Use of biomolecular interaction analysis to elucidate the regulatory mechanism of the cysteine synthase complex from Arabidopsis thaliana.

Real time biomolecular interaction analysis based on surface plasmon resonance has been proven useful for studying protein-protein interaction but has not been extended so far to investigate enzyme-enzyme interactions, especially as pertaining to regulation of metabolic activity. We have applied BIAcore technology to study the regulation of enzyme-enzyme interaction during mitochondrial cysteine biosynthesis in Arabidopsis thaliana. The association of the two enzyme subunits in the hetero-oligomeric cysteine synthase complex was investigated with respect to the reaction intermediate and putative effector O-acetylserine. We have determined an equilibrium dissociation constant of the cysteine synthase complex (K(D) = 25 +/- 4 x 10(-9) m), based on a reliable A + B <--> AB model of interaction. Analysis of dissociation kinetics in the presence of O-acetylserine revealed a half-maximal dissociation rate at 77 +/- 4 microm O-acetylserine and strong positive cooperativity for complex dissociation. The equilibrium of interaction was determined using an enzyme activity-based approach and yielded a K(m) value of 58 +/- 7 microm O-acetylserine. Both effector concentrations are in the range of intracellular O-acetylserine fluctuations and support a functional model that integrates effector-driven cysteine synthase complex dissociation as a regulatory switch for the biosynthetic pathway. The results show that BIAcore technology can be applied to obtain quantitative kinetic data of a hetero-oligomeric protein complex with enzymatic and regulatory function.

The interaction of proteins is essential for the function of living cells (1,2). Monitoring of biomolecular interaction analysis can be achieved by BIAcore technology that is based on surface plasmon resonance (SPR). 1 The physical phenomenon of SPR is widely used to visualize macromolecular interactions in real time with the advantages of no labeling requirements and the option to determine kinetic rate constants. It is mostly applied to quantification of antigen/antibody binding, receptor ligand screening, or characterization of protein modification. It has not been used so far to investigate enzyme-enzyme interactions, in particular with respect to regulation of the metabolic activity of multienzyme complexes.
In this study we have applied SPR to analyze the interaction of a hetero-oligomeric metabolic protein complex using the enzymes of cysteine biosynthesis from mitochondria of Arabidopsis thaliana. Serine acetyltransferase (SAT; EC 2.3.1.30) generates the activated sulfide acceptor O-acetylserine (OAS) from serine and acetyl-CoA. O-Acetylserine (thiol)-lyase (OAS-TL; EC 4.2.99.8) then inserts free sulfide in a ␤-replacement reaction to synthesize cysteine and acetate. One tetramer of SAT and two dimers of OAS-TL are presumed to form the hetero-oligomeric cysteine synthase complex (CSC) that was first described for Salmonella typhimurium (3,4). Proteinprotein interactions of such metabolic enzymes mostly serve to channel substrate intermediates between active sites of sequential reaction steps of a pathway. The advantages of these quaternary organizations include efficient catalysis rates and stabilization of intermediates by prevention of losses of intermediates into solution by diffusion (1,2). However, substrate channeling is not realized in the CSC, because a ready release of the reaction intermediate O-acetylserine (OAS) into the surrounding solution was observed for the bacterial complex (5).
The reason for the association of SAT and OAS-TL in the CSC is therefore unknown, despite the fact that the importance of this final step of sulfate assimilation is comparable with the fixation of ammonia by glutamine synthetase in nitrate assimilation (4,6,7).
In plants the synthesis of cysteine takes place in cytosol, plastids, and mitochondria. Nuclear encoded cDNAs of SAT and OAS-TL for each compartment-specific isoform have been cloned from several plants including A. thaliana (8,9). Expression analysis revealed semi-constitutive activity of the corresponding genes with only moderate up-regulation of mRNA steady-state levels in response to external factors such as sulfate deprivation or salt stress (10 -13). Metabolic regulation appears to be more important for the control of flux of reduced sulfur into cysteine. Feedback inhibition by cysteine of the cytosolic SAT isoform, but cysteine insensitivity of the plastid and mitochondrial isoforms from A. thaliana, is an element of a regulatory model for cysteine synthesis (6,8).
An additional regulatory mechanism may be provided by the reaction intermediate OAS. As suggested by feeding experiments, OAS is involved in the de-repression of genes encoding sulfate transporters, enzymes of sulfate reduction, and a seed storage protein during sulfate deficiency (6, 14 -16). Furthermore, incubation of the intact CSC with OAS dissociates the complex in vitro and results in changes of kinetic properties and activity of both enzymes; free SAT tends to aggregate, becomes unstable, and loses affinity to serine and acetyl-CoA; free OAS-TL, on the contrary, gains activity and increases its affinity to OAS and sulfide (3,17). These findings are corroborated by the identification of a bifunctional C-terminal domain of SAT from A. thaliana that is responsible for catalysis and protein-protein interaction with OAS-TL (18,19). The low activity of OAS-TL in the CSC apparently is the reason for diffusion of OAS into the surrounding solution, thus preventing any substrate channeling (5). An in vivo dissociation of the CSC may indeed occur, as suggested by the recent observations of increased OAS levels in response to sulfate starvation in A. thaliana (15,20). The association and dissociation of CSC in the absence or presence of OAS have been depicted in a preliminary reaction scheme in Fig. 1.
In this study we have applied SPR to elucidate the function of this protein complex. The homomeric subunits were found to behave similar to single polypeptide chains. This allowed modeling of interaction kinetics based on a simple A ϩ B N AB principle and demonstrated that SPR-based BIAcore technology (21) can be applied to quantitative kinetic analysis of a multimeric protein complex. The assessment of OAS effects on CSC dissociation eliminates any function in substrate channeling common to other metabolic multimeric complexes. Instead, the kinetics obtained provide the first evidence for a function of the dissociation state of a metabolic enzyme complex as a regulatory switch for the entire upstream biosynthetic pathway.

MATERIALS AND METHODS
Cloning Procedures-General cloning procedures, sequencing, and PCRs were carried out as described (22) using Escherichia coli XL1-Blue (Stratagene, La Jolla, CA) as a host. The coding region of SAT-A (X82888; see Ref. 23) without mitochondrial target peptide, reaching from bp 28 to 939, was amplified by PCR, verified by DNA sequencing, and cloned into pET28a (kanamycin resistance; Novagen, Madison, WI), yielding a fusion protein of 35 kDa with an N-terminal histidine tag for affinity purification. Mitochondrial OAS-TL-C (AJ271727; see Ref. 12) was expressed in pET3d (ampicillin resistance) as described (19) but without purification tag.
Protein Expression and Purification-SAT and OAS-TL were expressed singly or simultaneously in E. coli HMS 174 (Novagen) and purified following an optimized procedure of Droux et al. (17). Cells were grown overnight after induction (1 mM isopropyl-1-thio-␤-D-galactopyranoside) in LB medium supplemented with 100 g/ml ampicillin, 50 g/ml kanamycin, or both and 15 M thiamine and 10 M pyridoxine HCl. Cells were lysed by ultrasonication in elution buffer (50 mM HEPES/NaOH, pH 7.5, 300 mM NaCl, 0.05 mM EDTA, 0.005% Igepal CA-630) containing 40 mM imidazole, and protein supernatant was obtained by centrifugation (20 min, 48,000 ϫ g). OAS-TL-C alone was purified to apparent homogeneity by fast protein liquid chromatography essentially as described (17). SAT alone or the complex of SAT with OAS-TL was affinity-purified using a nickel-NTA-loaded HiTrap TM chelating column (volume 1 ml; Amersham Biosciences). Protein supernatant (60 -80 mg of protein) was directly applied to the column and washed with 10 ml of elution buffer containing 80 mM imidazole. Bound SAT⅐OAS-TL complex was eluted with elution buffer containing 400 mM imidazole in 5 fractions of 1 ml and yielded a total of 4 -8 mg of pure complex. In order to isolate SAT alone, bound complex was first treated with 5 ml of elution buffer containing 80 mM imidazole and 50 mM OAS. This caused complete dissociation of OAS-TL and was followed by a washing step with 10 ml of elution buffer containing 80 mM imidazole. SAT was subsequently eluted with imidazole as described above. Complex and free SAT in fraction 2 containing the highest protein concentration was diluted 10-fold immediately after elution with elution buffer lacking imidazole to avoid precipitation of protein.
Determination of Free and Complex-bound SAT and OAS-TL-SAT activity was determined via UV absorption as described (19). OAS-TL activity, with or without SAT, was assayed in a volume of 100 l containing assay buffer (50 mM Tris/HCl, pH 7.5), 40 -50 ng of purified protein, 10 mM O-acetyl-L-serine, 5 mM Na 2 S, and 2.5 mM dithiothreitol for 5 min at 25°C. Cysteine was determined as described (17). The dissociation state of the CSC in response to the presence of OAS was analyzed via increased OAS-TL activity according to the following assay sequence. 1) 40 -50 ng of purified CSC was pretreated at 25°C in a 10-l volume containing assay buffer, no sulfide, and 0 -1 mM OAS. 2) The assay was started by addition of 90 l of assay buffer containing 2.75 mM dithiothreitol with 11 mM OAS and 5.5 mM Na 2 S. 3) Incubation was for 5 min at 25°C, and cysteine determination was as above. Each measurement was carried out four times and repeated with three independent protein preparations. Kinetic analysis was performed using SigmaPlot software. Optimized fits resulted in hyperbolic functions based on the Michaelis-Menten equation Determinations of protein concentrations and denaturing SDS-PAGE were carried out following standard protocols (22). Biomolecular Interaction Analysis-Real time biomolecular interaction analysis was performed by surface plasmon resonance technology using a BIAcore X instrument equipped with a Sensor Chip NTA (24) from BIAcore AB (Freiburg, Germany) and the polyhistidine-tagged SAT. All experiments were carried out at 25°C with a constant flow rate of 5 l/min with 50 mM HEPES/NaOH, pH 7.4, 300 mM NaCl, 0.05 mM EDTA, and 0.005% Igepal CA-630 as running buffer. The cell without nickel loading served as a reference for nonspecific binding of proteins and chemicals. All solutions first passed this flow cell (one) and then the one with nickel charging (two). Net signal changes were calculated as differences of flow cell 2 minus flow cell 1. Flow cell 2 was charged by addition of 5 l of 0.5 mM NiCl 2 to the running buffer giving rise to a signal of 40 -50 relative units (RU). Purified SAT (25 ng/l) was immobilized by injection of 5-10 l of solution until a loading density equivalent to 1000 RU was reached. Kinetic analysis of protein interaction was carried out by injection of seven different OAS-TL concentrations in running buffer in the range of 20 -200 nM. The measurement series was repeated with three independent preparations of SAT protein and OAS-TL dilutions for each OAS-TL concentration. For OAS dissociation experiments, OAS-TL was injected at a concentration of 100 nM (ϳ400 RU) followed by injection of 100 l containing 14 different OAS concentrations from 0.5 to 600 M in running buffer with four repetitions. After each experiment the chip surface was regenerated by injection of the following: 1) 10 l of 0.5% SDS; 2) 50 l of protease K (100 g/l); 3) 5 l of 0.5% SDS; 4) 5 l of 0.35 M EDTA; 5) 5 l of 0.5 mM NiCl 2 (only flow cell 2). For data analysis the rate constants were calculated using the BIAevaluation software 3 without drift option during curve fitting (BIAcore AB, Freiburg, Germany) (25, 26). Data Assuming that the two binding sites on the SAT tetramer for the two OAS-TL dimers are independent from each other and show no cooperativity or anti-cooperativity, the kinetics observed can be simplified by a A ϩ B ó AB model with k 1 ϭ kЈ 1 and k Ϫ1 ϭ kЈ Ϫ1 (2). In the presence of OAS the SAT⅐OAS-TL complex can either dissociate directly (k Ϫ1 , here simplified for two identical binding sites according to 1) or a SAT⅐OAS-TL⅐OAS complex can be formed with cooperative binding of OAS and a higher dissociation rate k Ϫ2 for (OAS-TL) 2 . Note that the stoichiometry and sites of OAS binding within the complex and the (OAS-TL) 2 ⅐(OAS) n leaving group are hypothetical.
were fitted to a first order kinetic A ϩ B N AB model. The dissociation rate constant k d was determined from fitting the dissociation phase of the curves to the equation R ϭ R 0 e Ϫkd(tϪt0) . The association phase was fitted to the equation R ϭ R 0 /k s (1 Ϫ e Ϫks(tϪt0) ). The association rate constant k a was determined as the slope of k s plotted against C (k s ϭ k a C ϩ k d ). The equilibrium dissociation constant was determined by

Cysteine Synthase Complex Expression, Purification, and
Characterization-Detailed biochemical and biophysical analysis of the interaction of the SAT and OAS-TL subunits of the CSC was hampered in the past by limiting protein preparation procedures. To prepare sufficient amounts of purified complex, the co-expression and purification protocol for the mitochondrial CSC consisting of SAT-A and OAS-TL-C was optimized for the recombinant proteins from E. coli and could be completed within 2 h after harvest of cells. From 1 liter of culture a total of 60 -80 mg of soluble protein was obtained that yielded an average of 4 -8 mg of homogenous CSC after histidine tag affinity purification according to denaturing gel electrophoresis (Fig. 2). The hetero-oligomeric CSC bound to the nickel affinity matrix and was eluted by imidazole. In addition, OAS-TL was specifically and completely dissociated from column-bound SAT by the reaction intermediate OAS and regained full activity as a free homodimer. Thereafter, bound SAT could be eluted with imidazole. Some smaller protein bands were occasionally observed after purification and were verified as very minor degradation of both proteins as indicated by immunoblotting with specific antibodies (data not shown). The CSC itself as well as SAT and OAS-TL alone exhibited reproducible and high specific activities (Table I). SAT was active in the complex, but OAS-TL was strongly inactivated, displaying 6.5% activity compared with the same protein after dissociation from SAT. Isolated SAT protein remained fairly stable in vitro with about 50% loss of activity after 24 h at 4°C but proved to be sensitive to high concentrations of SAT protein at low temperatures. The advanced high yield and purity protocol thus provided functional CSC and its subunits.
Effect of O-Acetylserine on the Association State of the Cysteine Synthase Complex-The bifunctional role of OAS as a reaction intermediate and an effector was analyzed with respect to the dissociation equilibrium of the CSC. Because free OAS-TL was 15-fold more active than complex-bound OAS-TL, the ratio between these two activation states could be modulated by preincubation of the CSC with increasing OAS con-centrations followed by assay incubation at substrate-saturated conditions. No cysteine was formed during the preincubation period because of the absence of sulfide. Two minutes of preincubation were found to be sufficient to establish the corresponding equilibrium (data not shown). The activation of OAS-TL in response to OAS treatment followed a hyperbolic function that reached 80% saturation at 200 M OAS (Fig. 3). An optimized fit of the data followed Michaelis-Menten kinetics. A Michaelis-Menten constant was calculated that in this case describes an effector constant of OAS for OAS-TL activity. The obtained K m OAS ϭ 58 Ϯ 7 M corresponds to 50% of maximal OAS-TL activity and 50% dissociation state of the CSC. This indicates that, in the absence of sulfide, OAS concentrations of ϳ20 -200 M promote equilibrium states of almost complete association or dissociation of the CSC.
Biomolecular Interaction Analysis-Kinetic properties of the CSC were characterized by combination of SAT with independently purified OAS-TL using a BIAcore X instrument. For real time interaction analysis the histidine-tagged SAT subunit was immobilized as ligand on a Sensor Chip NTA (27). Because the assembled CSC consists of a hetero-oligomer of presumably one SAT homotetramer and two OAS-TL homodimers, special care was taken to ensure that complex stability was sufficiently reproducible to obtain binding SPR kinetics that could be correctly evaluated using established algorithms. Flow cell 1 was not charged with nickel. All solution had to pass this reference cell first before the nickel-charged flow cell 2 was incubated. Sensograms showed very minor nonspecific binding of SAT or OAS-TL in the range of 5% of the nickel-coated chip (Fig. 4A). The sample chip in flow cell 2 was generally loaded with nonsaturating amounts of SAT protein that corresponded to a signal of 1000 RU. Sometimes a small spike was observed following injection that might be a reaction of SAT oligomer rearrangement on the chip surface. The stable sensor signal observed during washes with running buffer (Fig. 4A, line C), however, indicated a high stability of the bound SAT, possibly because each SAT homo-oligomer provides several histidinenickel-binding sites. Injection of saturating amounts of OAS-TL (1000 nM) lead to a signal ratio of SAT:OAS-TL ϭ 2:1.
In SAT/OAS-TL association experiments the different OAS-TL concentrations applied correlated with increasing slopes of the association sensorgrams (Fig. 4B). Curve fits were performed under the assumption of an A ϩ B N AB binding model to calculate association kinetics where A would represent a SAT tetramer and B an OAS-TL dimer. The plot of k s versus concentration C [OAS-TL] yielded a linear function from which the association rate constant k a ϭ 5.1 ϫ 10 4 M Ϫ1 s Ϫ1 was calculated (Fig. 4C). Self-consistency of the data was confirmed by fitting data to the equation 1/R eq ϭ (K D , eq ϩ C)/(R max C), where R eq is the response at equilibrium and R max the response for complete saturation of SAT with OAS-TL (26). Application of this latter equation yielded an approximated K D value in the same range as before (not shown). The linear association kinetics over a 10-fold concentration range of the analyte OAS-TL strongly suggests that the simple binding model is applicable to describe the interaction between SAT and  OAS-TL oligomers as suggested in equilibrium 1 of Fig. 1.
A mono-exponential fit was performed to calculate an approximation of the dissociation rate constant. A value for k d ϭ 1.3 ϫ 10 Ϫ3 s Ϫ1 was obtained. Dissociation was independent of OAS-TL concentrations as expected. Although the dissociation phase showed some deviation from the A ϩ B N AB model in very late stages (Ͼ900 s), its kinetic analysis still provided a good approximation. An effect of SAT dissociation from the chip on the observed kinetics was negligible as shown by graph C in Fig. 4A. Thus, an equilibrium dissociation constant (K D ) in the absence of any effector could be calculated as K D ϭ 25 Ϯ 4 ϫ 10 Ϫ9 M (n ϭ 4).
SAT/OAS-TL Dissociation Kinetics for the Effector OAS-To determine the kinetic relationship between dissociation of the cysteine synthase complex and free OAS as depicted in Fig. 1, we applied different concentrations of OAS in the dissociation buffer (Fig. 5A). The OAS concentration-dependent change in k d , as determined from the initial phase of dissociation kinetics (900 s), showed somewhat stronger deviation from the ideal mono-exponential behavior than in the absence of effector. The plot of k d versus [OAS] resulted in a sigmoidal function (Fig.  5B). When modeled according to Hill, a Hill coefficient of 2.4 could be derived from these data, indicating high cooperativity for the dissociation of SAT and OAS-TL mediated by the effector OAS. The feasibility of this fit was justified by least square analysis (r 2 ϭ 0.99) compared with differential data fits using Michaelis-Menten kinetics (r 2 ϭ 0.94; dashed line in Fig. 5B). In particular the small standard errors obtained for the individual k d values at low OAS concentrations clearly support the sigmoidal slope (inset with initial reaction slopes in Fig. 5B). The k d OAS /2 corresponding to a half-maximal dissociation rate was calculated to 77 Ϯ 4.2 M OAS (n ϭ 4). Thus, the OAS concentration required to achieve 50% dissociation of the complex as determined via OAS-TL activation and 50% of the maximal dissociation rate as determined by biomolecular interaction analysis are in the same range. Due to micro-reversibility such an effect of OAS can also be expected for CSC formation. Both findings strongly indicate that changes of OAS concentrations in a narrow range of 50 -80 M result in significant changes of association state of the CSC. It is concluded that independent evidence derived from biochemical and biophysical approaches points to a double function for OAS not only as a substrate but also as an effector with the potential to regulate the association state of the CSC in vivo.

DISCUSSION
This study provides a quantitative kinetic description of the effect of the putative effector OAS on the enzyme-enzyme interaction in the mitochondrial CSC from A. thaliana. As a prerequisite for a detailed kinetic investigation, a preparation protocol for the mitochondrial CSC from A. thaliana and its free subunits, SAT and OAS-TL, was established that allows rapid purification at high yield, purity, and activity after expression in E. coli. Within less than 2 h CSC representing up to 10% of total soluble bacterial protein was purified to apparent homogeneity. This corresponds to an ϳ20-fold improvement of yield compared with previous preparations of recombinant CSCs (17,19,27). In contrast to free cytosolic SAT from A. thaliana (17), the mitochondrial SAT analyzed in this study proved to be quite stable in the absence of OAS-TL. This difference may be intrinsic for the different compartmental SAT isoforms from A. thaliana or a result of the different fusion tags and isolation procedures.
Our CSC preparation confirmed several characteristic features of the CSC that have been partly described for CSCs from bacteria (3,5,27) and plants (17,28) as follows: bound OAS-TL is much less active than the free enzyme; OAS can dissociate the CSC; OAS-TL released from the complex gains full activity as compared with free OAS-TL dimers. Based on these special properties an activity assay using OAS pretreatment of the complex with increasing OAS concentrations was developed that allowed us to correlate the activation of OAS-TL with the degree of dissociation of the CSC. The observed Michaelis-Menten constant for OAS of K m ϭ 57 M indicated a 50% dissociation already far below the millimolar OAS concentrations conventionally used to disrupt the complex in vitro (3,17,19). Furthermore, significant shifts in the equilibrium between CSC and free subunits can be predicted in response to even small changes of OAS concentrations in vivo.
A quantitative assessment of the kinetic behavior of SAT and OAS-TL interaction was required to gain evidence for a putative regulatory function of this metabolic protein complex system. However, dissociation kinetics of a hetero-oligomeric complex of 300 -350 kDa might be expected to be quite complicated because of the multitude of possible interaction sites between protein subunits. This study shows that biomolecular interaction analysis using SPR can be suitable for the analytical description of such mechanisms. Binding characteristics and binding stability of real time interaction as detected by BIAcore were carefully monitored and evaluated with respect to feasibility of a simple A ϩ B N AB model for association and dissociation of the CSC. In a first approximation, the binding kinetics observed indeed strongly suggest such tight SAT and OAS-TL homomers, respectively, that their interaction at the sensor surface can be sufficiently described with this model. Significant statistical deviation of the dissociation rate from an ideal mono-exponential function was only observed during long term monitoring (45 min). Intermediate steps during the dissociation reaction of the multimeric complex might be able to cause this difference or an overlay of complex dissociation with release of the SAT anchor protein from the nickel matrix of the chip. However, such bleeding of SAT was very minor (Fig. 4A). Whereas a SAT tetramer is assumed to bind two OAS-TL dimers (3,28,30), the loading of immobilized SAT with saturating concentrations of OAS-TL yielded a 2:1 ratio of SAT to OAS-TL. This may be caused by blocking of OAS-TL interaction domains on SAT subunits due to the fixed histidine-nickelbinding site, although such a steric inhibition was not observed during affinity tag isolations of the CSC (Fig. 2) (17,19). A 2:1 quaternary organization of the native complex cannot be excluded, because the SAT homomer from E. coli was recently reported (29) to consist of a dimer of trimers instead of a tetramer. It has to be considered, however, that these analyses were performed in the absence of OAS-TL.
The equilibrium dissociation constant for the interaction of mitochondrial SAT and OAS-TL in the absence of any effector (K D ϭ 25 Ϯ 4 nM) is in agreement with the K D value of 41 nM determined for the interaction between cytosolic SAT from A. thaliana and plastid OAS-TL from spinach (17). The application of the SPR-based biosensor method with an immobilized ligand thus confirms the analysis of the heterologous CSC that had been determined with soluble components in a velocity versus [OAS-TL] plot (17). The K D of the CSC indicates a very

FIG. 5. Dissociation of SAT and OAS-TL interaction by increasing concentrations of OAS as quantified by the BIAcore system.
A, OAS-TL was bound to SAT at non-saturating conditions. Proteinprotein interaction was analyzed at the OAS at concentrations indicated. The original sensorgrams were normalized to 100 RU. Shown are the first 300 s of the 900 s of dissociation phase used to calculate dissociation rate constants. B, dissociation rate constants k d were plotted against OAS concentrations to determine the Hill constant for cooperativity of complex dissociation. The inset represents a magnification of the marked square. stable complex that in fact is comparable with the interactions of antibody-antigen and receptor-ligand partners.
The effect of OAS on the dissociation of the CSC on the BIAcore chip surface was best described by a Hill plot that suggested strong cooperativity for the dissociation of SAT and OAS-TL. According to this fit, 77 M OAS is sufficient to achieve the half-maximal dissociation rate. This is in exactly the same range as the K m value of 58 M OAS required to adjust 50% equilibrium dissociation state of the complex. Changes of OAS concentrations in a very narrow range thus are bound to cause fast and almost complete association or dissociation of the CSC. It is important to note that the OAS concentrations relevant for complex dissociation are far below the substrate affinities of free and bound OAS-TL enzymes for OAS that range around 1 mM (9,28,30), whereas dissociation constants of OAS-TL dimers for OAS are below 10 M. 2 Dissociation effects can therefore become effective before substantial catalytic activity can take place. Taken together, such kinetic behavior provides an ideal molecular switch for regulation of downstream processes in a threshold-dependent manner.
Indeed, a regulatory function for OAS in regulation of sulfur metabolism has been suspected (7,14,31,32). OAS concentrations have been described to differ between 2 and 56 nmol/g FW, respectively, in rosette leaves of 9-week-old sulfur-sufficient and sulfur-limited A. thaliana plants (20). In sulfurdeprived potato leaves OAS levels increase from 3 to 6 nmol/g FW to 220 nmol/g FW within 8 days. 3 The cellular contents of OAS increase from 0.3 to 3.8 nmol/g FW in A. thaliana suspension cultures within 72 h after transfer to sulfate-depleted medium. 4 Assuming that OAS is not localized to the vacuole, a calculation based on relative volumes of cell compartments (33) yields cellular OAS concentrations between 10 and 60 M during sulfur-sufficient and 60 -200 M during sulfur-limited conditions. Intracellular OAS fluctuations would thus occur in the critical range of CSC dissociation.
Our findings therefore provide for the first time the quantitative basis for essential elements of a hypothesis for the function of the CSC (17,32). In this model SAT and OAS-TL are associated in the presence of sufficient sulfur in the cell, because sulfide stabilizes the complex and OAS levels are low (Fig. 6). OAS leaves the CSC because of the low affinity of bound OAS-TL and is consumed to produce cysteine by free OAS-TL dimers (3,17). If the cells encounter sulfate deficiency, sulfide levels will drop and OAS will accumulate to concentrations that are sufficient to effectively dissociate the CSC. Consequently, SAT would become less active by degradation or modification, and at the same time OAS could trigger the de-repression sulfate uptake and assimilation (14,16,31). Thereby the system becomes reversible, because sulfate enters the cell and after reduction sulfide reacts with the accumulated OAS via free OAS-TL dimers. Lowered OAS levels now pro-mote formation of the CSC from free OAS-TL and reactivated or newly synthesized SAT. It is concluded that the plant CSC functions as a molecular sensor system that monitors the sulfur status of the cell and controls sulfate assimilation and cysteine synthesis according to the availability of sulfate.