Dual temporal Role of Plastid Sigma Factor 6 in Arabidopsis Development

Plants contain nuclear-coded sigma factors for initiation of chloroplast transcription. The in vivo function of individual members of the sigma gene family has become increasingly accessible by knockout and complementation strategies. Here we have investigated plastid gene expression in an Arabidopsis thaliana mutant with a defective gene for sigma factor 6. RNA gel blot hybridization and real-time RT-PCR together indicate that this factor has a dual developmental role, with both "early" and "persistent" (long-term) activities. The early role is evident from the sharp decrease of certain plastid transcripts only in young mutant seedlings. The second (persistent) role is reflected by the up- and down-regulation of other transcripts at the time of primary leaf formation and subsequent vegetative development. We conclude that sigma 6 does not represent a general factor but seems to have specialized roles in developmental stage- and gene-specific plastid transcription. The possibility that plastid DNA copy number might be responsible for the altered transcript patterns in mutant vs. wildtype was excluded by the results of DNA gel blot hybridization. Re-transformation of the knockout line with the full-length sigma 6 cDNA further established a causal relationship between the functional sigma gene and the resulting phenotype.

, including those that associate transiently such as the sigma factors (Tiller and Link, 1993a, b).
Sigma factors are the principal regulators of transcription initiation in bacteria (Gruber and Gross, 2003;Borukhov and Nudler, 2003), and -in view of phylogenetic relationships (Martin et al., 2005) -it has come as no surprise that chloroplasts contain such factor(s).
Unlike the core subunits of the bacterial-type plastid RNA polymerase, the organellar sigma proteins are encoded by nuclear genes -in higher plants usually as a small gene family (Allison, 2000;Toyoshima et al., 2005). Arabidopsis contains 6 of these genes, AtSig1 to 6, all of which are unlinked and most of them on different chromosomes (Isono et al., 1997;Tanaka et al., 1997;Fujiwara et al., 2000;Hakimi et al., 2000;The Arabidopsis Genome Initiative, 2000). A common feature of these (highly split) genes is their coding region for the conserved C-terminal half of the derived protein (Hakimi et al., 2000), which contains the typical regions 1.2 to 4.2 for basic sigma functions (Gruber and Gross, 2003). In contrast, the N-terminal half consisting of the short transit peptide followed by considerable extrasequence of variable length is unconserved, suggesting that the latter might contain important determinants for specific properties of individual factors e.g. in development and stress response (Kanamaru and Tanaka, 2004).
While basic sigma functions have been tested to a large part using heterologous in vitro systems with authentic plastid or bacterially expressed sigma proteins and E. coli RNA polymerase (Hakimi et al., 2000;Hanaoka et al., 2003;Homann and Link, 2003), the availability of Arabidopsis sigma mutant lines has facilitated functional studies in vivo (Hanaoka et al., 2003, Privat et al, 2003Nagashima et al., 2004;Tsunoyama et al., 2004;Favory et al., 2005;Ishizaki et al., 2005 ). Here we investigate an A. thaliana knockout line

Characterization of the Arabidopsis Sig6-2 Mutant Allele
We have chosen the knockout mutant line sig6-2 that had been generated in the GABI-Kat program at the Max-Planck-Institute fuer Zuechtungsforschung, Cologne (242G06; Rosso et al., 2003). Sequencing of T-DNA borders identified the insertion site within exon 5 of the genomic Sig6 sequence on chromosome II (At2g36990) 1,022 nt downstream of the ATG initiation start codon, which would correspond to a derived protein that lacks all functional domains for sigma factor activity (Fig. 1A). Genomic PCR and Southern blot analysis of selfed progeny lines together verified the selection of a stable homozygous mutant line with a single-copy T-DNA insertion (data not shown).
Using RT-PCR and gene-specific primers (see Materials and Methods), transcripts from individual members of the Arabidopsis sigma gene family were assessed. As shown in The phenotype of sig6-2 differs from that of wildtype in a developmental stagespecific way. Homozygous mutant seedlings develop normally-shaped cotyledons, yet with increasing chlorophyll deficiency. Younger stages until approximately 4 d after sowing have pale green cotyledons, which then become yellowish and finally white during the next six to eight days (Fig. 1C). In contrast, the primary leaves and subsequent rosette leaves are seemingly unaffected. Except for the remainder of the white cotyledons, mutant plants are green and have normal morphological appearance (Fig. 1C). They tend, however, to be slightly smaller in size than wildtype of the same age, which could be related to delayed germination and/or seedling development.

Plastid Transcript Patterns in Wildtype vs. Sig6-2
Transcript levels of representative chloroplast genes at different times in development were assessed by northern blot hybridization. RNA samples from cotyledons (seedlings 4 to 10 days after sowing) or rosette leaves (plants 21 and 28 days after sowing) were fractionated and hybridized with gene-specific RNA probes (Fig. 2)

as described in Materials and
Methods. The results were grouped according to the observed expression mode in the mutant, taking into account the classification of plastid genes based on their transcription by PEP vs.
NEP (Hajdukiewicz et al., 1997;Shiina et al., 2005). Data for genes with an expression mode of typical (PEP-dependent) class I genes are presented in panel A and those for (PEP and NEP-dependent) class II genes in B. In panel C, patterns are shown for accD (an exclusively NEP-dependent) class III gene and clpP, a class II gene with a similar expression mode as accD.
The 1.3 kb psbA transcript ( Fig. 2A, top row) is a prominent band in all wildtype lanes from 4 to 28 d after sowing. In the mutant, this transcript is dramatically downregulated at 4 d, i.e. the youngest seedling stage analyzed, and is then rapidly restored to wildtype levels.
A similar time-course was noticeable both for the 1.9 kb rbcL transcript (second row) and the 0.6 kb precursor of the intron-containing trnV(UAC) gene (third row), again with a decrease only in the 4 d mutant sample. In addition, the trnV probe revealed a high-molecular (3.0 kb) signal in the mutant but not in wildtype. Unlike all other transcripts in Fig. 2A, this RNA species was not detectable before the 8 d stage.
In contrast to the genes in Fig. 2A, those in Fig. 2B did not give rise to transcripts with an early decrease at the 4 d mutant stage. The dicistronic (Sugita and Sugiura, 1996) atpB/E transcript at 2.0 kb (first and second row), the monocistronic atpE mRNA (0.7 kb; second row), and the tricistronic ndhC/K/J transcript (1.8 kb; third row) all were visible at almost constant intensity over the entire time span from 4 to 28 d. On the other hand, differences between wildtype and mutant patterns were evident in Fig. 2B, which were not noticeable for psbA and rbcL ( Fig. 2A). These included the gradual weakening of the 2.6 kb atpB/E band beginning with the first (4 d) mutant stage until complete loss after day 10, and the transient appearance of a large 4.8 kb species (8 and 10 d lanes). The latter, mutantspecific, RNA spans the entire atpB/E coding region and ends in a short distance downstream but has considerable extra-sequence on the 5´ side (data not shown). The ndhC probe detected a 3.0 kb transcript at day 8 and later. Both the time-course and size are reminiscent of the large band detected by the trnV probe ( Fig. 2A, third row), suggesting that it is the same transcript spanning these two adjacent genes (Sato et al., 1999).
Finally, as shown in Fig. 2C (first row), the 2.5 kb transcript of the accD gene was present in the mutant in amounts that were equal to (21 and 28 d) or higher (4 to 10 d), but never lower than those in wildtype. A similar pattern was also observed for the 1.2 kb clpP transcript (second row). To assess steady-state transcript levels of selected (monocistronic) genes more rigorously, quantitative real time PCR experiments were carried out (Fig. 3). Again, psbA and rbcL were found to give decreased transcript levels in 4, but not 10, day-old sig6-2 seedlings (Figs. 3A and B). For psbA (Fig. 3, panel A), the down-regulation compared to wildtype exceeded a factor four at the 4 d-stage and was less than 0.5 at 10 d, while for rbcL (panel B) the factors were larger than two (4 d) vs. less than 0.5 (10 d). In contrast, real time RT-PCR showed greater than threefold up-regulation for the clpP transcript at day 4, and twofold at day 10 ( Fig. 3, panel C). This pattern is in agreement with that observed for accD and clpP in the northern blot experiments (Fig. 2C). Because of their multiple overlapping transcripts, the polycistronic transcription units studied in Fig. 2B (atpB-E, ndhC-K-J) were not investigated by real time quantification.

Plastid/nuclear DNA Ratio is not responsible for altered Expression Patterns in Sig6-2
As changes in copy number of plastid DNA might contribute to the altered RNA patterns in the mutant, we tested this possibility by Southern hybridization with both plastid and nuclear probes. Total DNA was prepared from wildtype (WT) and mutant (sig6-2), either at the 4 d seedling stage (Fig. 4A) or from 28 d plants (Fig. 4B). After digestion of equal amounts of DNA with HindIII, followed by gel-fractionation and hybridization, a single signal at 7.5 kb was generated with the plastid psbA probe (left panels), and a 9.0 kb band with the nuclear 18S rDNA probe (right panels). The WT and sig6-2 lanes always revealed bands of equal intensity, suggesting that DNA copy number was not responsible for the different psbA transcript levels in mutant vs. wildtype that were observed at 4 but not 28 d (Fig. 2).

Rescue of SIG6 Gene Function by Complementation
To further confirm that insertional inactivation of the AtSig6 gene in the sig6-2 mutant is directly responsible for its phenotype, complementation experiments using the full-length cDNA were carried out. Following reverse transcription and amplification, the cloned AtSig6 specifically detected the T-DNA insertion resulting from the secondary transformation, with a signal visible only in the complemented mutant (C). Using the sig6 probe (right panel), a single 5.3 kb band was generated in the wildtype (WT), whereas a 3.5 kb band was noticeable in both the knockout (MT) and re-transformed plants (C). The latter also showed two additional bands at approx. 6.0 kb and 4.5 kb. As these two bands were consistently observed under a variety of experimental conditions, they probably indicate the presence of an additional EcoRI site adjacent to the insertion rather than partial digestion (not shown). In any case, none of them is visible in the WT and MT lanes, suggesting that they mark a single secondary insertion at a unique site.
We next analyzed the gene expression patterns of the complemented line (C) in comparison with those from wildtype (WT) and the sig6-2 knockout (MT). As shown in Fig

DISCUSSION
In the present work, we have characterized a new AtSig6 mutant allele, sig6-2, both at DNA and RNA level as well as by complementation with the intact cDNA. This reverse genetics strategy established a causal link between the introduced gene and the visual and molecular phenotype of the rescued transformants, both of which resembled that of the wildtype (Fig.   5). PCR and Southern blot analysis together established the gene-specific (single) T-DNA insertion both in the knockout mutant and in the complemented line (Fig. 1). Using the same techniques, evidence was obtained that plastid DNA copy number does not seem to be a significant factor responsible for distinct plastid RNA patterns of wildtype vs. mutant (Fig.   4). Unlike most mutants described for other Arabidopsis sigma factors (reviewed by Shiina et al., 2005;Toyoshima et al., 2005), those for sigma 6 reveal a developmental stagespecific phenotype. This was first shown for the mutant allele sig6-1 (Ishizaki et al., 2005), which has a pale green (chlorophyll-deficient) phenotype in 3-to 4-d seedlings and then regreens to wildtype levels until day 8. In addition, plastid gene expression at RNA level was affected in that mutant only in young (4 d) but not older (8 d) seedlings, which led the authors to conclude that AtSIG6 might have a function restricted to early seedling development (Ishizaki et al., 2005). The sig6-2 mutant allele analyzed in our present work has an even stronger phenotype than sig6-1, with cotyledons that are pale-green at day three to four and then become yellowish and finally white (day 10 to 12).

RT-PCR
The transcript patterns (Figs. 2 and 3) of sig6-2 were in agreement with those obtained for sig6-1 in at least some cases. This is evident for transcripts of the class I genes psbA and rbcL, each of which showed a sharp decrease in steady-state concentration at day 4 but not day 8 in sig6-1 (Ishizaki et al., 2005). In the sig6-2 line studied here by northern blot hybridization (Fig. 2) and quantitative real-time RT-PCR (Fig. 3), the psbA and rbcL transcript levels were strongly reduced at the earliest time-point (4 d) and rapidly recovered to almost wildtype levels by day 10. Hence, from the data obtained with class I genes, both the sig6-1 and sig6-2 mutant alleles are defective in a SIG6 function that plays a stagespecific critical role in early seedling development. Similar conclusions can be reached if the transcripts of the clpP (class II) and accD (class III) genes (Fig. 2C) are considered, although in these cases increased, rather than decreased, levels were found in the mutant as compared to wildtype.
A notable difference, however, is evident from the trnV(UAC) transcript pattern ( Fig.   2A, third row), consisting of two RNA species with different time-course during development. The smaller (0.6 kb) band shows the early decrease (4 d) as was seen for the class I transcripts psbA and rbcL (first and second row). The large 3.0 kb signal is visible only in the mutant, and only later throughout day 8 to 28. Neither effect was previously described for trnV in sig6-1 (Ishizaki et al., 2005). The presence and differential time-course of these two RNAs thus distinguishes the two mutant alleles and, furthermore, points to a role of SIG6 not only in seedlings but also in rosette-stage plants.
This view is strengthened by the data obtained with the polycistronic ndhC transcription unit (Fig. 2B, third row), which also results in two RNAs of different timecourse. The smaller (1.8 kb) species appears to be present in relatively constant amounts without a decrease at the 4 d-seedling stage. The (mutant-specific) 3.0 kb RNA is first visible at day 8 and then remains at constant level, i.e. both its size and time-course match those of the large trnV transcript ( Fig. 2A, row 3). As trnV and ndhC are immediately adjacent (Sato et al., 1999), it is likely that the 3.0 kb RNA detected in both cases is identical.
The atpB-E operon (Fig. 2B, first and second row) gives rise to several transcripts, none of which shows an early decrease comparable to that of the class I RNAs ( Fig. 2A): (i) The major 2.0 kb (atpB-E) and the 0.7 kb (monocistronic atpE) RNAs were both present in roughly constant amounts throughout development. (ii) The 2.6 kb RNA species was visible both in wildtype and sig6-2 at 4 d but was absent in the mutant at all subsequent stages. (iii) The mutant-specific 4.8 kb species accumulated transiently between 4 and 10 d and then completely disappeared (Fig. 2B, first and second row).
Together, the data presented in Fig. 2 indicate an unexpected complexity of SIG6dependent responses in Arabidopsis development. The model depicted in Fig. 6 suggests a dual role consisting of both an "early" and "persistent" (long-term) activity of the factor. An early decrease was seen for the transcripts of class I genes ( Fig. 2A), but also for the 0.6 kb trnV transcript ("expression mode I"). The opposite effect, i.e. the early increase of the accD and clpP transcripts (Fig. 2C), may be functionally related, although it could be due to efficient NEP transcription (Allison et al., 1996;Legen et al., 2002) of these genes in this situation in the mutant ("expression mode III"). Perhaps most notable, none of the mature transcripts in Fig. 2B revealed an early effect, indicating that a different gene-specific mechanism might be involved ("expression mode II"). At this early time point, another sigma factor might be able to substitute for SIG6 in the transcription of the mode II genes (Fig. 2B), but less efficiently, if at all, in the transcription of the mode I genes ( Fig. 2A). Likewise, the loss of the 2.6 kb atpB-E transcripts in the mutant is consistent with a second (long-term) role of SIG6, implying that it cannot fully be replaced by other factor(s) during late seedling development and rosette leaf formation. It is notable that none of the (monocistronic) class I genes psbA and rbcL showed any "persistent" effect such as mutant-specific transcripts of distinguishable size ( Fig. 2A and data not shown). Together, this would mean that, at least during the developmental stages and at the genes (promoters) investigated here, SIG6 seems to act as a specialized rather than general factor.
The transient 4.8 kb RNA of the atpB/E region (Fig. 2) may be a consequence of the fact that both the early and persistent (long-term) functions of SIG6 are absent in the mutant.
If not generated by an alternative sigma factor and PEP, this mutant-specific RNA could be the result of NEP-dependent transcription. A similar mechanism, i.e. formation of a large (polycistronic) transcript by usage of a NEP promoter in the absence of SIG6, could explain the 3.0 kb trnV (and ndhC) transcript. Furthermore, it was previously established that trnV is a PEP-dependent gene preferentially transcribed in the presence of SIG2 (Kanamaru et al., 2001;Hanaoka et al., 2003;Privat et al., 2003). The early decrease of the 0.6 kb RNA at day 4 ( Fig. 2A) suggests that SIG6, in addition to SIG2, may have a -temporally restricted -role in the transcription of this tRNA gene.
A question that emerges relates to the mechanism(s) involved in the functional overlap of plastid sigma factors, throughout development or only at certain times (Kanamaru andTanaka, 2004: Shiina et al., 2005). From in vitro studies using purified authentic (Tiller and Link, 1993a, b) or recombinant sigma proteins (Homann and Link, 2003) it appears that the phosphorylation state of these factors might be a critical determinant in transcription initiation activity. The protein kinase responsible for sigma phosphorylation (Baginsky et al., 1997(Baginsky et al., , 1999 has been cloned and characterized (Ogrzewalla et al., 2002). This plastid transcription kinase (PTK), a known CK2-type enzyme also termed cpCK2 (Loschelder et al.,

RT-PCR Detection of Sigma Factor Transcripts
Total RNA (2 g) from 6 day-old Arabidopsis seedlings was mixed with random primers (10 pM; Promega, Madison, WI), incubated at 70°C for 10 min, and chilled on ice for 1 min. After addition of 6 l AMV-RT buffer (Promega), 1 l RNasin (40 U/ l; Promega), 3 l dNTPs (0.25 mM each), and 3 l AMV reverse transcriptase (10 U/ l) to a final volume of 30 l, the reaction was incubated at 37°C for 90 min. Following heating to 95°C for 10 min, the mixture was chilled on ice for 1 min. One l RNase A (10 g/ l, Sigma) was then added and incubation continued at 37°C for 15 min. The cDNAs corresponding to each Arabidopsis sigma factor were amplified using Taq DNA polymerase (Promega).

RNA Isolation, Northern Blot Analysis
Cotyledon or rosette leave samples (100 mg) were frozen in liquid nitrogen and ground to powder. RNA was isolated by the acid guanidinium-phenol-chloroform (AGPC) method (Chomczynski and Sacchi, 1987). Briefly, the powder was resuspended in 1.

Real-Time PCR
Real-time one-step RT-PCR was carried out using the QuantiTect SYBR Green RT-

Genomic Southern Blot Hybridization
Genomic DNA for Southern blot analysis was prepared from cotyledons and rosette leaves by using the CTAB (cetyltrimethylammonium bromide) procedure (Doyle and Doyle, 1987).
Two to five g of total genomic DNA were electrophoresed through an 0.7% (w/v) agarose gel and blotted to positively charged nylon membrane (Roche). The membrane was then hybridized with either DIG-labeled DNA or RNA probes at 42°C or 50°C, respectively, according to the Roche manual. The DNA probes were generated using the PCR DIG probe synthesis kit (Roche). RNA probes were obtained by cloning of PCR amplified regions in pGEM-T Easy (Promega), followed by in vitro transcription using the DIG RNA labeling hybridization was carried out with probes that selectively detected either a chloroplast (psbA) or nuclear gene region (18S rDNA). The primer pairs for amplification were NorpsbA1 (5´-TTACCCAATCTGGGAAGCTG-3´), NorpsbA2 (5´-GCCTCAACAGCAGC TAGGTC-3´) as well as AT-18S-1 (5´-AAACGGCTACCACATCCAAG-3´) and AT-18S-2 (5´-GTACAAAGGGCAGGGACGTA-3´). sig6-2 mutant. Total RNA was prepared from 6-d seedlings, reverse-transcribed, and cDNA was amplified using the gene-specific primer pairs as described in Materials and Methods.     Two distinct components of SIG6 activity (indicated by perpendicular bars separated by dashed lines) together determine its total activity (heavy-lined curve): "Early" role in young seedlings and "persistent" (long-term) role during subsequent development of seedlings and rosette-stage plants. The suggested early role is based on the observation that transcripts of "expression mode 1" such as those of psbA, rbcL, and the 0.6 kb trnV(UAC) transcript are strongly downregulated in young mutant seedlings at 4 days. Thereafter, they recover to almost wildtype amounts in 8-to 10-day mutant seedlings (Figure 2A). The persistent role relates to the continuous presence or absence of mutant-specific transcripts of "expression mode 2", including those from the atpB/E (4.8 and 2.6 kb) and trnV/ndhC region (3.0 kb) ( Figure 2B). The trailing edge of the solid curve is thought to indicate overlap of functions.
The region above the curve reflects (sigma-dependent) transcription activity mediated by SIG1 to 5, but also (sigma-independent) transcription by NEP.