| |
|
|
A general overview of chloroplast protein phosphorylation and the potential role of calcium in regulation of chloroplast processes is given in the Introduction. Below I will focus on my experimental results concerning that subject.
The current set of experiments was inspired by recent purification of a putative kinase for the photosystem II D1 protein and its subsequent identification as a calcium-dependent protein kinase in A. K. Mattoo's laboratory (personal communication). The main goal was to check whether the thylakoid membrane possesses an intrinsic ability to phosphorylate D1 protein in a calcium-dependent manner. Another objective, which eventually became dominant, was to characterise calcium-dependent protein phosphorylation in chloroplast.
To my knowledge, till now there were only two reports in the literature about regulation of chloroplast protein phosphorylation by calcium. In one case, some preliminary data were mentioned without further elaboration (Hong, Takano et al. 1996). In the other, calmodulin was reported to participate in regulation of light-dependent phosphorylation (Li, Xiang et al. 1998). Indirect evidence, however, strongly suggests that calcium-dependent protein phosphorylation is involved in regulation of the expression of chloroplast genes (Grover, Dhingra et al. 1999). It should also be taken in account that free calcium concentration in chloroplasts is supposed to be higher than in the cytosol and exhibit light-dependent fluctuations of significant magnitude (Johnson, Knight et al. 1995). Thus there are preliminary indications for calcium-dependent protein phosphorylation/dephosphorylation in chloroplasts. The results presented below are probably the first attempt to investigate this phenomenon "in depth".
4.1.2 Characterisation of the set of thylakoid membrane proteins, undergoing Ca-dependent phosphorylation
Nicotiana tabacum was used in most of the experiments since it allows reproducible isolation of high quality intact chloroplasts and young plants are always available from the institute's greenhouse.
I used thylakoids obtained from tobacco intact chloroplasts to perform in vitro phosphorylation under different conditions. My results (Figures R-1 and R-3)
While for the described set of proteins phosphorylation is absolutely calcium-dependent, another group of polypeptides undergoes phosphorylation in the light in the absence of free calcium but to a significantly less extent than in it's presence. And this group in fact incorporates practically all the thylakoid phosphoproteins except those from the first set (Figure R-3). LHCII shows a particularly clean example of such behaviour. The fact that this phenomenon requires both light and calcium may be explained in the following way: the specificity of chloroplast calcium-dependent kinase allows it to phosphorylate all the proteins in question, but phosphorylation sites are exposed only in the light due, say, to conformational changes. Such a mechanism was recently shown to regulate phosphorylation of LHCII (Zer et al, 1999).
Additional related data are presented in two sections below ("Calcium-dependent phosphorylation of the thylakoid membrane proteins in different photosynthetic organisms" and "Intrinsic calcium-dependent phosphorylation in thylakoids vs. phosphorylation by recombinant CDPKs").
An attempt was made to identify proteins undergoing Ca-dependent phosphorylation in tobacco thylakoids. Thylakoid membranes, isolated from intact chloroplasts were subjected to in vitro phosphorylation in the dark in the presence of γ-32P-ATP (for detailed procedure see "Materials and methods"). Subsequent workflow and results are presented in Figure R-2.
Figure R-2). As to the other phosphopeptide (2.1), it contains much less phosphate and thus may be due to incomplete tryptic digestion (e.g. 2.1 incorporates 2.2). It may also represent a secondary phosphorylation site. Each of the proteins from 12/13 kDa doublet accounts roughly for 25% of thylakoid calcium-dependent phosphorylation. Four different phosphopeptides (3.1 - 3.4) originate from those polypeptides. The phosphate content of the proteins is approximately equal. The same is true for the derivative peptides. It is thus reasonable to suggest that both proteins have two equally available phosphorylation sites.The set of proteins described, possesses a number of unique features. One is that the obligatory requirement for their phosphorylation is a presence of micromolar concentration of free calcium in the surrounding media. Titration performed in order to determine minimal and saturating concentrations of calcium required for the calcium-dependent phosphorylation of the thylakoid proteins (Figure R-3) showed that the process is activated at 1 µM free Ca2+ in solution and saturated at 5 µM Ca2+.
In fact, the term "free" may not be completely correct. One can see (Figure R-3, Control lines) that if special precautions are not taken to remove all traces of calcium, the discussed set of proteins is still phosphorylated together with others (e.g. LHCII). Only inclusion of EGTA in the reaction buffer allows us to see the difference in their behaviour. This may mean that in vivo there is enough calcium on or near the thylakoid membrane to keep phosphorylation going and the metal is sufficiently strongly bound not to be lost during membrane isolation. Another possibility is that buffers routinely used for in vitro phosphorylation assay may themselves contain plenty of free calcium if no chelator is added. Metal is introduced as an impurity with buffer components. For instance, the buffer employed throughout this work for thylakoid phosphorylation might contain up to 12 µM free calcium in the absence of EGTA. (The figure is arrived at by assuming calcium is present at the top limit allowed by the manufacturers in all the buffer components). This is well above the activation threshold for calcium-dependent chloroplast protein phosphorylation, as shown by titration experiment (Figure R-3).
One of the distinguishing features of this newly discovered set of polypeptides is that their phosphorylation is not dependent on light and electron transport in chloroplasts. Figures R-1 and R-3 show that calcium-dependent protein phosphorylation in thylakoids proceeds with equal efficiency in the light and in the dark as well as in the presence of diuron. In other words presence of calcium, magnesium and ATP is not only necessary but also sufficient for phosphorylation of this class of proteins. This is rather unusual since most of the protein phosphorylation events in the chloroplasts are light/redox dependent.
Although direct light effects, like induction of conformational shift exposing the phosphorylation site, are sometimes involved (Zer, Vink et al. 1999), regulation mainly proceeds via the light-driven changes in redox potential of the surrounding medium. Some details of the molecular mechanism of kinase redox control were uncovered in recent investigations. It was shown for example that plastoquinol binding in the Qo pocket of the cytochrome b6/f complex is required for LHCII kinase activation (Zito, Finazzi et al. 1999). For the majority of chloroplast membrane proteins the phosphorylation index is increased concomitant with growth of redox potential of the environment and vice versa. A smaller group of polypeptides exhibits an opposite behaviour (Silverstein, Cheng et al. 1993). Yet, in both cases the extent of protein phosphorylation may be gradually adjusted to the changing conditions.
Calcium-dependent phosphorylation is subjected to a different, threshold type of regulation. It is thus likely to be involved in response to a certain stimuli rather than in adaptation to small environmental changes.
4.1.4 Influence of known inhibitors and enhancers on calcium-dependent protein phosphorylation in the thylakoid membrane
A potent activator of Ca/phospholipid-dependent protein kinase C phorbol-12-myristate-13-acetate (PMA, Sigma P-8139) at 100 ng/ml did not stimulate Ca-dependent protein phosphorylation in thylakoids.
Calmodulin antagonists W7 and trifluoperazine were used at 100 µM and compound 48/80 at 10 and 100 µg/ml. Their effect varied from one experiment to another, but significant inhibition of phosphorylation was never observed. It should be mentioned that compound 48/80 at 100 µg/ml caused lysis of intact chloroplasts. It thus probably may seriously alter the membrane structure and its influence on phosphorylation may not be clearly attributed to inhibition of calmodulin or calmodulin-like protein. Polyclonal antibodies against calmodulin (kindly provided by Prof. H. Fromm) also did not specifically inhibit Ca-dependent phosphorylation in thylakoids.
Another compound tested was n-propyl gallate (PGal), which is known to significantly interfere with chloroplast protein phosphorylation. Figure R-4 shows the influence of PGal on the calcium-dependent phosphorylation in chloroplasts in the dark.
The quantification results (right panel) demonstrate that PGal differentially inhibits phosphorylation of one protein subset over another. LHC phosphorylation is the most sensitive with an I50 < 50 µM, while calcium-dependent phosphorylation has an I50 of about 0.8 mM or at least 16 times higher.
4.1.5 Calcium-dependent phosphorylation of thylakoid membrane proteins in different photosynthetic organisms
Calcium modulated phosphorylation in thylakoids is a general phenomenon, rather than a special feature of tobacco. The experimental proof for this is presented below.
Chlamydomonas reihardtii, spinach, tobacco and Spirodela thylakoids were used in parallel experiments. They were each subjected to in vitro phosphorylation in the absence and in the presence of free calcium. In all the cases except Spirodela I was able to detect calcium-dependent light/redox independent protein phosphorylation. Results are presented on Figure R-5.
Spirodela. Moreover, the set of proteins undergoing that type of phosphorylation looks quite similar between species tested. Namely it includes ~12/13, ~22 and ~79 kDa proteins.The share of phosphorylated species in a total pool of a given protein, otherwise called a phosphorylation index, is determined by dynamic equilibrium between phosphorylation and dephosphorylation. This equilibrium is determined by the relative activity of two enzymes: kinase and phosphatase, as well as substrate availability (i.e. the phosphorylation site should be exposed). Thus changes in the phosphorylation index of the protein may be caused by different mechanisms. For example, an increase of phosphorylation may be attributed individually to increase in kinase activity, decrease of phosphatase activity or a combination of both things. In cases where calcium-dependent protein phosphorylation exists, one or more of the following conditions should be observed:
2. kinase is calcium insensitive, while phosphatase is active only in the absence of Ca2+ (e.g. directly or indirectly inhibited by calcium);
My experimental data show that all the thylakoidal proteins that undergo calcium-dependent phosphorylation behave in the same way. Their phosphorylation is induced and saturated at the same respective calcium concentrations, in the same way responds to inhibitors and is equally insensitive to light and electron transport in the chloroplast. All that strongly suggests that the underlying mechanism is also the same. From this point of view, calcium-induced conformational change seems to be a non-realistic option, since too many different proteins located in a specialised cell compartment could possess some common structural feature.
Chloroplast protein phosphatases are mainly redox-independent (Silverstein, Cheng et al. 1993; Elich, Edelman et al. 1997) and thus, potentially, might be involved. However, while there is strong evidence that calcineurin-like calcium stimulated protein phosphatases are present in plants (Bressan, Hasegawa et al. 1998; Kudla, Xu et al. 1999), there are no reports in the literature about calcium inhibited phosphatases in any organism.
Thus, the most probable (and simple) mechanism of calcium-dependent protein phosphorylation in thylakoids involves activation of the kinase by calcium. Here also a few alternatives exist but calcium-dependent calmodulin-independent protein kinase (CDPK or CPK) is the most likely candidate for reasons specified below. Dephosphorylation in that case may be conducted by any chloroplast phosphatase.
There are a few types of calcium-dependent kinases. They are phosphorylase-kinase, Ca2+/calmodulin-dependent protein kinase from brain or liver or protein kinase C. In addition to calcium, those enzymes need calmodulin or specific lipids for their activation (Edelman, Blumenthal et al. 1987). However these types of kinases were not isolated from plants up to now. Rather their functions in plants and algae are carried out by CDPKs and more recently described chimeric CCaM-kinases (Patil, Takezawa et al. 1995; Ramachandiran, Takezawa et al. 1997) that bind both calcium and calmodulin (the latter was not demonstrated to be required for the kinase activity per se). CCaM kinase is probably represented by a single gene and its expression is, so far, limited to flowers. So we are left with CDPKs.
CDPKs represent a unique enzyme family specific to plants and protozoa (for reviews see: Roberts and Harmon 1992; Satterlee and Sussman 1998). They combine all the functionality in one polypeptide chain1. This means that they are able to bind calcium ions directly and this event is sufficient to activate the kinase (Yuasa, Takahashi et al. 1995; Kim, Messinger et al. 1998). CDPKs are represented by gene families in all organisms examined. Sequence analysis (see below) shows that some CDPKs may be chloroplast targeted.
Accepting as a working hypothesis that CDPK is responsible for calcium-dependent protein phosphorylation in thylakoid membranes, the next step would be to try isolating the native enzyme. Two things complicate this task. The first is low abundance of regulatory enzymes, to which CDPK apparently belongs. That necessitates a large-scale biochemical procedure. The second is that CDPKs were shown to constitute multigene families in several organisms (Hrabak, Dickmann et al. 1996; Kim, Messinger et al. 1998; Lee, Yoo et al. 1998; Nishiyama, Mizuno et al. 1999). In Arabidopsis, for example, 12 genes are known at the present time. Recent data suggest that diversity on the protein level may be even higher than on the level of DNA due to alternative RNA splicing (Nishiyama, Mizuno et al. 1999). Similar phenomena were described for Ca2+/calmodulin-dependent protein kinase II isoforms from Drosophila where a single gene gives rise to several proteins with different specificity, stability and kinetic properties (GuptaRoy and Griffith 1996). CDPKs are ubiquitously expressed in various types of plant cells (Hong, Takano et al. 1996). On the subcellular level they were shown to be associated with the plasma membrane, cytoskeleton, chromatin, as well as the cytosol (Roberts and Harmon 1992). A corresponding activity was also detected in mitochondrial membranes (Pical, Fredlund et al. 1993) and chloroplasts (Hong, Takano et al. 1996; Li, Xiang et al. 1998). Thus, there is a high probability to isolate an incorrect isoform or mixture of isoforms.
I have chosen a different approach - to fish out a gene for chloroplast targeted CDPK based on known genes or proteins and then generate and characterise the corresponding recombinant enzyme.
4.2.2 Cloning and characterisation of the calcium-dependent protein kinases (CDPK) from Spirodela oligorrhiza and Arabidopsis thaliana
Thylakoid-associated protein kinase that phosphorylates a peptide mimicking the N-terminus of the D1 protein of photosystem II was recently isolated and characterized (A.K. Mattoo, personal communication). I used sequences of three out of five peptides presumably derived from that kinase for primer design. These three peptides showed high homology to protein kinases and had lead to isolation of the rice gene (A.K. Mattoo, in preparation). Table R-1 summarises the process of peptide selection.
Cloning of the gene for calcium-dependent protein kinase from Spirodela oligorrhiza included three PCR-based steps:
Total cDNA of Spirodela generated as described in "Materials and methods" was used as a template. The peptides from the putative rice D1 kinase and corresponding degenerate oligonucleotide primers used for the first and second PCR steps are listed in Table R-2.
Sequencing of the ~200 bp DNA fragment generated in the second round of PCR revealed that at least two closely related but distinctly different species were co-amplified. Both sequences belong to the CDPK gene family according to the results of the BLAST search. The corresponding genes were designated as SpCDPK1 and SpCDPK2 with "Sp" standing for Spirodela. Figure R-6 shows schematically how the internal fragments of Spirodela CDPKs were amplified.
The sequences obtained were used to design the primers for 5' and 3' RACE. The complete list of primers for RACE, full-gene amplification and sequencing is given in Table R-3. Techniques utilised are described in "Materials and methods" section above.
The 5'-portion of the SpCDPK1 gene obtained by RACE included an untranslated region of 318 base pairs, while the 3' portion contained one of about 130 base pairs along with part of the polyadenylation signal. An attempt to amplify the complete gene together with UTRs failed, however. The likely reason was the low amount of the full-length cDNA in the template used. Another set of primers was designed and used successfully to amplify the coding region only. The PCR product was cloned into pGEM-T vector to produce pGT-SpCDPK1 plasmid with a 1644 bp open reading frame of the SpCDPK1 gene under the control of the T7 transcription promoter.
Cloning of the second CDPK from Spirodela was less successful. I have isolated only the 5' portion of the gene (i.e. 5'-RACE product) at the present time and thus it wouldn't be further considered.
DNA sequences of SpCDPK1, rice CDPK2 and Arabidopsis CDPK7 (about the last one see below) are given in Figure R-7a. SpCDPK1 is highly homologous to other members of the family. On the DNA level SpCDPK1 shares 66% identity with the rice OSCPK2 gene.
Corresponding proteins have 81% identity and 91% homology (Figure R-7b). In addition to the serine/threonine kinase signature, scanning the sequence against PROSITE database revealed four EF hands (calcium-binding motives) in SpCDPK1 as in most CDPKs. A putative N-myristoylation site, found in the protein, is rather common for representatives of that family as well.
The presence of the SpCDPK1 gene sequence in the Spirodela genome was confirmed by southern hybridisation (Figure R-8, experimental details are given in the legend). A number of weak bands, in addition to the major ones (6 kb BamHI fragment and ~8 kb HindIII fragment), most likely indicates that a family of homologous CDPK genes exists in the Spirodela genome, as it is the case in other plants examined up to now (see "Introduction"). This idea is also supported by isolation of two different CDPK genes from Spirodela.
Data on calcium-dependent protein phosphorylation in thylakoids obtained thus far suggest that CDPK may be responsible for that phenomenon. If that is the case, there should be a chloroplast-targeted isoform of CDPK. A few dozen CDPK genes from different organisms are already in the database. So I decided first to select good candidates for chloroplast localisation from known genes. I used two different programs to predict possible intracellular localisation of CDPKs from Spirodela as well as from other organisms, based on the first 116 N-terminal amino acids of each sequence. The software utilised was ChloroP (Chloroplast Transit Peptide Prediction) program (Emanuelsson, Nielsen et al. 1999) at the Center for Biological Sequence Analysis, Department of Biotechnology, The Technical University of Denmark (http://www.cbs.dtu.dk/services/ChloroP/) and PSORT1 at the National Institute for Basic Biology, Okazaki, Japan (http://psort.nibb.ac.jp:8800/). Summarised results are presented in Table R-4.
ChloroP and PEPSORT predicted the presence of chloroplast transit peptide in different but significantly overlapping sets of proteins. It is seems reasonable to suggest that proteins having high scores in both sets are the best candidates to be chloroplast imported. At least five of the analysed plant CDPKs matched those criteria. They were: Solanum tuberosum CPK1, Arabidopsis thaliana calmodulin-domain protein kinase isoforms 6 and 7 (CPK6 and CPK7) and CDPK-related protein kinase (CRK), and Tortula ruralis calmodulin-like domain protein kinase. Arabidopsis thaliana CPK7 (NCBI accession U31836) was selected for cloning for two reasons: a high theoretical probability of chloroplast targeting and availability of cDNA libraries. Gene LHCP AB 180 for LHCII protein (NCBI accession X03908) was cloned to provide a reliable control for chloroplast import experiments.
The neural network score for each residue is determined. The higher this score, the more certain is the network that this residue is part of a cTP. A derivative of this score is used for finding the area in which the cleavage site is searched for - namely among the 40 residues surrounding the residue with the highest derivative score. Finally, the cleavage site score (CS-score) is presented for each residue. This score is calculated from a scoring matrix derived from an automatic motif finding algorithm called MEME (http://www.sdsc.edu/MEME/meme/website/intro.html). The CS-score is defined so that the predicted cleavage site is directly N-terminal of the highest scoring residue within the 40 residues.
Both genes were amplified in one PCR round and cloned into pGEM-T vector. Kim & Theologis lambda-ACT 2-hybrid library (CD4-22) obtained from Arabidopsis Biological Resource Center at Michigan State University (Columbus, OH) was used as a template.
I also tried to isolate CPK2 (NCBI gene accession number ATU31833) as negative control for chloroplast import (it has high probability nuclear localisation signal) and CDPK19 (NCBI gene accession number ATU20627), as it has high PSORT score for chloroplast import. However I failed in both cases. That may mean that CPK7 has a higher expression level than the latter two genes, or at least is more abundant in the cDNA library used.
The list of primers used for isolation of Arabidopsis genes is given in the Table R-5 below:
In vitro expression gives important information about a protein, such as electrophoretic mobility and the presence of particular posttranslational modifications. In the case of Spirodela CDPK1, some unusual properties of the enzyme were additionally discovered.
Rice OSCDPK2, Spirodela SpCDPK1 and Arabidopsis CPK7 genes were expressed in vitro in both reticulocyte lysate and wheat germ coupled transcription/translation systems. The experiments were aimed at:
Rice CDPK2 gene was kindly provided by Dr. Autar Mattoo as a complete cDNA clone in pZero vector (Invitrogen Co., Carlsbad, CA). For in vitro expression, the 2 kb EcoRI fragment from the original plasmid, containing the gene of interest, was subcloned into EcoRI-digested and dephosphorylated pBlueScript-KS vector (Stratagene, La Jolla, CA). The clone with the gene inserted under the control of the T3 transcription promoter (pBS-OSCDPK2 construct), was selected by restriction analysis.
Spirodela SpCDPK1 gene was excised as 1.6 kb ApaI/SpeI fragment from the pGT-SpCDPK1 plasmid (see above) and cloned into the corresponding sites of the pBlueScript-KS vector under the control of the T3 promoter to give pBS-SpCDPK1 plasmid.
Arabidopsis CPK7 and LHCP AB 180 genes were subcloned into the corresponding sites of pBlueScript-KS vector as 1.6 kb EcoRI/BamHI and 0.8 kb KpnI/PstI fragments, respectively. Both genes were inserted under the control of the T3 RNA-polymerase promoter to produce, correspondingly, pBS-ArCPK7 and pBS-ArLHCII plasmids.
Results of in vitro transcription/translation are presented in Figure R-9. Experimental details are given in the legend. Molecular weights of rice and Spirodela enzymes, as calculated from the electrophoretic mobility, are 58.5 and 64.4 kDa, while theoretical MWs are 58997 and 60956 kDa, respectively.
The Spirodela enzyme migrates about 3 kDa slower than expected. This usually means that the protein is more hydrophilic than average. That, in turn, is either a consequence of posttranslational modification, such as phosphorylation or glycosilation, or just an intrinsic property of the particular sequence. Recombinant CDPKs are capable of autophosphorylation (Figure R-15) and the translation reaction contains sufficient amounts of calcium, magnesium and ATP for them to operate with high efficiency. A direct check of whether and to what extent that phenomenon takes place was not conducted. If the mobility shift is due to phosphorylation, the protein band should split in two, as happens with D1 in vivo, but in the current case this is not observed.
It should be noted that boiling of the samples prior to electrophoresis caused the complete aggregation of the SpCDPK1 translation product while having virtually no effect on the rice CDPK2 protein (not shown). This suggests that Spirodela CDPK1 has reduced thermostability. This kind of aggregation is typical for hydrophobic polypeptides. Since rice and Spirodela CDPKs are both hydrophylic and have very similar hydropathy plots (Figure R-10)
, the question remains open as to which region(s) of the Spirodela protein determine its behaviour under high temperature and whether the observed difference has any physiological implications.
N-myristoylation of proteins plays a very important role in signal trunsduction, being responsible for protein-membrane and protein-protein interactions (Towler, Gordon et al. 1988; Taniguchi 1999). Sequence analysis predicts putative N-myristoylation sites (PROSITE ID PS00008) in Spirodela CDPK1, Arabidopsis CDPK7 and rice CDPK2 (presuming that N-terminal methionine is removed). To test, whether this site might be functional, I investigated the in vitro transcription/translation of the genes for all three kinases in presence of [9,10(n)3H]-myristic acid (for experimental details see "Materials and methods"). The wheat germ system was chosen since it is of plant origin and has a higher probability to contain a proper myristoyl-transferase. Results of the assay are presented on Figure R-11.
Arabidopsis became myristoylated. If this happens also in vivo, Arabidopsis CDPK7 should retain its N-terminal part. In that case, the putative signal peptide would not be cleaved off and the Arabidopsis protein would not be targeted to the chloroplast.The most direct way to determine whether any of the three CDPKs I am working with are chloroplast targeted is in vitro importation.
Import of in vitro expressed CDPKs into isolated pea chloroplasts was performed in collaboration with Tami Galperin and Prof. Zach Adam (Faculty of Agriculture, HUJ, Rehovot). Experimental details are described in "Materials and methods" section. Spirodela CDPK1, rice CDPK2 and Arabidopsis thaliana CDPK7 were expressed in vitro in a wheat-germ coupled transcription-translation system (TnTÆ, Promega Corp.) in presence of [35]S-methionine. Three experiments were performed (one failed due to poor quality of in vitro produced precursors). Figure R-12 shows the cleanest results obtained.
Experimental data suggest that Spirodela CDPK1 may be imported into isolated pea chloroplasts. Rate of import is much slower than that of LHCII, but comparable to a some other proteins, considered to be chloroplast targeted (Zach Adam, personal communication; see also for example (Reddy, Nair et al. 2001). The apparent size of the protein after import is 48 kDa. This would be the case if the cleavage site exactly precedes the second myristoylation site. Cleavage at the position predicted by ChloroP program however suggest that the size of imported protein should be 52 kDa. At present, it is not possible to exactly define the cleavage position of the transit peptide. As to myristoylation, experiments show that full length Spirodela CDPK1 is not modified. It is, however, possible that the protein truncated at the second myristoylation site would be a suitable substrate for myristoylation.
Rice CDPK2 and Spirodela CDPK1 were expressed and purified from E.coli for biochemical characterisation. In order to check whether the type of fusion has any influence on kinase activity and/or specificity, two kinds of expression constructs were used to produce the enzymes. In the first a 6-His tag was introduced at the protein N-terminus, while in the second the protein of interest was sandwiched between GST at the N-terminus and 6-His tag at the C-terminus. Thus in the case of 6-His tag, a relatively minor foreign fragment is introduced but a standard IMAC purification procedure does not yield a clean product; while for sandwich constructs a comparatively large stretch of unrelated sequence is introduced which facilitates easy purification of enzymes to near homogeneity.
Rice OSCDPK2 gene was amplified by PCR using pBS-RiceCDPK2 as a template and the primers shown in Table R-6.
The resulting PCR product was digested with SphI and SmaI and cloned into the pQE-30 vector (Qiagen Inc., Valencia, CA) digested with the same restriction enzymes to produce pRiceCDPK2-6H plasmid.
The SpCDPK1 gene was subcloned as a SphI/PstI fragment from pGT-SpCDPK1 into pQE-30 vector to produce pSpCDPK1-6H plasmid.
As a first step, an expression vector, pGEX-3X-6H, was constructed by cloning a synthetic oligonucleotide duplex into pGEX-3X vector (Pharmacia Biotech, Uppsala, Sweden) digested by BamHI and EcoRI (Figure R-13).
It should be mentioned that similar vectors based on pGEX-2T were designed previously (GeneBank accession numbers U84571 and U84572) but they are not commercially available. OSCDPK2 and SpCDPK1 genes were amplified by PCR using pBS-RiceCDPK2 and pBS-SpCDPK1, respectively, as templates and primers listed in Table R-7.
PCR products were cloned in pGEM-T vector. OSCDPK2 was then excised as SmaI/XhoI fragment and SpCDPK1 - as SalI/NotI fragment, and both genes cloned into pGEX-3X-6H digested with the corresponding enzyme pairs.
Details of the expression and purification procedure, and subsequent storage conditions are given in "Materials and methods" section above. Results are presented in Figure R-14.
The molecular weight of the "sandwiched" CDPKs, estimated by gel mobility, is ~92 kDa, which is in agreement with the calculated MWs of 86768 and 88809 Da for rice and Spirodela hybrid proteins, respectively. One liter of bacterial culture yielded 2-3 mg of clean functional enzyme. When stored at -20oC as specified above no significant loss of activity was observed for several months.
Calcium and magnesium ions are obligatory for the activity of the enzymes. Titration curves for Ca2+ and Mg2+ are shown on Figure R-15.
In order to reach maximal performance both kinases require at least 10 µM free Ca2+ and 6 mM free Mg2+ to be present in the reaction mixture. This is in good agreement with the data for previously described CDPKs (for refs. see Introduction/CDPK properties/Diversity and localisation).
Michaelis constants were determined for ATP and Myelin Basic Protein (MBP) according to (Harper, Huang et al. 1994). Experimental details are described in "Materials and methods" section and in the legend to Figures R-16.
Spirodela protein. Km values thus determined are given in the Table R-8. These constants are close to the previously reported for other recombinant CDPKs (Harper, Huang et al. 1994).
Determination of pH optimum using different buffering compounds gave conflicting results, probably due to buffer interference with enzyme activity. I can claim only that recombinant CDPKs are most active at neutral to alkaline pH (7.0-9.0). This makes chloroplast stroma, rather than acidic thylakoid lumen, an appropriate site of action for putative chloroplastic CDPK.
Previously, a number of CDPKs were shown to undergo autophosphorylation (Binder, Harper et al. 1994; Yoon, Cho et al. 1999). Both rice and Spirodela recombinant CDPKs possess the same property. Their rate of autophosphorylation is apparently lower than the rate of phosphorylation of the artificial substrate (MBP). Figure R-17 shows the data from one of the calcium titration experiments. The picture is identical for rice and Spirodela. Only that for rice is shown.
I tested the recombinant CDPKs for there sensitivity to two types of inhibitors: calmodulin antagonists (trifluoperazine, W7 and compound 48/80) and inhibitors of chloroplast protein phosphorylation (diuron and n-propyl gallate2). Figure R-18 shows that recombinant rice CDPK2-6His and Spirodela CDPK1-6His are relatively insensitive to trifluoroperazine and W7 at 100 µM concentration while they are almost completely inhibited by compound 48/80 at 100 µg/ml (~160 µM). This is in disagreement with reports in the literature regarding CDPKs isolated from plants, which were all sensitive to W7 and trifluoperazine with I50 ranging from 20 to 100 µM (Harmon, Putnam-Evans et al. 1987; Li 1991; Abo-el-Saad and Wu 1995; Li, Lee et al. 1998). n-Propyl gallate turns out to be a relatively potent CDPK inhibitor with I50 of about 100 µM for Spirodela enzyme and 400 µM for rice (Figure R-19). Diuron in micro-molar concentrations did not show any significant effect.
Comparison of the biochemical features of rice and Spirodela recombinant CDPKs with those of the thylakoid enzyme suggests that they might be proteins of the same nature.
Cation requirements and inhibitor sensitivity of CDPKs are described in detail in the corresponding subsection of "Introduction". Activation of calcium-dependent protein phosphorylation in thylakoids occurs at micromolar concentrations of free calcium in the presence of millimolar magnesium, exactly as it is observed for CDPKs (Figures R-3 and R-15). Calmodulin exhibits insignificant, and apparently unspecific, influence on the CDPK activity (Roberts and Harmon 1992) but since I haven't checked it on thylakoids, this criteria is not applicable. Phorbol-12-myristate-13-acetate (PMA) does not stimulate calcium-dependent phosphorylation of thylakoid proteins. Lipids do not enhance CDPK performance in many cases as well.
Sensitivity to known blockers of chloroplast protein phosphorylation looks similar for thylakoid-associated activity and known CDPKs. Diuron had no significant influence both on recombinant and thylakoid-associated enzyme. n-Propyl gallate suppressed the activity of both kinases to a similar extent with I50800 µM for native kinase and 100 -400µM for the recombinant ones (Figures R-4 and R-19 respectively).
The situation with calmodulin antagonists is more complicated. On the one hand, neither W7 nor trifluoperazine had any significant and reproducible influence on thylakoid protein phosphorylation. They also hardly decreased recombinant CDPK activity. On the other hand, compound 48/80 showed very significant inhibition of recombinant CDPKs (Figure R-18) but was not active in chloroplasts. Assessment of compound 48/80 effect however is complicated by it's putative interference with membrane structure and stability (see above).
Thylakoid proteins are phosphorylated mainly on threonine and to some extent on serine residues (for refs. see "Introduction"). CDPKs, based on their sequence and a number of experimental confirmations (Putnam-Evans, Harmon et al. 1990; Read and Mikkelsen 1990; Schaller, Harmon et al. 1992; Abo-el-Saad and Wu 1995; Barker, Templeton et al. 1998; Douglas, Moorhead et al. 1998; Ogawa, N. et al. 1998) also belong to serine/threonine protein kinases. SP-1 peptide from D1 protein, used in the current work as a substrate for recombinant CDPK (see below), also has only threonine and serine as potential phosphorylation sites.
4.3.6 Comparison of substrates phosphorylated by intrinsic and recombinant kinases in the thylakoid membrane
Thylakoid membranes isolated from tobacco intact chloroplasts were subjected to phosphorylation in the absence (1 mM EGTA added) or presence of free calcium (30 µM). Reactions were carried out in the light in the presence of 10 µM diuron. Recombinant rice and Spirodela CDPKs were tested. To check whether the type of fusion has any impact on kinase specificity and/or activity, both 6-His and GST/6-His fusions of each kinase were assayed. Approximately 1µg of recombinant kinase per µg chlorophyll was applied. Phosphorylation was carried out at an analytical scale (10 µg chlorophyll per reaction). Results were analysed by SDS-PAGE, 2-D electrophoresis and phosphoimaging (Figure R-20).
Comparison of the pattern of calcium-dependent protein phosphorylation in the thylakoid membrane by intrinsic kinase (lane 2), recombinant rice CDPK23 (lanes 3, 5), and Spirodela CDPK1 (lanes 7, 9) showed that their substrate specificities are similar but not identical. Indeed, both Spirodela recombinant enzymes and native kinase phosphorylated 14, 22, 45 and 79 kDa proteins and to some extent LHCII. In the case of rice CDPK2-GST-6His, phosphorylation of the 14 and 22 kDa proteins occured, but 40 and 62 kDa polypeptides were phosphorylated instead of 45 and 79 kDa ones. Stronger substrate phosphorylation by recombinant kinases may be explained by much higher enzyme-to-substrate ratios than in case of intrinsic kinase.
Two-dimensional fractionation revealed some differences in specificity between Spirodela CDPK and the thylakoid enzyme. Phosphoproteins generated by native kinase fractionate as relatively sharp spots, which means that they are modified in a uniform way: e.g. phosphorylated in one defined site. Proteins phosphorylated by recombinant kinase tend to fractionate in a few smeared interconnected spots (14 and 22 kDa region). This may be caused by non-uniform phosphorylation of one protein on multiple sites or, in addition, by phosphorylation of another protein of a very similar size. In the case of the 45 and 79 kDa proteins, smearing was not observed but additional phosphoproteins were detected. Their positions on the map (similar MW and slightly more acidic pI than that of "native" substrates) suggest that they might represent double phosphorylated isoforms of original kinase target proteins.
A general conclusion is that Spirodela kinase shows a good match of substrate pattern to that of the intrinsic activity. Substrate specificity is independent of fusion type (6His or GST-6His).
Another observation arising from comparison of silver stained gels and the corresponding phosphoimages is that the best CDPK substrates (the most efficiently phosphorylated ones) are probably not abundant proteins since there is hardly anything visibly stained in the regions of 14 and 22 kDa.
The data summarised in the current section promote CDPK as a good candidate for catayzing calcium-dependent thylakoid protein phosphorylation.
My first attempt to identify substrates of intrinsic thylakoid calcium-dependent phosphorylation failed. At the same time recombinant kinases, especially the Spirodela one, were shown to produce a very similar phosphoprotein pattern using thylakoids as substrate. I therefore decided to try characterising thylakoid proteins subjected to phosphorylation by recombinant Spirodela CDPK.
Thylakoids were phosphorylated by Spirodela CDPK1-GST-6His at a preparative scale (570 µg chlorophyll) under the same conditions as for analytical phosphorylation. Proteins were solubilized and separated by SDS-PAGE. Bands of interest (##2, 3, 4 and LHC4) were localised by autoradiography, excised and
subjected to alkylation-reduction and digestion by trypsin. Phosphate-containing peptides were isolated from the total mixture by chromatography on Fe-NTA resin (see "Materials and methods"). Phosphopeptides thus obtained were analysed by MALDI-TOF and RP-LPHC coupled to an electron-spray mass spectrometer. No data was obtained for peptides derived from bands #3 and #4 (14 and 22 kDa phosphoproteins) due to insufficient amount of material. Band #2 (~45 kDa) gave rise to three phosphopeptides. Two of them were identified as belonging to the photosystem II chlorophyll a/b binding protein (CP-47). Together with the apparent size of the initial protein this indicates, with high probability, that CP-47 is a CDPK substrate. The third phosphopeptide derived from band #2 could not be matched to any known protein based on MS/MS data alone. Combination of MS/MS with conventional sequencing gave the following oligopeptide sequence: PF(F/Y)FIVFPL. Unfortunately, this sequence alone is also insufficient to identify the original protein. One phosphopeptide was derived from the 29 kDa LHC region and, not surprisingly, was identified with high probability as a fragment of light-harvesting complex (LHCII) protein type I.
The possibility that CP-47 and LHC are not true CDPK substartes still may not be excluded. Peptides derived from them could mask ones originating from low-abundant proteins. Thus the question of the identity of CDPK substartes in thylakoid membrane remains opened.
Peptides derived from proteins known to contain in vivo phosphorylation sites were used as substrates in kinase assay. SP-1 and acetylated SP-1 peptides from D1 protein were used at 67 µg/ml, LHCII peptide was also used at 67 µg/ml, while wild type and mutant RuBisCO peptides were used at 10 µg/ml. Free calcium was either blocked by EGTA or adjusted to 100 µM. Results are presented in Figure R-21.
They show that from all the peptides tested only SP-1 is effectively phosphorylated by both rice and Spirodela recombinant CDPKs. Interestingly, this does not depend on whether SP-1 is acetylated or not. It should be noted, that SP-1 peptide contains both serine and threonine residues. I haven't performed phosphoamino acid analysis and thus do not know which amino acid(s) of the peptide undergo phosphorylation.
4.3.9 Possible research approaches to elucidating the Potential role of calcium-dependent protein phosphorylation in chloroplast physiology
The current data on calcium signalling in the chloroplasts are presented in "Introduction". Below, I would outline some approaches for further research in that field based both on previous reports and my data.
The overall goal is to determine the physiological significance of calcium-dependent phosphorylation in the chloroplast. It may be approached in two ways. One is to identify the substrates proteins. The current work suggests that one should look for relatively low abundant polypeptides. Chromatographic purification of intact phosphoproteins would be probably more appropriate than isolation of peptides. Another thing which may be done is construction of transgenic plants that express constitutively active chloroplast targeted CDPK. Similar experiments, but not connected with chloroplasts already showed CDPK involvement in stress responses (Sheen 1996).
Physiology and metabolism of the D1 protein remain a subject of intensive research after almost 30 years, but results presented below suggest that our knowledge of the subject is probably far from being complete.
The rate of synthesis of the photosystem II D1 protein is much higher than that of other thylakoid membrane proteins. This, together with the presence of nine methionine residues per molecule, allows practically exclusive labelling of D1 protein by a short pulse of [35]S-methionine. Depending on the length of the pulse different initial ratios of labelled precursor and mature protein can be produced. Protein labelling sufficient for subsequent detection by fluorography can be obtained in Spirodela after pulses of two minutes or less. Brief labelling and then chase in the presence of excess non-radioactive methionine was previously used to follow the conversion of the D1 precursor into mature-size protein (Reisfeld, Mattoo et al. 1982) and it's translocation from nonappressed stroma lamellae into the grana (Mattoo and Edelman 1987). D1 is known to undergo a number of other modifications in addition to C-terminal processing (Table R-9).
Since at least some of those modifications have the potential to change the isoelectric point of the molecule, it is not surprising that two-dimensional fractionation reveals the presence of several isoforms for both precursor and C-terminal processed polypeptides (Figures R-22 and R-23). In fact, three isoforms of precursor as well as of the mature protein are resolved with the fractionation system employed.
Pulse-chase experiments combined with the 2-D fractionation of proteins allow monitoring of the changes in ratio of the different D1 isoforms and, potentially, the real-time kinetics of different posttranslational modifications if knowledge of which isoforms correspond to which modifications is available.
Temporal dynamics of D1 protein isoforms was studied by pulse-chase and subsequent 2-D fractionation of the Spirodela labelled proteins. Experimental protocol normally included cycloheximide pretreatment of plants to reduce the background of cytosolic protein synthesis.
This set of experiments refers mainly to behaviour of the D1 protein after C-terminal processing. A typical set of results is presented on Figure R-22.
Quantitative analysis in the case of mature size protein is easier than for precursor from a technical point of view, and is not complicated by the necessity to account for label penetration rate. Figure R-22 clearly shows that M3 and M2 are transient intermediates in D1 maturation while M1 represents a relatively stable (functionally competent?) state of the protein.
The data are not sufficient to determine whether M3 is converted directly to M1 or there is a sequential reaction M3 M2 M1. It is also possible that both routes are functioning simultaneously.
From the data in previous sections it is evident that within 15 minutes from the start of labeling radioactivity is incorporated in all isoforms of both precursor and mature D1. In other words, 15 minutes is sufficient for D1 to go through all the stages on its way to maturation. I've tried to uncover the sequence of events during this first 15 minutes. To do that I performed very brief pulse labeling (less than a minute) followed by short chase with frequent sample collection. Results of two separate experiments of that type are shown on Figure R-23.
[35]S-methionine hadn't penetrated to the site of protein synthesis yet (Figure R-23, left panel). But already after ~60 seconds chase all the three precursor isoforms were present. In the repeat experiment, it took slightly more time to reach the same stage (Figure R-23, right panel). This is most likely due to the fact that the chase media was cooled on ice before use and, as a result, all enzymatic reactions and protein translocations were slowed down. In any case, all the precursor isoforms are probably formed in a few seconds after the synthesis of the polypeptide chain is completed. Thus, it is very difficult to access experimentally the sequence of their formation. One way could be to perform the labeling under decreased temperature to slow down the rate of protein synthesis and modification, although that approach may be complicated by overall decrease of label penetration into the chloroplasts.The data is not sufficient for quantitative assay of the changes of precursor isoform ratio with time of chase. The problem is that isoform conversion apparently happens on the same, or faster, time scale as penetration of labeled amino acids and protein synthesis inhibitors to their sites of action. For example, the total amount of radioactivity incorporated into D1 protein continues to increase for several minutes after the chase has already started. Even the reliable data on isoform kinetics do not allow unambiguous determination of the conversion route. Without knowing the exact nature of the isoforms one can not say what is the bottleneck in D1 processing, what is (if any) the dead end, is there dynamic equilibrium between some isoforms and so on. The only clear conclusion from the rapid pulse-chase experiments is that M-3 is the first form of C-terminal processed D1 protein that appears during the life cycle. Temporal dynamics of D1 isoforms in stroma or grana lamellae
It is well established that D1 precursor is confined to stroma lamellae. Thus, the question of isoform distribution between grana and stroma is related only to the mature-size protein. To determine whether particular Isoforms of C-terminal processed D1 are preferentially located in appressed or nonappressed regions of the thylakoid membrane I performed a pulse-chase experiment with subsequent fractionation of membranes into stroma and grana lamellae and 2-D mapping of radiolabelled proteins. Results are presented on Figure R-24.
It should be mentioned that while stroma lamellae were sufficiently clean in the experiment, grana lamellae still contained a small fraction of nonappressed regions. Thus, the picture for "grana lamellae" is not clean. M3 isoform shows a high transient increase confined mainly to the stroma lamellae. It is most likely that it is exclusively stroma-located species and its presence in the grana is due to insufficient quality of membrane separation. This is in line with the data from short pulse-chase experiments (Figure R-23) that show M3 as the immediate product of D1 precursor proteolytic processing, thought to occur in stroma. Another conclusion is that M2 is much more common for grana than for stroma lamellae. M1 presence in the stroma is evident. Assuming that it is the final "functional" D1, one may suggest that it is a part of stromal pool of photosystem II (Reisfeld, Mattoo et al. 1982).
As was mentioned above, diuron and n-propyl gallate are both able to prevent D1* formation. They both also decrease the rate of the D1 degradation (Ohad and Hirschberg 1992). Diuron was, in addition, shown to disrupt palmitoylation (Callahan et al, 1987) but n-propyl gallate was not previously tested in that regard. I decided to check the influence of these two compounds on the behavior of D1 isoforms. Figure R-25 shows the resulting fluorograms.
Figure R-25 suggest however that 15-20 minutes of treatment are sufficient to create a new steady state.If a completely denatured protein exists in a few different isoforms it means that it is covalently modified in a few ways, separately or in combination. On the one hand, knowledge of the pI of a certain isoform may give some clue about the nature of corresponding modification(s). On the other hand, a few modifications of D1 are known including phosphorylation and palmytoylation.
They may be directly monitored by two-dimensional mapping using an appropriate radioactive precursor and, thus, positions of the corresponding isoforms may be established.
Isoelectric points for isoforms were derived from Isoelectric pH calibration experiments, performed as described in "Materials and methods". The absolute values for pI of isoform differed by ~0.5 units in two experiments. However, the pI intervals between neighbouring isoforms, were highly consistent. The first experiment yielded pI values of 4.9±0.1, 5.1±0.1 and 5.3±0.1 for isoforms number 1, 2 and 3 of both precursor and mature size protein. The pI values from the second experiment were 5.4±0.1, 5.6±0.1 and 5.8±0.1 respectively.
Assuming that during isoelectric focusing D1 protein is completely denatured, it is possible to use the estimated pI of its isoforms to draw some conclusions about their nature.
Analysis of results is, however, complicated by the fact that different software used to predict pI based on primary structure (GCG, MacDNAsis and servers at http://www.expasy.ch/tools/ pi_tool.html and at http://www.up.univ-mrs.fr/~wabim/d_abim/ compo-p.html) all gave different results.
The analysis below was performed with the help of the GCG package using the Spirodela psbA gene sequence(Avni, Mehta et al. 1991). Theoretical pI values for D1 precursor and mature-size protein are 5.36 and 5.56 respectively assuming a free N-terminus (i.e. NH2 group). Figure R-23 suggests that P3 is the D1 precursor with a free terminal NH2, or some additional positively charged group in case the terminal NH2 is blocked. In fact, the D1 sequence initiates with N-formylmethionine, which is cleaved afterwards (since it is not present, at least in the phosphorylated form of D1; (Michel, Hunt et al. 1988). But whether this happens co-translationaly by analogy with other systems (Ben-Bassat, Bauer et al. 1987), or posttranslationally is not known. The D1 molecule with N-formylmethionine should have a pI about 0.2 units lower than that of P3. Thus, P2 may be N-formylated. The same pI may be obtained, however, after the cleavage of N-formylmethionine and subsequent acetylation of the Thr2 amino group. This would mean that threonine acetylation is not cotranslational, which is not a general rule for proteins not subjected to N-terminal proteolytic processing (Driessen, Jong et al.). The nature of the negative charge that distinguishes P2 and P1 remains an open question.
The known posttranslational modifications of D1 protein may, in theory, be followed by monitoring incorporation of radioactive precursors ([14]C-acetate, [3]H-palmitic acid, [32]P-phosphate). In practice, however, phosphorylation is the easiest case to analyse by 2-D fractionation. Experimental results are presented on Figure R-26.
2 group (removal of one positive charge), the phosphorylated isoform of D1 turned out to be the most acidic (M1). It is known that introduction of the phosphate group changes the mobility of D1 during SDS-PAGE fractionation causing the appearance of the D1*-band (Glaser and Melis 1987; Hoganson and Babcock 1988; Vermaas and Ikeuchi 1991). Formation of D1* can be prevented by DCMU (Glaser and Melis 1987), n-propyl gallate and some other substances. Pretreatment by those chemicals does not however eliminate the M1 isoform from the 2-D map (Figure R-25). On the other hand, if reagents blocking D1* formation are not used at all, M1 is the only isoform remaining after a few hours of chase (Figure R-22). It also incorporates about 90% of the radioactivity after 3 hours of labeling, while analysis of the same protein samples by means of regular SDS-PAGE shows that both D1 and D1* are present in a ratio close to 1:1. Those observations may be explained by the fact that 2-D maps have lower resolution in the second dimension than usual SDS-PAGE. All my attempts to resolve D1 and D1* as distinct spots in the second dimension were unsuccessful. Thus, I assume that M1 is in fact a mixture of two slightly different types of D1 molecules: M1 and M1* (by analogy with D1*). M1 and M1* are the most long-lived isoforms of the D1 protein.The position of the phosphorylated isoform of D1 was determined by a double-labelling experiment. One group of Spirodela plants was pulsed with 300 µCi/ml 35S-methionine for 15 minutes in white fluorescent light (50µE). Under those conditions, the three isoforms of the mature D1 protein are labeled approximately to the same extent. Another group of plants was pulsed with 500 µCi/ml NaH2[32]PO4 for three hours at the same light conditions. Proteins were extracted from both groups of plants and subjected to 2-D separation as described in "Materials and methods" section. Results are shown on Figure R-26 below. They suggest that the position of the phosphorylated D1 isoform corresponds to M1 - the most acidic species in the D1 pool.
1 To make the picture more comprehensive one should mention that a few more types of calcium-regulated kinases were discovered in plants. The first is CCaMK that contains both a visinin-like calcium-binding domain and a CaM-binding domain. It is subjected to regulation by both calcium and calmodulin but in a different fashion than animal CaMK (Patil, Takezawa et al. 1995; Ramachandiran, Takezawa et al. 1997). The second is mammalian CaMKII analogue containing a CaM-binding domain but lacking any calcium-binding sites (Watillon, Kettmann et al. 1993). Finally there are CDPK-related protein kinases (CRK), containing modified CaM-like domain but not dependent on calcium (Lindzen and Choi 1995; Furumoto, Ogawa et al. 1996).
2 n-Propyl gallate was shown also to inhibit human spleen protein tyrosine kinase (Lazaro, Palacios et al. 1995).
3 Rice CDPK2 6-His fusion (lane 3) gave a very low signal, almost the same as membranes alone, probably due to the loss of activity upon storage.
4 Band #1 was not analysed since it contained low amount of radioactive phosphate and thus could not be effectively tracked during purification.
|
alexander.raskind@weizmann.ac.il |
| |
|
|
.gif)
values for recombinant rice CDPK2 and Spirodela CDPK1
, µM
, µM
