcriptional activities of the different operons and compare them with steady-state RNA levels, we employed plastid run-on transcription assays. Hybridization of labelled run-on transcripts with plastid DNA fragments representing the entire plastid genome (Fig. 1) shows that essentially all fragments hybridize with C. reflexa run-on transcripts (Fig. 3). To achieve a resolution of hybridizations at the level of individual genes, labelled run-on transcripts or steady-state transcripts, from C. reflexa were hybridized to strand-specific probes of 15 genes dotted onto a nylon filter in a series of dilutions (Fig. 4). While the photosynthesis-related genes psbA, psbD, psaA and psaB, as well as genes for the transcription and translation apparatus (e.g. rrn16, rpoA, rpoB), show strong hybridization signals with labelled run-on transcripts, mainly rrn16 and the tRNAs are present at high levels among the steady-state RNA pool (Fig. 4), confirming the observations made with the tobacco plastid DNA filter shown in Fig. 2b.

Fig. 3.
figure 3

Analysis of the plastid run-on transcript pattern from C. reflexa. Run-on transcription assays with isolated chloroplasts from C. reflexa were carried out as described in Materials and methods. Hybridization of the α-32P-labelled run-on transcripts to a filter representing the entire tobacco plastid genome as DNA fragments (see Fig. 1) is shown. The positions of size standards are shown on the left

Full size image
Fig. 4.
figure 4

Comparison of run-on and steady-state transcripts from C. reflexa. α-32P-labelled RNA from run-on assays (top) or 3′-end labelling reactions (bottom) from C. reflexa was hybridized to identical dot blots containing in vitro-transcribed antisense RNA specific for 15 different chloroplast genes (see Materials and methods). The plasmid pBluescript (pBS) was used to assess the level of unspecific background hybridization. The designations of the gene probes are given above the blots

Full size image

A quantification of hybridization signals with a phosphoimager (data not shown) confirmed that the rrn16 gene is the major constituent of both run-on and steady-state transcript pools. mRNAs, in contrast, were found to contribute substantially only to the pool of run-on transcripts while tRNAs, on the other hand, accumulate highly in the steady-state RNA pool but exhibit low relative transcription rates.

Discussion

Using a clone library of BamHI fragments representing the entire tobacco plastid genome as a reference, we first estimated the relative sizes and coding capacities of the plastid DNAs of three parasitic Cuscuta species differing in their chlorophyll content and photosynthetic capacity. In addition, steady-state RNA patterns were compared between the three species and to the corresponding DNA pattern of the same species.

Based on the hybridization patterns shown in Fig. 2a, an estimate of the extent of gene losses in the individual regions of the plastid chromosome was undertaken and is depicted in Fig. 5. It is noteworthy that sequences found in the inverted repeat regions (IRA and IRB) of tobacco plastid DNA were largely conserved during the evolution of the three Cuscuta species. In contrast, the two single-copy regions, LSC and SSC, show a strong tendency towards loss or alteration of sequences, the extent of which differs substantially among the three species (Fig. 5).

Fig. 5.
figure 5

Approximate plastid chromosome sizes of different Cuscuta species in relation to N. tabacum. The loss of sequences from the four regions of plastid DNA, LSC, IRB, SSC and IRA, respectively, was estimated based on hybridization patterns with tobacco plastid DNA

Full size image

A loss of plastid genes has been previously described for the achlorophyllous species C. odorata. Van der Kooij et al. (2000) reported that this species does not possess the rbcL gene, and Krause et al. (2003) further demonstrated the lack of the rpoA and rpoB genes, coding for subunits of the plastid-encoded RNA polymerase (PEP). The present survey shows that out of the three species examined, C. odorata, which has lost the capacity for photosynthesis (van der Kooij et al. 2000), has suffered the most extensive deletions in the plastid genome compared to that of tobacco (Fig. 2a, Fig. 5). The coding capacity of this species, therefore, resembles that of other achlorophyllous parasites like Epifagus virginiana, Conopholis americana or Orobanche hederae (Morden et al. 1991; Wimpee et al. 1991; Delavault and Thalouarn 1994). Analysis of steady-state RNAs further shows that the entire reduced plastid genome of C. odorata is transcribed and produces stable RNAs (Fig. 2b). Like in other parasitic land plants, transcription of the plastid DNA of C. odorata is exclusively dependent on a nuclear-encoded RNA polymerase (NEP; Morden et al. 1991; Krause et al. 2003). Transcription by NEP requires that all genes in the genome have acquired a NEP promoter. Such changes in the promoter structure could indeed be demonstrated for the rrn16 and trnV genes of C. odorata (Krause et al. 2003). The stability of the individual transcripts of C. odorata was not yet under investigation but from the steady-state pattern shown in this study, it can be inferred that the transcripts are rather stable.

Although the plastid genome of C. gronovii has undergone extensive reductions (Fig. 2a), it still contains more genes than the plastid genome of C. odorata. Among the additional fragments that hybridize to labelled plastid DNA of this species and not to plastid DNA of C. odorata are those harbouring photosynthetic genes, e.g. psaA, psaB, psbC, psbD on the larger subfragments of clone B20 (see Fig. 2a). These genes are obviously expressed (Fig. 2), which is in line with the observation that C. gronovii is capable of performing photosynthesis (van der Kooij et al. 2000). Hence, in contrast to chloroplasts of non-parasitic higher plants, where the expression of photosynthesis-related genes seems to be dependent on the activity of the PEP, in C. gronovii photosynthesis-related genes are expressed despite the lack of PEP (Krause et al. 2003). Due to difficulties in obtaining sufficient amounts of intact plastids from C. gronovii, run-on transcription analyses that would provide in-depth information on relative NEP transcription rates were, unfortunately, not yet possible. In contrast to C. gronovii, tobacco mutants devoid of PEP activity are unable to photosynthesize (Allison et al. 1996; DeSantis-Macciossek et al. 1999), although NEP is unambiguously able to transcribe all genes in these plastids (Krause et al. 2000). Legen et al. (2002) were able to demonstrate that the accumulation of plastid-encoded polypeptides for photosynthesis-related functions is a result of transcript-specific stabilization and processing events that are altered in a PEP-deficient background (Krause et al. 2000; Legen et al. 2002), rather than being caused directly by a lack of PEP-gene transcription. We therefore conclude that the plastid DNA from C. gronovii must have acquired features of specific NEP-based transcription for the so-called PEP genes that enable translation of the corresponding transcripts.

The loss of the ndh genes and a few tRNAs and ribosomal proteins has already been described for the plastids of the green species C. reflexa (Haberhausen et al. 1992; Bömmer et al. 1993; Haberhausen and Zetsche 1994). Plastome-wide Southern analysis with tobacco plastid DNA as reference confirmed these previous findings (Fig. 2a). However, most of the subfragments of the other clones showed strong hybridization signals, demonstrating a high homology between the plastid genomes of C. reflexa and tobacco. In contrast, only a small subset of the largely intact plastid genome of C. reflexa is expressed at the level of stable steady-state RNA (Fig. 2). Dot blot analysis with gene-specific fragments shows that many mRNAs do not accumulate to appreciable levels while tRNAs and rRNAs are stabilized (Fig. 4). This result contrasts with findings from similar experiments performed with wild-type tobacco, where psbA, psbC, psbD and rbcL transcripts produced high hybridization signals (Legen et al. 2002). Thus, the RNA pattern of C. reflexa reveals the same parasite-specific characteristics observed for C. odorata and C. gronovii (see Fig. 2b; comparison of dot blot analysis not shown) and differs from that of photosynthetically active land plants.

This raises several questions: Why is C. reflexa nevertheless able to produce visible amounts of chlorophyll and to perform photosynthesis, and what are the differences in gene expression between C. reflexa and C. gronovii that could account for the physiological differences between these two species?

Analysis of relative transcription rates by plastid run-on assays revealed that the entire plastid DNA of C. reflexa is transcribed (Fig. 3), so that selective transcription of rRNA and tRNA genes as a reason for the observed steady-state pattern can be ruled out. Posttranscriptional mechanisms such as RNA degradation are known to play an important role in the regulation of plastid gene expression (Gruissem and Tonkyn 1993; Schuster et al. 1999). Transcript stability may vary in response to photosynthetic activity (Monde et al. 2000), and depends on whether a given transcript was synthesized by PEP or by NEP (Legen et al. 2002). Since photosynthesis rates in C. reflexa are very low compared to non-parasitic green land plants, it is likely that very low amounts of mRNAs are sufficient for the production of the subunits of the photosynthetic machinery. This assumption is supported by the work of Eberhard et al. (2002) who have recently demonstrated that the levels of two chloroplast mRNAs can be reduced by 90% without influencing the rate of synthesis of the corresponding proteins.

In contrast to the other two Cuscuta species, C. reflexa has still retained both the PEP enzyme and PEP promoters (Krause et al. 2003). Therefore, transcriptional activity of PEP could, directly or indirectly, be responsible for the higher photosynthetic activity and higher chlorophyll content of C. reflexa compared to C. gronovii.