Pyruvate decarboxylase is a key enzyme in organisms whose energy metabolism is based on alcoholic fermentation. The enzyme catalyses the nonoxidative decarboxylation of 2-oxo acids in the presence of the cofactors thiamine diphosphate and magnesium ions. Pyruvate decarboxylase species from yeasts and plant seeds studied to date are allosterically activated by their substrate pyruvate. However, detailed kinetic studies on the enzyme from Neurospora crassa demonstrate for the first time the lack of substrate activation for a yeast pyruvate decarboxylase species. The quaternary structure of this enzyme species is also peculiar because it forms filamentous structures. The complex enzyme structure was analysed using a number of methods, including small-angle X-ray solution scattering, transmission electron microscopy, analytical ultracentrifugation and size-exclusion chromatography. These measurements were complemented by detailed kinetic studies in dependence on the pH.
alcohol dehydrogenase from Saccharomyces cerevisiae
PDC from Saccharomyces cerevisiae
PDC from Zymomonas mobilis
Pyruvate decarboxylases (PDC) from several organisms such as yeasts, plant seeds and bacteria, have been isolated and studied in some detail in the past. Its simple preparation and outstanding allosteric behaviour qualify the species from Saccharomyces cerevisiae as a paradigmatic model of regulatory processes. The kinetic behaviour of several PDC species was analysed for the native substrate pyruvate and a number of?other 2-oxo acids, both aliphatic and aromatic [1,2]. The catalytic cycle and the mechanism of allosteric?substrate activation are understood in great detail [3–10].
Crystal structure analyses of various PDC species from yeast and bacteria have unravelled the architecture of the active and regulatory sites, the structure of domains, subunits and oligomers, and their corresponding interactions [11–17]. However, the oligomerization behaviour in solution was studied by means of small-angle X-ray scattering, varying the pH as well as the concentration of cofactors and other ligands. Based on these results [18–20], it has been possible to establish a network of structure–function relationships. Despite some quantitative differences, PDC species from various organisms show a great deal of similarity in protein structures and catalytic functions. The dominant oligomer state is the tetramer consisting of four identical subunits. The magnesium-ion-complexed cofactor thiamine diphosphate (ThDP) is bound to domains of two subunits within the same functional dimer. Consequently, the interaction areas of subunits within dimers are markedly greater than those between dimers within the tetramer. The dimer is the smallest catalytically active unit of PDC . Yeast PDC species present a more or less planar arrangement of dimers within the tetramer. Oligomer dissociation depends on the subunit concentration, as well as on the pH and the concentration of the cofactor ThDP. A dissociation constant of 8 μm was observed for PDC from S. cerevisiae (ScPDC) using analytical ultracentrifugation at pH 6.0 for the dimer–tetramer equilibrium . Extensive small-angle X-ray scattering studies on aqueous solutions of ScPDC, PDC from Kluyveromyces lactis (KlPDC) and PDC from pea seeds (Pisum sativum; PsPDC) at high protein concentrations have shown that significant tetramer dissociation is triggered at pH > 7.5. Dimers dominate at pH > 8 and significant amounts of monomers occur at pH > 9 [18–20]. pH-dependent oligomer dissociation is accompanied by the loss of cofactors. By contrast, tetramers are stabilised against rising pH by saturating concentrations of the cofactor up to pH 8.5.
In addition to dimers and tetramers, even octamers were found for PDC species from plant seeds. However, in 1990 Rosa et al.  reported a protein from the yeast Neurospora crassa with high sequence identity to PDC that forms cytoplasmatic filaments. In a subsequent study, the same authors demonstrated that the protein indeed acts as a PDC . However, the molecular mass of a typical filament or information about the molecular mass distribution for the filament population remains elusive.
Here, we present an improved purification procedure for wild-type PDC from N. crassa (NcPDC) and a steady-state kinetic analysis of the enzyme-catalysed reaction. In addition, the pH dependence of the structure of the purified enzyme is studied using analytic ultracentrifugation, small-angle X-ray solution scattering with synchrotron radiation and transmission electron microscopy – allowing us to establish structure–function relationships for this enzyme species.
A novel purification procedure was used based on the protocol of Rosa et al.  and consisted of two precipitation steps, four ultracentrifugation steps and one size-exclusion chromatography step. With the refined protocol, isolation of ∼ 15 mg homogeneous NcPDC from ∼ 80 g wet weight of N. crassa cells was possible (Fig. S1). The yield of NcPDC obtained using this procedure is approximately the same as the yields achieved with the established isolation protocols for the yeast PDC species ScPDC and KlPDC, respectively. The maximum catalytic activity observed in this study was 18 U·mg−1. Considering a molecular mass of 62 kDa for the NcPDC subunit, a kcat value of ∼ 18 s−1 per monomer is obtained. Compared with the corresponding values for other PDC species, this kcat value is rather low. For the wild-type forms of KlPDC, ScPDC and PsPDC values of 40–60 s−1 were determined [7,24–26].
The allosteric substrate activation of PDC species from yeasts and plants manifests itself in distinct lag phases in the progress curves of substrate conversion. Depending on the substrate concentration, these lag phases last several seconds and can be conveniently followed using the stopped-flow technique [24,27,28]. The extrapolated initial velocity is well below 5% of the steady-state velocity in the case of ScPDC and KlPDC. PDC species from plants possess significantly higher initial reaction velocities of catalysis (30% in case of PsPDC) . To date, Zymomonas mobilis PDC (ZmPDC) is the only species showing Michaelis–Menten steady-state kinetics. Correspondingly, lag phases are absent from the progress curves of ZmPDC at the seconds time scale. As illustrated in Fig. 1, the initial reaction velocity of NcPDC catalysis measured using the stopped-flow technique is of the same magnitude as the overall velocity (∼ 75%). Indeed, numerical differentiation of progress curves demonstrates a slight, but permanent, increase in the reaction rate over the whole reaction time (Fig. 1). However, a constant final reaction rate is not reached within the time window of our experiments. This makes determination of the steady-state rates difficult, which is why we used the initial rates of the progress curves measured using conventional spectrophotometers (Fig. S2) to determine the steady-state parameters. Kinetic measurements were performed in the pH range 5.0–7.0. Outside this range, NcPDC tends to form catalytically inactive aggregates. Strikingly, none of the obtained v/[S] plots shows any trace of sigmoidity (Fig. 2). Thus, an important hallmark of substrate activation is absent from the steady-state kinetics. By contrast, the v/[S] plots formally display negative cooperativity in the pH range 5–6. Hence, the Michaelis–Menten equation does not satisfyingly fit the steady-state data under these conditions. The deviation from hyperbolic behaviour vanishes progressively with increasing pH. This behaviour can be demonstrated by plotting the data according to Hanes . To evaluate our data we used Eqn (1), which consists of two independent Michaelis–Menten terms. This equation fits the empirical data well. A straightforward interpretation of this behaviour would imply the existence of at least two subspecies of NcPDC with differing KM and kcat values. In this case, KM1 and KM2 (Table 1) would refer to the KM values of different species. The same is true for the parameters a1 and a2, representing the ratios of the reaction rates of the subspecies to the overall reaction rate under saturation conditions. Thus, the sum (a1 + a2) is unity by definition. Such kinetic heterogeneity might well be linked to oligomeric equilibria. As shown below, structural heterogeneity is most pronounced at pH values between 5 and 6. It must be mentioned, however, that this ad hoc interpretation is by no means compelling because Eqn (1) is mathematically equivalent to Eqn (2), which is compatible with a vast number of alternative kinetic mechanisms.
Table 1. Kinetic parameters for the decarboxylation of pyruvate by NcPDC in dependence on pH. v/[S] data were fitted to Eqn (1). The S0.5 values were calculated according to and F = (KM1 + KM2) − 2·(a1·KM2 + a2·KM1) and A = KM1·KM2. kcat values were estimated from hyperbolic fits for all pH values. At pH ≥ 6.5 the Michaelis–Menten equation was used and S0.5 corresponds to KM. The Hill coefficients nH were drawn from the corresponding Hill plots. Slopes were taken in a substrate concentration range of 0.1–10 mm pyruvate.
a Buffer change from MES/NaOH to sodium phosphate.
The linearity of the data plotted according to Hanes  (plots of v/[S] versus [S], insets of Figs 2 and S3) accounts for the hyperbolic shape of the v/[S] plots at pH values > 6. Thus, steady-state data were evaluated according to Michaelis–Menten in the pH range 6.5–7. Table 1 summarizes the steady-state parameters obtained. The overall catalytic activity of NcPDC increases with decreasing pH, from 5.9 s−1 at pH 7.0 to 14.5 s−1 at pH 5.0.
The apparent affinity of the native substrate pyruvate for the enzyme NcPDC under conversion conditions decreases with increasing pH (Table 1). The lowest KM value obtained is 23 μm (double hyperbolic fit), which is well below the values found for other PDC species. However, around pH 6, the KM values are similar to those of other PDC species, 0.24 mm for KlPDC, 0.47 mm for ScPDC and 0.66 mm for PsPDC . With 216 mm−1·s−1, NcPDC shows the highest catalytic efficiency (kcat/S0.5) of all yeast PDC species. Only ZmPDC exhibits a higher value of 450 mm−1·s−1 .
The use of sodium phosphate instead of Mes/NaOH does not substantially alter the ionic strength of the buffer system. Nevertheless, steady-state parameters shifted significantly on changing the buffer, pointing to specific effects of the buffer species chosen. Interestingly, negative cooperativity is completely lost in phosphate at pH 6. For other yeast PDC species, phosphate acts as an allosteric inhibitor . As expected, no such effect is observed with NcPDC, because this species lacks substrate activation.
To date, protein crystal structure analysis of NcPDC has not been feasible. Although we succeeded in crystallizing the enzyme, the quality of the obtained crystals was not sufficient for data collection. Alternatively, we applied several methods to characterize the structure of the enzyme in aqueous solution, such as analytical ultracentrifugation, small-angle X-ray solution scattering, analytical size-exclusion chromatography and transmission electron microscopy. It was known from other studies that NcPDC forms filaments in the cytoplasm of N. crassa [22,23,32]. Therefore, NcPDC seems to be an essential and integral part of the cytoskeleton in this organism. The quaternary structure of the purified enzyme, however, was not analysed in earlier studies on this enzyme. Thus we subjected the purified material to several structural methods.
Our experiments using analytical size-exclusion chromatography qualitatively demonstrated the dominance of high-oligomer forms in the pH range 5.0–6.5. NcPDC eluted within the void volume of the column used under all conditions tested. At higher pH values the retention time of the enzyme corresponded to molecular masses of ∼ 220 kDa, indicating tetramers. Unfortunately, it was not possible to carry out any experiments at pH > 7.5, as explained below.
Accordingly, similar pH-dependent changes in the quaternary structure of NcPDC from high-oligomer forms at low pH to tetramers at pH 7.5 were detected using small-angle X-ray scattering. The scattering intensities at s-values close to zero decrease dramatically with increasing pH, reflecting a concomitant shift from higher to lower oligomeric states. This tendency is shown in Fig. 3 for the pH range 6.5–7.5. Estimated RG values > 5.0 nm at pH 6.5 are still well above values typical for PDC tetramers (3.9–4.0 nm) [18–20]. Molecular masses calculated from the ratio of the scattering intensities of aqueous solutions of NcPDC and BSA, respectively, point to hexamers and even to dodecamers. Even at pH 7.0, a certain tendency to form oligomer structures larger than tetramers is obvious from the scattering patterns at s < 0.2 nm−1 (Fig. 3). For pH values < 6.0, the observed scattering curves are very steep at low values of the scattering vector. Thus, the extrapolation of reliable I(0) values and, consequently, the calculation of molecular masses is not possible for pH < 6.0 (data not shown).
Analytical ultracentrifugation of NcPDC solutions yields sedimentation coefficients of 10.4 S at pH 7.5 and 12.6 S at pH 7.0 (Fig. S4). A sedimentation coefficient of 10 S is typical for the tetrameric state of PDC , whereas the higher value of 12.6 S at pH 7.0 points to a tetramer–octamer equilibrium. These results confirm the conclusions on pH-dependent oligomeric equilibria in NcPDC drawn from the other structural methods.
Finally, electron micrographs of NcPDC stained negatively with uranyl acetate clearly document the existence of bundles of filaments at pH 5.0 and 5.5 at magnifications of 30 000–85 000. Individual filaments are 120–200 nm long with diameters between 7 and 10 nm. A number of filaments is typically arranged side by side within the bundles. Strikingly, this type of superstructure is totally absent at pH 6.5 (Fig. 4).
It should be emphasized as a remarkable trait of NcPDC that the enzyme precipitated at pH values around 8 under all conditions tested. This process could not be prevented by the addition of excess ThDP. Even at lower pH, for example 6.8, no activity whatsoever could be recovered when starting with enzyme once exposed to pH > 8. This is in stark contrast to the behaviour observed for other yeast PDCs. These enzymes progressively lose activity at pH > 8.0 because of dissociation of the cofactor ThDP. However, the structures of their protein components are essentially stable up to pH 9.3 [7,25,26,33]. Thus, this loss of activity can be largely suppressed in the presence of excess ThDP. What is more, the activity of these species can be restored completely at pH values optimal for reconstitution (6.8 in the case of ScPDC). As a consequence, the deviating behaviour of NcPDC prevented any structural or kinetic characterization at pH values > 7.5.
The purified enzyme NcPDC is very sensitive to changes in the proton concentration in terms of quaternary structure and catalytic properties. Outside the pH range 5.0–7.5, the catalytic activity is completely lost and strong protein denaturation occurs. This outstanding dependence of structure and catalytic function on pH is unique among the PDC species studied to date, with a pH optimum for catalytic activity and stability in the range of 6.0–6.5. In contrast to NcPDC, other PDC species are structurally stable up to pH values around 9.3.
In many regards, NcPDC appears to be a very atypical yeast PDC. To the best of our knowledge, it is the first species from yeast that is not allosterically activated by its substrate pyruvate. Initial velocities around or above 75% of the steady-state velocities were determined throughout. Less than 5% are typical values for the allosterically activated species ScPDC and KlPDC, respectively, resulting in well-pronounced lag phases in their progress curves [4,12,34]. It should be pointed out that, unlike other yeast PDCs, the progress curves of NcPDC do not display well-defined lag phases on a time scale of seconds. This argues strongly against the existence of substrate activation in this species. Whether the very slow increase in activity seen over the observation window of 60–100 s represents a residual activation mechanism cannot be clarified at present. An alternative, and more likely, explanation of this phenomenon might consist in a substrate-triggered readjustment process between several oligomeric states. Significant initial velocities were found in progress curves for substrate conversion in case of PDC from P. sativum , an enzyme which also forms filamentous structures (Fig. S5), and α-keto acid decarboxylase from Mycobacterium tuberculosis . However, lag phases are still well defined for these species. Accordingly, the v/[S] plots for all the species mentioned above show distinct sigmoid-shaped profiles. This is not the case for NcPDC. The v/[S] plots obtained in our study display negative cooperativity and hence deviate from Michaelis–Menten kinetics in just the opposite way. Interestingly, the extent of this deviation decreases with increasing pH, as documented by the Hill coefficients in Table 1. At pH > 6, steady-state data conform closely to Michaelis–Menten kinetics. The finding of negative cooperativity might reflect the individual kinetic properties of the different oligomeric states of NcPDC that exist in equilibrium prior to substrate addition. This interpretation would imply that the kinetic heterogeneity seen in steady-state measurements mirrors the structural heterogeneity of highly oligomeric species also documented by electron microscopy. Kinetically, this model requires the existence of at least two enzyme subspecies, which differ in their respective KM values and whose interconversion proceeds significantly slower than catalysis. A concrete assignment of individual KM and kcat values to particular species is, however, not possible at this stage. Therefore, all kcat values given are weighted averages of the kcat values of individual oligomeric states. The exact distribution of these states is not known, however. As such, the negative cooperativity is an interesting finding and might be of regulatory significance. It must be mentioned that negative cooperativity has also been found for ScPDC at the very low pH of 4.5 . Although the activation behaviour of ScPDC in the pH range 4.5–5.5 is kinetically complex, progress curves obtained under these conditions can be explained using an extended activation model involving three species of different catalytic activity . Even under these conditions, the final steady state is attained after some 10 s. In the case of NcPDC, however, the reaction rate increases continuously over the observation window of some 100 s under all experimental conditions tested (Fig. 1). Hence, the molecular background of negative cooperativity in the case of NcPDC probably differs from that observed in ScPDC at extremely low pH values.
Within a very narrow pH range, the catalytic efficiency of NcPDC changes dramatically. It grows steadily from 2.7 mm−1·s−1 at pH 7.0 to 240 mm−1·s−1 at pH 5.0 without crossing an optimum. This catalytic efficiency is the highest ever determined for a PDC from yeast.
The insular position of NcPDC also manifests itself in its extraordinary molecular structure. Irrespective of the method used (analytical size-exclusion chromatography, analytical ultracentrifugation and small-angle X-ray solution scattering) high-oligomer structures were found at pH 5.0–6.5. Although the exact molecular masses of the species prevalent under these pH conditions remain elusive some conclusions can be drawn from electron microscopy studies. Rosa et al.  and Alvarez et al.  demonstrated the incorporation of NcPDC in cytoplasmic filaments using immunocytochemical methods. Applying transmission electron microscopy, we were able to show that purified NcPDC forms filaments with diameters of 7–10 nm and lengths of 120–200 nm at pH 5.0–6.0. Thus, the presence of other macromolecular species is not mandatory for the formation of NcPDC filaments. Given the dimensions of crystal and solution structure models of PDCs [15,17,20], a diameter of 7–10 nm corresponds well to the size of PDC tetramers. Hence, the smallest unit, visible as a darkly edged sphere in the electron micrographs may represent an NcPDC tetramer. Consequently, filaments typically contain 12–20 tetramers in total. This assumption is further strengthened by the fact that at the upper limit of the accessible pH range, i.e. pH 7.5, molecular masses of 220–240 kDa were determined for NcPDC solutions using analytical gel filtration, analytical ultracentrifugation and small-angle X-ray solution scattering. At pH 7.0 a mixture of mainly tetramers, and to a lesser extent, higher oligomers was found by analytical ultracentrifugation and small angle X-ray scattering. The scattering parameters at pH 6.5 clearly point to molecular masses of octamers and dodecamers. This distinct pH dependence of oligomer dissociation is unique among PDC species. At pH values > 7.0 other PDC species dissociate into dimers, > 8.5 even into monomers [18–20]. By contrast, neither of these oligomer forms was detected in the case of NcPDC. The tetramer is the typical oligomer state of other PDC species with the remarkable exception of PsPDC, which also forms filamentous structures at pH 6–6.5. This enzyme is the only other PDC species known to be active in a high oligomeric form . In contrast to other yeast PDC species, NcPDC tetramers display the lowest catalytic efficiency among all functional oligomeric forms of this enzyme. The catalytic efficiency obviously increases with the complexity of the NcPDC superstructure. The filament bundles visible in electron micrographs at pH 5.0 are the most efficient catalyst of pyruvate decarboxylation in N. crassa. Thus, the high-molecular filamentous structure of PDC in N. crassa cells is by no means merely a structural scaffold. Rather, it might be functional as a catalyst in vivo. Given its unusual structural and kinetic properties, NcPDC certainly deserves closer investigation.
Materials and methods
All chemicals were at least of analytical grade and purchased from AppliChem GmbH (Darmstadt, Germany), Carl Roth GmbH+ Co. KG (Karlsruhe, Germany), Sigma Aldrich Chemie GmbH (Hamburg, Germany), Merck KGAA (Darmstadt, Germany), VEB Laborchemie (Apolda, Germany), VWR International GmbH (Darmstadt, Germany) or Serva Electrophoresis GmbH (Heidelberg, Germany). Prepacked columns were from GE Healthcare (München, Germany) and the centrifugal concentrators were purchased from Viva Science Sartorius AG (Göttingen, Germany) group.
Cell cultivation and purification of NcPDC
Wild-type stems of N. crassa St. Lawrence 74-OR23-1A (987) and (988) were purchased from the Fungal Genetic Stock Center (FGSC, Kansas City, MO, USA). The fungus was kept on agar culture medium (3.5% (w/v) glucose, 1% (w/v) tryptone, 0.5% (w/v) yeast extract, 1.5% (w/v) agar). A single colony was transferred on the surface of the culture medium and was incubated 24 h at 30 °C. This stock could be stored at 4 °C for 4–6 weeks before recultivation. Vegetative cells were grown in the same medium without agar (6 × 1.5-L flasks), at 30 °C and 110 rpm for 48 h (preculture for 24 h). The mycelium was harvested by simple filtration through a large-pored cloth and washed 3–5 times with distilled water. The wet weight of the fungus amounts to 10–15 g cells·L−1 culture medium. For the elaboration of an improved enzyme preparation procedure, the basic protocol of Rosa et al.  was used as a starting point. The fresh cell material was suspended in 0.1 m sodium phosphate pH 6.0 (5% w/v glycerol, 1 mg phenylmethylsulfonyl flouride·L−1) and disrupted 10 × 30 s with 1-min interruptions for recooling in a bead beater (Biospec Products, Bartlesville, OK, USA) on ice. The cell debris was separated by centrifugation at 46 000 g for 20 min at 4 °C. NcPDC-containing protein was precipitated from the supernatant with 3.6% (w/v) poly(ethylene glycol) 6000, incubated for 1 h at 8 °C, and centrifuged at 142 000 g for 20 min at 4 °C. The pellet was resuspended in 0.1 m sodium phosphate, 0.4 m NaCl, pH 6.0. The high salt concentration induced the formation of NcPDC filaments. These filaments were precipitated by ultracentrifugation at 230 000 g for 25 min at 4 °C and suspended in a minimal volume of 25 mm Hepes/NaOH pH 7.5. The resuspension of the precipitated filaments in this buffer of very low ionic strength led to their disassembly. Nucleic acids were digested on addition of DNaseI (10 μg·mL−1) and 5 mm MgCl2. After 60-min incubation the solution was centrifuged again at 230 000 g for 20 min at 4 °C. For size-exclusion chromatography, the supernatant was loaded on a Superdex 200 column (26 × 600 mm) and eluted with 0.1 m sodium phosphate pH 6.0 at a flow rate of 1 mL·min−1. Fractions with the highest catalytic activity and homogeneity (as judged by SDS/PAGE) were merged. The purified enzyme was stored on ice.
Determination of protein concentration
Whenever possible (i.e. in the absence of excess ThDP in the samples) the protein concentration of aqueous enzyme solutions was determined spectrophotometrically from the UV spectrum at 280 nm (Jasco V-560 spectrophotometer, JASCO GmbH, Groß-Umstadt, Germany) using a calculated molar extinction coefficient of 70 800 m−1·cm−1 for one NcPDC subunit. Alternatively, the Bradford method  was applied using bovine serum albumin as standard protein.
The catalytic activity of NcPDC was measured on a Jasco V-560 spectrophotometer using the substrate pyruvate in a coupled optical test with yeast alcohol dehydrogenase (ScADH, 7 U·mL−1) and NADH as auxiliary enzyme system at 25 °C in 50 mm buffer (sodium phosphate or Mes/NaOH). Thus, the reaction product of the PDC catalysis, acetaldehyde, is reduced to ethanol. The decrease of NADH (0.3 mm initial concentration) was observed at 340 and 366 nm, respectively (εNADH 6220 and 3300 m−1·cm−1, respectively). To determine kinetic constants under different conditions (pH 5.0–7.5), the pyruvate concentration was varied from 0.01 to 40 mm. The corresponding v/[S] plots were treated according to the Michaelis–Menten equation or alternatively as a linear combination of two separate hyperbolic terms (Eqn 1). The program sigmaplot (Systat Software GmbH, Erkrath, Germany) was used for fitting the experimental data.
Stopped-flow measurements were performed to clarify whether the progress curves of NcPDC display distinct lag phases like other yeast PDC species. Lag phases are a hallmark of substrate activation in PDC species. The experimental conditions were similar to those used for steady-state kinetic measurements. One syringe contained 100 mm sodium phosphate pH 6.0, 14 U ScADH, 0.3 mm NADH, 0.1–40 mm pyruvate. The other syringe contained 10–20 μg NcPDC·mL−1 in the same buffer. The reaction was initiated by mixing the contents of both syringes in a volume ratio of 1 : 1. Time courses of the NcPDC catalysed reaction were recorded at a SX-18 MV spectrophotometer (Applied Photophysics Ltd., Leatherhead, UK). All measurements were carried out at 25 °C.
Analytical size-exclusion chromatography was used to determine the molecular mass of NcPDC in aqueous solutions as a function of the pH. A TSK G3000SW column (7.5 × 300 mm; TosoH Bioscience GmbH, Stuttgart, Germany) was equilibrated with the corresponding buffer – 0.1 m Mes/NaOH, sodium phosphate and Hepes/NaOH, respectively (pH 5–7.5). Samples of 100 μL at protein concentrations of 1–2 mg·mL−1 were applied on the column and eluted at a flow rate of 0.5 mL·min−1. For calibration, standard proteins with molecular masses in the range 29–669 kDa were used (protein calibration kit MW-GF-1000; Sigma).
Analytical ultracentrifugation was performed at 0.3 mg NcPDC per mL in 0.1 m sodium phosphate pH 6.0 and 6.5, and 0.1 m Hepes/NaOH pH 7.0 and 7.5, respectively, using an ultracentrifuge (XL-A; Beckman Instruments, Palo Alto, CA, USA) and a rotor (AN50Ti) with double sector cells. Samples were run at 116 500 g and 20 °C using the scanning absorbance optical system at 280 nm. The apparent sedimentation velocity was calculated according to a time derivative method using the software package provided by Beckman Instruments. Sedimentation velocity values of ScPDC were drawn from Killenberg-Jabs et al.  for comparison.
Small-angle X-ray solution scattering with synchrotron radiation
Data were collected at the EMBL beam line X33 at Hasylab, storage ring DORIS III, German Synchrotron Research Centre (DESY) at a camera length of 2.7 m and 10 °C using a MAR345 image plate detector (Marresearch GmbH, Norderstedt, Germany) and a 2D photon counting detector (Pilatus 300K-W; Dectris Ltd., Baden, Switzerland), respectively. The new vacuum sample cell allowed the reduction of sample volumes down to 50 μL. NcPDC solutions at concentrations of 1–2.6 mg protein·mL−1 and pH 4.5–7.5 (0.1 m sodium acetate, Mes/NaOH, and sodium phosphate, respectively, 5 mm dithioerythritol) were investigated. Buffer exchange and adjustment of protein concentration were carried out with centrifugal concentrators. The momentum transfer axis s (s = 4·π·sinθ/λ, where 2θ is the scattering angle and λ = 0.15 nm, the X-ray wavelength) was calibrated using collagen or tripalmitate as standards. Data were accumulated for 120 s in eight frames. Buffer solutions containing all components except the protein were measured before and after each run of a protein sample. Image file extraction during data collection, merging, intensity normalization for the transmitted flux and detector response, and scaling of the s-axis were realized using the data reduction program automar . The merged buffer scattering was subtracted from the protein scattering for each sample separately using the program primus-mar . The characteristic scattering constants I(0) (forward scattering intensity) and RG (radius of gyration) were calculated from the processed experimental data using the program gnom . The molecular mass of the enzyme samples in dependence on the experimental conditions (pH, protein concentration) was calculated from the ratios of the forward scattering intensities of the sample and of the molecular mass marker protein BSA.
Transmission electron microscopy
Enzyme solution (3 μL) was applied onto formvar-coated copper grids (diameter 3 mm). After 30 s adsorption, the excess of protein solution was sucked off. The air-dried grid was washed three times with 20 μL of distilled water. Negative staining of the protein was carried out with a 1% (w/v) aqueous uranyl acetate solution for 15 s. The excess of staining solution was removed carefully. The grids were observed with an EM900 transmission electron microscope (Zeiss SMT, Oberkochen, Germany) at an acceleration voltage of 80 kV. Electron micrographs were taken with a slow scan camera (Variospeed SSCCD camera SM-1k-120, TRS, Moorenweis, Germany).
The access to beam line X33 of the EMBL outstation at the storage ring Doris III at Hasylab c/o Desy Hamburg is acknowledged.