Evolutionary conservation of C-terminal domains of primary sigma(70)-type transcription factors between plants and bacteria.

Three different cDNAs coding for putative plant plastid sigma(70)-type transcription initiation factors have recently been cloned and sequenced from Arabidopsis thaliana. We have analyzed the evolutionary conservation of function(s) of the N-terminal and C-terminal halves of these three sigma factors by in vitro transcription studies using heterologous transcription systems and by complementation assays using Escherichia coli thermosensitive rpoD mutants. Our results indicate differences and similarities of the three plant factors and their prokaryotic ancestors. The functions of the N-terminal parts of the plant sigma factors are considerably different from the function of the N-terminal part of the principal sigma(70) factor of E. coli. On the other hand, the C-terminal parts have kept at least two characteristics when compared with their prokaryotic ancestors: 1) they can distinguish between different promoter structures, and 2) one of them is capable of fully complementing E. coli rpoD mutants, i.e. recognizing all essential E. coli promoters that are used by the E. coli principal sigma(70) factor. This shows for the first time in vivo a strong evolutionary conservation of cis- and trans-acting elements between the prokaryotic and the plant plastid transcriptional machinery.

Three different cDNAs coding for putative plant plastid sigma 70 -type transcription initiation factors have recently been cloned and sequenced from Arabidopsis thaliana. We have analyzed the evolutionary conservation of function(s) of the N-terminal and C-terminal halves of these three sigma factors by in vitro transcription studies using heterologous transcription systems and by complementation assays using Escherichia coli thermosensitive rpoD mutants. Our results indicate differences and similarities of the three plant factors and their prokaryotic ancestors. The functions of the N-terminal parts of the plant sigma factors are considerably different from the function of the N-terminal part of the principal sigma 70 factor of E. coli. On the other hand, the C-terminal parts have kept at least two characteristics when compared with their prokaryotic ancestors: 1) they can distinguish between different promoter structures, and 2) one of them is capable of fully complementing E. coli rpoD mutants, i.e. recognizing all essential E. coli promoters that are used by the E. coli principal sigma 70 factor. This shows for the first time in vivo a strong evolutionary conservation of cis-and trans-acting elements between the prokaryotic and the plant plastid transcriptional machinery.
The higher plant plastid genome is transcribed by two different transcriptional systems (reviewed in Ref. 1). One of the two systems is phage-like and its evolutionary origin is still unclear (2,3). The other system is of the prokaryotic type revealing the cyanobacterial origin of present-day chloroplasts (4 -6). All polypeptides that are necessary to build up the core enzyme of the prokaryotic-type plastid RNA polymerase are still encoded in the plastid genome (7,1). However, genes corresponding to sigma-like transcription initiation factors that are indispensable for the activity of this type of enzyme (reviewed in Ref. 8) are not found on the plastid genomes (7, 1). These genes have been transferred to the nucleus during evolution (9).
The existence of more than one sigma-like transcription factor in higher plant plastids had been suggested several years ago on the basis of biochemical approaches for spinach and mustard (10 -12). However, the corresponding genes have not been cloned. Only recently, six different cDNAs showing strong sequence similarity with genes coding for prokaryotic-like sig-ma 70 -type initiation factors have been cloned and sequenced from Arabidopsis thaliana (accession numbers: SIG1, dbj: AB004820 and emb: Y15362; SIG2, dbj: AB004821 and emb: Y14567; SIG3, dbj: AB004822; SIG4, gb: AF101075; SIG5, emb: Y18550; and SIG6, emb: AJ250812). For three of them, it has been shown that the corresponding proteins are transported into the chloroplasts (9,14), but the function of these putative transcription factors in the regulation of plastid gene expression is not clear. In particular, it is not clear why several sigma 70 -type factors have been conserved during evolution, despite an important reduction of (plastid) genome size.
Prokaryotic sigma factors can be classified into two families. Members of the first family are similar to the Escherichia coli sigma 70 factor, whereas members of the other family are similar to the E. coli sigma 54 factor. The sigma 70 family of transcription factors can be further subdivided into two or three groups. One primary sigma factor is required to ensure that "housekeeping functions" are performed, and alternative sigma factors direct transcriptional responses to changing environmental conditions (8). Although both groups of sigma 70 factors are similar with respect to their amino acid sequence, they recognize different sequences at the two promoter elements localized -10 and -35 base pairs upstream of the transcription start sites (reviewed in 8 and 15).
In the present study, we attempted to obtain information about the evolutionary conservation of components of the transcriptional machinery between prokaryotes and higher plant plastids. With this aim, we analyzed differences of the three plant sigma factors with respect to promoter recognition by in vitro transcription, and we determined the plant analogue to the primary sigma factor of E. coli by complementation of E. coli rpoD(ts) mutants. For the first time, we show in vivo complementation of a protein of the E. coli transcriptional machinery by a protein of the plastid transcriptional machinery.

Isolation of Clones Coding for A. thaliana Plastid Sigma Factors-A.
thaliana cDNA and/or genomic sequences, similar to prokaryotic sigma factors, have been searched for in currently available data banks. Six such sequences have been found. Four of them have been detected as partial cDNA sequences in the data base expressed sequence tag (dbEST), and two of them have been identified in the genomic sequences resulting from the Arabidopsis Genomic Initiative. The corresponding full-length cDNA clones were isolated from an A. thaliana Matchmaker cDNA library (CLONTECH) either by screening with the corresponding A. thaliana EST clones or by direct PCR 1 amplification (accession numbers: SIG1, Y15362; SIG2, Y14567; SIG5, Y18550; and SIG6, AJ250812).
Overproduction in E. coli and Purification of Recombinant Proteins-Sequences corresponding to full-length and truncated forms of SIG1, SIG2, and SIG3 polypeptides were PCR-amplified using the following primers at the 5Ј ends: SIG1-KLR, 5Ј-CGCGGATCCAAGCTTAGACT-ACCC-3Ј; SIG1-RLV, 5Ј-CGCGGATCCCGTCTTGTTGCGCAGGAAG-* This work was supported by the Region Rhone-Alpes in the frame of the programs Emergence and Biotechnologie and by the European Community in the frame of the Biotech Program BIO4-CT97-2245. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
After cleavage with the appropriate restriction enzymes, the PCR fragments were inserted into Bluescript KS vectors to give pMA plasmids.
For complementation studies E. coli strains UQ285 and CAG1 were obtained from the E. coli Genetic Stock Center (Yale University, New Haven, CT) and transformed with the corresponding plasmids. Transformants were selected on L-broth-ampicillin plates at 32°C and screened at 42°C.
In Vitro Transcription-In vitro transcription reactions were performed at 37°C in 25-l assays. The reaction mixture contained 40 mM Tris/HCl (pH 8.0); 20 M dithiothreitol; 14 mM MgCl 2 ; 40 M EDTA; 50 mM NaCl; 300 M each of GTP, ATP, and CTP and 5 M UTP, including 20 Ci of [␣-32 P]UTP; 71 nM template DNA; and 0.05 units of E. coli holoenzyme (Roche Molecular Biochemicals) or 26 nM of E. coli core enzyme (Epicentre Technologies). Sigma proteins were added to give final concentrations of 100, 200, or 300 nM. Reaction mixtures were preincubated for 5 min without template at 37°C. Transcription was started by addition of DNA template, and the reactions were stopped after 15 min by extraction with phenol/chloroform and precipitated with ethanol. Transcription products were analyzed on 8% acrylamide/urea gels. The template DNA was used in supercoiled form. It corresponded to the plasmid pTZ19 harboring the corresponding promoter regions and the E. coli threonine attenuator (2,4).
Western Blot Analysis of A. thaliana Proteins-Antibodies were produced in rabbits, and horseradish peroxidase-coupled secondary antibodies were detected using the ECL TM Western blotting detection system (Amersham Pharmacia Biotech).

Overproduction of Entire and Truncated Plant Sigma Factors in E. coli-
We have cloned and sequenced the cDNAs corresponding to six putative A. thaliana sigma factors as described under "Materials and Methods." We have analyzed the function of three of them (named SIG1, SIG2, and SIG3, according to Ref. 9) in more detail. To this end, selected fulllength clones have been used to amplify the coding regions and to clone them into two different expression vectors, pET28a and pQE30. Regions coding for supposed transit peptides have been eliminated in the cloning step, thus locating the N-terminal three amino acids to the sequences KLR, STE, and MLV, respectively, as indicated in Fig. 1A. Following transformation of E. coli BL21 or M15 and induction with isopropyl-1-thio-␤-Dgalactopyranoside, only one of the three transcription factors, SIG2, was expressed in sufficient amounts to allow purification and subsequent testing by in vitro transcription assays. SIG1 is expressed only at a very low level (not shown), and the expression of SIG3 seems to be toxic for the bacteria. This can be concluded from the analysis of growth curves of transformed bacteria (Fig. 1B). After induction with isopropyl-1-thio-␤-Dgalactopyranoside, the expression of SIG3 causes growth arrest of E. coli. Similar results were obtained using the pET28a constructions (not shown). In order to know which part of SIG3 is toxic for the bacteria, we subcloned the 5Ј and the 3Ј portions of the cDNAs. These two constructions give rise to truncated proteins corresponding to the N-terminal part (between MLV and GLR) and the C-terminal part (GLR up to the C terminus) of the factor. We found that the N-terminal part was sufficient to provoke growth arrest of E. coli (not shown). The truncated protein corresponding to the C-terminal part could be produced in large amounts without any harmful effects for E. coli.
Having thus shown that the N-terminal parts of two plant sigma-like factors might be incompatible with the E. coli transcriptional machinery, we recloned only the 3Ј portions of the three plant cDNAs that code for three essential functions of a sigma factor: promoter recognition (regions 2.4 and 4.2), DNA melting (region 2.3), and RNA polymerase-core interactions (regions 2.1 and 3.2, reviewed in Ref. 8). It has already been shown that truncated sigma factors retain DNA binding and core RNA polymerase binding activity (16 -18). Therefore, we assumed that it should be possible to use such truncated proteins to characterize these two functions by in vitro transcription assays. Sequence alignment of the amino acid sequences of the three plant sigma factors with the one of E. coli shows that the C-terminal regions, but not the N-terminal parts, of the proteins are highly conserved. Each of these truncated proteins (detailed in Fig. 1A) could be produced in large amounts in E. coli, indicating that the N-terminal parts of SIG1 and SIG3 are responsible for the failure of overproduction of the entire proteins.
Characterization of Truncated Sigma Factors by in Vitro Transcription-SIG2 was the only sigma factor that could be obtained as an entire protein by overproduction in E. coli. We used this protein to analyze whether our assumption was correct, i.e. that the truncated proteins could be used to characterize the two essential functions of a sigma factor: promoter recognition and RNA polymerase-core interaction. To this aim we compared the full-length SIG2 factor (SIG2-STE) with its truncated form (SIG2-VRG) in an in vitro transcription assay using three different plant plastid promoters with the E. coli core enzyme (Fig. 2). The sequences of the three different plastid promoters and their comparison to the E. coli consensus promoter sequence, which is recognized by the E. coli primary sigma factor, RpoD, are shown in Fig. 1C. These particular plastid promoters have been chosen for the following reason: The rbcL promoter represents one highly transcribed plastid promoter, regulating the transcription of the most abundant chloroplast protein and thus implying that it should be recognized by a primary sigma factor. The two other plastid promoters, rrn-P1 and rrn-P2, have plastid-specific regulatory functions in the expression of the rrn operon (4). Therefore, they represent good candidates to be recognized by an alternative sigma factor that fulfils specific functions in the down-regulation of rrn expression during plastid development.
The E. coli holoenzyme recognizes all three plastids promot-ers in in vitro transcription assays (Fig. 2, lanes 15 and 18), albeit with different efficiencies (rbcLϾrrn-P2Ͼrrn-P1; note that rrn transcription is exposed four times longer than rbcL transcription). The E. coli core enzyme produces a longer RNA than the one that is initiated at the -180 rbcL promoter (Fig. 2,  lanes 1 and 8, asterisk). We did not map the exact location of the 5Ј end of this RNA, but the production of this RNA can serve to verify the saturation by sigma factor of commercially available E. coli RNA polymerase preparations (see Fig. 6, lanes 1-3 and  18). Reconstitution of the E. coli core enzyme with either the full-length (Fig. 2, lanes 1-7; exposure times were 12 h for lanes 1-4 and 2 days for lanes 3Ј, 4Ј, and 5-7) or the truncated form of SIG2 (Fig. 2, lanes 8 -14; exposure time was 2 days) gave similar results. The nonspecific transcription product that was obtained when using the cloned rbcL promoter as template diminished gradually with increasing SIG2 concentrations, and very small amounts of the specific transcript (-180) were produced. The appearance of the specific transcript does not quantitatively correlate with the disappearance of the nonspecific transcript. We interpret this result to indicate that the factor has a high affinity for the E. coli core enzyme but a very low affinity for the -180 rbcL promoter. A slightly different result was obtained for the rrn promoter region. Neither of the two SIG2 proteins allows initiation, even at low levels, at either of the two E. coli-like rrn promoters (Fig. 2, lanes 5-7 and  12-14). If SIG2-VRG is added to transcription assays that are performed with the E. coli holoenzyme, the transcription that initiates at the rrn-P1 promoter is specifically diminished (Fig.  2, lanes 15-20). This suggests that of the three promoters tested, SIG2 recognizes only the rrn-P1 promoter efficiently. The full-length form of SIG2, SIG-STE, gave the same result as the truncated protein (not shown). Altogether, this indicates that SIG2 interacts with the E. coli core enzyme. This sigma factor can specifically recognize the rrn-P1 promoter, but it cannot activate transcription efficiently, not even when present as a full-length native protein.
In the case of SIG1, we analyzed two different truncated forms of the protein (RLV and RQR; see Fig. 1). SIG1-RLV corresponds in length to the truncated form of SIG2 and harbors a supposed 1.2 region at its N terminus. We found that only SIG1-RLV fulfils a sigma-like function in the in vitro transcription assay (Fig. 3, lanes 1-16; exposure time was 12 h). SIG1-RLV does not diminish the production of the nonspecific RNA (Fig. 3, asterisk) by the E. coli core enzyme, i.e. SIG1 has much less affinity to the E. coli core enzyme than  1-4), and 2 days (lanes 3Ј, 4Ј, and 5-14).
SIG2. However, activation of transcription at the -180 rbcL promoter by SIG1 is much stronger than that obtained by SIG2 (compare Fig. 2, lanes 1-4, with Fig. 3, lanes 13-16; note that all lanes were exposed for 12 h). SIG1 recognizes only the -180 rbcL promoter (Fig. 3, lanes 13-16, 12  Addition of SIG3 (GLR) to the in vitro transcription assays did not change the transcription pattern obtained with the E. coli core enzyme (not shown), suggesting that the truncated form of the factor does not interact with any of the three tested promoter regions or with the core RNA polymerase.

Analysis of the Promoter Specificity of Plant Plastid Sigmalike Factors by Complementation of E. coli 70 Mutants-To
study in vivo the function of the three plant sigma-like factors, we used two different E. coli strains harboring thermosensitive rpoD mutations, UQ285 (19) and CAG1 (20). In both cases, we obtained identical results. Fig. 4 shows the results obtained with the strain CAG1. To make the plant factors functionally comparable for its usage in the E. coli system, we fused the 3Ј ends of the plant cDNAs, coding for the regions 2-1 up to 4-2, to the 5Ј end of the primary E. coli 70 factor (Fig. 4A). These constructs result in the production of three hybrid sigma factors that are identical to the primary 70 factor of E. coli in their N-terminal part (region 1) but different in the C-terminal parts (regions 2-4). We found that of the three different hybrid factors (Fig. 4A), only one (pMA 55, NЈ-E. coli 70 -CЈ-plant SIG1; Fig. 4B) complements the E. coli rpoD mutant strain with the same efficiency as does the homologous E. coli 70 factor (pMRG1, Ref. 21), i.e. SIG1 represents a primary sigma factor when assayed in E. coli. As a control, we also attempted to complement the two E. coli mutations with the three fulllength plant sigma factors. All of these assays were negative, indicating that the plant-specific N-terminal part of SIG1 prevents the complementation. In Fig. 4B, only the result obtained with full-length SIG1 is shown (pMA 40).
SIG3 Might Contain an Inhibition Domain-The low, but obvious, complementation of the two E. coli rpoD mutants by SIG3 was surprising when compared with the complete absence of specific initiation that was observed by in vitro transcription. Therefore, we compared the amino acid sequences of two C-terminal parts that have been used for either in vivo complementation or for in vitro transcription in more detail. Sequence alignment with the prokaryotic sigma factors known thus far shows sequence similarity to the sigma K factor of Bacillus subtilis in a region of about 100 amino acids (Fig. 5). Sigma K is synthesized as a precursor protein with a 20-amino acid pro sequence. In the presence of this sequence, the factor is inactive, and its activation is brought about by proteolytic removal of the pro sequence (22). The two SIG3 constructs that have been used for in vitro transcription and in vivo complementation differ in the presence and absence, respectively, of the region that is similar to the inhibiting pro sequence (see Fig. 5). Therefore, low level of complementation could be related to the absence of this domain.
To test this hypothesis, we produced two additional N-terminal truncated proteins. One of them lacks only the supposed pro peptide (SIG3-ASL), and the other one lacks the entire N-terminal part that has amino acid sequence similarity to Sigma K (SIG3-HTR). We performed reconstitution assays of the E. coli core enzyme with the two different truncated forms of SIG3 using the same templates as already described. SIG3-ASL does not affect in vitro transcription (not shown). However, SIG3-HTR stimulates specific initiation at all three promoters, rrn-P1, rrn-P2, and rbcL (Fig. 6, lanes 4 -9; 12 h of exposure). Transcription of the cloned rDNA promoter produces many different transcripts. Therefore, we confirmed the P1-and P2-initiated RNAs by transcribing three different mutated promoter regions. These mutations have been previously constructed in our laboratory to analyze the regulation of plastid rDNA transcription by the transcription factor CDF2 (Ref. 4; Fig. 6, lanes 10 -17; 2 days of exposure). As already shown by transcription of these three templates using E. coli holoenzyme (4), the mutation in the -35 region of the P2 promoter diminishes initiation at P2 (Fig. 6, compare lanes 11 and 13). The mutation in the -10 region abolishes initiation at P2 (Fig. 6,  compare lanes 11 and 15), and the mutation in the CDF2 binding site abolishes initiation at the P1 promoter (Fig. 6,  compare lanes 11 and 17). Competition assays of SIG3-HTR with the E. coli holoenzyme confirm the recognition of all three promoters by SIG3-HTR (Fig. 6, lanes 18 -23; 2 h of exposure).
Activation of SIG3 Is Not Regulated by Proteolytic Cleavage of the Hypothetic Inhibition Domain-To analyze whether the activity of SIG3 might be regulated in vivo by proteolytic cleavage of the N-terminal region, we have prepared antibodies against the SIG3-HTR protein. These antibodies specifically recognize SIG3. They do not cross-react with SIG1 or with SIG2 (not shown). Western blot analysis of proteins isolated from Arabidopsis cotyledons revealed only one polypeptide of about 60 kDa, corresponding to the size of the full-length protein (Fig. 7). DISCUSSION Three different cDNAs coding for potential plastid-localized sigma 70 -type transcription initiation factors have recently been cloned and sequenced from A. thaliana (9,23), and the existence of a multigene family of sigma factors has been suggested for Zea mays (24). Meanwhile, the list of Arabidopsis sigma-  1-4, 9 -12, and 20 -22) or rbcL (lanes 5-8, 13-16, and 17-19) promoter regions have been transcribed by E. coli core enzyme (lanes 1-16) or E. coli holoenzyme (lanes 17-22). The transcription reactions were supplemented with increasing amounts of SIG1-RQR (lanes 2-4 and 6 -8) or SIG1-RLV ( lanes  10 -12, 14 -16, 18, 19, 21, and 22). The positions of the -180 rbcL promoter and of the rrn P1 and P2 promoters are indicated by arrows at the left and right. The asterisk indicates the position of an nonspecific transcript produced by the E. coli core enzyme from the rbcL plasmid. Exposure times were 12 h (lanes 1-16) and 4 h  (lanes 17-22). like factors has been extended to six (SIG1-SIG6; see accession numbers listed above). Considering the small size of the plastid genome compared with that of bacteria or cyanobacteria, the existence of several sigma-like factors in plastids is quite surprising. It suggests that these factors might play different roles in plastid gene expression.
The sigma 70 transcription factor family is well characterized in prokaryotes such as E. coli and B. subtilis. It is subdivided into two (8) or three (25) groups including primary and alternative sigma factors. In the present study, we analyzed whether the two-group system has been conserved in higher plant plastids, i.e. whether a functional specialization might explain the existence of several sigma-like factors in photosynthetically active plastids. We have used in vitro and in vivo approaches to study the evolutionary conservation of components of the transcriptional machinery between prokaryotes and higher plant plastids and to characterize the function(s) of three of the six different plant plastid sigma-like factors. Expression and complementation studies (Figs. 1 and 4) of fulllength and truncated plant plastid sigma polypeptides in E. coli indicate that the functions of the N-terminal parts of the plant sigma factors differ considerably from those of the corresponding portion of the principal sigma 70 factor of E. coli. High level production of the full-length protein in E. coli was only achieved for SIG2. However, SIG2 does not activate transcription in a heterologous system with E. coli core enzyme (Fig. 2). This might explain why it was possible to produce this protein in high amounts in E. coli.
To reveal and analyze specificity in promoter recognition of the three plant sigma-like factors, we used truncated polypeptides that harbor only the conserved regions 2-4. In this way, we avoid differences in promoter-RNA polymerase interactions that may arise from different functions of the different Nterminal sequences of these factors. In addition, it has been shown previously that the intact sigma 70 factor of E. coli does not bind to DNA but that the cleavage of the N-terminal part of the factor transforms it into a DNA-binding protein (16). Therefore, we supposed that it should be possible to reveal promoter specificity by using truncated sigma factors. We analyzed transcription factor-RNA polymerase-promoter interactions either by reconstitution of the truncated polypeptides with the E. coli core enzyme or by competition of the truncated polypeptides with the E. coli holoenzyme followed by in vitro transcription. These experiments clearly demonstrate differences in the recognition of three selected plastid promoters, rrn-P1, rrn-P2, and rbcL. SIG3 recognizes all three promoters, i.e. it is the least specific of the three regulatory proteins. SIG2 recognizes specifically the rrn-P1 promoter, and SIG1 recognizes only the rbcL promoter (Figs. 2, 3, and 6). The analysis of several truncated SIG3 proteins shows that SIG3 might contain an inhibition domain that is similar to the pro sequence of B. subtilis sigma K. However, we could not detect proteolytic cleavage of SIG3 in Arabidopsis. Nevertheless, our experiments suggest a function of this domain in the regulation of SIG3 activity. Instead of proteolytic cleavage, the activity of SIG3 might be regulated by posttranslational conformational changes that modify the accessibility of this putative inhibition domain.
If we compare the three plastid promoter structures with the consensus sequence recognized by the principal sigma 70 factor of E. coli, we find that the rbcL promoter has the highest similarity with the E. coli consensus (Fig. 1C). SIG1 recognizes specifically only the rbcL promoter, as analyzed by in vitro transcription, suggesting that SIG1 is the plant analogue to the primary sigma factor of E. coli. This hypothesis is further supported by the in vivo analysis of the hybrid sigma factors (Fig. 4). For this analysis, the N-terminal part of the E. coli primary 70 factor (region 1) was fused to the different Cterminal parts of the three plant sigma factors (regions 2-4). Region 1 had been shown to be important for open and ternary complex formation (26) and to induce conformational changes into the holoenzyme that are important for correct promoter recognition (27). Parts of regions 2 and 4 are important for the recognition of consensus -10 and -35 DNA sequences (for review, see Ref. 8). Thus, differences in promoter recognition and initiation of these hybrid sigma factors ought to be due to differences in the 3Ј plant-specific part of the constructs. Results show that of the three tested hybrid sigma factors, only SIG1 fully complements the E. coli thermosensitive rpoD mutants.
Our results indicate that the three higher plant plastid sigma 70 -like proteins have at least three characteristics in common with their prokaryotic ancestors: 1) they are composed of specific functional domains; 2) they distinguish between different promoter structures; and 3) only one of them (SIG1) is capable of recognizing all essential E. coli promoters that are recognized by the E. coli principal sigma 70 factor in vivo in E. coli. Therefore, we consider SIG1 to be the plant analogue to the primary sigma factor of E. coli. Interestingly, sequence alignment of all six Arabidopsis sigma factors to all five sigma factors localized on the Synechocystis genome (13) shows that SIG1 has the strongest sequence similarity and/or identity to all of the cyanobacterial sigma factors (Table I). Thus, SIG1 is the most prokaryotic-like plant sigma factor. The other plant sigma factors might have evolved in coordination with the transformation of a unicellular organism, the cyanobacterial ancestor, into an integrated part of a multicellular organism, the present-day plastid. The in vivo function of the SIG2specific rrn-P1 promoter is still unclear (4), 2 and the activity of SIG3 might be regulated by posttranslational modification(s). Therefore, our results suggest specific functions for SIG2 and SIG3 that are related either to plant development and/or changes of environmental conditions. Experiments are in progress in our laboratory to analyze the function of the three plant sigma factors during A. thaliana development using an antisense approach.    2, and 18 -23) supplemented with the truncated SIG-HTR protein (lanes 5, 6, 8, 9, 11, 13, 15, 17, 19, 20, 22, and 23). The positions of the -180 rbcL and of the rrn P1 and P2 promoters are indicated by arrows at the left and right, respectively. The asterisk shows the position of the nonspecific transcript that is produced by the E. coli core enzyme from the rbcL plasmid. -35, -10, and CDF2 correspond to the mutations 2, 3, and 1, respectively, as published by Iratni et al. (4). WT, wild type. Exposure times were 4 h (lanes 1-3), 12 h (lanes 4-9), 2 days (lanes 10 -17), and 2 h (lanes 18 -23).