G. Link, Plant Cell Physiology and Molecular Biology, University of Bochum, Universitaetsstr. 150,D-44780 Bochum, Germany. Fax: +49 234 3214 188, Tel.: +49 234 322 5495, E-mail: email@example.com
We have isolated and studied the cloned sigma factors SASIG1-3 from mustard (Sinapis alba). In functional analyses using both promoter and factor mutants, the three recombinant proteins all had similar basic properties but also revealed differences in promoter preference and requirements for single nucleotide positions. Directed muta- genesis of SASIG1 identified critical residues within the conserved regions 2.4 and 4.2 necessary for binding of the −10 and −35 promoter elements, respectively. SASIG1 and 2, but not SASIG3, each have a typical region 2.5 for binding of the extended −10 promoter element. SASIG3 has a pro-sequence reminiscent of σK from Bacillus subtilis, suggesting that proteolytic cleavage from an inactive precursor is involved in the regulation of plastid transcription. In addition, SASIG2 was found to be more abundant in light-grown as compared to dark-grown mustard seedlings, while the converse was true for SASIG3.
bacterial-type plastid RNA polymerase with core subunits encoded by organellar genes
Chloroplasts contain the photosynthetic machinery, which is built-up and maintained by gene-regulatory mechanisms both inside and outside the organelle. At the level of transcription this involves the participation of multiple RNA polymerases, at least two of which are located within the plastid compartment: (a) a single-subunit type enzyme related to those of T-odd phages and mitochondria; and (b) a multisubunit form resembling those of bacteria and eukaryotic nuclei (reviewed in ). The former is a product of nuclear gene(s) (nuclear-encoded phage-type plastid RNA polymerase; NEP); the latter, which is the primary enzyme for transcription of photosynthesis-related chloroplast genes , contains an organelle-encoded core of eubacterial α, β, and β′ homologues  and hence has been termed PEP .
The catalytic core of the PEP enzyme assembles with regulatory proteins that have been identified as functional equivalents of bacterial sigma factors and hence named SLFs (sigma-like factors) (reviewed in ). Given the central role of sigma in prokaryotic transcription initiation [6,7], and considering the endosymbiotic origin of chloroplasts , one might expect coding information for sigma-like protein(s) on plastid DNA. However, extensive sequence analysis has shown this not to be the case , suggesting that the SLFs might be nuclear gene products. Direct evidence for this was obtained by the cloning of sigma-like cDNA sequences from algae [9,10] followed by those from several higher plants (reviewed in ).
Our previous work on chloroplast transcription in mustard (Sinapis alba) had resulted in the purification and biochemical characterization of three SLFs [12,13]. As part of our efforts to clarify the role of this transcription factor family, we reported on the cloning of a first member, which we referred to as SASIG1 . Here we present SASIG2 and SASIG3, and we describe functional studies with all three recombinant factors.
Materials and methods
PCR and cDNA cloning
Oligonucleotides were derived from expressed sequence tags with sequence similarity to the coding regions of the AtSig2 and AtSig3 genes from Arabidopsis thaliana (GenBank accession numbers AC003981 and N97044) Two pairs, 5′-GAGAAACAAGTGATACGTTGGAGA-3′ and 5′-TTTCCTGAATGCAGATGACTCTAT-3′ from AtSig2 and 5′-GTAGAGATGAACTGGTGAAAAGCA-3′ and 5′-AAAACTTGTAACCCCTTGTGTGAT-3′ from AtSig3, were used for PCR-amplification from total S. alba DNA. The products were cloned into the EcoRV site of pBluescript (Stratagene), resulting in plasmids pBS/3C1Sig2 and pBS/4B2Sig3. The inserts were used for screening of a HybriZAP cDNA library . Clones pAD/3C1Sig2 and pAD/4B2Sig3 contained the full-length SaSig2 and SaSig3 cDNAs, respectively.
Isolation of proteins and Western blot analysis
Mustard (S. alba) seedlings were grown at 25 °C for 4 days either under continuous light (250 µmol photons·m−2·s−1) or in the dark. Cotyledons were harvested and whole-cell or plastid proteins were prepared as described [15,16]. Bacterially expressed recombinant proteins were isolated from inclusion bodies, followed by SDS/PAGE. The gel-purified proteins were used as antigens in a custom protocol for rabbit immunization (Eurogentec). For Western analysis, protein samples were separated by SDS/PAGE and transferred onto nitrocellulose membrane. Proteins were incubated with anti-SASIG2 and anti-SASIG3 antisera, respectively, and detected by nitroblue tetrazolium/5-bromo-4-chloro-3-indolylphosphate.
Cloned fragments of mustard chloroplast DNA [5,14,17]
Plasmid pSA05/H120 carries the psbA promoter (accession no. X04826); the 120-bp HinfI insert (H120) covers 68 bp upstream of the transcription start site and 52 bp noncoding 5′-sequence of the gene. A 450-bp EcoRI–HindIII fragment that carries the trnK (X04826) promoter was prepared from pSA364-EH450, a pSPT18-based derivative of pSA364. Bam0.5, the 460-bp BamHI fragment of pSA364-B0.5, represents intron sequences of the split trnK gene (X04826). Plasmid pSA364-ET0.2 contains the trnQ (X13558) promoter and flanking sequences in pSPT19. The 213-bp fragment resulting from cleavage by EcoRI and TaqI covers 56 bp downstream of the transcription start site. The pUC13-based plasmid pSA364-H018 carries the rps16 (X13609) promoter; its 188-bp HinfI insert extends 107 bp downstream of the transcription start point. The rbcL plasmid pBS1.4E/P (X73284) was constructed by cloning a 1.4-kb EcoRI–PstI fragment of pSA530 into pBSIISK. Using BamHI and HindIII digestion, an ≈ 500-bp subfragment containing the rbcL promoter and flanking sequences (164 bp downstream of the transcription start site) was generated. Plasmid pBS-1.7kb-B, including the promoter and both coding and noncoding regions of the ycf3 gene (AJ242660), was cleaved with DdeI and EcoRV; this resulted in a 273-bp fragment that covered 158 bp downstream of the ycf3 transcription start. The 205-bp insert of pBSKS/205TD carries the region upstream of the rrn16 gene (X04182) that contains the P1, PC and P2 promoters . All DNA fragments specified above were used as unlabelled competitors in electrophoretic mobility shift DNA binding assays (EMSA). In addition, the H120 fragment that carries the psbA promoter was used as a labelled probe and for construction of point mutants, and the EcoRI-linearized pSA05/H120 served as a transcription template (see below). The DNA fragments that carry mustard chloroplast promoters are summarized in Table 1 and details of the promoters are depicted in Figure 5A and 6A.
Table 1. Cloned DNA fragments containing mustard chloroplast promoters. Plasmids were digested using the indicated restriction enzymes. DNA fragments were gel-purified and then tested in competition EMSA experiments as described in Materials and methods.
Fragment size (bp)
Restriction enzyme(s) used
Portion downstream of transcription start (bp)
Bacterial expression and purification of recombinant factors
The SaSig1–3 cDNAs lacking the transit peptide region were each cloned in pQE30 (Qiagen). A truncated SaSig3 construct was made by using the PCR primers 5′-CACACAAGGGGTTACAAGTTCTCCACG-3′ and 5′-ACCA GCCAATTGGTTCCAAAAATCTATCT-3′ and ligation into pQE31 (Qiagen). Following transformation of M15, the recombinant proteins were purified on Ni-nitrilotriacetic acid agarose columns (Qiagen).
Gel shift DNA-binding assays
Recombinant sigma factors were incubated with 2.5 ng 32P-labelled H120 fragment and 0.5 µg E. coli core RNA polymerase in 50 µL of 30 mm Tris/HCl pH 7.0, 5 mmβ-mercaptoethanol, 0.5 mm EDTA, 5% (v/v) glycerol for 10 min at 25 °C. DNA–protein complexes were analysed on a native 5% (w/v) polyacrylamide gel (29 : 1 acrylamide/bisacrylamide) containing 0.5 m Tris/HCl pH 8.8. The gel was dried and then analysed using a Fuji BAS 2040-phosphoimager. Unlabelled DNA fragments representing various chloroplast promoters were prepared as summarized in Table 1. These fragments were used in competition EMSA (Fig. 5). Likewise, psbA promoter fragments carrying point mutations were used as competitors (Fig. 6).
Mutagenesis of the psbA promoter
The cloned 120-bp HinfI region containing the psbA promoter  (see Table 1) was used as the starting material for the construction of promoter point mutants. M-19 A/G, M-21 A/T, M-22 T/A, M-23 A/T and M-34 T/C were obtained by the M13-based technique previously described . All other point mutants were made by PCR using the QuickChange site-directed mutagenesis kit (Stratagene). Both the length and sequence outside the changed position were confirmed to be identical for each mutant fragment.
Mutagenesis and expression of SASIG1-300Q/H and SASIG1-455R/H
SASIG1 mutants were constructed by using the QuickChange mutagenesis kit (Stratagene). Primers for the substitutions 300Q/H and 455R/H were 5′-TATACTGGTGGATTCGACACGGTGTGTCAAGAGCATTAG-3′ and 5′-GAGAGAGAGGGTTCATCAGGTGGGGCTTGTGG-3′, respectively. The fragments were cloned into the BamHI/SalI sites of pMAL-c2x (NEB). The bacterially expressed mutant factors fused to maltose binding protein were purified on amylose affinity columns (NEB).
In vitro transcription
In vitro transcription reactions (25 µL) contained 50 mm Tris/HCl pH 8.0, 80 mm (NH4)2SO4, 10 mm MgCl2, 1 mm dithiothreitol, 600 µm each of ATP, GTP and CTP, 10 µm UTP, 20 µCi [α-32P]UTP (Amersham, 400 µCi mmol−1), 10 U RNaseOUT (BRL), 0.05 U of E. coli RNA poly- merase holoenzyme (Roche) or 25 nm core enzyme (Epicentre), and 1 µg double-stranded EcoRI-linearized DNA (pSA05/H120 carrying the psbA promoter; see Table 1 and section ‘Cloned Fragments’ above). Sigma proteins were added to give a final concentration of 100 nm. Following preincubation at 30 °C for 10 min without template, the latter was added and incubation was continued for 15 min. After phenol/chloroform extraction and ethanol precipitation the transcripts were electrophoresed on 6% (w/v) sequencing gels.
Characterization of cDNAs for putative sigma factors from S. alba
The full-length cDNAs for SASIG1  (accession number Y15899), SASIG2 (this work; accession number AJ276656) and SASIG3 (accession number AJ276657) were cloned from a mustard cDNA library. They translate into open reading frames for 481 (SASIG1), 575 (SASIG2) and 567 amino acids (SASIG3), respectively. As shown in Fig. 1, the C-terminal portion of each derived SASIG polypeptide resembles that of eubacterial σ70-type factors with their typical regions 1.2–4.2 . Within these regions, sequence elements can be located for which distinct functions have been assigned in bacterial systems. This is exemplified in Fig. 1 for region 4.2, which contains a helix-turn-helix (HTH) unit  involved in recognition of the −35 promoter element .
Compared to the C-terminal portion, the N-terminal half of the SASIG proteins is less conserved (Fig. 3A). Although a putative region 1.2 could be localized as depicted in Fig. 1A and 3A, no definite assignment of a region 1.1 was possible (sequence data not shown). The most proximal part of each N-terminal SASIG protein revealed characteristics that could be implicated with chloroplast targeting on the basis of chlorop, psort, targetp and pclr. The putative transit peptide was predicted to comprise 83 amino acids in SASIG1, 39 in SASIG2, and 74 in SASIG3 (Fig. 1B).
Plastid localization of SASIG proteins
Evidence for chloroplast targeting of SASIG1 was previously obtained . To find out if SASIG2 and 3 are likewise chloroplast localized, antibodies were raised against the recombinant proteins and used in immunoblotting experiments. In brief, cotyledons of light-grown mustard seedlings were harvested and protein extracts were prepared from either whole cells or purified chloroplasts. The extracts were then electrophoretically separated, immunoblotted, and probed with anti-SASIG2 and anti-SASIG3 sera. As shown in Fig. 2A, each antiserum detected one single band, at 61 kDa (SASIG2) and 65 kDa (SASIG3), respectively, both in the whole-cell (lanes 3 and 6) and chloroplast fractions (lanes 4 and 7). It was therefore concluded that both SASIG2 and SASIG3 are to a large part localized to the chloroplast.
To investigate if the expression of SASIG2 and SASIG3 was affected by light vs. dark growth conditions, mustard seedlings were grown for 4 days either under continuous light or in darkness. Whole-cell protein extracts were prepared from cotyledons and subjected to immunoblot analysis with antisera raised against SASIG2 or 3. As shown in Fig. 2B, the authentic plant protein corresponding to SASIG2 was found in higher relative amounts in the light (L) then in the dark (D). The converse was true for SASIG3, which accumulated to higher levels in etioplast-containing dark-grown tissue, indicating differential expression of the respective genes under these two growth conditions. Furthermore, additional distinct bands below the major signal were visible in the protein sample from dark-grown tissue (Fig. 2B, lane 4) following incubation with anti-SASIG3. Although cross-reaction of the antiserum with other (dark-specific) proteins cannot be ruled out, it appears more likely that these smaller extra bands resulted from proteolytic cleavage (see Discussion).
Sigma-like enzymatic properties of the cloned SASIG proteins
To help decide whether the cloned SASIG proteins had indeed properties consistent with a role as a sigma factor, we used gel shift DNA binding (Fig. 3B) and in vitro transcription assays (Fig. 4). In each case, the full reaction contained either of the recombinant proteins mixed with E. coli core RNA polymerase and the mustard psbA promoter [5,18]. The choice of the heterologous system was based on previous findings demonstrating that the (biochemically purified) chloroplast SLFs in combination with E. coli core enzyme were capable of faithful and efficient binding and transcription initiation at this promoter .
Fig. 3A gives a schematic view of the SASIG1–3 sequences derived from the full-length cDNAs. The recombinant proteins used in the in vitro experiments lacked the N-terminal putative transit peptide. Also included in this analysis was a more extensively truncated derivative of SASIG3, which represented only the C-terminal portion with position 374 as the first residue (SIG3-374) (Fig. 3A, bottom).
As shown in Fig. 3B, the full reactions containing DNA, core enzyme and one of the recombinant SASIG proteins always resulted in a band shift (‘b’ position) as compared to the free probe (‘f’ position; lanes 8–11). The binding signal was stable in the presence of excess nonpromoter DNA, i.e. the Bam0.5 intron fragment of the mustard chloroplast trnK gene  (lanes 12–15). The signal intensity became highly reduced, however, if unlabelled psbA promoter fragment was used as a competitor, as is exemplified in lane 16 for SASIG3 (see Figs 5 and 6 for the other factors). None of the controls consisting of the DNA probe alone (lane 1), or of probe plus recombinant proteins (lanes 4–7), gave a labelled band at the position of the binding signal. Although a band was visible at this position when the probe fragment was incubated with E. coli core enzyme alone in the absence of poly(dIdC) (lane 3), this unspecific binding signal  almost completely disappeared in the presence of the polynucleotide (lane 2). These data thus provided initial evidence that the SASIG proteins were able to confer specific DNA binding on the E. coli core polymerase.
In another set of experiments we tested psbA promoter-driven transcription by core polymerase in the presence or absence of the recombinant SASIG proteins. As shown in Fig. 4, multiple run-off transcripts were detected following synthesis with core enzyme alone (lane 2) and none were visible with any of the recombinant factors alone (data not shown). In contrast, a major (66-nt) transcript of the size expected for faithful initiation at the psbA promoter  was visible with E. coli RNA polymerase holoenzyme (lane 1). A single transcript of this size was detected also in the presence of core enzyme plus SASIG1 (lane 3), SASIG2 (lane 4), or SASIG3-374 (lane 6), indicating that each of these factors had mediated correct initiation. This band was not visible, however, in the presence of core enzyme plus full-length SASIG3 (lane 5).
Together, the EMSA (Fig. 3B) and transcription data (Fig. 4) suggest that the recombinant SASIG proteins confer promoter-specific binding and transcription initiation on the RNA polymerase, thus providing clues as to their functional roles as sigma factors . In the case of SASIG3, however, this seems true only for the truncated SASIG3-374, whereas the full-size protein directs DNA binding (Fig. 3B, lane 10) but not transcription initiation (Fig. 4, lane 5).
Relative affinity of SASIG proteins to chloroplast promoters
To further compare the promoter binding characteristics of the recombinant factors, they were tested in combination with a number of mustard chloroplast promoters as depicted in Fig. 5A. Competition gel shift experiments were carried out in the presence of E. coli core enzyme, labelled psbA promoter fragment, and excess unlabelled fragments. As shown in Fig. 5B, the latter varied in their efficiency to act as competitors, both compared to one another and depending on which sigma factor was present in the reaction mixture. The largest decrease in the intensity of the radioactive binding signal was always found when the psbA promoter fragment itself was used as competitor, which reflects the strength of this promoter  (Fig. 5A).
For the majority of other tested promoters, including those for trnK, trnQ, rps16 and rrn16, the competition patterns were similar with either SASIG1 or SASIG3 (including the truncated form SASIG3-374; data not shown), and they differed from that observed with SASIG2 (Fig. 5B). The only deviations from this general pattern were observed for the rbcL and ycf3 promoters (Fig. 5A). The rbcL promoter was completely ineffective in the presence of SASIG3 but showed slight competitor efficiency in the case of either SASIG1 or SASIG2. None of the three recombinant factors seemed to be able to mediate efficient binding to the ycf3 promoter as revealed by its almost complete inactivity as a competitor.
Mutational studies reveal common and distinct DNA-binding properties of SASIG proteins
To analyse the role of single nucleotide positions within the chloroplast psbA promoter for DNA-binding, competition gel shift assays were carried out (Fig. 6B). The psbA promoter mutants had base substitutions in either of the conserved regions, i.e. the −35-region, the TATA box-like element, the −10-region, or the extended −10-motif (Fig. 6A). The sigma factors that were used included the SASIG proteins as well as σ70 from E. coli.
Of the psbA promoter mutants that were altered in the −35-region, M-32 G/A (Fig. 6B, upper right panel) almost completely lacked competitor efficiency in the EMSA, suggesting that the G/A exchange at this position had rendered the psbA promoter inactive as a binding target with any of the four sigma factors. In contrast, the M-34 T/C mutant (upper row, middle) had a noticeable effect as a competitor, and even more so did M-30 C/A (second row, left). The latter revealed strength comparable to that of the wild-type promoter (WT; upper left) if either of the SASIG proteins was used. In the presence of σ70, however, this mutant had less competitor strength than the wild-type construct, which may indicate minor differences in binding requirements at this position for the plant vs. bacterial factors. Despite this, the results obtained with the −35 mutants support the view that bases within this region of the psbA promoter are common determinants of DNA binding strength.
In efforts to apply mutational strategies to other regions of the psbA promoter, we introduced four different single-base substitutions into the TATA-box like element (Fig. 6A). When these mutant constructs were tested, a partial loss of competitor strength became apparent, and this effect was more pronounced in the presence of SASIG1 and SASIG2 compared to SASIG3 and σ70. This was most evident in the case of M-19 A/G (Fig. 6B, third row, middle panel) and M-22 T/A (second row, right), whereas in the case of M-23 A/T (second row, middle) and M-21 A/T (third row, left) the decrease in competitor efficiency compared to wild-type (WT; upper left) was only marginal. These data suggested that the plant factors SASIG1 and SASIG2 were capable of recognizing bases within the spacer between the −10- and −35-regions. In contrast, SASIG3 and σ70 seemed to be less dependent on contact sites within this region of the psbA promoter.
When base changes within the −10-region (Fig. 6A) were tested in the EMSA (Fig. 6B, bottom row), they all caused a reduction in competitor strength compared to the wild-type psbA promoter, yet to different extents. Only a moderate decrease in competitor strength compared to wild-type (WT; upper left) was observed for M-5 T/C, a mutant in which the last base of the −10-element was altered (bottom row, right). More dramatic effects were noticeable with either the double mutant M-5–6 T/C–C/A (bottom row, middle) or with M-10 T/A (bottom, left).
In contrast with the −35 mutant M-32 G/A (upper right panel), none of these −10 mutations seemed to uniformly affect binding by all four sigma factors. In the case of M-5–6 T/C–C/A, the competitor strength of this mutant was still significantly stronger in the presence of SASIG1 than with any of the other factors tested. The −10 T/A mutation dramatically affected the promoter usage by SASIG1 and wild-type; SASIG2 (as well as σ70), but more weakly that by SASIG3 (Fig. 6B, lower left). The latter factor lacks the conserved Glu in the −10-recognition region (Fig. 7A), which participates in binding to the first base of the −10-element in E. coli.
The (5′-TG-3′) positions −13 and −12 of the mustard psbA promoter match those of the extended −10-motif of bacterial promoters, i.e. the known contact site for residues in the sigma region 2.5 . Conserved amino acids reminiscent of this region are present in SASIG1 and SASIG2 but not in SASIG3 (Figs 7A; H and E). It was therefore of interest to test the possible role of bases within this region of the chloroplast promoter. In contrast, the G/T transversion at −12 (Fig. 6A) resulted in almost complete loss of competitor efficiency in the presence of SASIG1, SASIG2 or σ70 (Fig. 6B, third row, right panel). With SASIG3, however, only a partial reduction in competitor strength compared to wild-type (upper left) was noticeable, which is consistent with the lack of a conserved region 2.5 in this factor (Fig. 7A).
In the case of σ70 the HTH-motif  in region 4.2 is involved in binding of the −35 promoter element . As shown in Fig. 7A, two conserved amino acids, E13 and R16, can be identified in all three SASIG proteins within the HTH-region. In σ70 the Arg corresponding to R16 (R588) is known to interact with the third base of the −35-element of prokaryotic promoters . Using SASIG1, we therefore converted R16 into a His and tested the effect of the resulting sigma mutant SASIG1–455R/H in competition EMSAs (Fig. 6). The maltose binding portion of the recombinant proteins did not interfere with core or promoter binding (data not shown).
As is evident from Fig. 7B (right panels; M-32 G/A), the mutant promoter M-32 G/A revealed almost wild-type competitor strength in the presence of the sigma mutant (455R/H), whereas it was largely inactive in the presence of wild-type SASIG1. None of the other −35-mutant promo- ters (Fig. 6A) showed any significant response to SASIG1–455R/H (data not shown). These results established the importance of residue(s) within the putative HTH-motif of the plastid sigma factor, which hence may be functionally related to that of bacterial σ70.
To study the interaction with the −10-region, we converted the Glu (Q300) in SASIG1 into a His and tested the promoter affinity of the resulting mutant factor SASIG1–300Q/H (Fig. 7B, panels on the left). Whereas the mutant promoter construct M-10 T/A almost completely lacked competitor efficiency in the presence of the unmodified SASIG1 protein (WT), it showed considerable strength (binding of the labelled psbA probe only 40% of that in the control) in the presence of the mutant factor (300Q/H). Hence this sigma mutant counteracted the effect of M-10 T/A. None of the other −10 promoter mutations (Fig. 6) were compensated by the Q/H exchange in SASIG1, nor was the psbA wild-type promoter affected to any appreciable extent (data not shown).
In the present work we have studied the three cloned mustard factors SASIG1-3. As shown in Figs 1 and 2, all three proteins have extensive sequence similarity to the conserved regions 1.2–4.2 of bacterial sigma factors (see, e.g. [7,29]). In contrast, a sequence that would closely match region 1.1, i.e. a known modulator of DNA binding by regions 2 and 4 , was not detected.
The N-terminal region of each SASIG protein exhibits stretches rich in Ser and Thr residues (Fig. 1B), which is a property of many chloroplast transit peptides (for review see ). and this region was tentatively identified as a transit peptide by several localization prediction programs. The predominant chloroplast localization of the authentic mustard proteins corresponding to SASIG2 and SASIG3 was verified by immunoblotting experiments (Fig. 2) and for SASIG1 this had previously been demonstrated by in organello import assays . We wish to note that it cannot be ruled out that a minor fraction might be targeted to different intracellular sites, as was shown recently for other chloroplast transcriptional proteins [1,32], including a putative sigma factor from maize . However, a predominant chloroplast localization of the mustard SASIG proteins is in agreement with the conclusions reached for the Arabidopsis putative sigma proteins [34–36].
Transcript analyses of sigma factor genes from monocot and dicot plant species showed most of these to be more actively expressed under light-grown as compared to dark-grown conditions in plant tissue [37–41], and to be almost silent in roots [34,36,42]. On the other hand, Western analyses have provided evidence for differential expression profiles at the protein level. Our previous results for SASIG1  and the present data for SASIG2 and SASIG3 together suggest that the three factors do not have uniform expression profiles. Both SASIG1 and SASIG2 (Fig. 2B, lanes 1 and 2) accumulate preferentially in green tissue of light-grown seedlings, whereas SASIG3 is more abundant in etioplast-containing dark-grown tissue (lanes 3 and 4). This situation is reminiscent of the protein accumulation profiles of two putative sigma factors from the monocot Zea mays. The ZMSIG3 protein, which has 36% sequence similarity to SASIG3 (data not shown), was mostly found in nongreen tissue, whereas ZMSIG1 (50% similarity with SASIG2) accumulated in green leaf tissue .
That the recombinant SASIG proteins have enzymatic characteristics consistent with a role as a sigma factor was demonstrated both in EMSA DNA binding (Fig. 3B) and in vitro transcription experiments (Fig. 4). In the gel shift assays each of the three proteins was found to interact with E. coli core enzyme, resulting in a functional RNA poly- merase holoenzyme that was capable of binding to the chloroplast psbA promoter (Fig. 3B, lanes 8–11). The specificity of DNA binding was indicated by the result that excess unlabelled psbA promoter fragment could function as a competitor (Figs 5 and 6), whereas complex formation was resistant to the addition of a nonpromoter fragment (Fig. 3, lanes 12–15). As shown in Fig. 3, there was no retarded signal with any of the sigma factors alone in the absence of core enzyme (lanes 4–7). This was true also for the N-terminally truncated factor SASIG3-374, in contrast to the situation for σ70, which upon cleavage of its N terminus becomes able to bind to promoter DNA in the absence of core enzyme . Furthermore, in native σ70 regions 1.1 and 4 are located in close proximity, with region 1.1 acting as an autoinhibition domain. Sigma–core interaction induces a conformational change that unmasks the DNA binding domains [43,44]. The lack of DNA binding by SASIG3-374 is consistent with the apparent absence of a functional region 1.1 in (full-length) SASIG3, although it should be noted that also region 1.2 and part of region 2 were removed during construction of SASIG3-374.
To further clarify the role of the SASIG proteins as initiation factors, in vitro transcription assays were carried out. The results depicted in Fig. 4 established this role for SASIG1 and SASIG2, but not for SASIG3. Transcripts of the size expected for correct initiation could only be detected with truncated SASIG3-374 but not with the full-length protein, indicating that the latter might be inactive because of inhibitory sequences at the N terminus.
Inspection of the SASIG3 sequence revealed a motif (residues 277–298) with strong similarity to the amino terminus of σK from B. subtilis(Fig. 8C). It has long been known that certain sporulation factors (σE, σK) are synthesized as inactive precursors that are converted into the active mature proteins by site-specific proteolysis of approximately 20 amino acids at the N terminus [46,47]. A transcription-inhibitory effect of the N-terminal region of the SASIG3 homologue from Arabidopsis thaliana, ATSIG3, has recently been described . Our present data confirm and extend these findings, suggesting that the SASIG3 N-terminal region inhibits transcription (Fig. 4) but not promoter binding (Fig. 3B).
If full-length SASIG3 represents the transcriptionally inactive ‘pro’ form of this plastid sigma factor, the cleavage and subsequent release of proteolytic fargment(s) might be a rapid and tightly regulated process involving one of the known chloroplast proteases in vivo. Our immunoblotting experiments (Fig. 2B, lane 4) suggest limited proteolytic cleavage of the authentic protein detected by antibodies against SASIG3. It is interesting to note, however, that smaller-sized bands were detected only with blotted proteins from dark-grown seedlings (lane 4), where SASIG3 seems to be present in higher relative amounts than in the fraction from light-grown material (lane 3). It is possible that the kinetics of the proteolytic cleavage differ in a plastid-type and/or developmental stage-specific way, with immediate consequences for the availability and function of SASIG3 in vivo as depicted in the model shown in Fig. 8.
This model is based on our observations that full-length SASIG3 efficiently binds to the promoter DNA but is unable to initiate transcription. Only after cleavage of the N-terminal region does the truncated factor (such as SASIG3-374) become transcriptionally competent. The enzymatic conversion of the pro-factor into a fully functional (truncated) SASIG3 protein not only has an effect on transcription under the control of this factor itself, but also on that driven by other plastid sigma factors. Based on the similarities in promoter usage in vitro by SASIG1–3 (Figs 3, 5 and 6), it is conceivable that these factors are capable of competing for one and the same promoter. Tight binding of one factor (e.g. the SASIG3 pro-factor) hence can result in inhibition of transcription by others.
Work on the in vivo expression of the sigma factors from other plants has established overlapping transcript patterns for individual members [34,37,40]. Moreover, genetic evidence is available from Arabidopsis, where a ATSIG2 knockout mutant has only a weak and stage-specific chlorophyll-deficient phenotype, and none of the chloroplast genes psbA, psbD and rbcL was reported to be significantly affected in its expression by this mutation . This suggests that the lack of one particular sigma factor may have less deleterious effects than the converse situation, where a factor physically interferes with transcription unless it is converted into its active form. Proteolytic cleavage (Fig. 8) is just one of a number of possible mechanisms to achieve this, considering the widespread occurrence of protein modifications that can affect the activity of transcription factors .
To address the question as to what extent the SASIG factors reveal selective promoter affinity, we carried out competition EMSA experiments using several different chloroplast promoters (Fig. 5). The strong psbA promoter is recognized by all SASIG factors, and with comparable affinity by each factor. Although the other plastid promo- ters tested in Fig. 5 acted more weakly in terms of competitor strength compared to the psbA promoter. In a relatively few instances there were noticeable differences in the promoter interaction of the individual plastid sigma factors e.g. in the case of the rbcL vs. ycf3 promoters (Fig. 5). This supports the view that, rather than acting in a strictly promoter-selective way, the multiple plastid sigma factors seem to be capable of substituting for each other in vivo.
Development and environmental cues are likely to play a role in the expression patterns of the SASIG factors, as is indicated, for example, by the complementary mode of light vs. dark accumulation of SASIG2 and SASIG3 (Fig. 2B). ATSIG5, one of the Arabidopsis factors, has been reported to reveal a blue-light activated mode of gene expression . Changes in sigma factor abundance and/or activity reflect the activity of interacting proteins, some of which have recently been described to play a role in the regulation of plastid transcription [52–55].
More detailed insight into the determinants for sigma factor–psbA promoter interactions was sought in the mutational studies shown in Figs 6 and 7. These experiments provided evidence that the three plant factors reacted similarly to the mutations in the −35-region but more diverse to the other regions of the promoter. Previous work using σ70 from E. coli had shown that mutations in the recognition helix of the HTH DNA-binding motif in region 4.2 led to altered interaction with the −35-region [28,56]. The mutation of the third base of the psbA−35-region (M-32 G/A) resulted in complete loss of promoter activity for all sigma proteins. The amino acid at position 16 of the σ70 HTH motif (R588) is known to interact with this specific base position .
In all three mustard sigma factors there is an HTH motif containing a conserved Arg within region 4.2 (Figs 1 and 7). Because of this similarity with σ70 we reasoned that Arg455 in SASIG1, Arg545 in SASIG2 and Arg542 in SASIG3 are the residues that might interact with the third base of the −35 element of the mustard psbA promoter. To address this question we substituted Arg455 of SASIG1 by a His and tested the resulting mutant factor SASIG1–455R/H for promoter binding activity. The results (Fig. 7B) showing that the effect of the promoter down-mutation was compensated by the Arg to His change in the recombinant factor suggest that Arg455 of SASIG1 is indeed involved in the interaction with the third position of the −35-region.
Using similar mutational approaches, evidence was obtained for a functional role of at least two other regions of the plastid sigma factor(s). One is the region 2.5 equivalent  found in SASIG1 and SASIG2 but not in SASIG3 (Fig. 7). With the M-12 G/T mutant of the psbA promoter that carried a TG to TT change of the extended −10-region, there was a high decrease in promoter affinity for either SASIG1, SASIG2 or σ70, as was expected from previous findings with σ70 and bacterial promoter mutants . For SASIG3, however, the decrease was much smaller, thus reflecting the lack of a conserved region 2.5 in this factor (Fig. 7).
Furthermore, evidence based on −10-mutants of the psbA promoter suggests a functional link with a −10 recognition region of a cloned plastid sigma factor (Fig. 7). Region 2 of σ70 forms an amphipathic helix , a sequence feature that is conserved in each of the plastid sigma factors (data not shown). This implies that the general −10 DNA-contacting mechanism might be comparable and this view has been strengthened by mutational studies. Conversion of the Glu at position 437 (region 2.4) of σ70 into a His had previously been shown to enhance the activity with a mutant promoter that carried a substitution at the first position of the −10 hexamer . As is evident from Fig. 7A, a conserved Glu is present in SASIG1 (Q300) and SASIG2 (Q394) but not in SASIG3, in agreement with the findings (Fig. 6) that SASIG3 reacted more weakly with the psbA mutant promoter M-10 G/T. These data suggested that Q300 of SASIG1 and Q394 of SASIG2 each were involved in the interaction with the first base of the −10 element of the psbA promoter. This notion was strengthened by the results obtained with SASIG1–300Q/H, which showed significant activity with the mutant promoter (Fig. 7B).
In E. coli, a second residue within region 2.4 of σ70 (T440) is known to interact with the first base of the −10-hexamer . As shown in Fig. 7A, none of the three plant factors contains this conserved Thr, which could reflect differences in DNA binding between the bacterial and plant sigma factors.
We would like to thank C. Berndt and M. Böckmann for their initial participation in the cloning of the plastid sigma factors. This work was funded by the Deutsche Forschungsgemeinschaft (SFB 480-B7).