In vitro NTPase activity of highly purified Pdr5, a major yeast ABC multidrug transporter

The ABC transporter Pdr5 of S. cerevisiae is a key player of the PDR network that works as a first line of defense against a wide range of xenobiotic compounds. As the first discovered member of the family of asymmetric PDR ABC transporters, extensive studies have been carried out to elucidate the molecular mechanism of drug efflux and the details of the catalytic cycle. Pdr5 turned out to be an excellent model system to study functional and structural characteristics of asymmetric, uncoupled ABC transporters. However, to date studies have been limited to in vivo or plasma membrane systems, as it was not possible to isolate Pdr5 in a functional state. Here, we describe the solubilization and purification of Pdr5 to homogeneity in a functional state as confirmed by in vitro assays. The ATPase deficient Pdr5 E1036Q mutant was used as a control and proves that detergent-purified wild-type Pdr5 is functional resembling in its activity the one in its physiological environment. Finally, we show that the isolated active Pdr5 is monomeric in solution. Taken together, our results described in this study will enable a variety of functional investigations on Pdr5 required to determine molecular mechanism of this asymmetric ABC transporter.

ATP-binding cassette (ABC) transporters are ubiquitous, primary active membrane proteins that are found in all kingdoms of life 1 . In general, they can be divided into two classes, depending on the direction of transport: ABC importers (including ECF transporters) 2 and exporters 3 . A functional unit of an ABC transporter consists of two nucleotide-binding domains (NBDs) that bind and hydrolyze ATP to energize the transport cycle, and two transmembrane domains (TMDs), which form the translocation pathway across the membrane 1 . Overexpression of ABC transporters that export toxic compounds is part of a phenomenon known as multidrug resistance and represents a main obstacle in chemo-therapeutic cancer treatment as well as bacterial infections 4,5 . In fungi and plants the overexpression of these drug efflux pumps is part of the pleiotropic drug resistance (PDR) network 6 . All PDR ABC transporters are, with the exception of Adp1, full-size transporter and possess a reverse topology of (NBD-TMD) 2 7,8 . The ABC transporter Pdr5 of Saccharomyces cerevisiae has been established as a model for fungal PDR proteins and studied for more than 25 years 9 . It confers resistance towards a broad range of structurally and functionally different substrates including azoles, ionophores, antibiotics and many others 10,11 . However, the nature of the physiological substrate(s) is not known. The expression of PDR ABC transporters is regulated through a complex regulatory network of transcription factors, of which the zinc finger regulators Pdr1 and Pdr3 are mainly responsible for Pdr5 regulation 12 . A mutation in Pdr1 (pdr1-3) is used for constitutive overexpression of the transporter 13 , which can also be used to overexpress other ABC transporter using this system 14 .
ABC transporters all share conserved motifs within their NBDs, namely the Walker A and B, the signature motif (or C-loop) as well as the D-and H-loop 15 . Pdr5, however, features substitutions of key residues in each motif, except of the D-loops, in one of its nucleotide binding sites (NBSs), which renders this NBS ATPase deficient, also known as degenerated. Therefore, Pdr5 belongs to the family of asymmetric ABC transporters amongst CFTR, MRP1, ABCG5/ABCG8 and others [16][17][18] . It remains elusive what the exact physiological function of the degenerated NBS is, but it is obvious that there is an essential function as restoring the canonical motifs leads to inactive Pdr5 19 . There are however, indications that the degenerated NBS is involved in the interdomain crosstalk between the NBDs and the TMDs. Single mutations within the deviant NBS did not impact the overall ATPase activity of the transporter while they severely affected the transport functionality 19,20 . The crystal structure of TM287/288 21 nicely demonstrated the consequence of one catalytic and one non-canonical site, e. g. only a bound nucleotide (ATP) at the non-canonical site. However, one has to stress that a gradient of asymmetry exists in (2019) 9:7761 | https://doi.org/10.1038/s41598-019-44327-8 www.nature.com/scientificreports www.nature.com/scientificreports/ asymmetric ABC transporters. The number of mutations in the catalytically relevant motifs resulting in the inactivation of the corresponding nucleotide-binding site ranges from a few (e.g. two mutations in ABCG5/ABCG8 16 or three in CFTR 22 and TM287/288 21 ) to all motifs forming the NBS (Pdr5 and its homologue in Candida albicans Cdr1 23 ). This obviously raises the questions whether or not a relation between number of disrupted motifs and the molecular mechanism of substrate transport exists.
Pdr5 from Saccharomyces cerevisiae was the first identified member of the PDR subfamily of asymmetric ABC transporters 9 . Due to the medical significance of mammalian homologues and the agricultural importance of plant and other fungal homologues, the yeast PDR system serves as a unique model to investigate their molecular mechanisms.

Results
Isolation and purification of Pdr5 in a functional form. In order to establish the purification of Pdr5 in a functional state at high purity and yield, we screened 20 different detergents for protein solubilization. In the course of these experiments, it turned out that PCC-α-M was the most suitable detergent for solubilization as well as for subsequent affinity purification and size exclusion chromatography. Figure 1 shows three selected examples of size exclusion chromatograms of Pdr5 after affinity purification in the presence of DDM, C 12 E 8 and trans-PCC-α-M. The protein yield in the case of DDM and the purity after the two-step purification procedure 26 was sufficient. However, the inhomogeneity of the sample as evident from the shape of the elution peak (Fig. 1a) shows that DDM does not fulfill the requirements for further, functional analysis of Pdr5. Additionally, when Pdr5 purified with DDM was assayed for ATPase activity, no activity was detected above background levels, although earlier work showed low remaining activity in DDM solubilized membrane fractions 27 . Therefore, a detergent screen was performed, using the oligomycin sensitive ATPase activity of solubilized plasma membranes containing Pdr5 as an indicator 25,27,28 (not shown). Besides the initially promising results for DDM, C 12 E 8 extracts showed rather high ATPase activity. Unfortunately, the following SEC showed again an inhomogeneous elution peak (Fig. 1b), which ruled out further use of this detergent.
We therefore focused on other members of the class of maltosides and an only recently commercially available member of this detergent class, trans-PCC-α-M (trans-4-(trans-4′-propylcyclohexyl)cyclohexyl-α-d-maltoside) 29 was chosen to be tested in all purification steps. Starting with the solubilization, trans-PCC-α-M showed nearly identical results to DDM with respect to the yield after solubilization. The yield of the subsequent affinity purification was even higher as compared to DDM (up to 1 mg of protein per L cell culture compared to 0.3 mg/L cell culture). Figure 1c shows the SEC chromatograms performed with trans-PCC-α-M for wild-type Pdr5 as well as the E1036Q mutant that behaved identically during solubilization and purification. Most importantly and in clear contrast to other tested detergents, the purification protocol using trans-PCC-α-M resulted in a solubilized Pdr5 protein in a highly homogenous state, without aggregates at high purity as evident from the SDS-PAGE shown in Fig. 2a, as well as in the Western blots using anti-Penta-His (Fig. 2b) and polyclonal, anti-Pdr5 antibodies (Fig. 2c). These findings were supported by mass spectrometry (MS) analysis, which demonstrated that more than 90% of peptides observed in the MS spectra were derived from Pdr5 (not shown). Additionally, with trans-PCC-α-M, purified Pdr5 possessed a preserved ATPase active during all steps of the purification (Fig. 3). Thus, with respect to our objectives, trans-PCC-α-M turned out to be the detergent of choice.

Pdr5 is a monomer in solution.
Pdr5 is a full-size ABC transporter comprising two NBDs and two TMDs encoded on a single gene 13 . Previously, a low resolution structure (25 Å), obtained by electron microscopy, of Pdr5 in a lipid bilayer showed a square-like arrangement of particles that led to the conclusion that Pdr5 may form a dimer of two full-size transporters 26 . To assess whether Pdr5 is monomeric or forms higher oligomeric species in solution, multi-angle light scattering in combination with SEC (SEC-MALS) analysis was performed 30 . As Pdr5 is detergent solubilized, it was necessary to first determine the value of dn/dc as a measure of Pdr5 bound trans-PCC-α-M to be able to distinguish between the mass of the micelle and the protein. The dn/dc value is the specific refractive increment that corresponds to the changes in the refractive index in relation to the change in concentration of the investigated macromolecule and thereby enables to distinguish between the mass of the protein and the bound detergent 30 . Using a batch method with different detergent concentrations 30,31 we . Pdr5-specific ATPase activity after each purification step. After solubilization, oligomycin-sensitive ATPase was measured (grey bar: overall activity, white bars: Pdr5-specific activity). The error bars represent the data of three independent measurements with the standard deviation reported as error.
www.nature.com/scientificreports www.nature.com/scientificreports/ obtained a value of dn/dc = 0.1392 ± 0.001 ml/g. As shown in Fig. 4, this resulted in a calculated protein mass of M W = 178.2 ± 0.2 kDa, close to the theoretical mass of 173.1 kDa, including the histidine tag and protease cleavage site, demonstrating that Pdr5 is indeed a monomer in solution.
NTPase activity of purified Pdr5. NTPase activity of purified wild type Pdr5 was performed by a colorimetric NTPase activity assay as described in detail by Baykov et al. 32 . The E1036Q Pdr5 mutant, which does not exhibit any significant ATPase activity 25 , purified by the same protocol, was used as a negative control. Plasma membrane embedded Pdr5 exhibits the highest ATPase activity at pH 9.5 25 . In order to validate whether this is the case for detergent purified Pdr5, we analyzed the pH dependency of the ATPase activity of Pdr5 in detergent solution. As summarized in Fig. 5, the activity has a broad range with a maximum at pH 9.5. Figure 6 shows the NTPase activity of the WT and E1036Q Pdr5 at different NTP concentrations. At saturating substrate concentrations the measured ATPase activity of WT Pdr5 was 208.5 ± 6.3 nmol/(min*mg) with a K m of 0.44 ± 0.05 mM while the E1036Q mutant did not exhibit any significant activity above background (see Figure 6a and Table 1). In the context of Golin et al. 33 , it was of interest to analyze the efficiency of other NTPs as it is suggested that the ABC transporter Pdr5 can utilize other nucleotides to fuel its transport activity. As summarized in Fig. 6 and Table 1, Pdr5 exhibits significant NTPase activity for all nucleotides tested, with the lowest efficiency for UTP and GTP (v max /K m of 0.0001 L/mg/min) followed by CTP (v max /K m of 0.0002 L/mg/min), and with the highest efficiency for ATP (v max /K m of 0.0005 L/mg/min). Although the v max of the CTPase is higher than the ATPase activity, only the latter displays a K m that is in respect to NTP concentrations within the physiological range (intracellular NTP levels: ATP: 1.51 ± 0.32 mM, GTP: 0.30 ± 0.02 mM, CTP: 0.21 ± 0.03 mM, UTP: 0.33 ± 0.08 mM 34 ) that allows the transporter to work always at saturating levels of nucleotide. The E1036Q mutation in the Walker B motif of NBD2 led to no significant activity in any of the performed assays (Figs 5 and 6), which demonstrates on the one hand that no detectable impurities are present and more importantly that in fact the degenerated NBS is ATPase  www.nature.com/scientificreports www.nature.com/scientificreports/ deficient as shown earlier 25 . As proposed by Golin et al. 33 , we show here that GTP can be used as an energy source for Pdr5. However, with the K m being above intracellular GTP levels, the significance of this nucleotide for active drug efflux under physiological conditions remains unclear.
Inhibitory effects of substrates and inhibitors on the ATPase activity of purified Pdr5. For plasma membrane vesicles it has been previously shown that Pdr5 is a strictly uncoupled transporter 25 , as none of the known substrates of Pdr5 were able to stimulate its ATPase activity. However, some substrates as well as inhibitors are able to reduce the ATPase activity up to complete inhibition 25 . This characteristic allows to validate whether detergent solubilized and purified Pdr5 shows the same or similar characteristics as embedded in its physiological environment, the lipid bilayer. This also opens up a suitable way to compare the functional properties of the transporter in solution with the one embedded in the membrane. We therefore measured the effect of several substrates and inhibitors on its ATPase activity. As shown in Fig. 7, all tested inhibitors and substrates are able to inhibit the purified Pdr5 transporter comparable to what was observed in plasma membranes. The respective K i of each effector-compound is summarized in Tables 2 and 3, respectively. Interestingly, in solution the concentration necessary to inhibit the ATPase activity of purified Pdr5 by its substrates rhodamine 6 G and ketoconazole is roughly 5 times higher than determined for Pdr5 in plasma membrane vesicles (Table 2). Moreover, for inhibitors the corresponding values are even up to 300-fold increased ( Table 3). All tested Pdr5 substrates and inhibitors are hydrophobic molecules and have similar partition coefficients to ketoconazole 35 (Tables 1 and 2), preferring the hydrophobic nature of a biological membrane. Thus, the local concentrations of the drugs in the  Table 1 www.nature.com/scientificreports www.nature.com/scientificreports/ membrane are several orders higher than in solution. Therefore, the increase in K i for purified Pdr5 compared to the membrane system is not surprising. Additionally, these values may reflect the more flexible environment for the protein within a detergent micelle compared to a lipid bilayer. Other tested substrates like cycloheximide did not inhibit the ATPase activity (not shown), similar to what was observed for Pdr5 in plasma membranes 25 . Together, the results of the Pdr5-ATPase inhibition assay confirm that detergent purified Pdr5 shows comparable properties and characteristics as in plasma membranes.
The asymmetric ABC transporter Pdr5 does not exhibit adenylate kinase activity. We tested whether Pdr5 possesses an adenylate kinase (AK) activity, as it was reported for other asymmetric ABC transporters like CFTR and TmrAB 36,37 . Therefore, the formation of ADP through the AK catalyzed reaction of ATP + AMP → 2ADP was assayed as a function of NADH consumption in the enzyme coupled ATPase assay. In the case of an AK activity, addition of AMP would increase the ADP concentration additionally to the formation   www.nature.com/scientificreports www.nature.com/scientificreports/ through the ATPase activity of the protein. The method used in this study was first described for CFTR 37 . Since Pdr5 exhibits a high basal ATPase activity that is not further stimulated by its substrates, as described in this study, other methods of assaying AK activity are in our opinion unsuitable. However, as seen in Fig. 8, the addition of AMP to the reaction did not increase the formation of ADP. Conversely, it led to a slight decrease in NADH consumption, which indicates a minor inhibition of the ATPase activity of Pdr5 with a maximum inhibition of about 20% at 2 mM AMP. We therefore can rule out that Pdr5 exhibits an adenylate kinase activity that was shown for other ABC transporter 36 .

Discussion
Pdr5 is the major ABC protein in the PDR network of S. cerevisiae 9,38 conferring resistance to many unrelated exogenous drugs or xenobiotics 12 . Although extensively studied for more than 25 years 9 , the molecular mechanism of the efflux pump remains mostly elusive, as it was not possible to investigate it in an isolated and functional state. This is most certainly due to the fact that mainly the detergent solubilization process causes an inactivation of the protein, leading to inactive protein after purification, as demonstrated in our present study. Previously, numerous mutational studies have provided important insights in residues involved in drug binding, interdomain crosstalk as well as the molecular diode function [38][39][40][41][42][43] of the unidirectional substrate transport of Pdr5. However in order to disclose at a molecular level how these mutations affect the different steps in the pump cycle, it is necessary to isolate the ABC transporter for these mechanistic studies in a functional state and, as a final goal, to determine the 3D-structure.
In plasma membranes vesicles, it was shown that Pdr5 exhibits a high basal ATPase activity over a broad pH range with a maximum at pH 9.5 that could not be further stimulated in the presence of substrates. However, some of its substrates inhibit the ATPase activity at high concentration, possibly due to non-competitive inhibition 25,27,33 . We tested whether purified Pdr5 shows similar behavior towards the well-studied substrates ketoconazole and rhodamine 6G and two inhibitors (oligomycin 11 and beauvericin 44 ) and could demonstrate that, indeed, the substrates and inhibitors affect the ATPase activity in the same way as it was shown for Pdr5 in plasma membranes. The measured K i s however differed in magnitudes, especially for the inhibitors. This finding however is not surprising, given the hydrophobic nature of the substances tested, which show a high preference for partitioning into the membrane as highlighted by the partition coefficients of these compounds. For beauvericin it was shown that only nanomolar concentrations are required to inhibit the ATPase and transport activity in plasma membranes, while in whole cell assays the concentration necessary to observe an impact on the Pdr5-mediated drug efflux was within the micromolar range 44 . This indicates that the drug cannot freely partition in the membrane presumably due to polar groups on the cell surface and/or the cell wall, which are not present in isolated plasma membranes. Therefore, higher concentrations were required to obtain inhibitory effects on Pdr5 compared to isolated plasma membranes. A similar effect is likely observed in our detergent-solubilized system, where such an increased drug accumulation within the protein-detergent micelle does not occur. Moreover, oligomycin is a known inhibitor of the Pdr5 ATPase activity 11 as well as of ATP synthase. Here, it binds to the membrane embedded F O subunit, which was shown by cryo-EM structure analysis 45 . Additionally, this effect is not specific to Pdr5. It was shown for the well-studied ABC transporter P-gp that EC 50 values for some hydrophobic compounds like verapamil with a logP value of 4.55 are well-above 300 times higher in a detergent system compared to P-gp embedded in a lipid bilayer 46 .
Pdr5 exhibits significant NTPase activity for all NTPs as shown in this study. Although ATP shows the highest efficiency with a K m below physiological ATP concentrations, we cannot rule out that the transporter might utilize other NTPs. In the light of the work of Golin et al. 33 , in which it was shown that substrate inhibition for GTP fueled transport takes place at higher concentrations compared to ATP mediated efflux and the 'kinetic substrate selection model' proposed by Ernst et al. 47 it can be speculated that the NTPs are utilized depending on the present substrate and its concentration as they exhibit different kinetic parameters. Additionally, it was shown for the multidrug transporter PatA/PatB that GTP is actually the preferred nucleotide to energize the translocation of its substrates compared to ATP 48 . This indicates that the choice of nucleotide might depend on several factors and it is not necessarily ATP inducing NBD dimerization and being hydrolyzed to energize the substrate efflux of ABC transporters, as it was also proposed in a study of a NBD from Methanocaldococcus jannaschii 49 . It is worth mentioning that the reported basal ATPase activity of Pdr5 in plasma membranes is in some cases up to 2 µmol/ Figure 8. Test for adenylate kinase activity of Pdr5 as described in 37 . The relative activity in percent corresponds to the formation of ADP as a function of NADH consumption in an enzyme coupled assay. An increasing amount of AMP at saturating ATP concentration (4 mM) was added and in case of adenylate kinase activity, an increase in activity should be observerd. Error bars represent 3 independent measurements. www.nature.com/scientificreports www.nature.com/scientificreports/ min/mg 25 . However, it is important to note that the catalytic efficiency (v max /K m ) of Pdr5 for ATP in detergent is 0.0005 and in the case of plasma membrane bound Pdr5 0.0008 L/mg/min (values taken from 19 ) which is a factor of 1.6, as the v max and K m are both lower compared to plasma membrane preparations. This reflects a v max /K m compensation. Additionally, the reported activity of Pdr5 in plasma membranes varies a lot as Golin et al. report an activity of about 150 nmol/mg/min 33 , which is within the range of activity reported in this study.The above described findings need further support from studies in a reconstituted system. However, it was not possible so far for us to reproducibly measure rhodamine 6G transport in Pdr5 containing proteoliposomes.
The low resolution structure of Pdr5 suggested that Pdr5 might function as a dimer in the membrane 26 . As Pdr5 is a full-size ABC transporter consisting of two NBDs and two TMDs within a single molecule 8 , these findings were quite surprising. However, as the author state in their study, dimerization was only observed upon detergent removal and no activity of the purified or reconstituted protein was reported 26 . It was shown for other ABC transporters like ABCA1 as well as the peroxisomal ABC transporters ABCD1 and ABCD2 that these proteins can actually form higher oligomeric functional units 50,51 . Nevertheless, for Pdr5 it is assumed that the functional unit is a monomer as shown for other full-size ABC transporters 52 . In our experiments, we used SEC-MALS analysis 30 of the purified functional Pdr5 in trans-PCC-α-M micelles to gain information of the oligomeric state of the protein in solution. Size-exclusion chromatography, coupled with multiangle light scattering, is a most widely used technique for determining the absolute molecular weight distribution and averaged molecular weights of proteins and natural polymers 53 . Here, we show with in vitro assays that we purified Pdr5 in a functional state and by SEC-MALS analysis that the functional, purified Pdr5 is monomeric in solution.
In summary, the above described purification protocol for Pdr5 in a monomeric physiological competent, high active form will allow further detailed functional and structural investigations on the molecular steps during the transport cycles of Pdr5. Additionally, the presented data and purification approach could be extended to the Pdr5 homolog Cdr1 of the clinically important pathogen C. albicans. Initial studies on Cdr1 purified with Triton X-100 resulted in rather low activity and might be improved using our protocol 54 .

Methods
Growth media and chemicals. Yeast strains were cultured in YPD medium (20 g/l tryptone/peptone, 10 g/l yeast extract and 20 g/l glucose). All chemicals, if not stated otherwise, were obtained from Carl Roth or Sigma-Aldrich. trans-PCC-α-M was purchased from Glycon Biochemicals. anti-Pdr5 antibody was purchased from Davids Biotechnology, anti-Penta-His antibody was obtained from Qiagen.
All steps of the membrane isolation were performed at 4 °C. Cells were resuspended in 50 mM Tris-HCl pH 8.0, 5 mM EDTA and two protease inhibitor tablets (EDTA-free, ROCHE). Lysis of the cells was performed with glass beads. The suspension was centrifuged twice at 1000 × g for 5 min (4 °C) and once at 3000 × g for 5 min (4 °C) to remove cell debris and the supernatant was centrifuged at 20,000 × g for 40 min (4 °C). The resulting pellet was resuspended in buffer A (50 mM Tris-HCl pH 7.8, 50 mM NaCl, 10% (w/v) glycerol) and adjusted to 10 mg/ml overall protein concentration. Solubilization of the membrane proteins was carried out with 1% (w/v) trans-PCC-α-M for 1 h under gentle stirring at 4 °C.
Affinity purification and size exclusion chromatography. The immobilized metal ion affinity chromatography (IMAC) of N-terminal 14x histidine tagged Pdr5 was performed as described in 14 . In short, solubilized proteins were separated from the non-solubilized fraction by ultracentrifugation at 170,000 × g for 45 min at 4 °C. A 1 ml HiTrap Chelating column loaded with Zn 2+ ions was equilibrated using low histidine buffer (50 mM Tris-HCl pH 7.8, 500 mM NaCl, 10% glycerol, 2.5 mM l-histidine, 0.003% (w/v) trans-PCC-α-M). Subsequently, the sample was loaded on the column, washing and elution was performed by a step gradient using low and high histidine buffer (50 mM Tris-HCl pH 7.8, 500 mM NaCl, 10% glycerol, 100 mM l-histidine, 0.003% (w/v) trans-PCC-α-M).
For size exclusion chromatography, the elution fractions were pooled and concentrated using a Vivaspin 6 (Sartorius) centrifugal concentrator (100 kDa MWCO). The size exclusion chromatography was performed on a Superdex 200 10/300 GL column (GE Healthcare) equilibrated with buffer A containing 0.003% (w/v) trans-PCC-α-M. Both purification steps were carried out on the Äkta protein purification systems (GE Healthcare).

SEC-MALS analysis.
To investigate the oligomeric state of purified Pdr5, multi-angle light scattering in combination with size exclusion chromatography (SEC-MALS) analysis was performed on an Agilent 1260 HPLC System. A triple-angle light scatter detector in combination with a differential refractive index detector (miniDAWN TREOS and Optilab rEX, respectively (Wyatt Technology Europe) were used and data analyzed with Astra 4 Software (Wyatt Technology Europe).
Released inorganic phosphate was determined by a colorimetric assay using Na 2 HPO 4 as a standard 55 .
ATPase inhibition assay. The inhibition assays were performed as the NTPase activity assays described above. The compounds were dissolved in an appropriate solvent (water, methanol or DMSO) and mixed with 750 mM Tris-glycine buffer to reach the desired stock concentrations. Purified Pdr5 was incubated with each substance for 5 min before the reaction was started by addition of 4 mM ATP. Detection of released inorganic phosphate was performed as described for the NTPase activity assay. The non-inhibited ATPase activity was set to 100%. The K i was determined using Equation (1) as described in 25 .
i Here, v corresponds to the relative ATPase activity, K i to the inhibitory constant in mol/L constant and [drug] to the drug concentration in mol/L. logP values of the tested compounds were calculated with the Chem3D program (Perkin Elmer) based on Molecular Networks' chemoinformatics platform MOSES (https://www.mn-am.com/moses).
Adenylate kinase activity assay. Adenylate kinase activity assay was performed as described in 37 . In brief, an enzyme coupled ATPase assay was performed in a 96-well plate at 30 °C and measured in a Tecan Infinite 200 PRO reader (Tecan). The reaction volume of 200 µl was composed of 300 mM Tris-glycine buffer pH 8, 0.004% (w/v) trans-PCC-α-M, 5 mM MgCl 2 , 4 mM PEP, 0.6 mM NADH, 3.5 µl pyruvate kinase/lactic dehydrogenase (PK/LDH) (Sigma-Aldrich). The reaction was started by the addition of 0-4 mM ATP. For detection of adenylate kinase activity, 0-2 mM AMP were added at saturating concentrations of ATP (4 mM). The absorbance of NADH was detected at 340 nm for 20 minutes. The decrease in NADH absorbance is proportional to the increase in ADP.