Transgenic expression of aminoglycoside adenine transferase in the chloroplast: a selectable marker of site-directed transformation of chlamydomonas.

Expression vectors for Chlamydomonas reinhardtii chloroplast transformation have been constructed with transcription and translation signals from chloroplast genes. The bacterial aadA sequence, coding for aminoglycoside 3" adenyl transferase, was inserted in these vectors and introduced into the C. reinhardtii chloroplast by particle gun transformation. The stable transgenic expression of this foreign protein in the chloroplast confers spectinomycin and streptomycin resistance to the transformed cells. This new marker can be used as a reporter of gene expression, and as a portable selectable cassette for chloroplast reverse genetics. Targetted gene disruption mutants of loci required for photosynthesis, tscA and psaC, were thus obtained. A gene disruption of an unidentified open reading frame, ORF472, remained heteroplasmic, suggesting that it has a vital function.


INTRODUCTION
The development of efficient chloroplast transformation in C. reinhardtii using microprojectile bombardment has opened the way for a systematic analysis of chloroplast gene function and expression. The initial report of chloroplast transformation with a particle gun made use of a photosynthetic marker for the selection of transformed cells (1). A deletion mutant of the chloroplast atpB gene was used as a host, and was bombarded with a fragment of wild-type DNA spanning the deletion. This provides a powerful selection, but has limitations for experiments where photosynthetic genes are being analyzed because only phototrophic cells can be selected. Similar strategies were used with other chloroplast photosynthetic genes such as rbcL or psbA (2). Another approach has been to use the mutations in the psbA gene which confer resistance to certain herbicides (3), or mutations in the ribosomal RNA genes which confer resistance to antibiotics such as spectinomycin, streptomycin or erythromycin (2). Co-transformation with such an rDNA drug resistance marker and a separate mutated atpB gene was further shown to allow the recovery of mutant strains bearing both markers (4). Transformation of the tobacco chloroplast has also been achieved using an analogous rDNA marker confering resistance to spectinomycin (5).
Here I describe transformation vectors that allow the expression of foreign genes in the C. reinhardtii chloroplast, as illustrated with the coding sequence of aadA, a bacterial gene that confers resistance to aminoglycosides such as spectinomycin and streptomycin. This is the first example of stable transgenic expression of a foreign protein in a chloroplast. The chimeric aadA gene provides a portable, dominant selectable marker which can be used for site-directed mutagenesis independently of photosynthetic function. The aadA sequence can also be used as a reporter to study chloroplast gene expression.

MATERIALS AND METHODS
Strains and media C. reinhardtii 137c and the chloroplast mutant H13 (6) were grown as described by Harris (7). Where necessary the media (Tris acetate phosphate medium (TAP) or high salt minimal medium (HSM)) were solidified with 2 % Bacto agar (Difco) and supplemented with spectinomycin (Sigma) or streptomycin (Calbiochem), taking into account the fact that these antibiotic preparations were only partially pure. For genetic crosses (7) the agar was washed with distilled water and dried before use. preparation). A 3.1 kb KpnI-Clal fragment from pR 12-3 (the Clal site is in the polylinker) was introduced into LSX cut with KpnI and Clal to provide the tscA marker in plasmid LSX-R 12.
To improve the translational fusion of inserted coding sequences, the 5'rbcL sequences in LSX were then replaced with the 5'rbcL or 5'atpA segments prepared by the polymerase chain reaction (PCR) as follows. Plasmid R15.0 (9) was linearized with EcoRI and HindHI and used as a template for S cycles of amplification as described previously (10), either with oligos LS3 and LS5 for 5'rbcL or oligos atpA3 and atpA5 for 5'atpA (table 1). The 5'rbcL fragment was first subcloned in pBluescript KSat the SmaI site, from which it was excised with ClaI and BamHI. The 5'atpA fragment was obtained by treating the products of the PCR reaction directly with the same enzymes. The 0.39 kb 5'rbcL and the 0.68 kb 5'atpA Clal -BamHI fragments were introduced into ClaI BamHI digested LSX-R12 to obtain the plasmid vectors rbcX and atpX (figure 1). The sequence of the expression site polylinker was verified.
The coding sequence of aadA was prepared by PCR as above using oligos AAD5 and AAD3 (table 1) and the 2.0 kb EcoRI fragment of pHP45-Omega (11) as a template. This coding segment starts at the Met codon at position 402 in the aadA sequence (12), which is probably the true start site (13). The aadA segment was subcloned at the SmaI site of pBluescript KS-, and the 0.81 kb NcoI-PstI fragment was excised and introduced into NcoI PstI digested rbcX or atpX to obtain plasmids rbcX-AAD and atpX-AAD (figure 1). The 1.9 kb EcoRI-Sacd fragment of atpX-AAD was further subcloned into EcoRI SacI digested pUC 18 to obtain the plasmid pUC-atpX-AAD (figure 1). The construction of the psaC: :AAD gene disruption is described elsewhere (14). The tscA::AAD disruption was obtained by partially digesting the plasmid pR12.3 (see above) with Scal, treating with Klenow DNA polymerase I fragment to generate blunt ends, isolating the full length linear molecules and introducing the EcoRV -SmaI aadA cassette from pUC-atpX-AAD. The E. coli transformants were selected on LA containing 50 gg/ml ampicillin and 25ug/ml spectinomycin.
The ORF472::AAD disruption was derived from pCM3 (15) by digesting with NsiI, treating with T4 DNA polymerase to obtain blunt ends and introducing the EcoRV -SmaI cassette as above.

Transformation
Transformation with a particle gun was as described previously (16,10). For spectinomycin selection the cells were carefully resuspended and split into three aliquots which were plated in 0.6% top agar on TAP medium containing 100 Ag/ml spectinomycin, and grown in dim light (approx 300 lux). After 2-3 weeks the colonies were restreaked on TAP + spectinomycin and grown for approximately one week. The cells were then resuspended and grown for 1 day in liquid TAP medium containing 20 tg/tml spectinomycin, diluted and plated on TAP + spectinomycin and on HSM minimal medium to obtain single colonies and to measure plating efficiency. These steps of restreaking and replating were repeated several times sequentially to obtain pure and when possible homoplasmic lines. Representative transformants were grown for DNA and RNA extraction and analysis (17).
Aminoglycoside adenyl transferase assay Cells were grown to a density of approximately 107/ml in TAP medium, harvested by centrifugation, washed in distilled water and finally resuspended in 1/40 th volume AAD-RB (25 mM Tris-HCl pH 8.0, 5 mM MgCl2, 100 mM NH4Cl, 0.5 mM DTT). The cells were broken by sonication, and the debris were removed by centrifugation for 30 min at 20000g. (The clear supernatants can be stored frozen with only a small loss in activity.).
AAD catalyses the transfer of an adenyl moiety from ATP to streptomycin; activity can be measured using radiolabelled ATP. The assay is based on the binding of streptomycin adenylate, but not ATP, to negatively charged phosphocellulose paper (18). The reactions were in 30 /O AAD-RB containing 15 ,ul crude sonic extract, 100 /LM rATP, 60 Atg/ml streptomycin and 0.1 uCi alpha32P-rATP. Blank reactions lacking either the crude extract or streptomycin were treated in parallel. After 30 min at 35°C, duplicate 10 ,tl aliquots were spotted on phosphocellulose paper filters (1.5 x 1.5 cm2, Whatmann P81) and allowed to adsorb for 30 seconds. The filters were immediately washed in water, first at 75°C, then 4 times at room temperature. The filters were dried and counted in toluene based scintillation fluid to measure the streptomycin adenylate formed. Total protein was measured using the Bio-Rad protein assay.

Expression vectors
Two vectors for the expression of polypeptide coding sequences in the chloroplast of C. reinhardtii have been constructed (figure 1). Both contain the promoter, the untranslated leader and the translation initiation ATG codon of C. reinhardtii chloroplast genes. Any coding sequence of interest can thus be placed under the control of homologous chloroplast transcription and translation signals. The first vector, rbcX, contains a 363 base pair (bp) fragment from the rbcL gene: 268 bp of upstream sequence, the 92 bp untranslated 5 '-leader and the ATG (9). The second one, atpX, contains a 653 bp fragment form the atpA gene: 221 bp of upstream sequence, the 429 bp 5'-leader and the ATG. Just The oligonucleotides used as primers for the PCR are listed in 5'-3' orientation with the bases matching the template in uppercase, and the linkers or site specific changes to improve codon usage in lower case. Numbering is as in the references for rbcL and atpA (9) and for aadA (12).
downstream of the ATG in these vectors, there is a set of cloning sites, followed by a 437 bp fragment containing the 3'-end of the rbcL gene (figure la). Both vectors also contain the tscA gene as a selectable marker that can restore the photosynthetic growth of tscA deletion mutants such as H13 (10; Choquet, Girard-Bascou and Rochaix, in preparation  (14). For the ORF472 disruption, a 232 bp NsiI fragment was replaced by the aadA cassette in pCM3, a 2.2kb MboI fragment from Ba7 cloned in the vector pJD2 (15). For the tscA disruption, the aadA marker was inserted into one of the Scal sites of R12. The hybridization probes used in figure 3 are indicated by brackets: R02 (EcoRI fragment) and R12KB (KpnI-BgII fragment).
Expression of the aadA gene The E.coli aadA gene encodes the enzyme aminoglycoside 3 "-adenyltransferase (AAD), and confers resistance to streptomycin and spectinomycin. Its coding sequence (12,13) was inserted in the two expression vectors to obtain the plasmids rbcX-AAD and atpX-AAD (figure 1). The plasmids were introduced into the C. reinhardtii host H13 by bombardment with a particle gun, and transformants were selected for photosynthetic growth by rescue of tscA. The transformants were tested for growth on media containing spectinomycin or streptomycin. All eight atpX-AAD transformants tested and seven out of eight rbcX-AAD transformants showed resistance to spectinomycin (150,g/ml). One rbcX-AAD transformed strain (5A) was not resistant. All the atpX-AAD transformants were also resistant to streptomycin (50 /kg/ml) but the rbcX-AAD transformants were not. These levels of resistance are well above those of the untransformed host H13, which is only resistant to approximately 50 jig/ml spectinomycin and 20 ,tg/ml streptomycin. The transformed strains are thus apparently expressing the foreign DNA. Drug resistance, particularly to streptomycin, is higher with the atpX vector than with rbcX. Seven strains were checked for the presence of the aadA sequences by Southern analysis (data not shown). Strain 5A, which was not resistant, did not have detectable aadA sequences. Three drug resistant rbcX-AAD transformants and three atpX-AAD transformants had incorporated the foreign DNA. These transformants had a complex pattern of integrated fragments suggesting multiple recombination events. This is probably because there are only short segments of homology between the host and the transforming DNA, because these segments are found at three different loci (tscA, rbcL and atpA) or because there are also regions of repetitive DNA in the sequences adjacent  Northern analysis of chimeric aadA transcripts. RNA from the untransformed host H 13 and from seven independent transformants was subjected to agarose gel electrophoresis, blotted to nitrocellulose and hybridized with a radioactive probe containing the E.coli aadA gene. Strains 5A, SB, 7A,and 7B are rbcX-AAD transformants; 9A, 9B and 9C are atpX-AAD transformants. Transformant 5A is phototrophic but is drug sensitive and lacks any detectable aad4 sequence by Southern analysis. Approximate sizes in kb are indicated alongside.
To confirm these results, the AAD enzyme activity was assayed directly in crude extracts of the transformed cells (table 2). The transfer of the 32P-labelled adenyl group from ATP to streptomycin or spectinomycin can be measured in a simple filter binding assay. The transformed strains did have AAD activity, confirming that the resistance to the drugs is due to the expression in Chlamydomonas of the foreign aadA DNA. The specific activity of AAD was on the average higher in the atpX-AAD than in the rbcX-AAD transformants, in rough agreement with their different degree of resistance to streptomycin. There were large variations in the levels of AAD activity in different transformants, possibly because in these strains there is a heterogeneous set of integration events. In the case of a simple integration at a single locus (psaC::AAD and ORF472::AAD, see below), the distribution of activities in different isolates seemed more narrow.
Total RNA was extracted from transformed cells and subjected to Northern analysis using the E. coli aadA gene as a probe (figure 2). A single band of approximately 1.1 kb hybridizes in the rbcX-AAD transformants, the size expected for a mRNA beginning at the same site in the rbcL promoter as the normal transcripts, extending through the aadA sequence and ending at the position of the rbcL 3'end (9). In the atpX-AAD transformants, two bands are observed: a 1.4 kb RNA, as expected for the atpA promoter and rbcL 3'end, and a smaller RNA (0.8 kb, the exact nature of this transcript and the reason for its appearance were not investigated). No aadA homologous transcripts are detected in the untransformed host H13 or in the transformant lacking aadA sequences (5A). The 1.1 kb and 1.4 kb mRNAs accumulate to similar levels in the rbcX-AAD and atpX-AAD transformants, suggesting that the difference in AAD expression and streptomycin resistance in the two sets of transformants are due to differences in the translation efficiency of the rbcL and atpA 5'-leader sequences. These differences could Extracts from the untransformed hosts (H13 and wild-type (WT)) and from a number of transformants were assayed in duplicate for AAD activity and for total protein content as described in Materials and Methods. The activity is expressed as cpm incorporated/4g protein.
be intrinsic to the 5'-leader sequences, or be due to the compatibility between the 5'-leaders and the aadA coding region of these chimeric genes.
Uniparental inheritance of the resistance marker Are the aadA sequences expressed in the chloroplast?: this is likely because the selection of the transformants is based on a chloroplast marker (tscA), because the chimeric aadA gene is under the control of chloroplast transcription and translation signals, and because the size of the transcripts suggests that these signals are used normally. However particle gun transformation can also deliver DNA to the nucleus (19,20), so that cotransformation of the chloroplast (tscA rescue) and of the nucleus (aadA expression) remains a possibility. In genetic crosses of Chlamydomonas reinhardtii, nuclear markers segregate in tetrads according to Mendelian genetics (2:2 in a backcross), but chloroplast genes are inherited uniparentally from the mt(+) parent in the majority of tetrads and segregate 4:0. Two atpX-AAD transformants (spcR, mt +) were crossed to the wild-type (spcS, mt -). In all of the 34 tetrads tested the segregation was 4:0 (spcR:spcS), indicating that the spectinomycin resistance is indeed inherited as a chloroplast marker (11 tetrads from a cross with transformant IOA and 23 tetrads from a cross with lOB). In a reciprocal cross (spcS, mt + x spcR, mt -), all 13 tetrads examined showed 0:4 segregation as expected.
Site-directed chloroplast gene inactivations Because the atpX-AAD chimeric gene shows higher resistance to the antibiotics, it was chosen as a selectable marker cassette. The chimeric aadA gene was transferred from atpX-AAD to the plasmid vector pUC18 from which it can be conveniently excised, in particular to yield a blunt-ended fragment (figure 1). The marker can thus be inserted into any target sequence; cloning is facilitated by the fact that the chimeric aadA gene is also expressed in E. coli and confers spectinomycin and streptomycin resistance to the bacteria. The chimeric atpX-AAD gene was inserted into three chloroplast genes: psaC (14), tscA, and ORF472 (figure 1). The psaC gene encodes a 9 kD subunit of photosystem I with the FA and FB iron-sulfur centers. The tscA gene is required for The initial transformed colonies were first restreaked on acetate spectinomycin medium, then grown for one day in acetate spectinomycin medium, and finally plated on minimal medium and on acetate spectinomycin medium. The plating efficiency after this first round of segregation is expressed as the ratio of the number of colonies growing on the minimal medium versus colonies growing on the acetate medium. The plating efficiency after two further rounds of restreaking, liquid culture and plating is shown in the column labelled 3d round.
trans-splicing of the psaA mRNA (21,6,10), and thus for the expression of one of the chl-a apoproteins of photosystem I. The function of ORF472 in Chlamydomonas is not known, it does not have similarity to any gene from plant chloroplasts (15,22). The plasmid DNAs were bombarded into wild-type C. reinhardtii, and transformants were selected for resistance to spectinomycin. The plates contained acetate as a source of reduced carbon and were placed in dim light because photosynthetic mutants are often light-sensitive. The initial transformed colonies were streaked on spectinomycin acetate medium and tested in two ways: fluorescence induction kinetics and plating efficiency on minimal medium (phototrophic growth).
As expected, all of the tscA::aadA and most of the psaC::aadA transformants tested had the high fluorescence yield and fluorescence induction curves typical of photosystem I deficient mutants (not shown). The plating efficiency on minimal medium was low, but still approximately 10-2 (table 3), suggesting that at least some of the cells still contained wild-type copies of the gene (each C. reinhardtii cell contains a single chloroplast with approximately 70 copies of the chloroplast genome). After two additional rounds of re-plating under selective conditions on spectinomycin acetate medium, the plating efficiency on minimal medium had dropped to less than l0-5. Thus the initial transformants were probably heteroplasmic and contained both wild-type and mutant (spectinomycin resistant) copies of the chloroplast genome that segregated during growth on selective medium, finally giving rise to strict acetate-requiring photosynthetic mutants. Southern analysis of the tscA::aadA mutants (figure 3) showed the presence of the interrupted genes (2.3 kb), but wildtype copies (1.7 kb) were not detectable. A similar analysis of the psaC::aadA mutants also showed that only interrupted mutant copies were present, but not wild-type ones (14).
One of the psaC::aadA transformants (1B2; mt +) was crossed with the wild-type, and the progeny were scored for acetate requirement, for mutant fluorescence induction kinetics and for spectinomycin resistance. Segregation was again 4:0 in all 14 tetrads examined, as expected for a chloroplast mutation.
The nature of the photosystem I deficiency in the tscA::aadA mutants was further characterized by Northern analysis ofpsaA mRNA maturation ( figure 4). The tscA gene is required for transsplicing of exons 1 and 2 (21,6). Its product is a small chloroplast RNA which probably base pairs with the two separate exon precursors and forms part of the catalytic core for the transsplicing of exons 1 and 2 (10; figure 5). Chloroplast and nuclear mutants defective in this step belong to class C (23, 6); they accumulate a trans-splicing intermediate containing exons 2 and 3. The tscA::aadA mutants failed to produce wild-type psaA mRNA, but accumulated this intermediate as expected (figure 4). DNA from the untransformed wild-type (WT) and from two tscA::AAD transformants (61A, 62A) was digested with KpnI and BglII, subjected to agarose gel electrophoresis, blotted to a nylon membrane and hybridized with the radioactive R12KB fragment (lanes 6-8; see figure lb). The wild-type 1.7 kb fragment is not detected in the transformants (even after extended exposure), but is replaced by two fragments of 1.3kb and 2.3kb. The latter contains the aadA sequence and also hybridizes to the AAD probe (lanes 9,10). (The R12KB probe contains repetitive sequences which also hybridize to larger fragments in the wild-type and in the transformants.)  Fig. 4. Northern analysis of psaA mRNA trans-splicing in tscA::AAD transformants. RNA from the wild-type (WT), from a tscA deletion mutant (H13), and from two tscA: :AAD transformants was subjected to agarose gel electrophoresis, blotted to a nylon membrane and hybridized with a probe specific for exon 3 of psaA. As diagrammed at the bottom, psaA mRNA is assembled from three separate precursors in two steps of trans-splicing, starting either with exons 1 and 2 or with exons 2 and 3 (23,6). The product of the tscA gene is a small RNA which is thought to assemble with exons 1 and 2 and is required for their trans-splicing (10). The tscA::AAD mutants, like the tscA deletion (H13), accumulate a 3.8 kb trans-splicing intermediate (0) containing exons 2 and 3 (and psbD, which is co-transcribed with exon 2) but fail to trans-splice exons 1 and 2, and thus lack the 2.7 kb wild-type mRNA (4). A precursor containing only exon 3 (0) is also present in the wild-type and in the mutants. The ORF472::aadA transformants had normal fluorescence induction kinetics, and had a high plating efficiency on minimal medium, even after two rounds of re-plating on spectinomycin medium (table 3). This could be interpreted in two ways: one possibility is that ORF472 is dispensable for photosynthetic growth, so that the mutant does not have an apparent phenotype under laboratory conditions. The other is that ORF472 is essential even on acetate medium and that a heteroplasmic mixture of wildtype and mutant genomes is maintained by balanced selection for viability and for spectinomycin resistance. To distinguish between these two possibilities, DNA from the transformants was subjected to Southern analysis (figure 3). Both wild-type and interrupted copies of the gene were observed. Thus ORF472 is apparently essential for cell viability on acetate medium. The AAD enzyme activity was lower than in the psaC::aadA mutants (table 2), which might be explained by the presence of a lower number of copies of the marker in these heteroplasmic strains.

DISCUSSION
The chimeric aadA gene as a reporter and a selection marker for C. reinhardtii chloroplast transformation The two expression vectors, rbcX and atpX, allow the expression of coding sequences driven by homologous transcription and translation signals in the C. reinhardtii chloroplast. A coding sequence can be fused to the inititator ATG with just three additional intervening codons. The tscA marker on these vectors provides a selection, in a tscA deletion host, for transformants which can then be tested for expression. Using this approach I have obtained the expression of the E. coli aadA coding sequence in the C. reinhardtii chloroplast. The cells contain chimeric aadA mRNA, have aminoglycoside adenyl transferase activity, are resistant to spectinomycin, and with the atpX vector, also to streptomycin. The resistance is inherited uniparentally from the mt+ parent as expected for a chloroplast gene. To my knowledge this is the first example of the stable expression of a foreign protein in a transgenic chloroplast. Transcripts encoding glucuronidase (GUS) and neomycin phosphotransferase (NPTH) have been obtained with chimeric constructs, but expression of the polypeptides or enzyme activity were not detected (24,25). Transient expression of chloramphenicol acetyl transferase and GUS in the chloroplasts of tobacco cultured cells were observed after particle gun transformation with chimeric constructs (26,27).
The chimeric aadA gene can serve both as a reporter and as a portable selectable marker for chloroplast transformation. The use of aadA as a reporter can be at the mRNA level (figure 2) or at the protein level since a very simple filter binding assay for AAD activity has been adapted to C. reinhardtii (table 2). It will probably prove important to use constructs that give identical transcripts for comparisons of AAD activity, because there is evidence that the structure of the transcript can influence its stability (25) and probably also its translation efficiency: although atpX-AAD and rbcX-AAD transcripts accumulate to similar levels in transformed cells, only atpX-AAD transformants are resistant to streptomycin, and AAD activity is on the average higher with the atpX-AAD construct.
Towards chloroplast reverse genetics The chimeric aadA gene can be inserted into chloroplast DNA in vitro and cloned in E. coli where the spectinomycin and streptomycin resistance are also expressed. This is consistent with previous findings that chloroplast promoters are often active in bacterial systems (28,29). The aadA construct is then introduced into the C. reinhardtii chloroplast by particle gun transformation. A process of homologous recombination can give rise to substitutions where the endogenous DNA is replaced by the aadA construct in some of the multiple chloroplast genomes. Provided that cells bearing the construct are viable under appropriate conditions, further growth under spectinomycin selection will give rise to homoplasmic strains. Using this approach we have generated site-directed gene disruptions of two chloroplast genes that are required for photosynthesis: psaC and tscA. The mutants are unable to grow phototrophically on minimal medium and are acetate-requiring. They have the expected fluorescence induction patterns for photosystem I deficiency. A more detailed analysis of the psaC::aadA mutants will appear elsewhere (14).
The tscA gene is required in trans for the trans-splicing of psaA mRNA, and is thought to interact with the separate exon 1 and exon 2 transcripts to form the characteristic core structure of group II introns (figure 5). In previous studies of tscA function, we have shown that insertions at certain sites in the tscA gene still allowed some degree of function, so that the tscA deletion host could be rescued by transformation and selection for photosynthetic growth (10). In our model, these insertions are in the periphery of the intron structure, in the loops of domains II and III. However one insertion (at a Scal site) failed to give rise to transformed colonies, and was assumed to inactivate tscA activity more completely. This insertion is in one of the core helices of the proposed structure and could prevent the interaction of the tscA RNA with the 3'part of intron 1 in the exon 2 precursor (figure 5, Scal site in helix IV). The aadA insertion described here is at the same site, and because the selection for spectinomycin resistance is independent of photosynthetic growth, the mutant strain can now be recovered. Its photosystem I deficiency and psaA trans-splicing defect confirm our previous inference.
The function of ORF472 in Chlamydomonas is not known. In contrast to the situation with the other gene disruptions, where the wild-type copy of the gene could be replaced by the mutant one, intact copies of ORF472 were always maintained. The persistance of a heteroplasmic mixture of wild-type and interrupted copies of this ORF in the transformed strains suggests that both may be under selective pressure: for cell viability on the one hand, and for spectinomycin resistance on the other. Such a balanced mixture of two genotypes in the chloroplast has been described for unstable suppressors of rbcL nonsense mutations. It was proposed that the suppressor alleles which restored rbcL function were also deleterious to the chloroplast, so that wildtype and suppressor copies were maintained by selection (30,31). Although it appears that ORF472 is essential for cell viability, the nature of this requirement is unclear. It is unlikely that ORF472 is directly involved in photosynthesis, because other photosynthetic mutants could be generated and maintained on medium containing acetate. The chloroplast probably ensures many other vital functions (e.g. in amino acid or lipid metabolism) in which ORF472 could participate either directly, for example by encoding an enzyme subunit, or indirectly by encoding a component generally required for gene expression (transcription, translation) or for DNA replication.
The tools described here should prove useful to study many aspects of chloroplast function. The aadA gene as a reporter with an easy assay could be used to study different aspects of gene expression at the transcriptional, post-transcriptional and translational levels. The chimeric aadA gene as a portable cassette and dominant selectable marker is a convenient tool for chloroplast reverse genetics. It could be used to interrupt and inactivate any gene whose function is dispensible under appropriate conditions. It should also be possible to introduce a large variety of mutations into the chloroplast genome by placing the selectable marker adjacent to cloned genes which would be subjected to more refined forms of site-directed mutagenesis. Transformation and selection for the aadA marker should give rise to a proportion of strains also carrying the adjacent, unselected mutation.