Current Genetics
Current Genetics (1981) 4:37-46
© Springer-Verlag 1981
Binding Sites of E. coli DNA-dependent RNA Polymerase on Spinach Chloroplast DNA Michael Zech 1 , Martin R. Hartley2 , and Hans J. Bohnert 1' 3 1 Botanical Institute, University of Diisseldorf,Federal Republic of Germany 2 Dept. BiologicalSciences,Universityof Warwick,Coventry, United Kingdom
Summary. Under stringent conditions E. coli DNA-dependent RNA polymerase holoenzyme binds selectively to some spinach chloroplast DNA fragments generated by restriction endonucleases. The strongest of these binding sites, as judged by the initial rate of complex formation, are located in the large single-copy DNA region (Crouse et al. 1978) of this molecule and correspond in map location with known protein coding sequences. Some of these binding sites have characteristics of complex formation comparable with those of the PR and PL promoters of phage lambda DNA. Binding sites located close to the rRNA operons on the chloroplast DNA bind polymerase less strongly than those described above. Since the rRNAs are the most abundant transcription products in vivo and in isolated chloroplasts (Hartley and Head 1979; Bohnert et al. 1977) this suggests that the E. coli and chloroplast enzymes do not recognize all of the major promoters in chloroplast DNA with the same efficiency of binding. We have investigated in detail one region of the chloroplast DNA from spinach which contains three strong binding sites. This region has been shown to contain at least the gene for a 32,000 dalton protein (Driesel et al. 1980) which is most probably the so-called photogene (Bedbrook et al. 1978). One of these three E. coli RNA polymerase binding sites is not more than approximately 150 bp apart from what, by hybridization studies using isolated mRNA, we know to be the coding sequence for this protein. The results suggest that for some genes on the chloroplast DNA the bacterial RNA polymerase may be used to search for transcription initiation sites.
3 Present address for correspondence: EMBL,Postfach 10 22 09, 6900 Heidelberg, FRG
Key words: RNA polymerase binding sites - Chloroplast DNA - Photogene
Introduction
The spinach chloroplast DNA is a circular molecule 143 Kb in size. Its gross morphology, like that of most other chloroplasts, is characterized by two repeat segments 22 Kb long in inverted orientation which are separated by a 80 Kb large single-copy DNA region and a 18.5 Kb small single-copy DNA region (Bedbrook and Kolodner 1979; Crouse et al. 1978). Among the genes known on this chloroplast DNA are rRNA and tRNA genes (Crouse et al. 1978; Whitfeld et al. 1978; Driesel et al. 1979), the gene for the LSU protein of the ribulose 1.5 diphosphate carboxylase (Whitfeld and Bottomley 1980) as well as for a 32,000 dalton protein (Driesel et al. 1980). While investigations to find the positions of more genes continue, the way in which chloroplast DNA transcription is regulated may already be studied with respect to known genes. The study of the mechanisms by which individual genes are transcribed may help to explain the way in which individual RNAs accumulate to differing extents within the chloroplast (Driesel et al. 1980; Matsuda and Surzycki 1980, Silverthome and Ellis 1980, Edelman and Reisfeld 1980). It is not yet known whether chloroplast mRNAs are transcribed as approximately gene-sized sequences, or are derived by processing larger transcripts. Only for the rRNAs a larger precursor molecule has been demonstrated which includes at least the 16S, 23S and 4.5S rRNA genes together with some transcribed spacer (Hartley and Head 1979, Bohnert et al. 1977). Similarly, little is known of the mechanisms whereby chloroplast mRNAs differ in their pool sizes. Do the dff0172-8083/81/0004/0037/$ 02.00
38
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA
ferences arise through transcriptional control or by different stabilities? The answer o f these questions would significantly contribute towards our understanding o f the regulation o f gene activity in chloroplasts. One way to investigate this problem is through the mapping of RNA polymerase binding sites on chloroplast DNA. It can then be established whether these sites are gene-promoter sites. The specificity of the binding, the spatial arrangement and the positions o f these sites in relation to known genes provide valuable preliminary information. The approach described in this paper was initially used to map bacteriophage promoters (Seeburg and Schaller 1975; West and Rodriguez 1980). Although several examples o f transcriptional control of chloroplast gene expression have been reported (Matsuda and Surzycki 1980; Bedbrook et al. 1978; Link et al. 1978) little is known about the mechanism involved. Chloroplast RNA polymerase is known to be a multisubunit enzyme (Smith and Bogorad 1974; Briat and Mache 1980), though no functions can yet be ascribed to individual subunits. Jolly and Bogorad (1980) have recently reported that maize plastids contain a protein which specifically enhances transcription of chloroplast DNA fragments in supercoiled chimeric plasmids by the purified chloroplast polymerase. Whether or not this "stimulator" factor was associated with the polymerase in vivo, and lost during purification or exists as a separate entity was not determined. The problem o f purifying chloroplast RNA polymerase with its in vivo regulatory functions still intact is a very real one and has probably not been solved completely. While we do not regard the use o f the E. coli enzyme in the context of the present study necessarily to be highly specific, the same is also true for the chloroplast enzyme because o f the uncertainties still connected with this enzyme. The initial work described here uses a heterologous system composed of E. coli DNA-dependent RNA polymerase and chloroplast DNA. The rationale for the use of this particular system is apparent when the similarities between bacterial systems and the chloroplast are considered. These similarities include details o f gene expression in both systems (Hartley and Head 1979; Bohnert et al. 1977; Young et al. 1979; Rochaix and Malnoe 1978), the organization o f rRNA cistrons (Nomura et al. 1977; Schwarz and K6ssel 1980), sequence conservation in rRNA genes (Schwarz and K6ssel 1980; Bohnert et al. 1980), and the similarity o f structure and antibiotic sensitivities between bacterial and chloroplast ribosomes (Nomura et al. 1977; Bottomley and Bohnert 1981). In addition, the gene for the LSU of the ribulose 1.5 diphosphate carboxylase appears to be transcribed and translated in a heterologous system derived from E. coli efficiently (Whltfeld and Bottomley 1980). These similarities suggest that the transciptional apparatus o f chloroplasts may be sufficiently similar to
that of E. coli to enable us to obtain meaningful data about promoter sites on chloroplast DNA b y using E. coli RNA polymerase. We found an overall correlation of the map o f polymerase binding sites with the map of chloroplast mRNAs which accumulate in vivo. No stich correlation was found, however, in the region of the rRNA genes since although these RNAs are most abundant in the chloroplast no binding site was located close to the genes. These data provide a basis to start looking for sequences that might act as promoters, and also for molecules which might act in the modification o f the activity (Jolly and Bogorad 1980) of the plastid's own RNA polymerase.
Material and Methods DNA Preparation. Spinach chloroplast DNA and chloroplast DNA fragments in pBR322 were isolated using standard procedures (Driesel et al. 1979; Whitfeld and Bottomley 1980, Bohnert et al. 1980). Phages T7, T5 and lambda were a gift from Dr. H. Bujard (Heidelberg) and DNA was isolated according to his laboratory protocol. Chemicals were always of the purest quality available. Polymerase Binding Reaction. The detailed protocol is presented in a recent paper (Zech et al. 1981). Restriction endonuclease digests of DNA were performed (Driesel et al. 1979) and 0.040.1 gmoles of DNA added to reaction mixtures containing a 10 to 80 fold molar excess of E. coli DNA-dependent RNA polymerase (Boehringer, Mannheim, and gifts of Drs. A. Dresel, Heidelberg and P. Whiffeld, Canberra) in 50 #1 incubations containing 20 mM Tris/HC1, pH 8.3, 40 mM KC1, 5 mM MgC12, 0.2 mM EDTA and 0.5 mM dithiothreitol. After incubation for 10 min (37 °C) 150 #1 heparin (100-500 ~g/ml) or poly rI (20250/zg/ml) in the above medium (37 °C) were added and the incubation was continued for 10 min. The solution was then passed through a pre-wetted nitrocellulose filter on a prewarmed (37 °C) filteration apparatus as described (Zech et aL 1981). The DNA fragments which passed through the filter, and those which adhered to the filter following elution (Zech et al. 1981) were precipitated by 2.5 volumes of ethanol at -80 °C and processed separately. Gel Electrophoresis. Samples were loaded onto either 0.6%, 1%, 1.8% horizontal agarose gels in NTE buffer (20 mM sodium acetate, 40 mM Tris/HC1, pH 7.6, lmM EDTA) or 3.5%, 5% and 7% vertical acrylamide gels in TBE buffer (90 mM: Tris/HC1 pH 8.2, 90 mM boric acid, 1 mM EDTA), respectively, and electrophoresed as described (Driesel et al. 1979; 1980; Bohnert et al. 1980). Phage !ambda DNA digested with HindlII, EcoRI, BamHI or pBR322 DNA digested with HpalI or Hind II served as molecular weight markers. Transcription in vitro. The assay is described for the chloroplast DNA fragment SalI-F (8,600 bp) which contains three strong polymerase binding sites. The fragment was isolated from SalI digested pSoc S15 by sucrose gradient centrifugation (Driesel et al. 1980). Of the essentially pttre DNA 0.5 pg was used per experiment. Assay contained the DNA in 20 mM Tris/HC1, pH 8.0, 10 mM MgC12, 0.1 mM EDTA, 140 mM KC1, 0.1 mM each of GTP, CTP and UTP, 0.005 mM ATP, 0.003 mM c~-32p-ATP
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA
39
(equivalent to 20-60/~Ci), 0.1 vg RNA polymerase (E. coli) and 1 /~g of E. eoli tRNA in a volume of 50 ~zl. After 20 rain at 37 °C rifampicin was added to a final concentration of 100/~g/ml and incubation was continued for 5 rain. The reaction was stopped by addin~ cold ATP to 2.5 mM and N-lauryl sarcosinate to 1%. Nucleic acids were precipitated, washed, and electrophoresed on 4% polyacrylamide gels in TBE buffer for 8 h at 100 Volts/20 cm gel. Autoradiography was performed on the wet gel for approximately 1 h (Kodak XR 5).
Miscellaneous Techniques. Isolation of mRNA from chloroplasts (Driesel et al. 1980), iodination of various RNAs (Bohnert et al. 1980), nick translation of DNA fragments (Whitfeld and Bottomly 1980), 5'-end labbelling or 3'-end labelling of DNAs (Maxam and Gilbert 1980); southern type DNA transfer (Bohnert et al. 1980), DNA/RNA and DNA/DNA hybridizations (Whitfeld and Bottomley 1980; Bohnert et al. 1980; Maxam and Gilbert 1980) were performed as alreadY described. Mapping of restriction endonuclease recognition sites on chloroplast DNA fragments was performed according to Smith and Birnstiel (1976).
Fig. I a-d. Comparison of filter bound and unbound restriction endonuclease-generated DNA fragments after incubation with RNA polymerase. Phage lambda DNA a) digested with BamHI x EcoRI; spinach chloroplast DNA digested with SalI b); BamHI c) and Saml d). b Lane with bound fragments; n Lane with unbound fragments. DNA/Polymerase ratios were 20 (a); 20 (b); 20 (c, left lanes); 80 (c, right lanes) and 15 (d), In a, at a ratio of 20 the influence of poly rI on the binding was tested (minus rI - 1 ; 25 ~g/ml - 2; 50 ~zg/ml - 3; 1 0 0 u g / m l - 4 )
Results
Phage l a m b d a D N A restricted w i t h B a m H I and E c o R I endonuclease served as t h e system to test the data which could be o b t a i n e d b y t h e binding o f R N A polymerase to
DNA. U n d e r stringent conditions (Seeburg and Schaller 1975; West and Rodriguez 1980), at a p o l y m e r a s e / D N A ratio o f 20, only three l a m b d a D N A fragments were b o u n d t o the f'flters, and this binding was u n a f f e c t e d b y t h e inclusion o f p o l y r 0 ) at 100 ~g/ml (Fig. la). O f this
40
M. Zech et al.: BacterialRNA PolymeraseBindingto ChloroplastDNA
BamHI/EcoRI digestion the DNA fragment C which contains the promoters PL, Prm, PR and Po (Szybalski and Szybalski 1979) as well as DNA fragments E, containing b2-region p'romoters, and G on which DNA fragment promoter PR is located (Szybalski and Szybaiski 1979) were bound efficiently. Only at a higher polymerase input do other DNA fragments become bound (results not shown). In other similar experiments using phage lambda, T5 and T7 DNAs the results obtained were as predicted from the literature (e.g. Szybalski and Szybalski 1979, as a review; West and Rodriguez 1980; Bujard 1980). Under comparable conditions the interaction of E.. coli RNA potymerase holoenzyme and spinach chloroplast DNA were studied at ratio 10 to 80. The Fig. lb and lc compare the bound and unbound DNA fragments after a SaiI and BamHI endonuclease digestion, respectively. They were selected from many similar experiments. In addition the restriction endonucleases PstI, XhoI and Sinai, for which physical maps exist (Crouse et at. 1978; Herrmann et al. 1980; Schmitt et al. 1981) were used, as well as the endonucleases EcoRI and HindlII which are only partially mapped. Under the conditions used no random binding of the bacterial enzyme to the chloroplast DNA was observed. Instead a preferential retention by the filter of DNA fragments from the single¢opy DNA regions (Crouse et al. 1978; Schmitt et al. 1981) of the molecule could be seen. This is especially obvious from Fig. 1d. In this experiment the endonuclease Sinai was used for restriction prior to binding. The enzyme cuts from the inverted repeat DNA five fragments of 5.0, 1.2, 3.4, 2.4 and 1.15 Md with the three last ones being either partially (3.4 and 1.15 Md) or totally (2.4 Md) part of the spinach chloroplast rDNA unit (Course et al. 1978; Hartley and Head 1979). All of these (ld, arrows) are very weak in binding. Even at polymerase/DNA ratio of 80 the DNA fragment SmaI-I (3.4 Md) which contains most of the 16S rDNA gene and a long region of more than 3 Kb upstream of this gene is bound only weakly. The DNA fragments SmaI-K (2.4 Md) containing the 16S-23S intracistronic spacer and most of the sequence of the 23S rRNA gene and SmaI-N (1.15Md) containing the 23S rRNA gene 3' end and the 4.5 S rRNA sequence and the 5S rRNA gene are noticeable in their inability to bind the bacterial enzyme. At a polymerase/DNA ratio of as low as 10, which is comparable to the experiments with phage lambda, efficient sites of binding were monitored on the DNA fragments SalI-B, C and F (Fig. lb) and on the BamHI generated DNA fragments A to E (Fig. 1c). Some slightly less efficient binding was observed with SalI-D and BamHI-A and HindHI-A. The positions of these binding sites on the map (Fig. 6) were studied in finer detail by carrying out similar experiments using DNA fragments generated by BamHI and HindlII digestion of the
chloroplast DNA (results not shown). The strongly binding fragments were mostly found in the large~ of the single-copy regions of the molecule (Crouse et al. 1978) although efficient binding to BamHI-A and HindlII-A indicates that the small single-copy DNA region also contains binding sites. This is shown in Fig. 6 by introducing the position of the largest HindlII generated DNA fragment (HFI in Fig. 6) which is part of both the largest BamHI and Sail generated DNA fragments (see also Schmitt et al. 1981). This binding to particular DNA fragments was not correlated with the length of the particular DNA piece (see e.g. SalI-F in Fig. lb). It was only observed when polymerase was present and the replacement of the enzyme by BSA, chloroplast RuDP carboxylase, or chloroplast CFI in comparable assays did not result in any fragment retention on the filter (data not shown). In order to shed some light on the significance of these data it was attemptet to determine whether transcripts of the DNA fragments are present in mature chloroplasts. Total chloroplast RNA (larger than 10S in order to exclude tRNA) was radio-iodinated and hybridized to DNA fragments on nitrocellulose (Fig. 2). With the possible exception of a very small Sail- generated fragment which was run off the gels, all SalI DNA fragments showed some hybridization. Strong hybridization was observed with Sail-B, C and F and less strongly with Sail-D, E, G, and H. The strong hybridization to SalI-A is predominantly to the rRNA. The small single-copy DNA region which is located on SalI-A hat to be mapped with other enzymes. By using the data obtained expecially with BamHI and HindlII, which are not yet published totally, it became apparent that transcripts exist from both singlecopy DNA regions. From these hybridizations it was also apparent that the sizes of populations of RNAs from different regions of the chromosome vary considerably since the hybridization intensity did not reflect the DNA stoichiometry. This is true even when the abundant transcription of the rRNA genes which comprise about 90% of the transcripts (Fig. 6, outer circle) is disregarded. Strongest hybridization was observed with SalI-F which carries the gene for a 32,000 dalton protein (Driesel et ai. 1980) (Fig. 2). This DNA fragment also carried the most efficient binding sites or the largest number of binding sites (Fig. 6, inner circle). The plasmid pSoc S15 contains the chloroplast DNA fragment SaiI-F (8.6 Kb) cloned in the SalI site of pBR322. When the inserted DNA from this plasmid is digested with EcoRI and RNA polymerase binding is performed, three subfragments bind strongly: 1.4, 1.28 and 0.8 Md in size (Fig. 3a). By increasing the KC1 concentrations in these experiments after binding (see Fig. 1 in Zech et al. 1981)it could be shown that binding to the 0.8 Md subfragment is strongly reduced, binding to the 1.4 Md subfragment is slightly
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA
41
Fig. 2 a and b. Hybridization of iodinated chloroplast RNAs to chloroplast DNA fragments, a Total chloroplast RNA to Sall generated total DNA fragments (A-K)comparison of different total RNA preparations (I-V). b mRNA enriched for the 32,000 dalton protein (14S RNA) to the insert (SalI-F) in pSoc S15 and to DNA fragments generated by Sall, BamHI and a double enzyme digest. In the BamHI digestion the fragments BamHI-F and BamHI-J carry the 23S and 16S rRNA genes, respectively. The two arrows point to the BamHI or BamHI x SalI fragments carrying the gene. It is apparent that the fragments BamHI-A to D are also transcribed
reduced, while the third site is not affected. This figure also demonstrates a comparatively less efficient binding of E. coli RNA polymerase to the pBR322 promoters which are located on the contaminating fragment named (pBR) in Fig. 3b. Under the conditions used, this DNA fragment remains unbound. The map of'this cloned piece of the chloroplast DNA, established mainly by the procedure of Smith and Birnstiel (1976), is shown in Fig. 3a. Besides looking at the stability of the complexes in high salt, initial experiments on the velocity of complex formation were performed. Binding of polymerase to all three EcoRI fragments of the insert in pSoc S15 was essentially complete after the shortest measurable time of 10-15 s at a polymerase/DNA ratio of 20 in the presence of 50/~g/ml poly rI (Fig. 3b). We focussed our interest mainly on the SalI-EcoRI subfragment approximately 2 Kb in length, since it most
probably contains the coding sequence for the so-called photogene (Driesel et al. 1980; Bedbrook et al. 1978). The smallest DNA fragment from this region to which a strong polymerase binding could be observed was a 440 bp HpalI/SmaI fragment which, upon digestion with HindlI, lost its ability to serve as a binding site (Fig. 4a). That this Hind II recognition site is close to the binding site is further substantiated by the fact that binding of RNA polymerase to the fragment prevents HindlI from cutting (Fig. 4b) at this position while the RNA polymerase is not inhibitory to the digestion by restriction endonucleases in general. Since mRNA for the 32,000 dalton protein gene hybridized to the 440 bp HpalI/SamI fragment (Driesel et al. 1980) and since the HpalI site in the gene is only approximately 300 bp apart from the HindlI site it can be estimated that the start of the mRNA hybridization
42
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA
pBR322
Hpan
1.4 pSocS15 57.21H i kb)n
Eco R! ~Ec~ RI1
J
._~
1.28
440bp ~ ~ LI d .'- ~ IS r nTa ] [(~
HpaE
~'~Eco RI
o.11/
0.58 0.85
/ 08 •
EcoR] EcoRI \ 10.19/ 0.37 "
Fig. 3 a and b. Map of pSoc S15 and binding of E. coil RNA polymerase to the chloroplast DNA insert, a The map shows some restriction endonuclease sites. The numbers give the fragment sizes in Md of the SalI/EcoRI or the EcoRI subfragments cleaved from SalI-F. Polymerase binding was observed to the fragments shown by the hatched areas, b Binding of RNA polymerase to the isolated SalI-F from Soc S15 after digestion with EcoRI. Binding was terminated by filteration after 15 s (1), 45 s(2), 1.5 min (3), 2.5 min (4), 4 rain (5). M is the marker DNA (pBR 322/Sali x PstI - upper two bands; pBR322]HpalI -- lower bands). The numbers give the molecular weight of SalI/EcoRI or EcoRI DNA subfragments of SalI-F in Md. A 1.6% agarose gel was used The DNA fragment pBR (arrows) represents some contaminating pBR322 (SalI/EcoRI, large fragment, 2.41 Md in size), b - lane with bound fragments; n - lane with not bound fragments
M. Zech et al.: Bacterial RNA Polymemse Binding to Chloroplast DNA
43
Fig. 4. a and b. RNA polymerase binding in the vicinity of the 32,000 dalton protein gene. a DNA fragment SalI-F was digested with HpalI (1) or Hpa II and HindlI (2). The 0.27 Md (440 bp) HpalI fragment (arrow) binds polymerase while in the double enzyme digest neither of the two subfragments binds. The fragments were separated on a 6% polyacrylamide gel; b = bound fragments; n = unbound fragments, b Protection of the HindlI site on the 0.27 Md (410) bp) HpalI fragment (or 1.28 Md Sal/EcoRI subfragment). 1 - marker DNA (pBR322/PostI - upper two bands; pBR322/HpalI lower bands 2 - SalI-F/Eco RI 3 - S a l I - F / E c o R I plus RNA polymerase followed by a HindlI digestion 4 - SalI-F/EcoRI x HindlI 5 - markerDNA(phagelambda/HindIII) The numbers give the molecular weight of the subfragments from the EcoRI/ SalI 1.28 Md fragment after HindlI digestion
is not more than 200, and most probably less than 1 O0 bp, apart from the gene coding sequence. By using the S a l I - F DNA fragment after two cycles of sucrose gradient centrifugation for in vitro transcription (Fig. 5) it could be shown that theE. coli R N A p o l y merase can recognize some signals on this DNA fragment. Although predominantly a heterogeneous population o f molecules was observed, superimposed in there were transcripts o f 105, 230, 290, 440, 630, 850, and 1,100 nucleotides in lenght (Fig. 5). The comparison o f the data on polymerase binding with the data on m R N A hybridizations to DNA fragments is shown in Fig. 6. The transcription map and the map of strong binding sites largely coincide, i.e. DNA fragments for which transcripts are present in the chloroplasts also possess polymerase binding sites. This correlation can be extended to the relative levels o f polymerase binding and hybridization. The fragment binding polymerase most strongly also hybridizes the most strongly with the presumptive m R N A for the photogene. This correlation does not hold true for the rRNA genes. The
4 Fig. 5. Transcription in vitro of purified DNA fragment SalI-F. Incubation at 37 °C of 0.5/zg DNA and 0.4 pg RNA polymerase in 140 mM KC1, 10 mM MgC12, 0.1 mM EDTA, 20 mM Tris/ HCI, pH 8.0, 0.1 mM DTT and 5% glycerol for 1 min (a); 3 min (b); 15 rain (c); In (d) RNase was added during the incubation for 15 min. The separation was on a 4% paa gel for 8 h at 100 V/20 cm gel. The numbers give sizes in nucleotides
44
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA
[]
RNA'polymnrasa
binding
L'--I transcripts i n - v i v a
~
G
BF3
LSU
H
BF51
K
F
32 Kd
I
~NA
r RNA~, Sal I - ~ Sma I Psi I
rRNAs
/
-
\
rRNAs
"iX,,,
Fig. 6. Map of DNA fragments binding RNA potymerase and map of transcripts found in chloroplast. On the inner circles the recognition sites for Sail, SmaI and PstI are included. The hatched areas represent SalI-DNA fragments and one HindlII-DNA fragment within SaU-A which bind RNA polymerase. On the outer circle the fragments for which transcripts present in viva could be seen are drawn. The height of the column over the respective circle gives a relative measure of the strength of binding or the intensity of radioactivity after hybridizing total high molecular weight chloroplast RNA to DNA fragments. Considerable variations o ccurred in the six experiments performed. The amount of transcript in the rRNAs should be approximately 90% of the total. Towards the centre the positions of some genes and locations of some BamHI and HindlII DNA fragments (BF3, BF5, HF1) are given. The arrows give the directions of transcription. The gene for the 32,000 dalton protein is transcribed towards the nearest rRNA "cistron" (Whitfeld and Bottomley, personal communication).
rRNAs comprise ca. 90% o f total chloroplast RNA, yet the rRNA genes do not possess binding sites for E. coli RNA polymerase (Fig. 6).
Discussion
Under defined conditions (e.g. Seeburg and SchaUer 1975; West and Rodfiguez 1980) the DNA-dependent RNA polymerase binds specifically to promoter sites. These sites can be distinguished from those showing nonspecific, adventitious binding o f the polymerase b y the inclusion of polyamines in the reaction mixture. The results obtained with phage lambda DNA (Fig. la) are consistent with this view (Szybal~ki and Szybalski 1979; Bujard 1980).
In the heterologous reactions between spinach chloroplast DNA and bacterial RNA polymerase, some parts o f the chromosome bind the enzyme specifically and strongly. Operationally we define a binding site as strong when at a polymerase/DNA ratio o f 10 at least half o f the DNA is specifically bound. Not all known phage lambda promoters would pass this selection criterion. Among total chloroplast DNA there are at least 6 SalI DNA fragments which fulfil this criterion. It has, however, to be admitted that such a strong binding might be simulated by the presence of several weak sites. The strongest binding DNA fragment generated b y SalI was indeed shown to contain at least three sites. Two o f these would still fulfil the criterion mentioned above. The use of a heterologous system, as a first attempt by us to look for signals on chloroplast DNA b y which
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA the expression o f genes might be regulated, seemed to be justified for several reasons. There are many similarities between the bacterial and chloroplast system concerning both the mechanism of transcription and translation (Nomura et al. 1977; Schwarz and K6ssel 1980; Bohnert et at. 1980, Bottomley and Bohnert 1981; Rochaix and Malnoe 1978). At least in one case these similarities have been exploited successfully since it could be shown that a coupled transcription/translation system from E. coli is able to express the gene for the LSU o f RuDP carboxylase (Whitfeld and Bottomley 1980; Bottomley and Whitfeld 1979) resulting in the authentic protein product. Following this consideration, there is a high likelihood that specific binding sites in chloroplast DNA can be recognized by E. coli RNA polymerase. We have previously reported (Driesel et al. 1980) that spinach chloroplast RNA sedimenting at 14S on sucrose density gradients contains several mRNAs, the most abundant o f which, as judged b y translation in reticulocyte lysates, codes for a polypeptide o f 32,000 dalton. The mRNA for this polypeptide hybridized strongly to SalI-F. The strongest binding site observed, located on Sail-F, has been shown to lie within less than 200 bp of one end of the 32,000 dalton protein gene. When the fragment S a l I , F was transcribed in vitro several distinct bands o f transcripts appeared. Whether these bands represented complete transcripts of genes was not determined (e.g. by introducing the RNA in a translation system). A strong possibility is that a least the very small RNAs originate by specific but unphysiological initiation or termination o f synthesis. A full-size transcript of the 32,000 dalton protein gene or its precursor (Driesel et al. 1980; Edelman and Reisfeld 1980) would be approximately 1,000 nueleotides long. Under appropriate conditions of light-induced chloroplast development, this mRNA constitutes an abundant transcript in a number of higher plants (Edelman and Reisfeld 1980; Silverthorne and Ellis 1980; Driesel et al. 1980; Speirs and Grierson 1978) and in the unicellular alga Chlamydomonas (Malnoe et al. 1979). The term "photogene" was coined by Bedbrook et al. (1978) to describe the photoregulated expression o f this gene. Our knowledge concerning the function of this protein is rapidly increasing since there are indications that it may have an important role as part of the primary electron acceptor in photosystem II (Steinback et al. 1981; Mattoo et al. 1981). Additional evidence for the presence o f a polymerase binding site close to the photogene has come from sequencing data. The region to which binding occurs is AT rich, i t shows excellent homology with known prokaryotic promoters, and from the sequence the start of a gene may be deduced nearby (G. Zurawski et al. unpublished results). Further sequencing studies on the DNA, RNA and the N-terminal sequence of the protein,
45
where a possible precursor should be taken into account, should be done to determine whether the E. coli polymerase binding site is a chloroplast promoter and functions with the chloroplast DNA-dependent RNA polymerase.
Acknowledgements: M. R. H. and H. J. B. want to acknowledge the hospitality of the Division of Plant Industry, CSIRO, Canberra where M. R. H. constructed some of the plasmids used in this study. We wish to thank Drs. Bedbrook, Bottomley, Chandler, Dennis and Whitfeld (Canberra) for their comments on the use of the heterologous system, and Christine Miehalowsld for technical assistance. - This work was supported by the Deutsche For schungsgemeinschaft.
References Bedbrook JR, Link G, Coen DM, Bogorad L, Rich A (1978) Proc Natl Acad Sci USA 75:3060-3064 Bedbrook JR, Kolodner R (1979) Annu Rev Plant Physiol 30: 593-620 Bohnert HJ, Driesel AJ, Herrmann RG (1977) In: Bogorad L, Well HJ (eds) Acides nucl~iques et synth~se des prot6ines chez les v6g~taux. CNRS, Paris, pp 213-217 Bohnert HJ, Gordon KHJ, Crouse EJ (1980) Mol Gen Genet 179:539-545 Bottomley W, Whitfeld PR (1979) Eur J Biochem 93:31-39 Bottomley W, Bohnert HJ (1981) In: Pirson A, Zimmermann MH (eds) Encyclopaedia of Plant Physiology, New Series. Springer, Berlin Heidelberg New York (in press) Briat JF, Mache R (1980) Eur J Biochem 111:503-509 Bujard H (1980) Trends Biochem Sci 5:274-278 Crouse EJ, Schmitt JM, Bohnert HJ, Gordon K, Driesel AJ, Herrmann RG (1978) In: Akoyunoglou G (ed) Chloroplast Development. Elesvier-North Holland, Amsterdam, pp 565572 Driesel AJ, (]rouse EJ, Gordon K, Bohnert HJ, Herrmann RG, Steinmetz A, Mubumbila M, Keller M, Burkard G, Weil JH (1979) Gene 6:285-306 Driesel AJ, Speirs J, Bohnert HJ (1980) Biochim Biophys Acta 610:297-310 Edelman M, Reisfeld A (1980) In: Leaver CJ (ed) Genome Organization and Expression in Plants. Plenum, New York, pp 353-362 Hartley MR, Head C (1979) Eur J Biochem 96:301-310 Herrmann RG, Whitfeld PR, Bottomley W (1980) Gene 8:179191 Jolly SO, Bogorad L (1980) Proc Natl Acad Sci USA 77:822826 Link G, Coen DM, Bogorad L (1978) Cell 15:725-731 Mainoe P, Rochaix JD, Chua NH, Spahr PF (1979) J Mol Biol 133:417-434 Matsuda Y, Surzycki SJ (1980) Mol Gen Genet 180:463-474 Mattoo AK, Pick U, Hoffman-Falk H, Edelman M (1981) Proc Natl Acad USA 78:1572-1576 Maxam AM, Gilbert W (1980) Methods Enzymol 65:499-560 Nomura M, Morgan EA, Jaskunas SR (1977) Annu Rev Genet 11:297-347 Rochaix JD, Malnoe P (1978) Cell 15:681-690 Schmitt JM, Bohnert HJ, Gordon KHJ, Herrmann RG, Bernardi G, Crouse EJ (1981) Eur J Bioehem 117:375-382 Schwarz Zs, KSssel H (1980) Nature 283:739-742 Seeburg PH, Schaller H (1975) J Mol Biol 92:261-277
46
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA
Silverthorne J, Ellis RJ (1980) Biochim Biophys Acta 607:319330 Smith HJ, Bogorad L (1974) Proc Natl Acad Sci USA 71:48394842 Smith HO, Birnstiel ML (1976) Nucleic Acids Res 3:2387-2398 Speirs J, Griers6n D (1978) Biochim Biophys Acta 521:619-633 Steinback KE, Pfister K, Arntzen CJ (1981) Z. Naturforsch 36C: 98-108 Szybalsld EH, Szybalski W (1979) Gene 7:217-270 West RW, Rodriguez RL (1980) Gene 9:175-193
Whitfeld PR, Herrmann RG, Bottomley W (1978) Nucleic Acids Res 5:1741-1751 Whitfeld PR, Bottomley W (1980) Biochem Intemat 1:172-178 Young R, Macklis R, Steitz J (1979) J Biol Chem 254:32643271 Zech M, Wolter FP, Bohnert HJ (1981) Experientia 37:537-539 Communicated b y C. P. Hollenberg Received May 25/July 16, 1981
Current Genetics (1981) 4:37-46
© Springer-Verlag 1981
Binding Sites of E. coli DNA-dependent RNA Polymerase on Spinach Chloroplast DNA Michael Zech 1 , Martin R. Hartley2 , and Hans J. Bohnert 1' 3 1 Botanical Institute, University of Diisseldorf,Federal Republic of Germany 2 Dept. BiologicalSciences,Universityof Warwick,Coventry, United Kingdom
Summary. Under stringent conditions E. coli DNA-dependent RNA polymerase holoenzyme binds selectively to some spinach chloroplast DNA fragments generated by restriction endonucleases. The strongest of these binding sites, as judged by the initial rate of complex formation, are located in the large single-copy DNA region (Crouse et al. 1978) of this molecule and correspond in map location with known protein coding sequences. Some of these binding sites have characteristics of complex formation comparable with those of the PR and PL promoters of phage lambda DNA. Binding sites located close to the rRNA operons on the chloroplast DNA bind polymerase less strongly than those described above. Since the rRNAs are the most abundant transcription products in vivo and in isolated chloroplasts (Hartley and Head 1979; Bohnert et al. 1977) this suggests that the E. coli and chloroplast enzymes do not recognize all of the major promoters in chloroplast DNA with the same efficiency of binding. We have investigated in detail one region of the chloroplast DNA from spinach which contains three strong binding sites. This region has been shown to contain at least the gene for a 32,000 dalton protein (Driesel et al. 1980) which is most probably the so-called photogene (Bedbrook et al. 1978). One of these three E. coli RNA polymerase binding sites is not more than approximately 150 bp apart from what, by hybridization studies using isolated mRNA, we know to be the coding sequence for this protein. The results suggest that for some genes on the chloroplast DNA the bacterial RNA polymerase may be used to search for transcription initiation sites.
3 Present address for correspondence: EMBL,Postfach 10 22 09, 6900 Heidelberg, FRG
Key words: RNA polymerase binding sites - Chloroplast DNA - Photogene
Introduction
The spinach chloroplast DNA is a circular molecule 143 Kb in size. Its gross morphology, like that of most other chloroplasts, is characterized by two repeat segments 22 Kb long in inverted orientation which are separated by a 80 Kb large single-copy DNA region and a 18.5 Kb small single-copy DNA region (Bedbrook and Kolodner 1979; Crouse et al. 1978). Among the genes known on this chloroplast DNA are rRNA and tRNA genes (Crouse et al. 1978; Whitfeld et al. 1978; Driesel et al. 1979), the gene for the LSU protein of the ribulose 1.5 diphosphate carboxylase (Whitfeld and Bottomley 1980) as well as for a 32,000 dalton protein (Driesel et al. 1980). While investigations to find the positions of more genes continue, the way in which chloroplast DNA transcription is regulated may already be studied with respect to known genes. The study of the mechanisms by which individual genes are transcribed may help to explain the way in which individual RNAs accumulate to differing extents within the chloroplast (Driesel et al. 1980; Matsuda and Surzycki 1980, Silverthome and Ellis 1980, Edelman and Reisfeld 1980). It is not yet known whether chloroplast mRNAs are transcribed as approximately gene-sized sequences, or are derived by processing larger transcripts. Only for the rRNAs a larger precursor molecule has been demonstrated which includes at least the 16S, 23S and 4.5S rRNA genes together with some transcribed spacer (Hartley and Head 1979, Bohnert et al. 1977). Similarly, little is known of the mechanisms whereby chloroplast mRNAs differ in their pool sizes. Do the dff0172-8083/81/0004/0037/$ 02.00
38
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA
ferences arise through transcriptional control or by different stabilities? The answer o f these questions would significantly contribute towards our understanding o f the regulation o f gene activity in chloroplasts. One way to investigate this problem is through the mapping of RNA polymerase binding sites on chloroplast DNA. It can then be established whether these sites are gene-promoter sites. The specificity of the binding, the spatial arrangement and the positions o f these sites in relation to known genes provide valuable preliminary information. The approach described in this paper was initially used to map bacteriophage promoters (Seeburg and Schaller 1975; West and Rodriguez 1980). Although several examples o f transcriptional control of chloroplast gene expression have been reported (Matsuda and Surzycki 1980; Bedbrook et al. 1978; Link et al. 1978) little is known about the mechanism involved. Chloroplast RNA polymerase is known to be a multisubunit enzyme (Smith and Bogorad 1974; Briat and Mache 1980), though no functions can yet be ascribed to individual subunits. Jolly and Bogorad (1980) have recently reported that maize plastids contain a protein which specifically enhances transcription of chloroplast DNA fragments in supercoiled chimeric plasmids by the purified chloroplast polymerase. Whether or not this "stimulator" factor was associated with the polymerase in vivo, and lost during purification or exists as a separate entity was not determined. The problem o f purifying chloroplast RNA polymerase with its in vivo regulatory functions still intact is a very real one and has probably not been solved completely. While we do not regard the use o f the E. coli enzyme in the context of the present study necessarily to be highly specific, the same is also true for the chloroplast enzyme because o f the uncertainties still connected with this enzyme. The initial work described here uses a heterologous system composed of E. coli DNA-dependent RNA polymerase and chloroplast DNA. The rationale for the use of this particular system is apparent when the similarities between bacterial systems and the chloroplast are considered. These similarities include details o f gene expression in both systems (Hartley and Head 1979; Bohnert et al. 1977; Young et al. 1979; Rochaix and Malnoe 1978), the organization o f rRNA cistrons (Nomura et al. 1977; Schwarz and K6ssel 1980), sequence conservation in rRNA genes (Schwarz and K6ssel 1980; Bohnert et al. 1980), and the similarity o f structure and antibiotic sensitivities between bacterial and chloroplast ribosomes (Nomura et al. 1977; Bottomley and Bohnert 1981). In addition, the gene for the LSU of the ribulose 1.5 diphosphate carboxylase appears to be transcribed and translated in a heterologous system derived from E. coli efficiently (Whltfeld and Bottomley 1980). These similarities suggest that the transciptional apparatus o f chloroplasts may be sufficiently similar to
that of E. coli to enable us to obtain meaningful data about promoter sites on chloroplast DNA b y using E. coli RNA polymerase. We found an overall correlation of the map o f polymerase binding sites with the map of chloroplast mRNAs which accumulate in vivo. No stich correlation was found, however, in the region of the rRNA genes since although these RNAs are most abundant in the chloroplast no binding site was located close to the genes. These data provide a basis to start looking for sequences that might act as promoters, and also for molecules which might act in the modification o f the activity (Jolly and Bogorad 1980) of the plastid's own RNA polymerase.
Material and Methods DNA Preparation. Spinach chloroplast DNA and chloroplast DNA fragments in pBR322 were isolated using standard procedures (Driesel et al. 1979; Whitfeld and Bottomley 1980, Bohnert et al. 1980). Phages T7, T5 and lambda were a gift from Dr. H. Bujard (Heidelberg) and DNA was isolated according to his laboratory protocol. Chemicals were always of the purest quality available. Polymerase Binding Reaction. The detailed protocol is presented in a recent paper (Zech et al. 1981). Restriction endonuclease digests of DNA were performed (Driesel et al. 1979) and 0.040.1 gmoles of DNA added to reaction mixtures containing a 10 to 80 fold molar excess of E. coli DNA-dependent RNA polymerase (Boehringer, Mannheim, and gifts of Drs. A. Dresel, Heidelberg and P. Whiffeld, Canberra) in 50 #1 incubations containing 20 mM Tris/HC1, pH 8.3, 40 mM KC1, 5 mM MgC12, 0.2 mM EDTA and 0.5 mM dithiothreitol. After incubation for 10 min (37 °C) 150 #1 heparin (100-500 ~g/ml) or poly rI (20250/zg/ml) in the above medium (37 °C) were added and the incubation was continued for 10 min. The solution was then passed through a pre-wetted nitrocellulose filter on a prewarmed (37 °C) filteration apparatus as described (Zech et aL 1981). The DNA fragments which passed through the filter, and those which adhered to the filter following elution (Zech et al. 1981) were precipitated by 2.5 volumes of ethanol at -80 °C and processed separately. Gel Electrophoresis. Samples were loaded onto either 0.6%, 1%, 1.8% horizontal agarose gels in NTE buffer (20 mM sodium acetate, 40 mM Tris/HC1, pH 7.6, lmM EDTA) or 3.5%, 5% and 7% vertical acrylamide gels in TBE buffer (90 mM: Tris/HC1 pH 8.2, 90 mM boric acid, 1 mM EDTA), respectively, and electrophoresed as described (Driesel et al. 1979; 1980; Bohnert et al. 1980). Phage !ambda DNA digested with HindlII, EcoRI, BamHI or pBR322 DNA digested with HpalI or Hind II served as molecular weight markers. Transcription in vitro. The assay is described for the chloroplast DNA fragment SalI-F (8,600 bp) which contains three strong polymerase binding sites. The fragment was isolated from SalI digested pSoc S15 by sucrose gradient centrifugation (Driesel et al. 1980). Of the essentially pttre DNA 0.5 pg was used per experiment. Assay contained the DNA in 20 mM Tris/HC1, pH 8.0, 10 mM MgC12, 0.1 mM EDTA, 140 mM KC1, 0.1 mM each of GTP, CTP and UTP, 0.005 mM ATP, 0.003 mM c~-32p-ATP
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA
39
(equivalent to 20-60/~Ci), 0.1 vg RNA polymerase (E. coli) and 1 /~g of E. eoli tRNA in a volume of 50 ~zl. After 20 rain at 37 °C rifampicin was added to a final concentration of 100/~g/ml and incubation was continued for 5 rain. The reaction was stopped by addin~ cold ATP to 2.5 mM and N-lauryl sarcosinate to 1%. Nucleic acids were precipitated, washed, and electrophoresed on 4% polyacrylamide gels in TBE buffer for 8 h at 100 Volts/20 cm gel. Autoradiography was performed on the wet gel for approximately 1 h (Kodak XR 5).
Miscellaneous Techniques. Isolation of mRNA from chloroplasts (Driesel et al. 1980), iodination of various RNAs (Bohnert et al. 1980), nick translation of DNA fragments (Whitfeld and Bottomly 1980), 5'-end labbelling or 3'-end labelling of DNAs (Maxam and Gilbert 1980); southern type DNA transfer (Bohnert et al. 1980), DNA/RNA and DNA/DNA hybridizations (Whitfeld and Bottomley 1980; Bohnert et al. 1980; Maxam and Gilbert 1980) were performed as alreadY described. Mapping of restriction endonuclease recognition sites on chloroplast DNA fragments was performed according to Smith and Birnstiel (1976).
Fig. I a-d. Comparison of filter bound and unbound restriction endonuclease-generated DNA fragments after incubation with RNA polymerase. Phage lambda DNA a) digested with BamHI x EcoRI; spinach chloroplast DNA digested with SalI b); BamHI c) and Saml d). b Lane with bound fragments; n Lane with unbound fragments. DNA/Polymerase ratios were 20 (a); 20 (b); 20 (c, left lanes); 80 (c, right lanes) and 15 (d), In a, at a ratio of 20 the influence of poly rI on the binding was tested (minus rI - 1 ; 25 ~g/ml - 2; 50 ~zg/ml - 3; 1 0 0 u g / m l - 4 )
Results
Phage l a m b d a D N A restricted w i t h B a m H I and E c o R I endonuclease served as t h e system to test the data which could be o b t a i n e d b y t h e binding o f R N A polymerase to
DNA. U n d e r stringent conditions (Seeburg and Schaller 1975; West and Rodriguez 1980), at a p o l y m e r a s e / D N A ratio o f 20, only three l a m b d a D N A fragments were b o u n d t o the f'flters, and this binding was u n a f f e c t e d b y t h e inclusion o f p o l y r 0 ) at 100 ~g/ml (Fig. la). O f this
40
M. Zech et al.: BacterialRNA PolymeraseBindingto ChloroplastDNA
BamHI/EcoRI digestion the DNA fragment C which contains the promoters PL, Prm, PR and Po (Szybalski and Szybalski 1979) as well as DNA fragments E, containing b2-region p'romoters, and G on which DNA fragment promoter PR is located (Szybalski and Szybaiski 1979) were bound efficiently. Only at a higher polymerase input do other DNA fragments become bound (results not shown). In other similar experiments using phage lambda, T5 and T7 DNAs the results obtained were as predicted from the literature (e.g. Szybalski and Szybalski 1979, as a review; West and Rodriguez 1980; Bujard 1980). Under comparable conditions the interaction of E.. coli RNA potymerase holoenzyme and spinach chloroplast DNA were studied at ratio 10 to 80. The Fig. lb and lc compare the bound and unbound DNA fragments after a SaiI and BamHI endonuclease digestion, respectively. They were selected from many similar experiments. In addition the restriction endonucleases PstI, XhoI and Sinai, for which physical maps exist (Crouse et at. 1978; Herrmann et al. 1980; Schmitt et al. 1981) were used, as well as the endonucleases EcoRI and HindlII which are only partially mapped. Under the conditions used no random binding of the bacterial enzyme to the chloroplast DNA was observed. Instead a preferential retention by the filter of DNA fragments from the single¢opy DNA regions (Crouse et al. 1978; Schmitt et al. 1981) of the molecule could be seen. This is especially obvious from Fig. 1d. In this experiment the endonuclease Sinai was used for restriction prior to binding. The enzyme cuts from the inverted repeat DNA five fragments of 5.0, 1.2, 3.4, 2.4 and 1.15 Md with the three last ones being either partially (3.4 and 1.15 Md) or totally (2.4 Md) part of the spinach chloroplast rDNA unit (Course et al. 1978; Hartley and Head 1979). All of these (ld, arrows) are very weak in binding. Even at polymerase/DNA ratio of 80 the DNA fragment SmaI-I (3.4 Md) which contains most of the 16S rDNA gene and a long region of more than 3 Kb upstream of this gene is bound only weakly. The DNA fragments SmaI-K (2.4 Md) containing the 16S-23S intracistronic spacer and most of the sequence of the 23S rRNA gene and SmaI-N (1.15Md) containing the 23S rRNA gene 3' end and the 4.5 S rRNA sequence and the 5S rRNA gene are noticeable in their inability to bind the bacterial enzyme. At a polymerase/DNA ratio of as low as 10, which is comparable to the experiments with phage lambda, efficient sites of binding were monitored on the DNA fragments SalI-B, C and F (Fig. lb) and on the BamHI generated DNA fragments A to E (Fig. 1c). Some slightly less efficient binding was observed with SalI-D and BamHI-A and HindHI-A. The positions of these binding sites on the map (Fig. 6) were studied in finer detail by carrying out similar experiments using DNA fragments generated by BamHI and HindlII digestion of the
chloroplast DNA (results not shown). The strongly binding fragments were mostly found in the large~ of the single-copy regions of the molecule (Crouse et al. 1978) although efficient binding to BamHI-A and HindlII-A indicates that the small single-copy DNA region also contains binding sites. This is shown in Fig. 6 by introducing the position of the largest HindlII generated DNA fragment (HFI in Fig. 6) which is part of both the largest BamHI and Sail generated DNA fragments (see also Schmitt et al. 1981). This binding to particular DNA fragments was not correlated with the length of the particular DNA piece (see e.g. SalI-F in Fig. lb). It was only observed when polymerase was present and the replacement of the enzyme by BSA, chloroplast RuDP carboxylase, or chloroplast CFI in comparable assays did not result in any fragment retention on the filter (data not shown). In order to shed some light on the significance of these data it was attemptet to determine whether transcripts of the DNA fragments are present in mature chloroplasts. Total chloroplast RNA (larger than 10S in order to exclude tRNA) was radio-iodinated and hybridized to DNA fragments on nitrocellulose (Fig. 2). With the possible exception of a very small Sail- generated fragment which was run off the gels, all SalI DNA fragments showed some hybridization. Strong hybridization was observed with Sail-B, C and F and less strongly with Sail-D, E, G, and H. The strong hybridization to SalI-A is predominantly to the rRNA. The small single-copy DNA region which is located on SalI-A hat to be mapped with other enzymes. By using the data obtained expecially with BamHI and HindlII, which are not yet published totally, it became apparent that transcripts exist from both singlecopy DNA regions. From these hybridizations it was also apparent that the sizes of populations of RNAs from different regions of the chromosome vary considerably since the hybridization intensity did not reflect the DNA stoichiometry. This is true even when the abundant transcription of the rRNA genes which comprise about 90% of the transcripts (Fig. 6, outer circle) is disregarded. Strongest hybridization was observed with SalI-F which carries the gene for a 32,000 dalton protein (Driesel et ai. 1980) (Fig. 2). This DNA fragment also carried the most efficient binding sites or the largest number of binding sites (Fig. 6, inner circle). The plasmid pSoc S15 contains the chloroplast DNA fragment SaiI-F (8.6 Kb) cloned in the SalI site of pBR322. When the inserted DNA from this plasmid is digested with EcoRI and RNA polymerase binding is performed, three subfragments bind strongly: 1.4, 1.28 and 0.8 Md in size (Fig. 3a). By increasing the KC1 concentrations in these experiments after binding (see Fig. 1 in Zech et al. 1981)it could be shown that binding to the 0.8 Md subfragment is strongly reduced, binding to the 1.4 Md subfragment is slightly
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA
41
Fig. 2 a and b. Hybridization of iodinated chloroplast RNAs to chloroplast DNA fragments, a Total chloroplast RNA to Sall generated total DNA fragments (A-K)comparison of different total RNA preparations (I-V). b mRNA enriched for the 32,000 dalton protein (14S RNA) to the insert (SalI-F) in pSoc S15 and to DNA fragments generated by Sall, BamHI and a double enzyme digest. In the BamHI digestion the fragments BamHI-F and BamHI-J carry the 23S and 16S rRNA genes, respectively. The two arrows point to the BamHI or BamHI x SalI fragments carrying the gene. It is apparent that the fragments BamHI-A to D are also transcribed
reduced, while the third site is not affected. This figure also demonstrates a comparatively less efficient binding of E. coli RNA polymerase to the pBR322 promoters which are located on the contaminating fragment named (pBR) in Fig. 3b. Under the conditions used, this DNA fragment remains unbound. The map of'this cloned piece of the chloroplast DNA, established mainly by the procedure of Smith and Birnstiel (1976), is shown in Fig. 3a. Besides looking at the stability of the complexes in high salt, initial experiments on the velocity of complex formation were performed. Binding of polymerase to all three EcoRI fragments of the insert in pSoc S15 was essentially complete after the shortest measurable time of 10-15 s at a polymerase/DNA ratio of 20 in the presence of 50/~g/ml poly rI (Fig. 3b). We focussed our interest mainly on the SalI-EcoRI subfragment approximately 2 Kb in length, since it most
probably contains the coding sequence for the so-called photogene (Driesel et al. 1980; Bedbrook et al. 1978). The smallest DNA fragment from this region to which a strong polymerase binding could be observed was a 440 bp HpalI/SmaI fragment which, upon digestion with HindlI, lost its ability to serve as a binding site (Fig. 4a). That this Hind II recognition site is close to the binding site is further substantiated by the fact that binding of RNA polymerase to the fragment prevents HindlI from cutting (Fig. 4b) at this position while the RNA polymerase is not inhibitory to the digestion by restriction endonucleases in general. Since mRNA for the 32,000 dalton protein gene hybridized to the 440 bp HpalI/SamI fragment (Driesel et al. 1980) and since the HpalI site in the gene is only approximately 300 bp apart from the HindlI site it can be estimated that the start of the mRNA hybridization
42
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA
pBR322
Hpan
1.4 pSocS15 57.21H i kb)n
Eco R! ~Ec~ RI1
J
._~
1.28
440bp ~ ~ LI d .'- ~ IS r nTa ] [(~
HpaE
~'~Eco RI
o.11/
0.58 0.85
/ 08 •
EcoR] EcoRI \ 10.19/ 0.37 "
Fig. 3 a and b. Map of pSoc S15 and binding of E. coil RNA polymerase to the chloroplast DNA insert, a The map shows some restriction endonuclease sites. The numbers give the fragment sizes in Md of the SalI/EcoRI or the EcoRI subfragments cleaved from SalI-F. Polymerase binding was observed to the fragments shown by the hatched areas, b Binding of RNA polymerase to the isolated SalI-F from Soc S15 after digestion with EcoRI. Binding was terminated by filteration after 15 s (1), 45 s(2), 1.5 min (3), 2.5 min (4), 4 rain (5). M is the marker DNA (pBR 322/Sali x PstI - upper two bands; pBR322]HpalI -- lower bands). The numbers give the molecular weight of SalI/EcoRI or EcoRI DNA subfragments of SalI-F in Md. A 1.6% agarose gel was used The DNA fragment pBR (arrows) represents some contaminating pBR322 (SalI/EcoRI, large fragment, 2.41 Md in size), b - lane with bound fragments; n - lane with not bound fragments
M. Zech et al.: Bacterial RNA Polymemse Binding to Chloroplast DNA
43
Fig. 4. a and b. RNA polymerase binding in the vicinity of the 32,000 dalton protein gene. a DNA fragment SalI-F was digested with HpalI (1) or Hpa II and HindlI (2). The 0.27 Md (440 bp) HpalI fragment (arrow) binds polymerase while in the double enzyme digest neither of the two subfragments binds. The fragments were separated on a 6% polyacrylamide gel; b = bound fragments; n = unbound fragments, b Protection of the HindlI site on the 0.27 Md (410) bp) HpalI fragment (or 1.28 Md Sal/EcoRI subfragment). 1 - marker DNA (pBR322/PostI - upper two bands; pBR322/HpalI lower bands 2 - SalI-F/Eco RI 3 - S a l I - F / E c o R I plus RNA polymerase followed by a HindlI digestion 4 - SalI-F/EcoRI x HindlI 5 - markerDNA(phagelambda/HindIII) The numbers give the molecular weight of the subfragments from the EcoRI/ SalI 1.28 Md fragment after HindlI digestion
is not more than 200, and most probably less than 1 O0 bp, apart from the gene coding sequence. By using the S a l I - F DNA fragment after two cycles of sucrose gradient centrifugation for in vitro transcription (Fig. 5) it could be shown that theE. coli R N A p o l y merase can recognize some signals on this DNA fragment. Although predominantly a heterogeneous population o f molecules was observed, superimposed in there were transcripts o f 105, 230, 290, 440, 630, 850, and 1,100 nucleotides in lenght (Fig. 5). The comparison o f the data on polymerase binding with the data on m R N A hybridizations to DNA fragments is shown in Fig. 6. The transcription map and the map of strong binding sites largely coincide, i.e. DNA fragments for which transcripts are present in the chloroplasts also possess polymerase binding sites. This correlation can be extended to the relative levels o f polymerase binding and hybridization. The fragment binding polymerase most strongly also hybridizes the most strongly with the presumptive m R N A for the photogene. This correlation does not hold true for the rRNA genes. The
4 Fig. 5. Transcription in vitro of purified DNA fragment SalI-F. Incubation at 37 °C of 0.5/zg DNA and 0.4 pg RNA polymerase in 140 mM KC1, 10 mM MgC12, 0.1 mM EDTA, 20 mM Tris/ HCI, pH 8.0, 0.1 mM DTT and 5% glycerol for 1 min (a); 3 min (b); 15 rain (c); In (d) RNase was added during the incubation for 15 min. The separation was on a 4% paa gel for 8 h at 100 V/20 cm gel. The numbers give sizes in nucleotides
44
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA
[]
RNA'polymnrasa
binding
L'--I transcripts i n - v i v a
~
G
BF3
LSU
H
BF51
K
F
32 Kd
I
~NA
r RNA~, Sal I - ~ Sma I Psi I
rRNAs
/
-
\
rRNAs
"iX,,,
Fig. 6. Map of DNA fragments binding RNA potymerase and map of transcripts found in chloroplast. On the inner circles the recognition sites for Sail, SmaI and PstI are included. The hatched areas represent SalI-DNA fragments and one HindlII-DNA fragment within SaU-A which bind RNA polymerase. On the outer circle the fragments for which transcripts present in viva could be seen are drawn. The height of the column over the respective circle gives a relative measure of the strength of binding or the intensity of radioactivity after hybridizing total high molecular weight chloroplast RNA to DNA fragments. Considerable variations o ccurred in the six experiments performed. The amount of transcript in the rRNAs should be approximately 90% of the total. Towards the centre the positions of some genes and locations of some BamHI and HindlII DNA fragments (BF3, BF5, HF1) are given. The arrows give the directions of transcription. The gene for the 32,000 dalton protein is transcribed towards the nearest rRNA "cistron" (Whitfeld and Bottomley, personal communication).
rRNAs comprise ca. 90% o f total chloroplast RNA, yet the rRNA genes do not possess binding sites for E. coli RNA polymerase (Fig. 6).
Discussion
Under defined conditions (e.g. Seeburg and SchaUer 1975; West and Rodfiguez 1980) the DNA-dependent RNA polymerase binds specifically to promoter sites. These sites can be distinguished from those showing nonspecific, adventitious binding o f the polymerase b y the inclusion of polyamines in the reaction mixture. The results obtained with phage lambda DNA (Fig. la) are consistent with this view (Szybal~ki and Szybalski 1979; Bujard 1980).
In the heterologous reactions between spinach chloroplast DNA and bacterial RNA polymerase, some parts o f the chromosome bind the enzyme specifically and strongly. Operationally we define a binding site as strong when at a polymerase/DNA ratio o f 10 at least half o f the DNA is specifically bound. Not all known phage lambda promoters would pass this selection criterion. Among total chloroplast DNA there are at least 6 SalI DNA fragments which fulfil this criterion. It has, however, to be admitted that such a strong binding might be simulated by the presence of several weak sites. The strongest binding DNA fragment generated b y SalI was indeed shown to contain at least three sites. Two o f these would still fulfil the criterion mentioned above. The use of a heterologous system, as a first attempt by us to look for signals on chloroplast DNA b y which
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA the expression o f genes might be regulated, seemed to be justified for several reasons. There are many similarities between the bacterial and chloroplast system concerning both the mechanism of transcription and translation (Nomura et al. 1977; Schwarz and K6ssel 1980; Bohnert et at. 1980, Bottomley and Bohnert 1981; Rochaix and Malnoe 1978). At least in one case these similarities have been exploited successfully since it could be shown that a coupled transcription/translation system from E. coli is able to express the gene for the LSU o f RuDP carboxylase (Whitfeld and Bottomley 1980; Bottomley and Whitfeld 1979) resulting in the authentic protein product. Following this consideration, there is a high likelihood that specific binding sites in chloroplast DNA can be recognized by E. coli RNA polymerase. We have previously reported (Driesel et al. 1980) that spinach chloroplast RNA sedimenting at 14S on sucrose density gradients contains several mRNAs, the most abundant o f which, as judged b y translation in reticulocyte lysates, codes for a polypeptide o f 32,000 dalton. The mRNA for this polypeptide hybridized strongly to SalI-F. The strongest binding site observed, located on Sail-F, has been shown to lie within less than 200 bp of one end of the 32,000 dalton protein gene. When the fragment S a l I , F was transcribed in vitro several distinct bands o f transcripts appeared. Whether these bands represented complete transcripts of genes was not determined (e.g. by introducing the RNA in a translation system). A strong possibility is that a least the very small RNAs originate by specific but unphysiological initiation or termination o f synthesis. A full-size transcript of the 32,000 dalton protein gene or its precursor (Driesel et al. 1980; Edelman and Reisfeld 1980) would be approximately 1,000 nueleotides long. Under appropriate conditions of light-induced chloroplast development, this mRNA constitutes an abundant transcript in a number of higher plants (Edelman and Reisfeld 1980; Silverthorne and Ellis 1980; Driesel et al. 1980; Speirs and Grierson 1978) and in the unicellular alga Chlamydomonas (Malnoe et al. 1979). The term "photogene" was coined by Bedbrook et al. (1978) to describe the photoregulated expression o f this gene. Our knowledge concerning the function of this protein is rapidly increasing since there are indications that it may have an important role as part of the primary electron acceptor in photosystem II (Steinback et al. 1981; Mattoo et al. 1981). Additional evidence for the presence o f a polymerase binding site close to the photogene has come from sequencing data. The region to which binding occurs is AT rich, i t shows excellent homology with known prokaryotic promoters, and from the sequence the start of a gene may be deduced nearby (G. Zurawski et al. unpublished results). Further sequencing studies on the DNA, RNA and the N-terminal sequence of the protein,
45
where a possible precursor should be taken into account, should be done to determine whether the E. coli polymerase binding site is a chloroplast promoter and functions with the chloroplast DNA-dependent RNA polymerase.
Acknowledgements: M. R. H. and H. J. B. want to acknowledge the hospitality of the Division of Plant Industry, CSIRO, Canberra where M. R. H. constructed some of the plasmids used in this study. We wish to thank Drs. Bedbrook, Bottomley, Chandler, Dennis and Whitfeld (Canberra) for their comments on the use of the heterologous system, and Christine Miehalowsld for technical assistance. - This work was supported by the Deutsche For schungsgemeinschaft.
References Bedbrook JR, Link G, Coen DM, Bogorad L, Rich A (1978) Proc Natl Acad Sci USA 75:3060-3064 Bedbrook JR, Kolodner R (1979) Annu Rev Plant Physiol 30: 593-620 Bohnert HJ, Driesel AJ, Herrmann RG (1977) In: Bogorad L, Well HJ (eds) Acides nucl~iques et synth~se des prot6ines chez les v6g~taux. CNRS, Paris, pp 213-217 Bohnert HJ, Gordon KHJ, Crouse EJ (1980) Mol Gen Genet 179:539-545 Bottomley W, Whitfeld PR (1979) Eur J Biochem 93:31-39 Bottomley W, Bohnert HJ (1981) In: Pirson A, Zimmermann MH (eds) Encyclopaedia of Plant Physiology, New Series. Springer, Berlin Heidelberg New York (in press) Briat JF, Mache R (1980) Eur J Biochem 111:503-509 Bujard H (1980) Trends Biochem Sci 5:274-278 Crouse EJ, Schmitt JM, Bohnert HJ, Gordon K, Driesel AJ, Herrmann RG (1978) In: Akoyunoglou G (ed) Chloroplast Development. Elesvier-North Holland, Amsterdam, pp 565572 Driesel AJ, (]rouse EJ, Gordon K, Bohnert HJ, Herrmann RG, Steinmetz A, Mubumbila M, Keller M, Burkard G, Weil JH (1979) Gene 6:285-306 Driesel AJ, Speirs J, Bohnert HJ (1980) Biochim Biophys Acta 610:297-310 Edelman M, Reisfeld A (1980) In: Leaver CJ (ed) Genome Organization and Expression in Plants. Plenum, New York, pp 353-362 Hartley MR, Head C (1979) Eur J Biochem 96:301-310 Herrmann RG, Whitfeld PR, Bottomley W (1980) Gene 8:179191 Jolly SO, Bogorad L (1980) Proc Natl Acad Sci USA 77:822826 Link G, Coen DM, Bogorad L (1978) Cell 15:725-731 Mainoe P, Rochaix JD, Chua NH, Spahr PF (1979) J Mol Biol 133:417-434 Matsuda Y, Surzycki SJ (1980) Mol Gen Genet 180:463-474 Mattoo AK, Pick U, Hoffman-Falk H, Edelman M (1981) Proc Natl Acad USA 78:1572-1576 Maxam AM, Gilbert W (1980) Methods Enzymol 65:499-560 Nomura M, Morgan EA, Jaskunas SR (1977) Annu Rev Genet 11:297-347 Rochaix JD, Malnoe P (1978) Cell 15:681-690 Schmitt JM, Bohnert HJ, Gordon KHJ, Herrmann RG, Bernardi G, Crouse EJ (1981) Eur J Bioehem 117:375-382 Schwarz Zs, KSssel H (1980) Nature 283:739-742 Seeburg PH, Schaller H (1975) J Mol Biol 92:261-277
46
M. Zech et al.: Bacterial RNA Polymerase Binding to Chloroplast DNA
Silverthorne J, Ellis RJ (1980) Biochim Biophys Acta 607:319330 Smith HJ, Bogorad L (1974) Proc Natl Acad Sci USA 71:48394842 Smith HO, Birnstiel ML (1976) Nucleic Acids Res 3:2387-2398 Speirs J, Griers6n D (1978) Biochim Biophys Acta 521:619-633 Steinback KE, Pfister K, Arntzen CJ (1981) Z. Naturforsch 36C: 98-108 Szybalsld EH, Szybalski W (1979) Gene 7:217-270 West RW, Rodriguez RL (1980) Gene 9:175-193
Whitfeld PR, Herrmann RG, Bottomley W (1978) Nucleic Acids Res 5:1741-1751 Whitfeld PR, Bottomley W (1980) Biochem Intemat 1:172-178 Young R, Macklis R, Steitz J (1979) J Biol Chem 254:32643271 Zech M, Wolter FP, Bohnert HJ (1981) Experientia 37:537-539 Communicated b y C. P. Hollenberg Received May 25/July 16, 1981
