Targeted disruption of the plastid RNA polymerase genes rpoA, B and C1: molecular biology, biochemistry and ultrastructure


*For correspondence (fax +49 89 171 683;
†These authors have equally contributed to this work.


The plastid encoded RNA polymerase subunit genes rpoA, B and C1 of tobacco were disrupted individually by PEG-mediated plastid transformation. The resulting off-white mutant phenotype is identical for inactivation of the different genes. The mutants pass through a normal ontogenetic cycle including flower formation and production of fertile seeds. Their plastids reveal a poorly developed internal membrane system consisting of large vesicles and, occasionally, flattened membranes, reminiscent of stacked thylakoids. The rpomaterial is capable of synthesising pigments and lipids, similar in composition but at lower amounts than the wild-type. Western analysis demonstrates that plastids contain nuclear-coded stroma and thylakoid polypeptides including terminally processed lumenal components of the Sec but not of the ΔpH thylakoid translocation machineries. Components using the latter route accumulate as intermediates. In striking contrast, polypeptides involved in photosynthesis encoded by plastid genes could not be detected by Western analysis, although transcription of plastid genes, including the rrn operon, by the plastid RNA polymerase of nuclear origin is found as expected. Remarkably, ultrastructural, sedimentation and Northern analyses as well as pulse experiments suggest that rpoplastids contain functional ribosomes. The detection of the plastid-encoded ribosomal protein Rpl2 is consistent with these results. The findings demonstrate that the consequences of rpo gene disruption, and implicitly the integration of the two plastid polymerase types into the entire cellular context, are considerably more complex than presently assumed.


Plastid chromosomes encode four RNA polymerase genes, designated rpoA, B, C1 and C2, that resemble the three RNA polymerase core genes of eubacteria (reviewed in Igloi & Kössel 1992), commensurate with the endocytobiotic origin of the organelle. The genes for σ-type subunits of such enzymes, which play a crucial role in transcription initiation conferring specificity to σ70 promoter selection, are of nuclear origin (summarised in Link 1996). Polypeptides resembling such subunits have been described previously ( Lerbs et al. 1988 ; Tiller et al. 1991 ) and cDNAs encoding σ subunits have recently been isolated from the rhodophycean alga Cyanidium caldarium ( Liu & Troxler 1996; Tanaka et al. 1996 ) as well as from higher plants ( Isono et al. 1997 ; Kestermann et al. 1998 ; Tanaka et al. 1997 ; Tozawa et al. 1998 ). The eubacterial-like RNA polymerase of plastids is therefore of dual genetic origin as various other complex structures of the organelle stroma, the thylakoid system and the inner envelope membrane (summarised in Herrmann 1997).

One of the most striking recent results in organelle biology, as in the biology of eukaryotes in general ( Herrmann 1997), was the demonstration of a nuclear-coded, second plastid-located RNA polymerase (gene: rpoY) resembling phage-type and mitochondrial enzymes (gene: rpoZ;Hedtke et al. 1997 ). This enzyme appears to be the result of an internal duplication of rpoZ that was accompanied by the gain of appropriate promoter and transit peptide sequences. Several lines of molecular genetic and biochemical evidence had suggested the existence of such an (or more) additional enzyme(s): (i) the loss of the entire rpoB/C1/C2 operon (plus photosynthesis genes) in the plastid chromosome of the non-photosynthetic higher plant parasite Epifagus virginiana ( Morden et al. 1991 ), although translocation of rpo genes to the nucleus has not been rigorously excluded in this case; (ii) the demonstration of a phage-type RNA polymerase in chloroplast fractions ( Lerbs-Mache 1993); (iii) work on ribosome-deficient plastids ( Falk et al. 1993 ; Hess et al. 1993 ) which showed that this phenotype does not preclude the transcription of the rpo gene cluster and of other loci; and (iv) recent comparable results with an rpoB-deficient tobacco mutant generated by targeted gene disruption ( Allison et al. 1996 ). The latter material was reported to have segregated to the homoplastomic state, a result that could not be achieved in the corresponding experiment with Chlamydomonas ( Rochaix 1995). The analysis of plastid transcript patterns has suggested three categories, as well as a complex design of promoters, which may bear multiple initiation sites for these enzymes ( Allison et al. 1996 ; Kapoor et al. 1997 ; Vera & Sugiura 1995). Two types of promoters appear to operate with only one of the two enzymes, and a third class with both.

The existence of at least two distinct RNA polymerases, encoded in different cellular compartments and of probably different phylogenetic origin, and the different classes or the complexity of promoters suggest that the promoters of plastid chromosomes, as those of mitochondria and translocated nuclear genes, have undergone substantial qualitative changes during evolution ( Herrmann 1997), and that transcription in the organelle is significantly more complex than in its prokaryotic counterparts. Obviously, knowledge of the ontogenetic and phylogenetic impact of this design, for instance the co-operation of the two enzymes to ensure accurate transcription initiation in plastids in time and space, is crucial for understanding the integration of the organelle’s genetic machinery into the overall cellular context, but also of eukaryotism and its evolution ( Herrmann 1997). These aspects, the observation that rpoB-deficient material yields a stable phenotype ( Allison et al. 1996 ), and that information about decoding of transcripts in this genetic background is lacking, has prompted us to generate additional rpo-deficient mutants by targeted gene disruption and to study their phenotypes in greater detail. In this paper we present data that substantially extend our knowledge on material lacking the plastid-encoded RNA polymerase.


PEG-based plastid transformation: rpo phenotypes

In PEG-based plastid transformation, protoplasts are treated with polyethylene glycol (PEG) in the presence of suitable plasmid vectors ( Fig. 1; for details see Experimental procedures) and regenerated to small colonies of 10–20 cells without selection. Further culture is performed under selective pressure and resistant cell lines can be detected as fast growing green, macroscopically visible colonies, while untransformed cell lines grow slowly and are of a brownish colour. Green colonies formed during the selection period from all samples treated with any of the transformation vectors used. The frequency varied between 22 and 130 independent lines per 106 treated protoplasts. The first resistant colonies appeared 6 weeks after transformation, and segregation of transformed and wild-type sectors became apparent in calluses and regenerated shoots after an additional culture period without selection for approximately 3 more weeks ( Fig. 2a–c). Excised white shoots grew into healthy plants ( Fig. 2f,g) which were morphologically indistinguishable from wild-type material, except for their off-white colour. Some of the shoot cultures developed flowers (in vitro flower formation is occasionally observed with the wild-type as well). The flowers showed pink corollas and set seed when pollinated with wild-type or mutant pollen. The fertility of mutant pollen was also evident from the occurrence of seed production by spontaneous self-pollination in vitro, and verified by the efficient production of seed after pollinating the cytoplasmic male sterile tobacco line Nta(big)S ( Kofer et al. 1991 ; data not shown).

Figure 1.

Construction of Nicotiana tabacum (a) rpoA (b) rpoB and (c) rpoC1 mutants.

Targeted mutation of rpoA, rpoB and rpoC1 were derived by insertion of the aadA cassette into unique restriction sites within the respective genes as described in the Experimental procedures. Figures shown in (a), (b) and (c) correspond to transformation vectors prpoAI, prpoBI, prpoC1–I, respectively, which differ from prpoAII, prpoBII, prpoC1–II in the orientation of the aadA-cassette.

Figure 2.

Phenotype of rpo mutants of Nicotiana tabacum.

(a–c) Segregation after 10 weeks of culture in calluses grown on medium without antibiotics for 4 weeks (a) rpoA; (b) rpoB ; (c) rpoC1 ; (d) white and green shoots regenerating from a callus after 6 weeks without antibiotics (rpoB ); (e) wild-type and mutant (rpoA) sectors in a leaf of a regenerated plantlet; (f) pure white regenerant after dissection from the callus (rpoC1 ); (g) vigorously growing shoot culture of an rpoB  mutant.

rpo genotypes

For molecular analysis, the mutant lines had undergone at least three cycles of shoot regeneration from leaf explants. They were maintained on medium (modified VBW) inducing multiple shoot formation for more than 20 months. In addition, using a rapid and efficient system for regeneration of plants from protoplasts, which allows shoot formation within less than 2 weeks ( Dovzhenko et al. 1998 ), colonies and small shoots were regenerated with high efficiency from more than 107 mutant protoplasts. Fertile seeds capable of germinating were produced from all of the mutant lines except for rpoC1 mutants, where so far all seeds generated failed to germinate. Green sectors were never observed to segregate in any of these tissue culture lines. This suggested the complete absence of wild-type plastome copies.

Correct integration of the aadA cassette was first monitored by Southern and sequence analyses (data not shown). Corresponding wild-type signals were not visible (detection level approximately 0.1%). To enhance sensitivity, the highly efficient Qiagen PCR analysis of DNA from seedling-derived plantlets was performed with primers to either side of the aadA cassette, one of which was designed to bind to one of the homologous flanking regions of the transformation construct, and the other one to a plastome sequence outside the transformation vector insert. Amplification signals from transformed plastome template DNA were found as expected. Surprisingly, however, signals expected from the wild-type plastome were also noted. To investigate whether such signals were generated from plastome-derived sequences in the nuclear and/or mitochondrial genomes (promiscuous DNA; Brennicke et al. 1993 ; Ellis 1982), the amplified DNA fragments were sequenced including a total of more than 800 bp outside the transformation vector insert. No deviation from vector and neighbouring plastome sequences was detected. This seemed to indicate that wild-type plastome copies are present in the mutant plastids, since promiscuous plastid DNA fragments are generally not fully identical with the corresponding plastome DNA. On the other hand, the substantial reduction of the wild-type PCR-signal in the DNA from Percoll gradient-purified mutant plastid, copy number calculations based on ethidium bromide-stained agarose gels, and differences in the amplification kinetics between wild-type and mutant rpo gene sequences suggest that the number of wild-type plastome copies, if any, must be significantly lower than the number of plastids per cell (see Discussion).

Ultrastructural analysis

Wild-type tobacco chloroplasts of shoot cultures grown in vitro are expectedly lens-shaped, relatively large, up to 12 μm in diameter and up to 4 μm in thickness. Their matrix is densely packed with ribosomes, and numerous stroma and grana thylakoids are prominent. Starch grains are common, while few evenly distributed plastoglobules are detected ( Fig. 3a).

Figure 3.

Electron micrographs of ultra thin sections and cross-fractured, frozen hydrated probes from leaf tissue of tobacco wild-type and various rpo mutants.

(a) Typical lens-shaped wild-type chloroplast with numerous stroma and grana thylakoids. The matrix is densely packed with ribosomes. Plastoglobules show a uniform distribution. (b) Typical rpo mutant plastid from leaf tissue. The shape of the organelle is formed by a single huge ‘vesicle’. Clusters of osmiophilic plastoglobules (Pg) are predominant and typical thylakoids are not observed. (c) Cryo-scanning electron microscopy of a cross-fractured and etched chloroplast from tobacco rpo mutants. Four plastid ‘vesicles’ are fractured, three of them show a solid matrix. Remarkable difference in etching depth is seen comparing the central cellular vacuole (V) and plastid ‘vesicles’. The ‘vesicle’ membrane, which is slightly undulated can be examined from the inside (c; asterisk). (d) In early stages of leaf development and in shoot apex tissue the small lens-shaped rpo mutant plastids show numerous vesicular invaginations of the inner plastid membrane (arrows) and a tubular network. (e) Starch grains (S) can only be detected in rpo mutant plastids of guard cells (e) and root tips (ultrathin section). (f) Plastid of dark grown rpo leaf tissue. Membranous structures as vesicles (arrows) or a reticulum (arrowhead) are frequently observed as well as electron translucent plastoglobules (Pg) and osmiophilic deposits (asterisk). Starch grains (S) are common.

The plastids of all rpo mutants investigated vary in size and shape ( Fig. 3b–f). They are usually roughly ellipsoidal or spherical in shape and 4–6 mm in diameter, bounded by a double membrane and contain a fairly homogeneous matrix with few structures resembling ribosomes (see below). Depending on the developmental stage of the leaf, a variety of plastid stages can be observed. In tissues of the shoot apex and in younger leaves, lens-shaped plastids are predominant. They show three characteristic features: a tubular network, vesicular invaginations extending from the inner envelope membrane ( Fig. 3d), and large ‘vesicles’. In fully developed leaves, plastids possess either a single or a few large ‘vesicles’ ( Fig. 3b,c) which appear to cause the roughly spherical shape of the organelle ( Fig. 3b). The vesicles are membrane-surrounded spaces with fine granular material of proteinaceous appearance in their interior ( Fig. 3b,c). Compared to apex tissue and younger leaves, the tubular network of the mutants is reduced in fully developed leaves and invaginations are only rarely observed. In addition, light-grown material may contain flattened membranes which in some cases may even be attached to each other or to vesicles (see below and Discussion). Plastids from dark-grown material occasionally displayed structures reminiscent of prolamellar bodies. Clusters of osmiophilic plastoglobules are found at all stages of plastid development ( Fig. 3b). Mutant plastids can form starch grains, in spite of the absence of photosynthesis, yet these are restricted to stomatal guard cells ( Fig. 3e) and root tips. Plastids of dark-grown rpo leaf tissue are much smaller (approximately 2 μm) and more or less spherical ( Fig. 3f). Typically, they contain starch grains and appear to lack large ‘vesicles’. Membranous structures in the form of small vesicles ( Fig. 3f; arrows) or a reticulum ( Fig. 3f; arrowhead) are frequently observed as well as electron translucent plastoglobules ( Fig. 3f).

That the plastid ‘vesicles’ are true membrane-engulfed spaces and not invaginations from the cytosol was proven by the fact that they are surrounded by a single unit membrane as well as by cryo-scanning-electron microscopy (frozen hydrated) ( Fig. 3c). The cross-fractured structures display a solid matrix with a lower sublimation rate than the central cellular vacuole ( Fig. 3c), either due to a higher osmolarity or a higher protein content (see below).

Northern analysis

All rpo-deficient mutants contained reduced transcript levels for constituent polypeptides of the photosynthesis machinery (psaA, psaB, psbA, psbB, psbC, psbD, petB, rbcL) and rRNAs whereas mRNAs for functions not related to photosynthesis (i.e. accD) accumulate to wild-type levels (data not shown). This is in accordance with the observations made with rpoB-deficient plants ( Hajdukiewicz et al. 1997 ). Low amounts of processed plastid rRNAs could also be detected in the polysomal fraction of rpo-deficient plants ( Fig. 4). This and the data below demonstrate that the mutant plastids contain functional plastid ribosomes.

Figure 4.

Detection of cytosolic 18S rRNA and plastid 16S rRNA in the polysomal fraction of wild-type and rpo-deficient plants by Northern analysis.

The images on the left show polysomal RNA isolated from wild-type and rpoC1 mutant plants electrophoretically separated in 1% agarose-formaldehyde gels. The images on the right show corresponding nylon membranes hybridized against specific probes.

Serology, pulse experiments, ribosomes and ribosomal activity

Approximatly 20 monospecific, polyclonal antisera, representative of plastid- and nuclear-coded polypeptides and the major organelle spaces, were chosen for Western analysis with total cellular extracts to assess the status and nature of the rudimentary organelle and its vesicles. All components of nuclear origin studied could be detected in substantial amounts ( Fig. 5). The proteins were terminally processed and their amounts often indistinguishable from those of the wild-type on a per fresh weight basis. It is relevant to mention that plastocyanin and the 33 kDa polypeptide of the oxygen-evolving complex, which are both located in the thylakoid lumen, are terminally processed as well, while the lumenal 23 kDa and 16 kDa polypeptides of the oxygen-evolving complex accumulate as processing intermediates (see Discussion). No signals could be detected for the plastid-coded components of the photosynthetic machinery in Western blots, but pulse-labelling experiments in the presence of anti-metabolites of protein synthesis (cycloheximide and/or chloramphenicol) indicated low residual translational activity for some of these genes, e.g. rbcL, atpA and atpD, as judged from co-migration in different SDS and SDS/urea containing gel systems (data not shown). Moreover, an antiserum raised against the plastome-encoded rpl2 gene product verified that translation takes place in rpo-deficient plastids ( Fig. 5).

Figure 5.

Western analysis of total proteins from tobacco wild-type and rpo mutants.

The identities of the genes for the respective proteins are indicated at the margin (see also Experimental procedures).

Lipid and pigment analysis

Young leaves of rpoA-, rpoB- and rpoC1-deficient T0 and T1 material are faintly coloured, whereas fully developed material appears completely bleached. In accordance with this observation, corresponding extracts contained traces (≤ 1% of wild-type amounts) of chlorophyll (illustrated in Fig. 6a for the rpoB mutant). No chlorophyll could be detected in the extracts of older leaves. This suggests light-dependent bleaching of chlorophyll. To monitor whether the capacity of chlorophyll biosynthesis was impaired in the transformants, leaf sections of rpoA, rpoB and rpoC 1 mutants were incubated with 5-aminolevulinate. After 20 h incubation in darkness, significant amounts of protochlorophyllide were found in all mutant lines (shown in Fig. 6b for the rpoC1 mutant). This implies that the enzymes of tetrapyrrole biosynthesis, at least the steps from 5-aminolevulinate to protochlorophyllide, are operating in the rpo material. Protochlorophyllide which accumulated after incubation with 5-aminolevulinate in the tissues of wild-type and mutants was expectedly not phototransformable. In contrast, protochlorophyllide which accumulated in etiolated wild-type seedlings without incubation with 5-aminolevulinate, is phototransformable (data not shown). Etiolated seedlings of rpoA  and rpoB  mutants did not contain any protochlorophyllide (detection limit: 2–3% of wild-type levels).

Figure 6.

(a) Absorption spectra of the pigment extract in ethyl acetate of young leaves of wild-type (A) and of the rpoB  mutant (B).

(b) Absorption spectra of the pigment extract from the rpoC1  leaf sections infiltrated for 20 h with 5-amino-levulinate in darkness (A) before and (B) after 10 min irradiation.

The leaves of all rpo mutants contained reduced levels of total carotenoids, about 1–2% of the carotenoid content of green wild-type leaves ( Table 1). Based on fresh-weight, the carotenoid amount of the mutants was in the same order of magnitude as that of etiolated wild-type leaves. The mutants contain the customary principal components, such as lutein, β-carotene, violaxanthin and neoxanthin. Relatively high levels of antheraxanthin and zeaxanthin were detected in the mutants but not in the wild-type ( Fig. 7). These carotenoids are constituents of the xanthophyll cycle involved in the dissipation of energy (non-photochemical quenching; Horton & Ruban 1992). Transformation of violaxanthin into antheraxanthin and zeaxanthin is a typical response to light stress. As shown in Table 1, this reaction was found in all rpo mutants, but not in wild-type plants, already under a moderate increase of light intensity. This observation supports the suggestion of chlorophyll bleaching outlined above.

Table 1.  Relative amounts (%) of carotenoids in wild-type and rpo mutants under ‘low light’ (LL, 80 μE m–2 s1) and ‘high light’ (HL, 200 μE m–2 s1) conditions 1
NeoxanthinViolaxanthinAntheraxanthinLutein + zeaxathinβ-carotin

  1. 1 The peak areas of the HPLC elution profile (see Fig. 7) were evaluated (total carotenoids = 100%). Zeaxanthin forms a shoulder at the peak of lutein, therefore the sums of both are given. The increase in this peak area upon transfer from LL to HL is mainly due to the increase of zeaxanthin as shown by the increase of the shoulder.

Figure 7.

HPLC of the pigments from (a) green wild-type, (b) etiolated wild-type, and (c) light-grown rpoC1 plants.

The pigment pattern of rpoA and rpoB  plants (not shown) is similar to that of rpoC1  plants. The pigment extract was separated on an RP-18 column with an acetone/25 m m NH4-acetate gradient as described in Experimental procedures. Detection was by a diode array instrument, set at 350–750 nm. The pigments were identified by absorption spectra of the single peaks and by fluorescence as follows: (1) neoxanthin, (2) violaxanthin, (3) antheraxanthin, (4) zeaxanthin, (5) lutein, (6) chlorophyll b, (7) chlorophyll a, (8) β-carotene.

The lipid content of the mutants was only 1–2% that of the wild-type plants (based on equal leaf areas).Two-dimensional chromatography of total lipids from mutant and wild-type leaves uncovered the presence of all thylakoid-specific lipids described by Bratt & Åkerlund (1993), albeit in reduced amounts (data not shown).


We have employed a transplastome approach to selectively eliminate the plastid-located, phylogenetically ancient RNA polymerase by gene disruption. The data confirm, revise and extend findings obtained with the rpoB-deficient phenotype ( Allison et al. 1996 ). All six mutant strains were investigated in parallel with the approaches chosen. No significant difference was noted between the individual disrupted genes and with the different polarities of cassette insertion (see also Serino & Maliga 1998). The resulting lines are genetically stable, off-white, and can be propagated remarkably well under heterotrophic conditions. The plants grow vigorously ( Fig. 2), they flower, are fertile, and even set viable seed.

One of the surprising findings of our study is the presence of a minute fraction of wild-type plastome in phenotypically stable material in which segregating green sectors were never observed, not even in more than 107 protoplast-derived clonal lines. Although long-term maintenance of wild-type plastome copies in transplastomic tobacco lines has been noted before ( Kofer et al. 1998 ), their possible presence under the outlined conditions is unexpected and would be in discordance with earlier findings in tobacco ( Allison et al. 1996 ). However, it would be consistent with those in Chlamydomonas reinhardtii where a homoplastomic rpoB strain could not be obtained ( Rochaix 1995). In principle, the detection of wild-type sequences could either represent genuine plastid chromosomes or promiscuous DNA sequences in the nucleus and/or mitochondria. Under the chosen conditions, the PCR detection limit was only 30 template molecules, which is equivalent to contaminating DNA from only seven nuclei, if a single promiscuous DNA copy were present per haploid genome. A considerably lower degree of contamination would be sufficient if such sequences were of mitochondrial origin. The data clearly show that most, if not all, of the ‘wild-type’ PCR signal is not derived from plastids and therefore from ‘promiscuous’ DNA. Although the presence of a minute fraction of wild-type plastomes in the phenotypically homogenous material cannot be rigorously excluded at present, and hence even if some of the wild-type PCR signals observed with Percoll-purified rpo plastids were of plastid origin, these cannot account for the biochemical findings outlined in this study. The aspect of homoplastomy cannot be finally settled with the presently available methods and deserves further study.

It had been noted that in rpoB-deficient material plastid chromosomes are transcribed from a distinct set of promoters and that the transcripts are processed ( Allison et al. 1996 ). This suggested that the nuclear-coded polymerase is involved in the expression of only a distinct set of genes, predominantly of those contributing to establish the organelle’s genetic system, but not of those for the photosynthetic machinery (see Hajdukiewicz et al. 1997 ). Employing ultrastructural, Western, Northern, pigment and ribosome analyses and pulse-labelling experiments in the presence of compartment-specific inhibitors of protein synthesis, we have checked the overall biogenetic potential of the residual organelle including the presence of functional ribosomes in rpo plastids, and have predominantly addressed three aspects of general interest: (i) are plastid transcripts decoded in the mutant material and, if so, are the products stable; (ii) how are the nuclear and plastid genetic machineries, including the two RNA polymerase activities, integrated temporally and spatially into the regulatory network of the cell; and (iii) what is the ontogenetic pattern and origin of the thylakoid system? Our data show that the functional consequences of plastid rpo gene disruption are considerably more complex than previously anticipated.

The capability of chlorophyll and lipid biosynthesis and the presence of fully processed stroma and thylakoid polypeptides of nucleo-cytosolic origin, demonstrate that the nuclear-coded components for the organelle are basically provided by the cell. The nuclear-coded polypeptides include lumenal components that, in the case of plastocyanin and the 33 kDa protein of the oxygen-evolving system but not of the 23 kDa and 16 kDa proteins of the same assembly, were terminally processed ( Fig. 5). This implies that the former proteins have crossed three membranes, thylakoid membranes in addition to envelope membranes, suggesting that the vesicles observed represent residual thylakoids. These two proteins are translocated via the Sec-like route across the thylakoid membrane. On the other hand, the accumulation of the processing intermediates of the 23 kDa and 16 kDa polypeptides ( Fig. 5) suggests that the ΔpH route, which these proteins use ( Robinson & Klösgen 1994), does not operate, probably because no proton gradient is established across the residual thylakoid membranes. Furthermore, the synthesis of some of the major thylakoid polypeptides in amounts similar to those found in the wild-type and the formation of starch grains suggest a massive energy transfer to the residual organelle most likely from a respiratory source. These findings have an interesting analogy to etioplasts which virtually do not possess a functional thylakoid system yet are capable of synthesising a substantial fraction of organelle polypeptides regardless of their genetic origin and including all lumenal proteins studied ( Herrmann et al. 1992 ). However, there are significant differences between etioplasts and rpo plastids, apart from the finding that the proteins of the ΔpH route are terminally processed in the former.

Differing from etioplasts which generally synthesise substantial amounts of plastome-encoded thylakoid proteins (e.g. of the cytochrome complex and ATP synthase), no plastid-encoded polypeptides of the photosynthetic machinery could be detected by Western analysis ( Fig. 5). On the other hand, pulse labelling experiments and the presence of the plastome-encoded ribosomal protein Rpl2 ( Fig. 5) clearly revealed residual translational activity, commensurate with ultrastructural work ( Fig. 3) and the finding of sedimentable ribosome-like particles ( Fig. 4). Independent, although indirect, evidence that rpo-deficient plastids possess translational activity was also obtained by investigating the processing status of appropriate individual plastid-encoded transcripts, notably editing sites in ndhB ( Karcher & Bock 1998) and the intron of rpl2 transcripts. The rpl2 intron is excised in the rpo-deficient material (data not shown) but not in ribosome-deficient plastids from various mutant lines which had suggested that a plastid-encoded product is involved in its splicing, presumably the reading frame ycf14 (matK) ( Hess et al. 1993 ). If correct, splicing of the rpl2 transcripts in the mutants would, therefore, also stand for translation of ycf14 in an rpo-deficient background.

The organelle ribosome, similar to the thylakoid membrane, is of dual genetic origin and approximately one-third of the ribosomal proteins originates in plastid genes. This suggests that the organelle ribosomes have been maintained during the process of plastid segregation after transformation rather than that the material is not homoplastomic. In barley, rpl2 has been deduced to be exclusively transcribed by the nuclear-coded RNA polymerase ( Hübschmann & Börner 1998). However, some ribosomal proteins are derived from heterogenic operons encoding components of the photosynthetic machinery as well, such as the psaA-psaB-rps14 operon for instance. Therefore, such operons or part of them should be transcribed by both enzymes. Three crucial questions emerge from these findings: (i) what is the mode and level of regulation that leads to the perpetuation of functional ribosomes in the absence of the plastid-derived RNA polymerase; (ii) how can such a defect cause a differential translational and/or a group-specific protein stability response, in spite of the presence of nuclear-coded subunits of thylakoid polypeptide complexes; and (iii) how do the two kinds of RNA polymerase interact in ontogeny? Further work is required to establish the causal relationships between these intriguing observations which demonstrate that regulation is not merely transcriptional.

Another striking finding of this study is the existence of a residual thylakoid system in the absence of the plastid-encoded enzyme, assumed to read loci and operons encoding components for the photosynthetic machinery (see Hajdukiewicz et al. 1997 ). Obviously, the rpo-deficient genetic background and the resulting rudimentary organelle do not abolish a basic membrane flow of the thylakoid system. This includes not only membrane flattening and stacking-like structures but also vesicles apparently detaching from the inner envelope. There is considerable uncertainty regarding the origin of thylakoid membranes in a multi-cellular plant. At least two non-exclusive hypotheses have been proposed: either the thylakoid system forms from pre-existing prothylakoids, or it emanates and detaches from the inner envelope membrane of the organelle (e.g. Menke 1960).

Although dynamic projections are inherently difficult to deduce from ultrastructural approaches alone that present a static view of membranes only, the temporal succession of different rpo plastid morphotypes indicates a transient appearance and disappearance of vesicles depending on the developmental stage and the prevailing light conditions. Furthermore, rpo etioplasts can form prolamellar-like bodies that evaginate some thylakoid-like structures after illumination. In the initial cells constituting the apical meristem, vesicles and protrusions are not or only rarely seen in plastids of rpo material. Subsequently, vesicle budding must be a frequent event at certain stages of organelle development, as judged from both an increase in budding structures and vesicle population. Although it is not yet clear what parameters and mechanisms are involved in vesicle formation and membrane flow, it is evident that genes transcribed by the plastid-encoded RNA polymerase apparently do not play a significant role in this process. Consequently, the underlying regulatory and mechanistic key steps leading to the basic membrane pattern should all be nuclear-controlled. The rpo phenotype, because of its genetic stability and the lack of the potentially intermingling organelle decoding system for polypeptides involved in photosynthesis, offers a unique opportunity to dissect the overall process of thylakoid biogenesis into principal blocks of events, i.e. without integration of a functional photosynthetic machinery. It will be highly interesting to explore the succession of events, how all the different structural components are brought together and thus produce an active thylakoid system.

Experimental procedures

Protoplast isolation

Growth of donor plants and protoplast isolation essentially followed the procedures described by Koop & Kofer (1995) and Koop et al. (1996) . Shoot cultures of Nicotiana tabacum L. cv. Petit Havanna were axenically grown from seedlings at 25°C with 8/16 h dark/light cycles at 0.5–1 W m–2 (Osram L85 W/25 Universal White fluorescent lamps) in jars with agar solidified B5-medium ( Gamborg et al. 1968 ). Fully expanded leaves were collected for protoplast isolation after approximately 5 weeks of culture.

Plastid transformation

For each transformation construct (see below) two aliquots of 105 protoplasts in 100 μl PTN ( Koop et al. 1996 ) were treated with 50 μg plasmid DNA (purified by Qiagen columns) in 25 μl TE-buffer, pH 5.6, for 10 min. Transformation was initiated by the addition of 125 μl 40% PEG-solution, prepared by dissolving 0.41 g Ca(NO3)2·4 H2O, 1.27 g mannitol, and 10 g PEG 1500 (Merck, Darmstadt, Germany) in bi-distilled water to give a total volume of 17.5 ml. After 7.5 min, 125 μl, and after a further 2 min, 2.625 ml protoplast culture medium were added (final volume 3 ml).

Alginate embedding and protoplast culture

A substantially improved culture protocol ( Dovzhenko et al. 1998 ) using polypropylene grids rather than stainless steel mesh for supporting thin alginate layers was employed. In brief, 3 ml of protoplasts treated for transformation were mixed with an equal volume of 2.8% alginate (alginic acid, sodium salt, ‘low viscosity’, (Sigma, Deisenhofen, Germany) dissolved in 10 m m MES, pH 5.7, 10 m m MgCl2, mannitol to 550 mOsmol). Aliquots (1 ml) of this mixture were applied to the centre of an agar layer (1% in 10 m m MES, pH 5.7, 20 m m CaCl2, mannitol to 550 mOsmol), and a polypropylene grid (12 × 12 meshes, 2 × 2 mm mesh size, Scrynel PP2000, K. H. Büttner GmbH, Wasserburg. Germany; Golds et al. 1992 ) was immediately inserted into the protoplast/alginate mixture (see also Kuchuk et al. 1998 ). After 30 min the meshes carrying the solidified alginate layers were carefully removed from the agar layer and transferred (upper side down) to a 6 cm Petri dish containing 3 ml of PCN culture medium ( Koop & Kofer 1995; Koop et al. 1996 ). Approximately 2 × 105 untreated feeder protoplasts were included in the liquid culture medium during the initial 3 week culture period. Protoplasts were grown at 25°C in the dark for 1 day and then transferred to the light with 8/16 h dark/light cycles at 0.5–1 W m–2 (Osram L85 W/25 Universal White fluorescent lamps).

Selection and regeneration of transplastomic plants

Selection with 500 mg l–1 of each spectinomycin and streptomycin started after 3 weeks of culture. Appearing green colonies were transferred to RMOP medium ( Svab et al. 1990 ) lacking antibiotics to allow for segregation of wild-type (green) and transformed (white) sectors. Leaf explants (1–2 mm) excised from white sectors ( Fig. 2d) and white shoots regenerated on RMOP medium ( Fig. 2e) were transferred to 720 ml glass jars containing VBW medium ( Aviv & Galun 1985), modified by the addition of 0.5% casein hydrolysate (vitamin-and salt-free; ICN Biochemicals, Cleveland, OH, USA). Transplastomic lines were kept as shoot cultures on the same medium with transfer of shoot tips at 3–4 week intervals.

Plasmid construction, DNA and RNA analyses

The recombinant plasmids pTB7 and pTBa1 encoding rpoB, C1 and C2 and rpoA, respectively, from tobacco ( Sugiura et al. 1986 ) were used as the source for the construction of the transformation vectors ( Fig. 1). The aadA cassette ( Koop et al. 1996 ) was inserted into the unique BstEII restriction site within rpoB of plasmid pTB7 (nucleotide positions 18 942–29 830 of the complete tobacco chloroplast genome sequence; GenBank accession number Z00044; S54304). The resulting recombinant plasmids, differing in the polarity of cassette insertion, were designated prpoBI and prpoBII, respectively. The segment of the tobacco plastid chromosome containing rpoC1 was excised from pTB7 as a PstI/SalI fragment (position 20,287–26 722). The fragment was ligated into Bluescript KS vector, restricted with PstI and SalI, generating plasmid p-rpoC1. The aadA cassette was then inserted into the unique Bpu1102I restriction site of p-rpoC1, yielding plasmids prpoC1–I and prpoC1–II, respectively. The rpoA-containing segment of the tobacco plastid chromosome was excised as a PstI fragment (nucleotide positions 79 333–83 298) from plasmid pTBa1. The fragment was ligated into Bluescript vector generating plasmid p-rpoA. Insertion of the aadA cassette into the unique restriction site PmII of the plasmid resulted in plasmids prpoAI and prpoAII, respectively. Figure 2–7 as well as Table 1 show results derived from plants transformed with plasmids prpoAI, prpoBI and prpoC1–II, respectively.

Southern analysis was carried out using standard procedures described in Sambrook et al. (1989) . Gene-specific DNA and cDNA fragments were amplified using PCR primer pairs 5′-TCCTGGCTCAGGATGAACG-3′ and 5′-AACGATGGCAACTAAACACG-3′ for 16S rDNA, 5′-GGAGAATTAGGGTTCGATTC-3′ and 5′-ACTTCTCCTTCCTCTAAATG-3′ for 18S rDNA, 5′-GCTAAATAAATCAATGGGCAG-3′ and 5′-AATGGGATCTAGAGAGACC-3′ for accD, 5′-GATAATGAAGAACCATAAGA-3′ and TACTACCACTTGGATCTGGA-3′ for psaA, 5′-CTGTGATAAGTTGTGAAGAG-3′ and 5′-CGGTTTACGCACTAATGAAG-3′ for psaB, 5′-GGAGCAATGAACCTATTTGA-3′ and 5′-ATGGAAAGAACAGGTTCAAA-3′ for psbC, 5′-GTCGGTCCGGTCTATTGCTC-3′ and 5′-TCCAAGCGCGAATACCTTCG-3′ for psbD, 5′-GAGACTAAAGCAAGTGTTGG-3′ and 5′-GATCATTTCTTCGCATGTACC-3′ for rbcL, 5′-GAGCTTCGATAGCAG- CTAGG-3′ and 5′-ATCCCTACCTTATTGACGGC-3′ for psbA, 5′-TGTATTTCCGGAGTATGAG-3′ and 5′-CTCTATAAAGGCCCAGAAA-3′ for petB, 5′-ACCCGGGTTATTCTATTC-3′ and 5′-CCAAACGGACCTCCCCAGATGG-3′ for rpl2, 5′-AGTATGTAAATGTATTC- ATTTCC-3′ and 5′-GGGATGTAACTCCTATGCC-3′ for rpoA, 5′-TCCAAAACTTGAGATAATGGG-3′ and 5′-TGATTGGAATAGAAAATTTCGG-3′ for rpoB, 5′-CCATTAGATGGGGCTCGC-3′ and 5′-GACTTTTTGGTTAGACGCG-3′ for rpoC1, and 5′-TTGCCAACTACCTTAGTGATC-3′ and 5′-GAAGCGGTTATCGCCGAAG-3′ for aadA. PCR was performed using the Taq DNA polymerase system from Qiagen (Hilden, Germany) according to the supplier’s instructions. Using this system we routinely amplified sequences from as few as 30 template copies in a 30 μl reaction volume. Double-stranded DNA probes were prepared by 32P-random priming of PCR-generated fragments. For sequence determinations, amplification products were precipitated with two volumes of ethanol in the presence of 0.4 m ammonium acetate, dissolved in water and directly sequenced using the BigDyeTM Terminator cycle sequencing ready reaction kit (PE Applied Biosystems, USA). Sequencing products were analysed using an ABI PRISM 377 DNA Sequencer (PE Applied Biosystems, USA).

Polysomes were prepared as described by Barkan (1993). Total leaf and polysomal RNA was isolated using TRIzol reagent (Gibco/BRL, USA). RNA was electrophoretically separated in 1% agarose-formaldehyde gels and transferred onto nylon membranes according to Towbin et al. (1979) . Hybridisation with double-stranded DNA probes was carried out in Rapid Hybridisation Buffer (Amersham, Braunschweig, Germany) overnight at 60°C. DNase I-treated RNA preparations were reversely transcribed with a hexanucleotide random primer mixture using RNase H Moloney Leukemia Virus reverse transcriptase (SuperscriptTM, Gibco/BRL, USA) following the manufacturer’s instructions.

Polyacryl amide gel electrophoresis, Western analysis and protein labelling experiments

SDS-PAGE was performed in 10% (w/v) gels as described by Laemmli (1970). Samples were heated for 5 min at 70°C, mixed with 1/10 volume of glycerol-dye solution and immediately applied to the gel. Electrophoresis was conducted at a constant current of 20–25 mA for 2 h at room temperature.

Proteins separated in SDS polyacryl amide gels were electrophoretically transferred to nitrocellulose filters (Schleicher and Schüll, Dassel, Germany). The filters were probed with monospecific polyclonal antisera and developed with goat anti-rabbit IgG serum conjugated with horseradish peroxidase. The peroxidase activity was detected by reaction with luminol and H2O2 (Amersham). The antisera used were elicited against the large subunit of ribulose bisphosphate carboxylase/oxygenase (gene: rbcL), thioredoxin m (trxA), ferredoxin-thioredoxin oxidoreductase (ftrA), ferredoxin NADP+ oxidoreductase (petH), the cytochromes f (petA), b6 (petB) and b559 (psbE), subunits IV (petD) and the Rieske protein (petC) of the cytochrome complex, LHCII (lhcb), the 33, 23 and 16 kDa polypeptides of the oxygen-evolving system (psbO, psbP and psbQ, respectively), the D1 protein (psbA), subunit CP43 of photosystem II (psbC), plastocyanin (petE), subunit delta (atpD) of the ATP synthase, the protein encoded by the plastid reading frame ycf3 and the ribosomal proteins L2 (rpl2) and L21 (rpl21).

Total leaf proteins were labelled by inserting the petiole into 4 μl 35S-methionine (15 μCi μL–1; 1000 Ci mmol–1) in the presence of cycloheximide (200 μg ml–1) to inhibit translation on cytosolic ribosomes or in the presence of chloramphenicol (10 μg ml–1) to inhibit translation in plastids (L. Eichacker, personal communication). Labelled proteins separated by SDS-PAGE as described above were detected by autoradiography of dried gels.

Pigment and lipid analysis

For the preparation of etiolated tobacco seedlings (wild-type), sterilised tobacco seeds were germinated in complete darkness for 13 days at 27°C on agar medium. Half of the seedlings were irradiated with white light (36 μE m–2 s–1) for 10 min, kept in darkness for 20 min, and then immediately frozen in liquid nitrogen. Control seedlings were collected and frozen without illumination.

Leaves of rpo transformants were harvested and cut longitudinally into equal halves. Duplicate samples were incubated for 20 h at room temperature with 1.5 ml of 5 m m 5-aminolevulinic acid (ALA) in 35 m m phosphate buffer, pH 7.5, or without ALA. Two samples, one with and one without ALA, were given light and dark treatments, and their pigment composition was analysed as described below. The remaining two samples were not illuminated prior to pigment analysis. All samples were frozen in liquid nitrogen.

The plant material was pulverised in liquid nitrogen and extracted three times with 80% acetone. The pigments were quantitatively transferred into ethyl acetate and spectra were recorded with a diode array instrument (DAD 2010, J & M, Aalen, Germany). Chlorophylls, protochlorophyllide and total carotenoids were determined according to Lichtenthaler (1987) with modifications from Porra (1991). For HPLC analysis, the ethyl acetate was removed with a continuous stream of dry nitrogen. The residual material was dissolved in acetone and analysed on a Hypersil ODS (5 μm) reverse-phase column (Shandon, 200 × 4 mm) with a gradient of A (100% acetone) and B (50% acetone, 50% 25 m m ammonium acetate) and a flow rate of 1.0 ml per min as follows: 30% A/70% B for 5 min, followed by a linear gradient to 100% A within 20 min, and at least a further 7 min at 100% A. Chlorophylls were detected by absorption with the diode array instrument (see above) and by fluorescence (RF 551 Shimadzu) with the excitation and emission wavelengths set at 425 and 635 nm, respectively. Carotenoids were detected by absorption at 450 nm.

Lipids were extracted from the leaves of wild-type with chloroform/methanol (1:1, v:v). The extract was washed with water, applied to HPTLC (Merck KG60) and developed with chloroform/methanol/water (65:25:4, v:v:v) in the first, and with chloroform/methanol/acetic acid/water (85:15:10:3.5) in the second direction. The chromatogram was sprayed with a ferrous sulfate/potassium permanganate solution that was prepared by combining the following solutions: 2 g FeSO4 in 120 ml H2O, 180 mg KMnO4 in 80 ml H2O, and 6 ml conc. H2SO4 (G. Henkelmann, personal communication), and developed at 120°C for 10 min.

Electron microscopy

Small pieces of the leaves, shoot tips or root tips were fixed immediately after collection with 2.5% glutardialdehyde in 75 m m sodium cacodylate, 2 m m MgCl2, pH 7.0, for 2 h at room temperature. Subsequently, the material was rinsed several times in fixative buffer and post-fixed for 1 h with 1% osmium tetroxide in fixative buffer at room temperature. After two washing steps in distilled water, the tissue pieces were stained en bloc with 1% uranyl acetate in 20% acetone for 1 h. Dehydration was performed with a graded acetone series. Tissue samples were then infiltrated and embedded in Spurr’s low-viscosity resin ( Spurr 1969). After polymerisation, ultra thin sections with thickness between 50 and 70 nm were cut with a diamond knife and mounted on collodion-coated copper grids. The sections were post-stained with aqueous lead citrate (100 m m, pH 13.0). All micrographs were taken with EM 912 or EM 109 electron microscopes (LEO, Oberkochen, Germany).

Small sectors of leaf tissue were rapid-frozen in nitrogen slush. A frozen specimen was transferred to the cryo preparation chamber (DT1500 HF Cryotrans; Oxford Instruments, UK) fractured, etched and coated with a magnetron sputter coater in the cryo preparation chamber. It was then transferred to the cold stage in the field emission SEM (LEO 1500 with Gemini column). All images were taken at 2.6 kV at a working distance of 4 mm.


The skilful technical assistance of Ms Martina Reymers, Ms Silvia Dobler and Ms Petra Winterholler is gratefully acknowledged. Plasmids pTB7 and pTBa1 were kindly provided by Prof. M. Suguira, the spray reagent for lipid visualisation by Dr H. Henkelmann. We thank A.C. Robins (Oxford Instruments, UK) for cryo-preparation and Dr H. Jaksch (LEO Elektronenmikroskopie GmbH, Oberkochen Germany) for cryo-scanning electron microscopy. Dr L. Eichacker kindly made the pulse labelling protocol available prior to publication. Some of the antisera used were supplied by Dr J. Feierabend (Frankfurt/Main), Dr J.-D. Rochaix (Geneva, Switzerland), Dr R. Scheibe (Osnabrück), Dr A. Subramanian (Tucson, USA), and Drs T. Börner and W. Hess (Berlin). This work was supported by Deutsche Forschungsgemeinschaft (SFB 184, Ko 232/13–1), and the Fonds der Chemischen Industrie.