Gene-specific trans-Regulatory Functions of Magnesium for Chloroplast mRNA Stability in Higher Plants*

In higher plant chloroplasts the accumulation of plastid-encoded mRNAs during leaf maturation is regulatedvia gene-specific mRNA stabilization. The half-lives of chloroplast RNAs are specifically affected by magnesium ions.psbA mRNA (D1 protein of photosystem II),rbcL mRNA (large subunit of ribulose-1,5-bisphosphate carboxylase), 16 S rRNA, and tRNAHis gain stability at specific magnesium concentrations in an in vitrodegradation system from spinach chloroplasts. Each RNA exhibits a typical magnesium concentration-dependent stabilization profile. It shows a cooperative response of the stability-regulatedpsbA mRNA and a saturation curve for the other RNAs. The concentration of free Mg2+ rises during chloroplast development within a range sufficient to mediate gene-specific mRNA stabilization in vivo as observed in vitro. We suggest that magnesium ions are a trans-acting factor mediating differential mRNA stability.

In higher plant chloroplasts the accumulation of plastid-encoded mRNAs during leaf maturation is regulated via gene-specific mRNA stabilization. The half-lives of chloroplast RNAs are specifically affected by magnesium ions. psbA mRNA (D1 protein of photosystem II), rbcL mRNA (large subunit of ribulose-1,5-bisphosphate carboxylase), 16 S rRNA, and tRNA His gain stability at specific magnesium concentrations in an in vitro degradation system from spinach chloroplasts. Each RNA exhibits a typical magnesium concentration-dependent stabilization profile. It shows a cooperative response of the stability-regulated psbA mRNA and a saturation curve for the other RNAs. The concentration of free Mg 2؉ rises during chloroplast development within a range sufficient to mediate gene-specific mRNA stabilization in vivo as observed in vitro. We suggest that magnesium ions are a trans-acting factor mediating differential mRNA stability.
The efficiency of gene expression depends on the stability of mRNAs by determining the pool of templates available for synthesis of the respective gene products. Gene regulation in many systems is accomplished by changing the stability of a certain message during the course of a developmental program or in response to environmental changes (1). In chloroplasts of higher plants differential mRNA stability is responsible for controlling mRNA accumulation during chloroplast maturation (2,3). In spinach as well as in barley it has been shown that during leaf development mRNAs are stabilized in a gene-specific manner (4,5). The mRNA that is stabilized to the highest extent encodes the D1 reaction center protein of photosystem II (psbA).
In recent years considerable progress has been made in elucidating the mRNA degradation mechanism in chloroplasts (6,7). In higher plants plastid mRNA degradation is initiated by endonucleolytic cleavages (8). The resulting proximal fragments are polyadenylated as a tag for rapid exonucleolytic decay (9,10). This mechanism implies that initiation of mRNA decay is the crucial process by which the stability of a certain message is regulated. The accessibility of the primary cleavage sites for endonucleases determines the proportion of molecules to be degraded. However, only limited information is available so far about the molecular mechanism mediating the regulation of degradation initiation, the cis-regulatory elements and trans-regulatory factors involved. Up to now, only a few cisregulatory elements of mRNAs have been described using genetic approaches in higher plants and in the green alga Chlamydomonas reinhardtii. In Chlamydomonas several nuclear mutants have been isolated that affect the stability of a variety of chloroplast-encoded mRNAs; those mutations each interfere with the accumulation of a single defined chloroplast mRNA (11). In the nuclear mutation nac2-26, for example, the stability of the chloroplast psbD mRNA is dramatically decreased (12). Chloroplast transformation with constructs of the psbD leader fused to a reporter gene showed a destabilized chimeric transcript in the mutant background and normal accumulation in the wild type, indicating that the 74-nucleotide leader of the mRNA includes a determinant for psbD mRNA degradation (13,14). Chloroplast lysates from wild type and mutant cells as an in vitro degradation system for the analysis of synthetic RNA transcripts reflect the observations made in vivo. The primary cleavage sites could be detected on these RNA transcripts. cis-Regulatory elements for the degradation of rbcL mRNA in Chlamydomonas have also been analyzed (15). The 63 nucleotides of the rbcL 5Ј leader fused to the Escherichia coli ␤-glucuronidase gene (gus) as a reporter confer instability to the chimeric transcripts in the light. The addition of the 257 nucleotides from the adjacent coding region prevented this destabilization. In Chlamydomonas the role of the 5Ј untranslated region of the petD mRNA for RNA stability and translation was studied by extensive mutational analysis. It was demonstrated that sequences essential for translation, as well as sequences that directly or indirectly affect RNA stability, reside within the 5Ј untranslated region of the petD mRNA. The finding that in all mutants where translation was compromised petD mRNA accumulated to a lower level than in wild type strains indicated that mRNA stability may be linked to translatability (16).
In chloroplasts of higher plants a correlation of polysome association and mRNA stability was suggested for rbcL mRNA from studies of nuclear mutants in maize. Mutants in which many chloroplast mRNAs are associated with abnormally few ribosomes showed that the level of rbcL mRNA was reduced 4-fold, indicating that the rbcL mRNA is destabilized as a consequence of its decreased polysome association (17). Further indications that cis-regulatory elements in higher plant chloroplast mRNAs are involved in RNA accumulation come from studies with transformed chloroplasts of tobacco. Fusions of different 3Ј untranslated regions to the E. coli ␤-glucuronidase gene, which in these constructs is preceded by the psbA promoter and the psbA 5Ј untranslated region, suggested that the accumulation of the chimeric RNA is not changed dramatically by the different 3Ј ends but is more likely influenced by other elements of the RNA (18). Direct evidence for cis-regulatory functions of the 5Ј region of higher plant chloroplast mRNAs comes from transplastomic plants carrying fusions of the gus gene with rbcL promotor/leader fragments. gus mRNA accumulation is independent of light as long as a certain element of the 5Ј untranslated region is included in the construct. Lower rates of rbcL transcription in the dark were compensated by increased mRNA stability (19). In tobacco, the 5Ј untranslated region of psbA mRNA alone seems not to be sufficient to confer the stability of the intact mRNA to a reporter fusion construct (20).
Most chloroplast mRNAs are flanked by a stem-loop structure in their 3Ј untranslated region that participates in the processing of the mature 3Ј end (3). In addition, these elements are important for impeding the progress of processive exoribonucleases (21). In transformed Chlamydomonas chloroplasts in vivo, it has been shown that partial or complete deletions of the stem-loop of the atpB gene leads to a decrease in mRNA accumulation, whereas the transcription rate of this gene remains unaffected (22). The stem-loop structure can be replaced in vivo by a sequence of 18 guanosines, which also serves as a barrier for a 3Ј35Ј exonuclease in vitro. Strains containing the polyguanosine tract instead of the stem-loop structure within the 3Ј untranslated region accumulate nearly wild type levels of atpB transcripts and the ATPase ␤-subunit protein (23).
The stem-loop structures of the 3Ј untranslated regions are known to bind chloroplast proteins. The petD 3Ј untranslated region forms a complex with 55-, 41-, and 29-kDa RNA-binding proteins. An 8-nucleotide AU-rich sequence motif downstream of the stem-loop, termed box II, appears to be essential for RNA-protein complex formation in vitro (24). In addition, the stem-loop itself is necessary for protein binding. The AU-rich box is also recognized by a 57-kDa protein, which possibly forms a stable complex together with a 33-kDa protein (25). These proteins may either be involved in mRNA processing or mediate stabilization against degradation. Beyond those proteins, most RNA-binding proteins that so far have been isolated, cloned, and characterized from chloroplasts of higher plants cannot be assigned to specific mRNAs but exhibit general functions. The 28-kDa ribonucleoprotein is involved in 3Ј end processing of several mRNAs (26); the 100-kDa ribonucleoprotein is the exonuclease polynucleotide phosphorylase (27); the ribosomal protein S1 (28) and the nuclease CSP41, which had previously been identified as a 3Ј end binding protein (29), could also be detected as part of a complex binding to the 5Ј untranslated region of psbA mRNA. 1 In this work we provide data showing that, in addition to proteins, the divalent cation Mg 2ϩ (as a non-proteinaceous factor) is required not only for the formation of chemically stable and functional RNA in a general manner but also for gene-specific differential stabilization of chloroplast RNAs. Furthermore, we show that the concentration of free magnesium ions rises during chloroplast development within a range sufficient to confer gene-specific mRNA stability regulation. Preparation of in Vitro Degradation Extracts-Intact chloroplasts were isolated according to Ref. 33. For preparation of in vitro degradation extracts, chloroplasts were resuspended in 20 mM HEPES, pH 7.9, 60 mM KCl, 20 mM EDTA, 2 mM dithiothreitol, and 20% (v/v) glycerol and lysed by 10 -15 strokes using a potter with pestle S. Extracts were adjusted to approximately 5 mg/ml protein as determined by Bradford assay. The extracts were stored at Ϫ70°C after freezing in liquid nitrogen. Chlorophyll concentrations were determined spectrophotometrically (34).
In Vitro Degradation Assay-In vitro degradation experiments were performed with 100 l of degradation extract (5 mg/ml protein) per time point. The extract was thawed on ice, and the mixture was transferred to 25°C for incubation. Reactions were stopped by adding 50 l of 6.0 M urea, 1.0% SDS, and 200 l of phenol/chloroform (8). After phenol/ chloroform extraction and a subsequent chloroform extraction, nucleic acids were recovered by ethanol precipitation. Concentrations of nucleic acids were determined spectrophotometrically at A 260 .
Northern Analysis-For Northern analysis 2 g of chloroplast RNA per lane were separated on 1.2% agarose-formaldehyde gels according to Ref. 32. Hybridization conditions were as published (4) except that oligonucleotide probes were used. Hybridization temperatures were as follows: psbA-3, 60°C; rbcL-1/rbcL-2 mixture, 50°C; 16 S, 55°C; tRNA His , 50°C. Filters were washed three times for 30 min at the respective hybridization temperatures in 5ϫ SSC, pH 7.0, 0.1% SDS. Filters were exposed to Kodak X-AR x-ray films. An excess of the probe was always examined by a dilution series of total spinach RNA on each blot to ensure the quantitative Northern analysis.
High Resolution Northern Analysis-For high resolution Northern analysis 2 g of chloroplast RNA per lane were separated on denaturating 5% sequencing polyacrylamide gels containing 8 M urea, 0.5ϫ TBE (10ϫ: 89 mM Tris, 89 mM boric acid, 1 mM EDTA). Gels were pre-run at 100 W for 20 min to allow heating and run at 100 W for 60 min. RNA samples were dissolved in 90% formamide containing 0.05% bromphenol blue, 0.05% xylenecyanole, and 1 mM EDTA, heated to 85°C for 5 min, and chilled on ice prior to loading onto the gel. Gels were transferred to a Biodyne A nylon membrane (Pall Europe Limited, Portsmouth, United Kingdom) in 5ϫ SSC overnight using the setup for agarose gels. The membrane was UV-treated to covalently couple the RNA (120 mJ, Stratalinker; Stratagene GmbH, Heidelberg, Germany). The hybridization procedure was performed as described above. The probe for high resolution Northern analysis was the psbA-1/psbA-2 mixture, which was hybridized at 40°C.
Electropotentiometric Determination of Magnesium Concentration-Intact chloroplasts were isolated as described above, except using buffers devoid of magnesium and EDTA. Lysis was performed by resuspension in 20 mM HEPES, 60 mM KCl, 15 mM KOH (to pH 7.9), and 20% (v/v) glycerol and 10 strokes each in a potter with first a light pistil and then a tight fitting pistil. The stromal fraction was isolated as the supernatant of a 30-min centrifugation at 20,000 rpm (Beckman JA20.1 rotor). The chloroplast stroma was filtrated through Centricon TM tubes (Amicon/Millipore, Bedford, MA) with an exclusion mass of 30 kDa. The resulting solution was directly subjected to Mg 2ϩ determination. The concentration of free Mg 2ϩ was measured using the magnesium-selective macroelectrode ETH 7025 as described (35). Briefly, the magnesium electrode was calibrated in the "background buffer" used for chloroplast extract preparation (20 mM HEPES, 60 mM KCl, 15 mM KOH (to pH 7.9), 20% (v/v) glycerol) followed immediately by measuring up to 11 extract samples and a subsequent additional calibration to assay for drift. The calibration curves were linear from 20 -0.15 mM Mg 2ϩ , with regression coefficients Ͼ0.98.
Determination of Chloroplast Volumes-Isolated intact chloroplasts were analyzed for their dimensions using a Zeiss Photomikroskop III. Photographic prints were scanned, and the images were evaluated using the ScionImage software (Scion Corp., obtained as freeware). The length l and width w of chloroplasts were determined, and the volume V was approximated by a rotational ellipsoid.
Quantification of Northern Blots and Determination of Hill Coefficients-X-ray films were quantified by densitometric analysis using a Hewlett Packard ScanJet 4c/T scanner. Scans were evaluated as TIFF images using the ScionImage software. Linearity of the scanned hybridization signal was ensured by dilution series of total RNA coprocessed with each quantitative Northern analysis. Relative RNA stability for magnesium-dependent stability profiles was determined by normalizing the data points obtained after 180 min of incubation to the "0" time point. To plot the relative RNA stability against concentrations of free Mg 2ϩ , the data were fitted using a modification of the Hill algorithm for binding of small ligands to a macromolecule (36). Here the mathematics was applied to describe the stabilizing effect of magnesium ions, i.e. binding is replaced by the stabilizing effect in the analysis. The original Hill function, where f ϭ fraction of binding sites bound, n ϭ Hill coefficient (n Ͼ 1 for cooperative binding), [L] ϭ free ligand concentration, and K ϭ apparent dissociation constant for interacting sites, is therefore modified to fit the stabilization profiles, where S ϭ RNA stability in relative units as a function of magnesium concentration, S 0 ϭ stability at 20 mM EDTA (no magnesium added to correct for the offset), S max ϭ maximal RNA stability, n ϭ Hill coefficient, and K ϭ [Mg 2ϩ ] of half-maximum RNA stability. Fits were performed using the Origin software (Microcal Corp.).
In Vitro RNA Synthesis-The rbcL mRNA 5Ј untranslated region had been cloned as a 216-base pair PCR fragment containing the T7 promotor fused to a BglII restriction site, and the rbcL 5Ј untranslated region (Ϫ180 -6, Ref. 30) was cloned into BamHI/HincII sites of pUC18. The insert sequence was verified by sequencing. Radiolabeled RNA transcripts were synthesized using the T7 in vitro transcription system (37).
UV Cross-linking Analysis-For analysis of protein binding, label transfer experiments were performed according to Ref. 38. 12.5% polyacrylamide/SDS gels were used according to Ref. 39. The gels were stained with silver nitrate (40), dried, and exposed to Kodak XAR x-ray films.

Magnesium Ions Stabilize Chloroplast
RNAs-To study chloroplast mRNA degradation we recently established an in vitro degradation system that faithfully reflects mRNA degradation in terms of cleavages and processing of degradation fragments after cleavage (8,9,10). This system consists of isolated, lysed chloroplasts and allows the observation of the internal mRNA that is complexed with proteins as in the native state as well as the analysis of additionally added transcripts. Lysis of the chloroplasts enables us to vary degradation conditions externally by variation of the buffer constituents. Earlier work already showed the effect of magnesium ions on RNA stability, an effect that is destabilizing in chloroplast extracts from Chlamydomonas (13). In degradation extracts of spinach chloroplasts magnesium ions have the opposite effect. They induce stabilization of internal RNAs. In the experiment shown in Fig.  1 a degradation extract was prepared in the presence of 20 mM EDTA to complex the endogenous magnesium ions of the chloroplast. The Mg 2ϩ content was reconstituted by adding MgCl 2 to a concentration of 25 mM. After incubation of the extract at room temperature for different periods of time as indicated in Fig. 1, total RNA was prepared and subjected to Northern analysis using gene-specific probes. As a control for ionic strength effects NaCl was added to another series of degradation experiments to a final concentration of 50 mM. We analyzed two mRNAs, psbA and the rbcL mRNA (psbA: D1 reaction center protein of photosystem II; rbcL: large subunit of the ribulose-1,5-bisphosphate carboxylase). In spinach, psbA mRNA is stabilized during leaf development, whereas no dra-matic changes can be observed in the stability of rbcL mRNA (4). Two structural RNAs, 16 S ribosomal RNA and tRNA His , which are supposed to be stable in all stages of chloroplast development, were studied. As shown in Fig. 1 all RNAs analyzed are stabilized by the addition of Mg 2ϩ , whereas the addition of NaCl has no effect. The finding that NaCl has no stabilizing effect on chloroplast RNAs in vitro indicates that not merely an increase in ionic strength is involved here. Still the stabilizing effect of Mg 2ϩ may be general for all RNAs.
To approach the question of specificity, titration experiments were performed to analyze the stabilization profiles as well as the "midpoint stabilizing" Mg 2ϩ concentration of different RNAs. Chloroplast extract was incubated at room temperature for 180 min in the presence of increasing concentrations of magnesium ions as indicated in Fig. 2 and compared with extract representing the time point 0. Total RNA was prepared from each sample and subjected to quantitative Northern analysis using the same gene-specific probes as described above. The results depicted in Fig. 2 show that magnesium ions have specific effects on the RNAs analyzed. Ribosomal 16 S rRNA and tRNA His are already stabilized at low added Mg 2ϩ concentrations (2.5-5 mM), whereas psbA mRNA requires concentrations of 10 -15 mM Mg 2ϩ added to the 20 mM EDTA extract to gain stability. rbcL mRNA shows an intermediate characteristic, becoming stabilized at concentrations of 5-10 mM added magnesium ions. Besides the specific concentrations of magnesium ions that result in stabilization of RNAs, the stabilization profiles as a function of Mg 2ϩ seem to be different.
Magnesium-dependent Stabilization Profiles Are Specific for Different RNAs-To gain a more detailed insight into those stabilization profiles that have been described in a rough qualitative manner above, the data were quantitated. However, before a quantitative evaluation could be made, the concentrations of free magnesium ions at each point of the titration experiment had to be determined. This was facilitated using an electropotentiometric procedure with magnesium-specific electrodes (35). The advantage of this method compared with atom absorption spectroscopy is that only the concentration of free ionized magnesium is determined, whereas total magnesium would be measured by atom absorption spectroscopy. In chloroplasts, total concentrations of magnesium ions as high as 20 -30 mM were determined earlier (41). However, it is very likely that only a portion of these ions is present as free ionized Mg 2ϩ and thereby is available for interactions with RNA. To perform the determination of free Mg 2ϩ , chloroplast degradation extracts had to be cleared from the membrane fraction by centrifugation, and the resulting supernatant had to be filtered through a Centricon TM tube with a size exclusion mass of 30 kDa. Otherwise, certain components of the extracts induced a drift at the electrode that did not allow for precise measurements. In Fig. 3A the calibration curve is shown in which measured voltage is plotted against the negative logarithm of To analyze the magnesium-dependent stabilization profiles of the different RNAs, x-ray films were densitometrically quantified. RNA amounts were expressed as percent of RNA remaining after 180 min compared with the amount of RNA at time 0 set to 100%. The percentage of remaining RNA is plotted against the concentration of free [Mg 2ϩ ] as shown in Fig. 4. The qualitative observations made before are very much supported by the quantitative evaluations. tRNA His is stabilized at low [Mg 2ϩ ] free , showing a very steep saturation curve. The form of the curve, with its steep slope at low [Mg 2ϩ ] free , reflects the known high affinity of tRNA to magnesium ions (42). The ribosomal 16 S RNA also exhibits a steep slope, although it is not as pronounced as the one for tRNA. The analyzed mRNAs deviate from each other in their stabilization characteristics. Whereas rbcL mRNA also shows the course of a saturation curve, psbA mRNA stabilization follows a sigmoidal stabilization profile, indicating the cooperative action of several magnesium ions. The addition of Mg 2ϩ has no effect on mRNA stabilization up to 4 mM [Mg 2ϩ ] free ; maximum stabilization is reached at a concentration of 7-8 mM [Mg 2ϩ ] free , with a midpoint of 6 mM resulting in 50% stabilization. At 6 mM [Mg 2ϩ ] free tRNA His has already reached maximum stabilization. This is not the case for rbcL mRNA and 16 S rRNA, which do not show a steep correlation of stability with [Mg 2ϩ ] free .
To analyze the degree of cooperativity involved in stabilizing  4. Quantitation of magnesium-dependent RNA stability profiles. X-ray films obtained in the experiment described in Fig. 2 were quantified by densitometric analysis. Relative RNA stability for magnesium-dependent stability profiles was determined by normalizing the data points obtained after 180 min of incubation to the 0 time point. Relative mRNA stability is plotted against [Mg 2ϩ ] free , as determined in Fig. 3. The data were fitted using a modification of the Hill algorithm for binding of small ligands to a macromolecule (36) in which binding was replaced by relative RNA stability. The Hill function modified to fit the stabilization profiles is shown in Equation 3. chloroplasts by [Mg 2ϩ ] free , we applied the mathematical Hill model originally used for the description of ligand binding. Here, binding is replaced by the biological effect, i.e. mRNA stabilization. Table I summarizes the data. psbA mRNA requires 7 (Ϯ 3) molecules of Mg 2ϩ per mRNA molecule interacting in a cooperative manner to mediate stabilization in the extract. In contrast, the independent effect of Mg 2ϩ ions mediates stabilization in the case of rbcL mRNA and 16 S ribosomal RNA and tRNA His , as represented by Hill coefficients of 1.4 (Ϯ 0.7) and 0.4 (Ϯ 0.4), respectively.
The data presented above show that Mg 2ϩ is required for the formation of stable chloroplast RNAs in vitro. Each RNA analyzed so far has a typical concentration of [Mg 2ϩ ] free to gain stability. In contrast to tRNA His , 16 S ribosomal RNA, and rbcL mRNA, psbA mRNA exhibits a cooperative stabilization profile, whereas the other RNAs gain stability by the independent action of magnesium ions.
Cleavage Sites on the psbA mRNA Are Protected by Magnesium Ions-To gain information on the mechanism of RNA stabilization by Mg 2ϩ , we analyzed the degradation products of psbA mRNA that are formed during the decay process in the absence and the presence of free Mg 2ϩ . Fig. 5 shows the high resolution Northern analysis of psbA degradation intermediates that appear during in vitro degradation kinetics using oligonucleotides complementary to the 5Ј untranslated region. Some fragments can already be detected at the 0 time point, indicating that they are already present at measurable steady state levels in intact chloroplasts. In the absence of free Mg 2ϩ two effects can be observed. (i) Various additional degradation fragments accumulate compared with a decay course in the presence of magnesium, and (ii) the intensity of signals derived from degradation intermediates that are present in both experiments is significantly lower in the presence of free Mg 2ϩ ions, except for signals that are already detectable at 0 min. These data indicate that the accessibility of the cleavage sites is affected by the presence of magnesium ions. Some of the cleavage sites are fully protected under the experimental conditions employed; the others are less active. By inducing protection of the psbA mRNA molecule against endonucleolytic cleavage, magnesium ions play a central role in the formation of stable mRNA, which has half-lives in vivo up to more than 10 h in spinach (4) and 40 h in barley (5).
Biological Relevance of Magnesium-mediated RNA Stabilization-To analyze the biological relevance of the Mg 2ϩ -mediated RNA stabilization in chloroplasts, the stromal concentration of [Mg 2ϩ ] free was determined. Chloroplasts were fractionated (43) using buffers devoid of Mg 2ϩ and EDTA. The resulting stromal fraction was filtered through Centricon TM tubes with an exclusion size of 30 kDa to separate it from substances interfering with the potentiometric Mg 2ϩ determination. The stromal fraction was normalized so that 1 ml represented the stromal volume of 8.5 ϫ 10 8 chloroplasts. To elucidate developmental changes of [Mg 2ϩ ] free , chloroplasts from 10-day-old seedlings grown under constant light/dark conditions were analyzed and compared with those isolated from etiolated seedlings (7 days) after 12 h of light treatment. Table   II summarizes the results. The measurements show a significant increase of [Mg 2ϩ ] free comparing data obtained from chloroplasts illuminated for 12 h and chloroplasts grown during a constant light/dark cycle. Microscopic analysis shows that chloroplasts illuminated for 12 h already have the same volume as those from light-grown leaves (cf. Table II). The same observation was made earlier in barley (44). Concentrations of free magnesium ions in the stromal space were determined by measuring the concentration of [Mg 2ϩ ] free in the normalized stromal extract. Together with the spectroscopic determination of chlorophyll concentration, the concentration of free magnesium ions in the stromal fraction of intact chloroplasts was estimated using correlations made earlier for mature chloroplasts of 23 l of stroma per mg of chlorophyll (45). We extrapolated from mature chloroplasts to estimate the magnesium concentration in chloroplasts illuminated for 12 h assuming the same volume of stroma, which probably results in a slight overestimation of [Mg 2ϩ ] free because the space percentage of thylakoid membranes and lumen is smaller than in mature chloroplasts. We observe a 2-to 3-fold increase of [Mg 2ϩ ] free during leaf development from 3-4 to 8 -10 mM, as summarized in Table II. These concentrations convincingly correspond to the RNA stabilization profiles. They allow us to transfer the magnesium-dependent RNA stability profiles to the situation in vivo.
Protein Binding to the mRNA Untranslated Region Is Affected by Magnesium Ions-To elucidate whether magnesium ions interfere with protein binding to chloroplast mRNAs, we analyzed the cross-linking pattern of proteins from lysed chloroplasts prepared in the presence of 20 mM EDTA reconstituted with increasing concentrations of Mg 2ϩ . Fig. 6 shows the results for the rbcL 5Ј untranslated region, which are the same as  5. Magnesium-dependent changes in psbA mRNA degradation intermediates. In vitro degradation of internal mRNA was performed in extracts of lysed chloroplasts (5 mg/ml protein) from mature leaves at 22°C prepared in the presence of 20 mM EDTA and after reconstitution with 25 mM MgCl 2 . After the indicated times RNA was purified, and 2 g per sample was analyzed on 5% polyacrylamide sequencing gels containing 8 M urea, 0.5ϫ TBE run at 55°C. The RNA was transferred to a nylon membrane and hybridized using a 32 Pradiolabeled oligonucleotide complementary to the 5Ј untranslated region of psbA mRNA. Sizes labeled in the figure are derived from subsequent hybridizations of the DNA marker pBR322/HinfI. the results obtained for the psbA 5Ј untranslated region (data not shown). In the presence of 20 mM EDTA an ensemble of several proteins can be detected to bind the chloroplast mRNA; the major species have molecular masses of 120, 80, 41, 43, 45, 28/29, and 24 kDa. This cross-linking pattern remains similar after the addition of 2.5 and 5 mM MgCl 2 , which result only in micromolar changes in concentration of [Mg 2ϩ ] free , as determined from the measurement of [Mg 2ϩ ] free in chloroplast extracts (cf. Fig. 3B). With the addition of 7.5 mM MgCl 2 the cross-linking pattern changes. Whereas binding to the 120-kDa species is only barely detectable, interaction with a 100-kDa species, which is very likely to be the chloroplast polynucleotide phosphorylase (PNPase) 2 homolog, can be observed. The weak binding to the 80-kDa protein is replaced by binding to a 67-kDa peptide that was described earlier to cross-react immunologically with E. coli RNase E antibodies (27). Both PNPase and RNase E seem to require Mg 2ϩ for their binding to chloroplast mRNAs. The PNPase shows maximum binding between 20 and 22.5 mM added Mg 2ϩ , which corresponds to ϳ10 mM [Mg 2ϩ ] free (cf. Fig. 3B). Within the family of RNA-binding proteins with molecular masses of 41-45 kDa, a change in intensity between the 45-and 43-kDa proteins is observed upon adding 7.5 mM MgCl 2 . Whereas binding to the 41-kDa species can only be detected in the presence of 20 mM EDTA, the affinity of the 43-kDa protein seems to be reduced in favor of binding to the 45-kDa species. We identified the 45-kDa protein as CSP41, 3 which is an endonuclease described earlier as being involved in chloroplast mRNA 3Ј end-processing (29). Binding affinity of the CSP41 also is reduced with increasing concentrations of Mg 2ϩ , whereas the 43-kDa species we identified as the ribosomal protein S1 (28) is not affected in the range of 7.5-22.5 mM added Mg 2ϩ . The low molecular mass species, 28/29 and 24 kDa, show an ionic strength dependence of binding that results in a gradual loss in binding activity. These results show that the binding pattern of proteins to chloroplast mRNAs is dependent on the concentration of MgCl 2 , but we could not detect gene-specific differences.

Stabilization of Chloroplast RNAs by Magnesium
Ions-Divalent cations like magnesium are a central component for the structural integrity and biological activity of RNA. This has been extensively studied for catalytic RNAs like the Tetrahymena group I intron, the hammerhead ribozyme, the hepatitis delta ribozyme, and the M1 RNA of RNase P (42,46,47). In this work we show that magnesium ions also affect the physical half-lives of RNAs in a gene-specific fashion. In spinach chlo-roplasts, magnesium specifically stabilizes RNAs. Each of the RNAs we analyzed, 16 S ribosomal RNA, tRNA His , and psbA and rbcL mRNA, requires a specific concentration of Mg 2ϩ to adopt a stable form. In addition, the stabilization profile of each of the RNAs shows a typical feature indicating the specificity of the Mg 2ϩ effect. In the case of tRNA His , very low concentrations of free magnesium ions already are sufficient to confer stability to the RNA, as indicated by the steep slope of the curve. This low concentration of required magnesium ions probably reflects the high affinity of tRNA to Mg 2ϩ . For example, the binding constant of Mg 2ϩ to the anticodon loop of yeast tRNA Phe has been determined as 2 ϫ 10 3 M Ϫ1 at pH 7.1 and 0.1 M sodium concentration (48,49). tRNA has between three and six Mg 2ϩ binding sites depending on the tRNA species, which are independent. This is also reflected by the observed stabilization profile, which follows a saturation curve. The

in chloroplasts
Chloroplasts were isolated from dark-grown seedlings that had been illuminated for 12 h (young chloroplasts) and from seedlings grown for 10 days in a 12-h light/dark cycle (mature chloroplasts). Sizes and magnesium concentrations were determined as described under "Experimental Procedures"; n ϭ number of chloroplasts analyzed for volume determination. A, B, C, and D represent independent preparations of chloroplasts. For mature chloroplasts the stromal space was approximated with 23 l/mg chlorophyll (45). Chlorophyll concentrations of mature chloroplasts were as follows: A, 3.0 mg/ml; B, 3.3 mg/ml, C, 3.3 mg/ml; D, 3.2 mg/ml. The stromal space of young chloroplasts was extrapolated based on the total volumes determined. tions analyzed (50). The two additional cross-links affected by magnesium are located in the decoding region of the 30 S subunit and are present only at 3 mM and higher concentrations. Thus, in E. coli ribosomes there are regions of greater conformational freedom and responsiveness to magnesium than in the rest of the subunit. Provided that chloroplast ribosomes resemble those of E. coli, the magnesium-mediated stabilization of 16 S RNA may result from those accessible regions of the RNA whose conformation is influenced by interaction with magnesium. A direct effect of mono-or divalent cations on mRNA halflives has not yet been described. The destabilizing effect of magnesium in the Chlamydomonas in vitro degradation system (13) indicates that in this system Mg 2ϩ ions act as an activating cofactor for the ribonuclease. In the spinach system it is likely that Mg 2ϩ ions influence the degradation substrate. Earlier work reports a positive effect of Mg 2ϩ ions on the stability of a petD 3Ј end precursor RNA in in vitro transcripts added to a processing extract (51). In our work we showed that different chloroplast mRNAs respond to magnesium in a specific manner. The stability-regulated psbA mRNA follows a sigmoidal magnesium-dependent stabilization profile, whereas rbcL mRNA, which is stability-regulated only to a minor extent, shows a stabilization profile similar to that of tRNA and 16 S RNA but with a flat slope. The determination of Hill coefficients reveals that 7 (Ϯ3) magnesium ions have to act cooperatively to confer mRNA stability to psbA mRNA, in contrast to rbcL mRNA, which requires the independent effect of magnesium ions. The strong cooperativity of the magnesium effect on psbA mRNA stability allows a sensitive switch-like response to changes of the stromal free magnesium concentration within far less than an order of magnitude. A similar role of calcium ions for the regulation of mRNA stability is rather unlikely because the level of free Ca 2ϩ in the stromal space of chloroplasts has been determined in vivo to be less than 150 nM (52). Binding constants of Ca 2ϩ to RNA have been determined to be in the millimolar range (53), which is four orders of magnitude higher than [Ca 2ϩ ] free . Therefore, a direct interaction of Ca 2ϩ with the chloroplast RNAs is very improbable.
The finding that the pattern and concentrations of degradation fragments are dependent on the presence of magnesium as shown for the psbA mRNA indicates that the accessibility of cleavage sites is changed by magnesium ions. There are several possibilities to accomplish this type of interference. (i) The interaction of magnesium ions is a direct interaction with the RNA and induces a conformation that is resistant to chloroplast endonucleases. Chloroplast nucleases have a high preference for single-stranded substrates 3 ; therefore the stabilization of double-stranded regions within the mRNA would be protective against endonucleolytic degradation. (ii) Structural changes of the RNA induced by the presence of magnesium can result in a change in the capability of the psbA mRNA to form complexes with protective proteins. Such an interpretation would be supported by results described earlier showing that there are conformers of the psbA mRNA 5Ј untranslated region that are inactive in complex formation with RNA-binding proteins (37). (iii) As a third model, the ions may interact with RNA-binding proteins to induce a protein conformation active in RNA binding. However, this last interpretation is quite unlikely because the magnesium-dependent RNA stabilization is specific for different RNAs in terms of the required concentration and stabilization profile. Our findings that protein binding patterns to chloroplast mRNA are dependent on the concentration of [Mg 2ϩ ] free but are comparable between different mRNAs support this interpretation. According to our data the putative nucleases RNase E and PNPase require Mg 2ϩ for binding to RNA. The finding of enhanced nuclease binding in combination with RNA stabilization can be explained as a consequence of protein structural organization and cleavage specificity. In E. coli RNase E the N-terminal catalytic domain is separate from the central RNA binding domain. The catalytic function of RNase E requires a free 5Ј end of the RNA substrate and an internal single-stranded cleavage site (reviewed in Ref. 54). Therefore, the protein probably binds to the RNA without showing high cleavage activity when the 5Ј end or the cleavage site are structurally sequestered. A similar scenario can be discussed for PNPase, which in E. coli has two RNA binding domains and attacks the RNA at the 3Ј end. Within the degradosome complex those 3Ј ends are newly formed by RNase E (54). The chloroplast nuclease CSP41 is not as well characterized, but its response to magnesium may be mechanistically similar.
We propose that the stabilizing effect involves the interaction of magnesium with RNA rather than with proteins. Structural changes of RNA as a result of an interaction with magnesium ions, which are induced only by a change in ionic strength, in general are more likely to be changes in tertiary folding than in secondary structure. Preliminary experiments using Pb 2ϩ cleavage to analyze for magnesium binding sites reveal a prominent binding position close to the major cleavage site at position 86 in the coding region of the psbA mRNA that we published earlier (8). This indicates that the stabilizing effect of magnesium ions results from the direct binding to the RNA. The role of mRNA structure in RNA metabolism in vivo has been elucidated in C. reinhardtii. Structure is important for translation (55,56) as well as for RNA stability (57) in the alga system. Still, it has not yet been shown how far structural changes are a component of the regulatory machinery of gene expression.
Magnesium as a trans-Regulatory Factor for RNA Stability during Leaf Development?-In the classical view trans-regulatory factors are regarded as proteins. We like to discuss a similar role for magnesium ions in respect to the regulation of RNA stability in higher plant chloroplasts. Rising concentrations of magnesium ions result in the stabilization of chloro- plast RNAs in a specific manner. Each RNA analyzed requires a defined and specific magnesium concentration to gain stability, and each RNA follows a typical stabilization profile. Our determination of magnesium concentrations reveals a concentration of 3-4 mM free magnesium ions in the soluble compartment of young chloroplasts in the light and of 8 -10 mM free magnesium ions in mature illuminated chloroplasts. Comparison of these magnesium concentrations to the stability profiles that we determined shows that psbA mRNA is unstable under magnesium conditions of young chloroplasts and gains an ϳ4fold higher amount of remaining RNA in vitro in the presence of the magnesium concentration corresponding to mature chloroplasts. The ratio between the RNA remaining after 3 h in extracts reflecting young and mature chloroplasts, respectively, is 1 for tRNA His , 1.4 for 16 S ribosomal RNA, and 1.5 for rbcL mRNA. These data fit nicely with changes of mRNA half-lives during leaf development, which have been determined previously for psbA and rbcL mRNA in vivo in spinach (4). In that work a relative increase of mRNA half-lives of psbA mRNA of 2-3-fold was observed relative to 16 S rRNA during leaf development, and a moderate half-life increase of 1.1-fold relative to 16 S rRNA was determined for rbcL mRNA. Comparing the magnesium concentrations of young and mature leaves, the stabilizing effect of magnesium ions for chloroplast mRNAs when calculated relative to 16 S rRNA in vitro is 2.8-fold in the case of psbA mRNA and ϳ1.1-fold for rbcL mRNA. As summarized in Fig. 7, the extent of stabilization by magnesium ions that we observed in vitro precisely matches the mRNA stability data obtained in vivo. Thus, our data show that the changes in magnesium concentrations are sufficient to result in gene-specific changes in mRNA half-lives.
Our findings prompted us to propose a new concept of development-dependent RNA stability regulation, which is mediated by physiological changes resulting from the chloroplast maturation process. According to our data the magnesium concentration of the stroma changes during leaf development, which may be a result of the increasing physiological activity of maturing chloroplasts. The magnesium ions interact with chloroplast RNAs to stabilize them according to the specific response of the individual RNA species to magnesium. The reactivity of different RNAs is specified by its capability to interact with magnesium via specific binding sites that are encoded in the sequence and structure of the RNA.