Quantitative analysis of transcription and RNA levels of 15 barley chloroplast genes. Transcription rates and mRNA levels vary over 300-fold; predicted mRNA stabilities vary 30-fold.

Higher plant plastid genomes encode rRNAs, tRNAs, and protein subunits of the RNA polymerase, ribosomes, and the photosynthetic apparatus which vary over 1000-fold in abundance. Quantitative analysis of transcription and RNA levels was carried out on 15 plastid genes which are located in 14 different transcription units covering 50% of the barley plastid genome. Transcription of 16S rRNA, trnfM-trnG, and trnK was high relative to most other plastid genes. Transcription of trnfM-trnG was 5 times greater than trnK indicating that differences in tRNA levels in plastids could be due, in part, to differences in transcription. Among the protein coding genes, mRNA levels varied over 900-fold and transcription over 300-fold. The gene showing the lowest transcription rate and mRNA level, rpoB, is located in a gene cluster which encodes subunits of the plastid RNA polymerase (rpoB-rpoC1-rpoC2). RpoA, which encodes the alpha subunit of the RNA polymerase, was located in a gene cluster encoding ribosomal proteins (rpl23, rps19, rpl16) and infA. RNA from this gene cluster is 30-fold more abundant than rpoB mRNA, suggesting that expression of rpoA is regulated at the level of translation or protein stability. Polycistronic operons encoding subunits of the photosynthetic apparatus (psbB-psbH-petB-petD; psbK-psbI-psbD-psbC; atpB-atpE; psaA-psaB) had higher transcription rates and correspondingly higher mRNA levels than genes which encode ribosomal proteins or RNA polymerase subunits. RbcL and psbA, which are located in separate transcription units, exhibited the highest transcription rates and mRNA levels. Correspondence between transcription rate, mRNA level, and protein abundance indicates that transcription is a primary determinant of barley plastid gene expression. In addition, a 30-fold variation in predicted mRNA stability was observed which further increases the dynamic range of plastid mRNA abundance.


Quantitative Analysis of Transcription and RNA Levels of 15 Barley Chloroplast Genes
TRANSCRIPTION RATES AND mRNA LEVELS VARY OVER 300-FOLD; PREDICTED mRNA STABILITIES VARY 30-FOLD* (Received for publication, March 16, 1992) Jeffrey C. Rapp, Brian J. Baumgartner, and John Mullet$ From the Department of Biochemistry and Biophysics,Texas A&M University,College Station,Texas 77843 Higher plant plastid genomes encode rRNAs, tRNAs, and protein subunits of the RNA polymerase, ribosomes, and the photosynthetic apparatus which vary over 1000-fold in abundance. Quantitative analysis of transcription and RNA levels was carried out on 15 plastid genes which are located in 14 different transcription units covering 50% of the barley plastid genome. Transcription of 16s rRNA, trnfM-trnG, and trnK was high relative to most other plastid genes. Transcription of trnf"trnG was 5 times greater than trnK indicating that differences in tRNA levels in plastids could be due, in part, to differences in transcription. Among the protein coding genes, mRNA levels varied over 900-fold and transcription over 300-fold. The gene showing the lowest transcription rate and mRNA level, rpoB, is located in a gene cluster which encodes subunits of the plastid RNA polymerase (rpoB-rpoC1-rpoC2). RpoA, which encodes the (Y subunit of the RNA polymerase, was located in a gene cluster encoding ribosomal proteins (rp123, rpsl9, rp116) and infA. RNA from this gene cluster is 30-fold more abundant than rpoB mRNA, suggesting that expression of rpoA is regulated a t the level of translation or protein stability. Polycistronic operons encoding subunits of the photosynthetic apparatus (psbB-psbH-petB-petD; psbK-psbI-psbD-psbC; atpB-atpE; psaA-psaB) had higher transcription rates and correspondingly higher mRNA levels than genes which encode ribosomal proteins or RNA polymerase subunits. RbcL and psbA, which are located in separate transcription units, exhibited the highest transcription rates and mRNA levels. Correspondence between transcription rate, mRNA level, and protein abundance indicates that transcription is a primary determinant of barley plastid gene expression. In addition, a 30-fold variation in predicted mRNA stability was observed which further increases the dynamic range of plastid mRNA abundance.
Plastids of most higher plants contain multiple copies of a circular DNA molecule which ranges in size from 120 to 217 * This work was supported by Grant DCB-8916199 from the National Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ T o whom correspondence should be addressed. Tel.:  kbp,' depending on plant species (for review, see Palmer (1990)). Variation in plastid genome size is due primarily to differences in the size of an inverted DNA repeat and small differences in gene content (Palmer, 1990;Baldauf and Palmer, 1990;Gantt et al., 1991;Shinozaki et al., 1986;Hiratsuka et al., 1989). Plastid DNA contains over 120 genes which encode tRNAs, rRNAs, and proteins. Plastid rRNAs (16S, 23S, 4.5S, 5 s ) are co-transcribed from an operon located in an inverted DNA repeat in those genomes having this DNA structure (Strittmatter and Kossel, 1984;Keus et al., 1984). Genes encoding tRNAs are distributed throughout the genome. Some tRNA genes are transcribed individually, others are located within tRNA gene clusters, two are located within the rrn transcription unit, and some are located within or at the end of larger transcription units which contain other genes. The plastid-encoded proteins include subunits of an RNA polymerase (encoded by rpoA, B,Cl,C2), plastid ribosomes (12 rps, 9 rpl genes in rice), a putative NADH oxidoreductase complex (ndhA, B,C, D,E, F,G), and proteins involved in photosynthesis. This latter group of genes encodes the large subunit of Rbu-P2 carboxylase (rbcL), subunits of photosystem I (PSI) (psaA, B,C), photosystem I1 (PSII) (psbA, B,C, D,E, F, G, H,I, K,L), the ATP synthase (atpA, B,E, F,H, I), and the cytochrome b6-f complex (petA, B,D) (for review, see Umesono and Ozeki (1987) and Ohyama et al. (1988)). Plastid protein complexes accumulate to dramatically different levels. Chloroplasts contain approximately lo7 Rbu-P2 carboxylase complexes (Dean and Leech, 1982), lo6 PSI, PSII, ATP synthase, and cytochrome be-f complexes , lo4 to lo5 ribosomes , and less than lo4 RNA polymerase complexes per organelle.
Each of these complexes contains proteins which are nuclearencoded. Therefore, plastid and nuclear genes encoding subunits of a common protein complex need to be expressed a t similar levels, while genes encoding subunits of different complexes are often needed at very different levels.
Previous studies have shown qualitatively that the transcription activity and RNA levels of different plastid genes varies considerably (ie. Deng and Gruissem, 1987;Mullet and Klein, 1987). However, quantitative data on these parameters has been collected only for psbA and rbcL . Both of these genes are highly expressed and encode proteins involved in photosynthesis. In this paper, we have quantitated the transcription activity and RNA levels of 15 plastid genes at one stage of chloroplast development in barley. This analysis revealed over 300-fold variation in transcription activity and more than 900-fold variation in RNA level for different plastid genes. In general, variation in transcription activity was paralleled by transcript abundance indicating that at this developmental stage, transcription activity is a primary determinant of RNA level for most of the genes examined. In addition, among the RNAs analyzed, a 30-fold variation in predicted mRNA stability was observed.

MATERIALS AND METHODS
Plant Growth, Plastid, and RNA Isolation-Barley seedlings were grown in controlled environmental chambers as described by Baumgartner et al. (1989). Plastids were isolated from apical 3-cm sections of primary leaves of 4-day-old dark-grown seedlings as previously described (Baumgartner et al., 1989). RNA was extracted from isolated plastids as described in Orozco et al. (1985). To ensure maximum recovery of plastid RNA, no more than 2 X lo8 plastids were used per 400 p l of extraction buffer.
Polymerase Chain Reaction Procedures-DNA fragments containing portions of barley chloroplast genes were prepared by the polymerase chain reaction procedure. For known barley genes, oligonucleotide primer sequences were derived from published barley chloroplast gene sequence data (Sexton et al., 1990b;Boyer andMullet, 1988 Zurawski et al., 1984;Reverdatto et al., 1989;Oliver and Poulsen, 1984;Neumann, 1988). Genes for which the sequence from barley was unknown, sequences from rice, maize, and tobacco chloroplast DNA were aligned, and regions of 100% homology (12-18 nucleotides) were chosen for use as PCR primers (Shinozaki et al., 1986;Hiratsuka et al., 1989). Additional nucleotides which encoded restriction endonuclease cleavage sites were added to the ends of the primers to facilitate cloning of the PCR amplified fragments. Thirty PCR cycles were used for amplification with a 1-min denaturation a t 94 "C, 2 min annealing a t 48 "C, and polymerization for 45 s a t 72 "C.
Cloning of Chloroplast DNA Gene Fragments, DNA Sequence Analysis, Gene Organization-DNA restriction fragments containing portions of barley chloroplast genes, or PCR-generated fragments digested with the appropriate restriction endonucleases were cloned into pBluescript vectors (Sambrook et aL, 1989). DNA sequence analysis was carried out using the dideoxy chain-termination procedure with the Sequenase Kit from U. S. Biochemicals. Sequence data was analyzed using programs from Intelligenetics. RpoA, rpoB, rpoC1, rpoC2, ndhA, rp116, infA, rpsl6, and rp123 were localized by Southern blotting and PCR amplification reactions. Probes for rpoA, rpoB, rp116, and ndhA were generated by PCR amplification (Table I).
2.8-kbp between the inverse compliment of the rpl16 primer and rpoA Analysis of Plastid RNA Transcript Abundance-RNA extracted from isolated plastids was fractionated on formaldehyde-agarose gels as previously described . After electrophoresis, RNA was transferred onto Genescreen Plus nylon membranes via vacuum (Bio-Rad) or capillary transfer. RNA gel blots were hybridized with in vitro generated antisense RNA probes. RNA probes radiolabeled with [a-"PIUTP were synthesized from recombinant pBluescript expression vectors with T7 or T3 RNA polymerase according to the conditions recommended by the supplier (Stratagene, Co.). Prehydridization, hybridization, and wash conditions were as previously described for RNA probes (Sexton et al., 1990a).
Linearized CsC1-purified plasmid DNA (10-20 pg/lOO-pI reaction) was used for in vitro transcription reactions (Stratagene Co.). After 2-h reactions, samples were treated with 5 units of RQ1 DNase I (Promega) for 40 min at 37 "C, extracted with pheno1:CHCln:isoamyl alcohol (1:1:0.04), the aqueous phase was adjusted to 0.3 M sodium acetate and 2.5 volumes of ethanol was added to precipitate the RNA. RNA was concentrated by centrifuging 15 min in a microcentrifuge, washed with 70% ethanol, and dried 2 min in a speed vac. The dried RNA pellet was dissolved in 100 pl of T E (made with diethylpyrocarbonate-treated water) and 300 p1 of a 6.33 M CsCl solution was added. To obtain purified in vitro RNA, the samples were transferred to Beckman LA100.2 polycarbonate centrifuge tubes and centrifuged a t 100,000 rpm, 15 "C for 4 h in a Beckman TL-100 tabletop ultracentrifuge. After centrifugation the supernatant was discarded, the RNA pellet was washed with 70% ethanol, and dried in a speed vac. The RNA was resuspended in 50-100 pI of T E and residual CsCl removed using a Sephadex G-50-80 spin column. RNA was separated on formaldehyde-agarose gels, stained with ethidium bromide, and RNA integrity visualized using a UV transilluminator. The concentration of RNA was determined by measuring the absorbance at 260 nm (1 absorbance unit equal to 40 pg of RNA ml-l).
RNA isolated from a known number of plastids was dissolved in RNA loading buffer (0.04 M MOPS, pH 7.0, 50% formamide, 6% formaldehyde, 0.001 M EDTA, 0.01 M Na acetate), heated 10 min a t 65 "C, and applied in duplicate to a Genescreen Plus membrane with a Schleicher and Schuell Minifold 1 (Delrin) dot blot apparatus. Known amounts of in vitro synthesized RNA standards were applied to the same membrane. Membranes were baked for 2-4 h a t 80 "C. The blots were then cut into strips, each containing plastid RNA and a different RNA standard dilution series. Individual strips were hybridized to RNA probes complementary to the RNA standard; hybridization and wash conditions were the same as for Northern analysis. The amount of radiolabeled probe hybridized to plastid RNA and the standards was quantitated with a Betagen Betascope blot analyzer or by scintillation counting of excised dots.
Plastid Run-on Transcription Assays-Plastid run-on transcription assays were carried out as described by Mullet and Klein (1987), with some modifications. ATP, CTP, and GTP were used a t 0.5 mM, 2 X lo7 plastids were used in 200-pl reactions, and 250 pCi of [L~-~'P]UTP were used in 5-min assays. Radiolabeled transcripts from 5-min plastid run-on transcription assays were extracted and hybridized to dot blots containing 1 pmol/dot of in vitro synthesized antisense RNA transcripts for each of the 14 genes being analyzed. Antisense RNAs were applied in duplicate for 16 S RNA, rbcL, psbA, trnK, t r n f " trnG, rpsl6, psbD, psaA, and atpB; antisense RNAs for genes with low transcription rates (rpoA, rpoB, ndhA, petB, and rp116) were applied in quadruplicate. Hybridization and wash conditions were the same as those used for the Northern analysis except that hybridizations were carried out for 48 h. The amount of radioactivity hybridized was determined by cutting out the radioactive spots and counting by scintillation spectrometry.
Estimation of Hybridization Efficiency-Percent hybridization of RNAs from run-on transcription assays to antisense RNA fixed on dot blots was estimated as follows. A known amount of radiolabeled in vitro synthesized sense RNA was hybridized, in the presence of unlabeled plastid RNA from 2 X lo7 plastids, to dot blots containing in vitro synthesized antisense RNA transcripts (1 pmol/dot, 2 or 4 dotslgene). The same hybridization conditions were employed (equal time and volume) as those used for the hybridization of run-on transcripts. After washing, radioactivity hybridized to antisense RNA was determined by cutting out the dots and counting in a scintillation counter. The average percent hybridization per pmol of antisense RNA (1 dot) from 2 experiments was determined for each RNA; rp116 = 7.6 k 1, rpoA = 11.3 f 1, petB = 9.6 * 1, rbcL = 13.6 f 2,psaA = 8.2 i 0.3, rpoB = 10.6 f 2, trnfM-trnG = 7.2 & 0.5, psbD = 12.9 2 0.6, rpsl6 = 11.8 f 2, trnK = 12.4 k 1, psbA = 9.3 i 0.4, 16 S rRNA = 10.4 i 0.2, ndhA = 7.7 k 1.
Mung Bean Nuclease Protection Assay-Run-on transcripts were hybridized to antisense RNA dot blots as described above. Following hybridization and washing (including two additional 23 "C, 5-min washes with 2 X SSC, 0.5% Triton X-100 to remove SDS), radioactivity hybridized was determined with a Betascope. Blots were then t.reated with 670 units/ml mung bean nuclease in 0.03 M Na acetate, pH 4.5, 0.05 M NaC1, 0.001 M ZnC12, 5% glycerol, and 0.001% Triton X-100 for 15 min at 37 "C. Reactions were stopped by washing the membranes with 2 X SSC, 0.5% SDS. Filters were air-dried and recounted with a Betascope.

RESULTS
Barley Plastid Genes Selected for Analysis-Fifteen barley plastid genes were selected for analysis of transcription and RNA level. The group of genes to be analyzed encode representative rRNA, tRNA, and proteins found in the plastid RNA polymerase, ribosomes, the putative ndh complex and each of 5 protein complexes involved in photosynthesis (Rbu-P1 carboxylase, PSI, PSII, ATP synthase, cytochrome be-f).
Eleven of the genes to be studied had previously been located on the barley plastid genome (psbA, trnK, rpsl6,psbD, t r n w , trnG, psaA, atpB, rbcL, petB, 16s rRNA). Probes for the previously characterized barley plastid genes were obtained from existing subclones by amplifying selected regions of the genes using PCR or subcloning restriction fragments (Table  I). All gene-specific probes were cloned into Bluescript vectors and verified by DNA sequencing. Four genes selected for study had not been previously located on the barley plastid genome (rpoA, rpoB, ndhA, rp116) although they are present in the plastid genomes of other higher plants (i.e. Shinozaki et al., 1986;Hiratsuka et al., 1989). Sequences of these genes from rice, tobacco, and maize were aligned and 12-18-base pair regions with 100% identity were used to design PCR primers for amplification of barley plastid DNA ( Table I). The resulting amplified DNA was cloned and a minimum of 150 base pairs of each insert was sequenced. The sequences of the inserts showed greater than 90% sequence homology with the corresponding genes in rice (data not shown). In order to map the four genes, insert DNA was radiolabeled and used to probe a n ordered bank of PstI DNA fragments which spans the barley plastid genome (Poulsen, 1983). In some cases, PCR primers were tested for their ability to amplify DNA from clones containing specific PstI subclones of the barley genome, The organization of rpoA, rpoB, and rp116 relative to several other genes in the plastid genome was next determined by a series of genome walking experiments (using PCR amplification of overlapping regions, see "Materials and Methods"). In other higher plant plastid genomes, rpoB is located Barley plastid gene probes Several DNA fragments were amplified from chloroplast DNA by the polymerase chain reaction (PCR) as described under "Materials and Methods." PCR primer coordinates are followed by EMBL accession numbers. When DNA fragments were cloned as restriction fragments, the restriction enzymes used are indicated. Clones and plastid DNA sequences were obtained from the following sources; 1, Poulsen, 1983;2, Oliver and Poulson, 1984;3, Boyer and Mullet, 1988;4, Sexton et al., 1990b;5, Zurawski et al., 1984;6, Neumann, 1988;7, Berends et al., 1987;8, Zurawski and Clegg, 1984;9, Reverdatto et al.. 1989: 10, Hiratsuka et al.. 1989. in the gene cluster rpoB-rpoC1-rpoC2. Likewise, rp116 and rpoA are part of a large conserved gene cluster found in other plastid genomes (rpl23-rpl2-rpsl9-rpl22-rps3-rpll6-rpll4-rps8-infA-rp136-rpsll-rpoA) (Shinozaki et al., 1986Hiratsuka et al., 1989. To test if these genes were also clustered in barley, PCR primers homologous to regions in rpoB, rpoC1, and rpoC2 and rp116, rpoA, rp123, and rpsl9, infA were used to amplify regions of the barley plastid DNA (data not shown).
Based on this analysis, we found that the organization of rpoB-rpoC1-rpoC2 and rp123-rps19-rp116-infA-rpoA in barley is similar to other higher plant plastid genomes. Northern Blots and Identification of Co-transcribed Genes-Each of the gene-specific RNA probes described in Table I was used to probe Northern blots of plastid RNA (Fig. 1). This was done to test the quality of plastid RNA used for quantitation (described below), and to determine the complexity of the RNA populations which hybridize to each probe. The abundance of RNAs detected by the different probes varied greatly. Therefore, blots were exposed to x-ray film for different lengths of time so that RNAs from all the genes could be detected. The 16s rRNA probe detected a.predominant 1.6-kb RNA which corresponds to mature 16s rRNA (Fig. 1, lane 1 ). In contrast, the probe for trnf"trnG hybridized to RNAs 1.0 kb and less than 0.1 kb in size (Fig. 1, lane 2) which correspond to an unspliced precursor and tRNA size RNAs as previously described (Oliver and Poulsen, 1984). The trnK gene contains an intron which encodes an open reading frame of 504 amino acids (Sexton et al., 199Ob;Boyer and Mullet, 1988). The trnK probe detects the unspliced precursor (2.6 kb) RNA but not spliced tRNA(Lys) (Fig. 1, lane 3). Six genes encoding proteins involved in photosynthesis were analyzed. The probe for rbcL, the gene encoding the large subunit of Rbu-P2 carboxylase, hybridized to a 1.6-kb primary transcript of minor abundance and a 1.4-kb RNA (Fig. 1, lane 4 ) Poulsen, 1984). PsbA encodes protein Dl, a reaction center subunit of photosystem 11. Northern analysis showed that the psbA probe hybridized to a 1.2-kb RNA as previously described (Boyer and Mullet, 1988). PsbD, like psbA, encodes a photosystem I1 reaction center protein (D2). This gene is located in a complex operon which includes psbK-psbI-psbD-psbC-orf62-trnG (Berends et al., 1987). The Northern blot obtained with the psbD probe reflects this complexity (Fig. 1,  lane 6). PsaA encodes a subunit of the photosystem I reaction center. This gene is co-transcribed with psaB in barley to yield a 5.3-kb RNA (Fig. 1, lane 7) (Berends et al., 1987). Genes encoding subunits of the ATP synthase are located in 2 gene clusters (atpB-atpE; @I-atpH-atpF-atpA). The atpB-atpE gene cluster has been shown to be co-transcribed in several plants and in barley, the RNA hybridizing to the atpB probe (2.2 kb) is sufficiently large to include atpE as well (Fig.  1, lane 8). Two of the 3 plastid genes encoding subunits of the cytochrome bG-f complex are located in the psbB-psbH-petB-petD gene cluster. In other higher plants these genes are co-transcribed and complex RNA processing events yield a heterogeneous population of RNA (Westhoff and Herrmann, 1988;Tanaka et al., 1987). Not surprisingly, the petB probe hybridized to numerous RNAs ranging in size from 4.8 to 1.0 kb (Fig. 1, lane 9). Seven genes encoding subunits homologous to the proteins of the mitochondrial NADH oxidoreductase are located in tobacco and rice plastid genomes (Shinozaki et al., 1986;Hiratsuka et al., 1989), although the function of this complex in higher plant plastids is at present unknown. One representative of this group of genes (ndhA), which contains an intron in rice (Matsubayashi et al., 1987), was chosen for analysis. Northern analysis using the ndhA probe revealed RNAs of 4.0, 2.2, and 1.3 kb (Fig. 1, lane 10). Two genes encoding ribosomal proteins were analyzed (rps16, rp116). Rpsl6 is located upstream from trnK, is independently transcribed, and contains an intron (Sexton et al., 1990b). Northern blots using the rpsl6 probe detected unspliced precursor RNAs of approximately 1.2 kb and spliced RNA of 0.6 kb (Fig. 1, lane 11). Rp116 in contrast, is located in a conserved gene cluster with several other ribosomal protein encoding genes, infA, and rpoA (Fig. 1, lanes 12 and 13). The rp116 probe hybridized to RNAs ranging from 0.8 to 3.0 kb and the rpoA probe hybridized to RNAs ranging in size from 1.2 to 5.1 kb. Finally, RNAs hybridizing to rpoB are shown in Fig.  1, lane 14. The rpoB RNA is 6.0 kb suggesting that in barley, rpoB may be co-transcribed with rpoC1. In spinach, co-transcription of rpoB, rpoC1, and rpoC2 was detected (Hudson et al., 1988).
The 15 genes selected for study are located in 14 different transcription units. With the exception of the 16s rRNA gene, all are single copy genes. While the 15 genes represent only about 10% of the total number of plastid genes, we estimate that the transcription units encoding these genes represent approximately 70 kbp of transcribed DNA or about 50% of the barley plastid genome.
Quuntitation of Plastid RNA Leueki-The abundance of the RNAs corresponding to each selected gene was determined using Northern dot blots prepared under the same conditions used for the Northern blots in Fig. 1. In preliminary experiments, RNA from lo5 to lo7 plastids was applied to nylon membranes and blots were probed with radiolabeled antisense RNA from each of the genes. The blots were analyzed to determine the plastid RNA concentration where hybridization signals were proportional to input RNA. Control experiments were carried out with known amounts of sense RNA synthesized in uitro from each of the selected genes to determine the concentration range needed for the standard curves (data not shown). Linear regression analysis showed that hybridization of the probes to the RNA standards increased linearly with increasing input of RNA with correlation coefficients greater than 0.98. Next, plastid RNA and a set of RNA standards for one gene was applied to a blot. The blot was hybridized with the corresponding gene-specific RNA probe and the resulting hybridization signals quantitated so that the level of hybridizing RNA could be calculated. The data from this analysis is shown in Table 11.
Quuntitation of Plastid Gene Transcription-Variation in plastid RNA levels for different genes could be due to differences in transcription or RNA stability. Therefore, the transcription activity of the 15 genes under study was assayed using a quantitative run-on transcription assay previously developed for barley Klein and Mullet, 1990). To carry out the assay, plastids were isolated from chilled leaves to maintain the in uiuo distribution of RNA polymerase on plastid DNA. A known number of plastids were lysed in a transcription buffer which contained nucleotide triphosphates, including [32P]UTP, and transcription was allowed to proceed for 5 min. Heparin was added to the transcription reactions to prevent soluble RNA polymerase from reinitiating on the DNA and to block RNA degra-

Barley Chloroplast
Transcription and RNA Abundance  dation . The number of plastids per reaction volume was maintained at lO"/pI where transcription activity is proportional to plastid number. After the 5-min assay, the radiolabeled RNAs were phenol extracted and hybridized to antisense RNA prepared from each of the 15 genes being studied. Preliminary experiments showed that for many DNA regions both DNA strands were transcribed making the use of antisense RNA probes essential. Following hybridization and washing, the radioactivity hybridized to each probe was determined by scintillation counting. This data was then corrected for hybridization efficiency and probe size and finally expressed as femtomole of UMP incorporated/ kb of DNA/5 x lo6 plastids/5 min (Table 11).
Hybridization efficiency was determined by synthesizing radiolabeled sense RNA for each selected gene and hybridizing a known amount of the radiolabeled probe to antisense RNA using the same conditions described above for total plastid transcription assays. Antisense RNA (1 pmol) was applied to 2 dots for a gene transcribed at high rates or 4 dots for a gene transcribed at low rates. This assay showed that between 7 and 13% of each input probe hybridized per dot of antisense RNA (1 pmol). These values were used to correct the data obtained from plastid run-on transcription assays for variation in percent hybridization. The validity of correction for probe size was tested and confirmed by hybridizing radiolabeled run-on transcripts to two different sized probes from the same gene (i.e. rbcL, 0.65-kb probe = 490 cpm; 1.3 kb probe = 1119 cpm; psbA, 0.55-kb probe = 1638 cpm; 1.4-kb probe = 4139 cpm).
Plastid transcription units are often arranged in tandem with little nontranscribed DNA separating each transcription unit. This situation introduces two potential sources of error into the determination of transcription activity. First, it is possible that transcription of the gene being assayed could be influenced by transcription from an upstream gene if normal termination events are disrupted in uitro. This source of error could be significant if the upstream gene is transcribed at a much higher rate than the gene being analyzed. We tried to minimize this source of error by keeping the assay time to a minimum, but in one case (rpsl6) we believe that transcription rates were overestimated due to read-through from trnQ upstream of rpsl6. A second source of error could occur if transcription begins within the region of the gene being assayed and continues downstream of the probe for a significant distance during the assay. The resulting transcripts would hybridize to the probe and contain substantial single stranded regions which extend beyond the probe. These latter single stranded RNAs would be susceptible to mung bean nuclease. T o test the magnitude of this potential error, run-on transcripts hybridized to the 14 antisense RNAs were treated with mung bean nuclease. The amount of radioactivity removed by mung bean nuclease treatment ranged from 33% (psbA,psbD) to less than 10% (data not shown). In mock experiments using completely homologous radiolabeled sense probes with no overlap, approximately 10% of the radiolabel was removed from the blots. This experiment indicates that read-through transcription influences the run-on assay results to only a small extent.

DISCUSSION
Barley Plastid Gene Organization-The barley plastid genome is approximately 133 kbp in size and contains an inverted repeat similar to most other higher plant plastid genomes. Approximately 25% of the barley plastid genome has been sequenced and over 40 genes identified (Chakhmakhcheva et al., 1989;Neumann, 1988;Reverdatto et al., 1989, Sexton et al., 1990bSogaard and Wettstein-Knowles, 1987). During the course of this study, 4 additional genes, rpoB, rpoA, rp116, and ndhA, were identified and located on the barley plastid genome. Subclones of each of these genes were obtained using PCR primers and partial sequence analysis used to verify gene identity. In addition, rpoB was located in the rpoB-rpoC1-rpoC2 gene cluster and rpoA and rp116 located within a gene cluster which also contains infA, rpsl9, and rp123 as has been found in other higher plant genomes (Shinozaki et al., 1986;Hiratsuka et al., 1989;Hudson et al., 1988). Interestingly, the arrangement of identified genes in the plastid genome of barley is nearly identical to rice (Hiratsuka et al., 1989). This may not be too surprising because rice and barley are both C3 monocots which diverged from C4 monocots within the last 40 million years (Wolfe et al., 1989). In contrast, monocots and dicots diverged about 100-200 million years ago and flowering plants from liverwort about 400 million years ago (Wolfe et al., 1989). Even so, dicot, monocot, and liverwort plastid genomes contain many similar co-transcribed gene clusters (for review, see Baldauf and Palmer, (1990) and Palmer (1985)). Even more remarkable are plastid gene clusters which appear to be remnants of operons found in Escherichia coli (ie. S10-spc-a). The plastid gene clusters often encode proteins exclusively for photosynthesis or transcription/translation suggesting that segregation of functionally related genes into common transcription units is advantageous to the plant. In this paper we report that expression of different gene clusters varies dramatically at a single stage of barley chloroplast development.
Expression of Genes Encoding rRNA and tRNA-The transcription and accumulation of 16s rRNA was analyzed during this study because the production of rRNAs is a key regulatory point in the production of ribosomes, and ribosome level can potentially limit translation capacity, growth, and development (Nomura et al., 1984;Bendich, 1987). Plastid ribosomes contain stoichiometric amounts of 16S, 23S, 4.5S, and 5s rRNAs which are co-transcribed from an operon which also encodes trnI and trnA. In barley, these genes are duplicated because they are located in the inverted DNA repeat. Analysis of 16s rRNA levels in chloroplasts of 4-day-old dark-grown plants showed the 16s rRNA to be the most abundant RNA analyzed. If one 1 6 s rRNA is present in each ribosome, then plastids at this stage of development contain approximately 150,000 ribosomes. Previously it was estimated that barley plastids contain 20,000 to 200,000 ribosomes depending on developmental stage . Transcription of the genes encoding the 16s rRNA was very active relative t o other genes consistent with our previous observations . A high capacity for rRNA synthesis is important if rRNA synthesis limits ribosome accumulation and the rate of chloroplast biogenesis as proposed by Bendich (1987).
The transcription of three genes encoding tRNAs was directly examined (trnK, trnfM, trnG). Information on the expression of several other genes encoding tRNAs was obtained because they are co-transcribed with other genes being analyzed (16S-trnI-trnA-23S-4.5S-5S; psbI-psbK-psbD-psbC-orf62-trnG; possiblypsbA-trnH). The trnfM-trnG gene cluster was one of the most actively transcribed DNA regions analyzed. A similar level of transcription may occur for trnH if this gene is co-transcribed with psbA in barley as found for spinach (Thomas et al., 1988;Gruissem and Zurawski, 1985b). Transcription of trnI and trnA which are located within the rrn transcription unit, was about 1.5-fold less active than trnfM-trnG and transcription of trnK 6-fold lower. Transcription of the psbD-psbC operon is similar to trnK suggesting that genes encoding tRNAs within this operon (trnG) may also be synthesized at rates lower than trnfM-trnG. It is possible that the different rates of tRNA synthesis result in differential accumulation of the tRNAs. In bean plastids, tRNA abundance varied 5-fold and was correlated with codon usage (Pfitzinger et al., 1987). Unfortunately, the Northern blots used for RNA analysis in this study are not suitable for tRNA quantitation. The trnK probe hybridized primarily to unspliced precursor RNAs and the trnfM-trnG probe to precursor and tRNA sized RNAs. Further analysis will be required to determine the relationship between transcription of the genes encoding tRNAs and the relative abundance of the corresponding tRNAs.
Overall Transcription and RNA Abundance of Protein Coding Genes-The transcription and RNA levels of 11 plastid genes which encode proteins were analyzed. These genes encode subunits of the RNA polymerase (rpoA, rpoB), ribosomes (rpsl6, rp116), the putative NADH oxidoreductase (ndhA), and the photosynthetic complexes (Rbu-Pp carboxylase (rbcL), PSI1 (psbA, psbD), PSI (psaA), cytochrome bs-f (petB), ATP synthase (atpB)). The abundance of these protein complexes in plastids varies approximately 1000-fold. Analysis of RNA abundance revealed over a 900-fold range of mRNA level from a high of 45 fmol/5 X lo6 plastids for rbcL to 0.05 fmol for rpoB. Similarly, transcription of the 11 protein coding genes varied over 300-fold with psbA the most actively transcribed and rpoB the least actively transcribed. Overall, there was a strong correlation between gene transcription activity and RNA abundance. Transcription and mRNA levels of genes encoding proteins for photosynthesis was greater than genes encoding proteins for ribosomes or the RNA polymerase. The difference in transcription activity among transcription units could help explain the maintenance of gene clusters which encode proteins of related function. For example, if a gene which encodes a protein involved in photosynthesis was moved into the rpoB-rpoC1-rpoC2 transcription unit, the transcription rate in this region might be insufficient to support normal biosynthesis of the protein unless unusual transcript stability or translation efficiency were developed. I n vitro studies using a limited set of spinach genes showed that the relative strength of the psbA, rbcL, and atpB promoters was 1.6/1.0/0.7 (Gruissem and Zurawski, 1985a). The run-on transcription assays described here demonstrate that transcription of psbA was 6-fold greater than rbcL and 11fold greater than atpB. In other words, the run-on assay revealed a 12-fold variation in transcription activity between these genes, whereas in vitro transcription assays revealed only a 2-fold difference. This result could be due to differences between barley and spinach promoters for these genes. Alternatively, the DNA templates or protein extracts used for in vitro studies may have lacked important elements which specify relative promoter activity. Genes with the highest transcription activity (psbA, rbcL) have transcription promoter elements located -10 and -35 base pairs from the transcription initiation site which resemble prokaryotic a-70 transcription elements (for review, see Hanley-Bowdoin and Chua (1987)). Promoter elements for plastid genes transcribed a t very low rates, such as rpoB, have not been characterized. It is possible that variation in promoter sequences or spacing between elements could account for the observed differences. Alternatively, the differences in transcription could be due to the presence of different classes of promoters in the plastid genome which require specific CT-factors (Tiller et al., 1991) or different RNA polymerases (Greenberg et al., 1984).
Expression of Genes Encoding Proteins Involved in Photosynthesk-The abundance of mRNAs which encode proteins involved in photosynthesis varied 12-fold. The most abundant mRNA of this group corresponded to rbcL. This is consistent with greater abundance of the large subunit of Rbu-P2 carboxylase which is encoded by rbcL compared to the electron transport units (IO7 uersus lo6 units/plastid). The next most abundant mRNA analyzed corresponds to psbA. This gene encodes protein D l , one of two reaction center protein subunits of photosystem I1 (Mattoo et al., 1989). D l is one of the least stable proteins of the thylakoid membrane because it is damaged and turned over during the course of normal photosynthetic activity (for review, see Mattoo et al. (1989)). Therefore, it is not surprising that psbA mRNA levels are higher than mRNAs which encode more stable proteins (i.e. subunits of PSI encoded by psaA-psaB). There are small differences in the abundance of psbD, petB, psaA, and atpB mRNAs. Although the protein complexes in which these proteins are localized are usually present in stoichiometric amounts (for exceptions, see Chow et al. (1990)), the number of subunits per complex varies (psbD = 1; petB = 4; atpB = 3; atpE = 1; psaA = 1). This indicates that the final accumulation of each protein subunit is regulated by the abundance of mRNA as well as at other levels of expression.
Seven plastid genes encode proteins with homology to subunits of the mitochondrial NADH oxidoreductase (ndh genes) (Shinozaki et al., 1986;Hiratsuka et al., 1989;Matsubayashi et al., 1987). At present, activity of this putative complex has not been reported in higher plant plastids. The abundance of transcripts hybridizing to ndhA was lower than the genes encoding photosynthetic proteins. This does not discount a role for the NADH oxidoreductase in photosynthesis, but suggests that this complex may be present at lower levels than the other photosynthetic electron transport complexes.
Expression of Genes Encoding Ribosomal Proteins and RNA Polymerase Subunits-The abundance of rpsl6 and rp116 mRNA differed 10-fold even though the ribosomal proteins encoded by these genes probably accumulate in stoichiometric Barley Chloroplast Transcription and RNA Abundance amounts in ribosomes as in E. coli (Lindahl and Zengel, 1982).
The difference in abundance of translatable mRNA for these two genes may be even greater because a significant amount of unspliced rpsl6 mRNA was detected. This suggests that a difference in translation efficiency or protein stability accounts for equal end-product formation. The gene with higher mRNA level, rp116, is co-transcribed with several other ribosomal protein genes in a gene cluster which resembles the S10-spc-a operon of E. coli (Zhou et al., 1989;Ruf and Kossel, 1988). Synthesis of ribosomal proteins from this gene cluster in E. coli is autoregulated translationally by free ribosomal proteins subunits (for review, see Lindahl and Zengel(1982)). Perhaps a similar type of regulation occurs in plastids. Similarly, rpoA mRNA levels are 30-fold greater than rpoB mRNA levels, yet these subunits probably accumulate in a 2:1 ratio a s in E. coli (Burgess, 1969;Rajasekhar et al., 1991). RpoA is the terminal gene in the S10-spc-a operon in E. coli and is also autogenously regulated at the translational level (Lindahl and Zengel, 1982). Therefore, it is possible that the rpoB-rpoC1-rpoC2 mRNA and corresponding RNA polymerase subunits are produced in limiting amounts and rpoA translation is modulated by a feedback mechanism. Predicted RNA Stability-Quantitative analysis of RNA leveIs and transcription rates provides a method to predict relative mRNA stability by comparing the ratio of RNA abundance/transcription rate for different genes. Using this approach, the relative stabilities of RNAs hybridizing to 14 of the genes analyzed in this study were calculated (Table 11). As expected, 1 6 s rRNA had the highest predicted relative stability consistent with this RNA being a component of the ribosome, a relatively stable structure. In contrast, the predicted stability of RNAs hybridizing to trnf"trnG and trnK was much lower. This is consistent with the observation that the probes used for these genes hybridized primarily to precursor RNAs and not mature tRNAs.
Among the mRNAs there was a 30-fold difference in predicted RNA stability as compared to a 300-fold variation in transcription rate. If only the genes encoding proteins involved in photosynthesis are considered, a 9-fold variation in predicted RNA stability was observed compared to a 35-fold range in transcription rate. Stability did not correlate with level of gene expression because abundant mRNAs (rbcL) and those which accumulate at lower levels (rpoA) had similar predicted stabilities. In general, while significant variation in RNA stability was predicted, the influence of RNA stability on mRNA levels a t this stage of barley chloroplast development appears to be a less important determinant of plastid mRNA levels than transcription. This conclusion of course must be qualified by noting that for many of the genes analyzed, several different RNAs were detected. Therefore, our analysis represents the sum of these RNAs and individual RNAs within the complex populations detected could have different stabilities.
Recent discussions have emphasized the importance of RNA stability in determining plastid RNA levels (Klaff and Gruissem, 1991;Mullet and Klein, 1987;Stern et al., 1991;Krupinska and Apel, 1989) which may appear at variance with our results. A role for modulated RNA stability has been invoked in several studies, including our own , where light induced changes in plastid transcription and RNA levels were analyzed. In these cases, transcription rates for specific genes changed to a greater extent than RNA levels. For example, illumination of 8-day-old darkgrown barley for 4 h caused a 4-fold increase in psbD-psbC transcription but only a 1.5-fold change in mRNA level (Sexton et al., 1990a). Likewise, a %fold decrease in psbA transcription during light-induced chloroplast maturation is ac-companied by a smaller and delayed decrease in psbA mRNA abundance . We initially interpreted these results to indicate that the stability of psbD-psbC and psbA mRNAs was changing. However, these results might also be explained, in part, by rapid light-induced changes in transcription rates which are followed by slow changes in RNA levels if the RNAs have long half-lives. In other words, following a rapid change in transcription rate, the levels of stable mRNAs require an extended period of time to reach a new steady state. The stability of rbcL mRNA was measured in plants (Klaff and Gruissem, 1991) and in cell culture (Ngernprasirtsiri et aE., 1990). In these studies, rbcL mRNA half-life was estimated to be between 5 and 12 h, suggesting that plastid mRNAs are long lived in sharp contrast to mRNAs in E. coli. Using the estimated half-life of rbcL mRNA as a reference, we predict that most of the mRNAs examined in this study are relatively stable with half-lives between 1 and 12 h.