PPTases (phosphopantetheinyl transferases) are of great interest owing to their essential roles in activating fatty acid, polyketide and non-ribosomal peptide synthetase enzymes for both primary and secondary metabolism, as well as an increasing number of biotechnological applications. However, existing techniques for PPTase characterization and development are cumbersome and technically challenging. To address this, we have developed the indigoidine-synthesizing non-ribosomal peptide synthetase BpsA as a reporter for PPTase activity. Simple co-transformation allows rapid assessment of the ability of a PPTase candidate to activate BpsA in vivo. Kinetic parameters with respect to either CoA or BpsA as variable substrate can then be derived in vitro by continuously measuring the rate of indigoidine synthesis as the PPTase progressively converts BpsA from its apo into holo form. Subsequently, a competition assay, in which BpsA and purified carrier proteins compete for a limited pool of CoA, enables elucidation of kinetic parameters for a PPTase with those carrier proteins. We used this system to conduct a rapid characterization of three different PPTase enzymes: Sfp of Bacillus subtilis A.T.C.C.6633, PcpS of Pseudomonas aeruginosa PAO1, and the putative PPTase PP1183 of Ps. putida KT2440. We also demonstrate the utility of this system for discovery and characterization of PPTase inhibitors.
- non-ribosomal peptide synthetase
- phosphopantetheinyl transferase (PPTase)
- polyketide synthase
PPTases (phosphopantetheinyl transferases) catalyse the post-translational attachment of a 4′-PP (4′-phosphopantetheine) co-factor derived from CoA to the carrier proteins of NRPS (non-ribosomal peptide synthetase), PKS (polyketide sythetase) and FAS (fatty acid synthetase) enzymes [1–3]. This modification is essential for the activity of these synthetases, as the 4′-PP group is the site of substrate and intermediate linkage and allows transfer of intermediates between the spatially discrete domains and modules of these complex enzymes [4–7]. PPTases can be divided into three classes . The first class, the AcpS-type PPTases, have a monomer size of approximately 120 amino acids function as a homotrimer of discrete α/β-fold subunits and serve primarily to activate the ACPs (acyl carrier proteins) found in type II FAS [1,9–11]. The second class of PPTases are integrated into the α-subunit, and activate the ACPs, of fungal type I FAS enzymes . The third class, the Sfp-type PPTases, are monomers of approximately 240 amino acids and are composed of two AcpS-like sub-domains [13–15]. Sfp-type PPTases are frequently associated with secondary metabolite biosynthetic clusters, and serve to activate the ACPs, PCPs (peptidyl carrier proteins) and/or ArCPs (aryl carrier proteins) of NRPS and PKS enzymes contained therein [1,2,16]. Broad-specificity Sfp-type PPTases such as PcpS of Pseudomonas aeruginosa , PspT of Ps. syringae  and the family prototype Sfp of Bacillus subtilis [1,18] can activate a wide range of carrier proteins of both primary and secondary metabolic pathways. Their role in primary metabolism makes many broad-specificity Sfp-type PPTases essential for viability of the host organism, and several commentators have noted that such enzymes are promising targets for drug development [19–21]. In contrast, some Sfp-type PPTases such as EntD of the Escherichia coli enterobactin synthesis pathway are specialized and appear to act only on carrier proteins within a particular secondary metabolite cluster .
Neither the substrates nor the products of PPTase-catalysed 4′-PP attachment can be directly measured using simple spectrophotometry, and PPTase activity has traditionally been monitored by HPLC or radiolabelling [1,2,13,14,22,23], both of which are technically challenging and time consuming. More recently, FRET (fluorescence resonance energy transfer) [20,21] and fluorescence polarization  assays that utilize fluorophore-labelled CoA conjugates have been developed. These assays allow PPTase activity to be monitored spectrophotometrically and are amenable to high-throughput screening applications; for example, the FRET-based technique has been used to screen the LOPAC1280 compound library for inhibitors of B. subtilis Sfp . However, these approaches are also technically challenging and do not enable measurement of PPTase kinetic parameters with natural substrates or evaluation of the relative activity of different PPTase/carrier protein combinations. We present an alternative approach for evaluation of PPTase kinetic parameters and substrate specificity that is also compatible with high-throughput applications. Our approach does not employ artificial CoA conjugates, but rather uses the single module NRPS BpsA as a reporter. BpsA catalyses the conversion of two molecules of L-glutamine into indigoidine, a pigment that can be readily detected either in vivo or in vitro owing to its characteristic blue colour and strong absorbance at 590 nm . Similar to all NRPS, BpsA activity is dependent on PPTase-mediated activation from an apo to a holo form . As the endogenous PPTases of E. coli are ineffective in this role, recombinant (His6-tagged) BpsA purified from this host is found almost exclusively in the inactive, apo form . In the present study, we sought to design and implement a novel assay platform in which apo-BpsA acts as a reporter for PPTase activity both in vivo and in vitro. In vitro, this system enables derivation of kinetic parameters for a PPTase with respect to CoA or BpsA as a variable substrate by continuously measuring the rate of indigoidine synthesis by BpsA as it is progressively converted from apo into holo form by that PPTase. Importantly, we show that conducting this reaction in the presence of alternative purified carrier proteins that compete for a limited pool of CoA enables kcat/Km to be estimated for a candidate PPTase with respect to a wide variety of carrier protein substrates. Furthermore, we demonstrate that this simple and flexible assay system readily lends itself to chemical screening to identify and evaluate chemical inhibitors of a target PPTase.
Media and chemicals
LB (Luria–Bertani) broth and agar were used for routine growth and maintenance of all E. coli strains used in this study. Where necessary for plasmid selection and maintenance, media were supplemented with 50 μg/ml kanamycin and/or spectinomycin. For observation of indigoidine synthesis, in vivo strains expressing BpsA and an activating PPTase were streaked on to ZYP-5052/glutamine agar [10 g/l N-Z amines, 5 g/l yeast extract, 0.5% glycerol, 0.05% D-glucose, 0.2% α-lactose, 100 mM L-glutamine, 25 mM (NH4)2SO4, 50 mM KH2PO4, 50 mM Na2HPO4 and 1.5% (w/v) agar] and grown for 14–16 h at 37 °C. Plates were then placed at room temperature (21 °C) and incubated for a further 6–12 h, to facilitate indigoidine production.
To create the plasmids used in this study, target sequences were amplified by PCR using primers incorporating unique restriction sites to enable directional cloning. PPTases, pPCP (PCP domain of the first module of the Ps. aeruginosa NRPS PvdD) and bPCP (BpsA PCP domain) were ligated between the NdeI and HindIII sites of pET28a(+). BpsA was ligated between the BamHI and XhoI sites of pCDFDUET. mPCP (PCP domain of Microcystis aeruginosa NRPS/PKS MycG) was ligated between BamHI and XhoI sites of pET30b. npArCP (ArCP from the Nostoc punctiforme protein HetM) was ligated between NcoI and XhoI sites of pET30b. The identity of all the inserts was confirmed by sequencing.
Protein expression and purification
Expression of all proteins was conducted using E. coli BL21(DE3) as a host strain, in LB broth supplemented with plasmid-appropriate antibiotics. For PcpS, Sfp and PP1183, 500 ml expression cultures were inoculated to a D600 of 0.1 and grown (37 °C, 200 rev./min) until a D600 of 0.6–0.9 was reached. Cultures were then chilled in an 18 °C water bath for 20 min before induction by addition of IPTG (isopropyl β-D-thiogalactopyranoside; 0.5 mM final concentration). Expression was then allowed to proceed for 24 h (18 °C, 200 rev./min) before harvest by centrifugation (4000 g, 15 min, 4 °C). Following harvest, cell pellets were resuspended in binding buffer [5 mM imidazole, 0.5 M NaCl, 25% (v/v) glycerol and 20 mM Tris/HCl, pH 7.9] and lysed by sonication (10×10 s bursts at 10s intervals with cells maintained on ice using a Biologics model 150 V/T. sonicator) PPTases were then purified from the soluble fraction using standard Ni-NTA (Ni2+-nitrilotriacetate) chromatography (Novogen) at 4 °C except that 25% (v/v) glycerol was added to the wash and elution buffers. Collected fractions (1.5 ml) were then assessed for protein concentration using a Bio-Rad DC® protein quantification kit and the two fractions containing the highest concentration were pooled and immediately desalted using a GE Hi-Trap® desalting column. Buffer composition was then adjusted to 43% (v/v) glycerol and 25 mM Tris/HCl, pH 7.5, and aliquots were stored at −80 °C. Expression and purification of BpsA was conducted as outlined for PPTases, except that binding and elution buffers contained 12.5% (v/v) glycerol, the soluble fraction was halved and applied to two separate 1.5 ml His·Bind® columns, and these columns were washed with 45 ml of binding buffer each with the wash buffer step omitted. The first 6 ml eluted from each column in was pooled, and buffer was exchanged for 50 mM sodium phosphate buffer and 12.5% (v/v) glycerol, pH 7.8, using a 100 kDa molecular-mass cut-off Millipore Amicon® ultra-15 centrifugal filter unit. Buffer composition was then adjusted to 43% (v/v) glycerol and 50 mM sodium phosphate buffer, pH 7.8, and aliquots stored at −20 °C. Expression and purification of bPCP and pPCP was achieved as outlined for PcpS except binding, wash and elution buffer contained 12.5% (v/v) glycerol. Expression and purification of mPCP and npArCP was achieved as outlined for bPCP, except that expression was conducted at 30 °C for 8 h. Following buffer exchange, all protein preparations were assessed for purity by SDS/PAGE and quantified using a Bio-Rad DC® kit.
Kinetic analysis of BpsA
To bring about conversion of apo- into holo-BpsA, a pre-activation mix was established that contained 3.4 μM apo-BpsA, 0.25 μM PP1183, 10 mM MgCl2, 100 μM CoA and 50 mM buffer (sodium phosphate buffer, pH 7.8, or Tris/HCl, pH 8.0). This was incubated for 20 min at 30 °C. Triplicate 2-fold serial dilution series of L-glutamine were established in a final volume of 50 μl in a 96-well microplate. Reaction mix (100 μl of 5 mM ATP, 15 mM MgCl2, 8 mM L-glutamine and 75 mM sodium phosphate buffer, pH 7.8) was then added to each well. Reactions were initiated by addition of 50 μl of pre-activation mix to each well, followed by mixing at 1000 rev./min for 10 s. A590 measurements were then taken every 6–10 s for 10 min using an Envision® EnSpire™ microplate reader. The resulting data was visualized and velocities were derived using the slope function of Microsoft Excel. Kinetic parameters were derived from velocity values using GraphPad Prism®. For examination of the relationship between BpsA concentration and reaction velocity, a triplicate 2-fold serial dilution series of pre-activated BpsA was established in a 96-well microplate in a final volume of 50 μl, and 100 μl reaction mix without L-glutamine or ATP was then added to each well. Reactions were initiated by addition of 50 μl mixture of 8 mM L-glutamine and 5 mM ATP to each well, followed by mixing at 1000 rev./min for 10 s. A590 values were then recorded and data were analysed as above. Although not optimal for BpsA function (~61% maximum activity), pH 7.8 was chosen for kinetic analysis owing to the fact that the data generated from kinetic analysis of BpsA would ultimately be applied to analysis of PPTases, and PcpS  and Sfp  still retain some activity at this pH.
Kinetic analysis of PPTases
For examination of CoA as the variable substrate, triplicate 2-fold serial dilution series of CoA were established in a final volume of 50 μl in a 96-well microplate. Reaction mix (100 μl of 5 mM ATP, 20 mM MgCl2, 8 mM L-glutamine, 0.08–0.25 μM PPTase and 100 mM sodium phosphate buffer, pH 7.8) was then added to each well. Reactions were then initiated by addition of 50 μl of 3.33 μM apo-BpsA to each well followed by mixing at 1000 rev./min for 10 s. A590 values were then recorded as described above. For examination of BpsA as the variable substrate, triplicate 2-fold serial dilution series of BpsA were established in a final volume of 50 μl in a 96-well microplate. Reaction mix (100 μl of 5 mM ATP, 20 mM MgCl2, 8 mM L-glutamine and 100 mM Tris/HCl, pH 8.0) was then added to each well. Reactions were then initiated by addition of 50 μl of 0.32–1.0 μM PPTase to each well followed by mixing at 1000 rev./min for 10 s. A590 values were recorded as described above. For derivation of PPTase velocity values, the slope between every two to eight data points was measured across the entire data range using the slope function of Microsoft Excel with the resulting values forming a new data set (BpsA velocity). The slope between every two to eight points in the BpsA velocity data set was then determined using the slope function of Microsoft Excel and the maximum value from the resulting data set for each reaction taken as a measure of PPTase velocity. It was necessary to vary the number of points between which slope values were taken due to minor variations in reaction velocity. For fast reactions, smaller increments could be used; however, for slow reactions, plate reader background noise became a significant factor in point-to-point variation, yielding artificial fluctuations in measured slopes if too few points were used. PPTase reaction velocities were initially obtained with the units ΔA590/s2, these were converted into the standard units of amount of BpsA modified/s, which is the same as amount of CoA consumed/s, based on the established linear relationship between holo-BpsA concentration and reaction velocity.
Conversion of Vmax units for PPTases
Since the apparent rate of indigoidine synthesis catalysed by BpsA is directly proportional to the concentration of holo-BpsA in a reaction, the change in indigoidine synthesis rate over time, i.e. acceleration, is dependent on the rate of PPTase-mediated conversion of apo-BpsA into holo-BpsA. Acceleration of indigoidine synthesis is therefore a measure of the velocity of the PPTase reaction. Initially, maximum velocity values obtained using the BpsA coupled assay have the units A590/s2, or ΔV0, where V0 has units A590/s. To convert these values into [holo-BpsA] formed/s, we simply use the relationship between [E] and V0 as defined by the Michaelis–Menten equation: (1) where [E]=[holo-BpsA] and V0 is the apparent rate of indigoidine and has the units A590/s. Rearranging this gives: (2) since Km, Vmax and [S] are all known for the conditions employed in the assay, we can substitute these terms into the above equation and convert ΔV0 values into Δ[E] values.
Carrier protein competition assay
A 2-fold serial dilution series of carrier protein was established in a final volume of 50 μl in a 96-well microplate. An aliquot (50 μl) of a solution containing 3 μM apo-BpsA, 750 nM CoA, 40 mM MgCl2 and 400 mM Tris/HCl, pH 8.0, was added to each well. PPTase reactions were then initiated by addition of 50 μl of diluted PPTase solution per well [0.08–0.25 μM PPTase in ddH2O (double-distilled water)]. Plates were mixed (1000 rev./min, 30 s) and incubated for 15–30 min at 30 °C to allow PPTase reactions to proceed to completion. Subsequently, indigoidine synthesis was initiated by addition of 50 μl of a mixed solution of 4 mM L-glutamine and 4 mM ATP to each well, followed by mixing at 1000 rev./min for 10 s. A590 values were recorded as previously described. Maximum velocity values for indigoidine synthesis were derived using the slope function of Microsoft Excel and converted into percentage maximum velocity values relative to the fastest reaction recorded for a single triplicate experiment. For generation of IC50 curves for each PPTase/carrier protein combination, data from two independent triplicate experiments was pooled and four parameter dose–response curves were fitted using the non-linear regression function of GraphPad Prism®. IC50 values were converted into estimates of kcat/Km as described below.
Conversion of competition assay IC50 values into estimates of kcat/Km
The competition assays described consist of two phases. The first of these is the competition phase, in which BpsA and the subject carrier protein compete for a limited pool of CoA. The second phase is the measurement phase, in which a relative measure of holo-BpsA produced during the competition phase is determined. Since the apparent indigoidine synthesis rate in aqueous solution, as determined by change in A590, is directly proportional to [holo-BpsA], a relative measure of [holo-BpsA] can be achieved by measuring the rate of indigoidine synthesis upon addition of the L-glutamine and ATP. The amount of holo-BpsA produced is dependent on the average velocity of BpsA modification (VB) compared with the average velocity of carrier protein modification (VC) during the competition phase. The IC50 value for a given carrier protein is the concentration that results in a 50% reduction of indigoidine synthesis velocity, and therefore a 50% reduction in the amount of holo-BpsA formed during the competition phase of the assay. Therefore when [CP]=IC50 (where [CP] is concentration of carrier protein), VB=VC and 50% of the available CoA will be incorporated into BpsA, with the remaining 50% used in modification of the subject carrier protein. Since PPTases are known to obey the Michaelis–Menten model, the situation when [CP]=IC50 can be expressed as: Substituting in carrier protein and BpsA concentrations and cancelling the [PPTase] term gives: Substituting the known kcat and Km values with respect to BpsA for each PPTase results in a known constant (KBpsA) for each PPTase:
To allow derivation of an estimate, the assumption IC50≪KmC is made, which allows simplification to:
Assessment of PPTase inhibition by 6-NOBP (6-nitroso-1,2-benzopyrone)
Triplicate 2-fold serial dilution series (250–0.12 μM) of the previously identified PPTase inhibitor 6-NOBP were established in a final volume of 50 μl of 10% DMSO in a 96-well microplate. 100 μl of reaction mix (5 or 20 μM CoA, 5 mM ATP, 20 mM MgCl2, 8 mM L-glutamine, 1.66 μM apo-BpsA and 100 mM Tris/HCl, pH 7.8) were then added to each well. Reactions were initiated by addition of 50 μl 0.32–1.0 μM PPTase in water to each well followed by mixing at 1000 rev./min for 10 s. A590 values were then recorded and PPTase velocities were determined as previously described. For generation of IC50 curves, four-parameter dose–response curves were fitted using the non-linear regression function of GraphPad Prism®. The resulting IC50 values were then converted into Ki values using the Cheng–Prussoff equation. The data presented in Table 4 was generated using the equation for competitive inhibition, as this yielded the most consistent Ki values between data sets for each of the three PPTases.
Co-expression of BpsA and PPTase candidates
It has previously been shown that the endogenous PPTases of E. coli are not effective in activating the PCP-domain of BpsA and that co-expression of a cognate PPTase is necessary for production of the blue pigment indigoidine . This enabled us to rapidly test whether three candidate PPTases (Sfp of B. subtilis subsp. spizizenii, PcpS of Ps. aeruginosa PAO1 and the putative PPTase PP1183 of Ps. putida KT2440) were able to activate BpsA in vivo through co-expression from IPTG-regulated constructs in E. coli. PcpS has previously been well characterized both in vivo  and in vitro , but neither PP1183 nor Sfp from B. subtilis subsp. spizizienii have been directly studied, although the latter enzyme shares 96% amino acid identity with the extensively characterized Sfp of B. subtilis subsp. subtilis [1,15]. We observed that indigoidine exhibits low-level toxicity to E. coli, such that strains co-expressing BpsA and either Sfp, PcpS, or PP1183 exhibited severe growth retardation when sub-cultured on plates containing IPTG. To circumvent this problem, we developed and tested auto-induction agar plates, prepared by addition of 1.5% agar to Studier ZYP-5052 auto-induction medium (a liquid growth medium designed to induce expression from lac-based promoters only after the available glucose has been exhausted) . We reasoned that this medium would induce indigoidine production only at later stages of colony growth, and found that auto-induction agar plates enabled both rapid colony growth and efficient induction of pigment production following extended incubation at room temperature. Under these conditions, co-expression of BpsA with any of the three candidate PPTases resulted in substantial production of indigoidine in vivo (results not shown).
Kinetic analysis of BpsA
To facilitate development of more sensitive in vitro assays for PPTase evaluation, we tested whether synthesis of indigoidine by purified BpsA obeys the Michaelis–Menten model. Formation of the product indigoidine was monitored at 590 nm in a 96-well microtitre plate format for replicate reactions in real time. Consistent with previously published results , the catalytically active holo form was not detected in His6-tagged BpsA preparations purified from E. coli, as indicated by complete inability of the purified protein to synthesize indigoidine in vitro (results not shown). Addition of purified Sfp, PcpS or PP1183 together with CoA and Mg2+ was sufficient for conversion of apo- into holo-BpsA, as indicated by initiation of indigoidine synthesis in the presence of L-glutamine and ATP. MALDI–TOF (matrix-assisted laser-desorption ionization–time-of-flight) analysis of trypsin-digested BpsA which had been previously incubated with PP1183 in the presence or absence of CoA was also conducted. For the zero CoA condition, an ion corresponding to the predicted tryptic fragment containing the peptidyl carrier protein active site was present (expected mass 2524.240; observed mass 2524.237). By contrast, for reactions in which CoA was included, this peak was completely absent (not shown). These results indicate complete conversion from apo- into holoBpsA.
A pH profile of the relative initial reaction rates for phosphopantetheinylated holo-BpsA revealed a pH optimum between 8.5 and 8.8, with the enzyme retaining >50% of maximal activity within the range 7.5–9.0 (Figure 1A). We also noted that BpsA activity was sensitive to buffer type, with reactions established using Tris/HCl as the buffering agent displaying ~2-fold higher velocity than identical reactions established using sodium phosphate buffer at the same pH and concentration (results not shown). In these and all subsequent experiments, we consistently observed that the indigoidine concentration in vitro reached a peak value and then declined (e.g. Figure 2A). This appears to be a consequence of gradual precipitation of indigoidine in aqueous solution; removal of the supernatant followed by resuspension of the water-insoluble precipitate in an equal volume of DMSO yielded a deep blue solution with an absorbance maximum at 590 nm.
Analysis of the BpsA reaction rate over a replicate 2-fold serial dilution series (from 0.78 μM to 16 mM) of L-glutamine was conducted at both pH 7.8 (sodium phosphate buffer) and pH 8.0 (Tris/HCl). Non-linear regression analysis of the resulting maximum velocity values (apparent rate; ΔA590/s) revealed an excellent fit to the Michaelis–Menten equation (Figure 1B). Consequently, the apparent rate was expected to be directly proportional to the concentration of holo-BpsA present in a reaction mixture. This assumption was verified experimentally by monitoring the apparent rate at ten different concentrations of holo-BpsA (from 360 nM to 1.14 μM) at constant L-glutamine/ATP concentrations. As shown in Figure 1(C), the relationship between enzyme concentration and velocity is linear within this range (R2=0.989). Kinetic parameters for BpsA are presented in Table 1; these parameters were subsequently employed for the derivation of PPTase kinetics as outlined in subsequent sections. Maximum velocities are presented with the units A590/s, as the extinction coefficient at 590 nm for indigoidine in water is unknown.
Kinetic analysis of PPTases using a BpsA coupled assay
Having established a linear relationship between the concentration of holo-BpsA and the apparent rate of indigoidine synthesis in aqueous solution, we next tested whether it was possible to quantify PPTase activity using acceleration in indigoidine synthesis as a measure of the rate of 4′-PP attachment to apo-BpsA. Replicate solutions containing candidate PPTases and all of the substrates necessary for both phosphopantetheinylation and indigoidine synthesis were established across a 2-fold serial dilution series of CoA (0.024–50 μM) in 96-well microtitre plates. Reactions were initiated by addition of 0.83 μM apo-BpsA to each well and A590 was measured continuously. Under these conditions, the pigment synthesis reaction undergoes an initial acceleration phase as apo-BpsA is converted into holo-BpsA by the PPTase (Figure 2A). Assuming that maximal acceleration in indigoidine synthesis occurs at the point of maximal PPTase velocity (i.e. when conversion of BpsA into the holo form is occurring most rapidly), the slope of the velocity curve at this point represents the maximal velocity of the PPTase-catalysed reaction (Figure 2B). This data enabled Michaelis–Menten parameters to be derived for each candidate PPTase in this study (Figures 2C and 2D, Table 2) (see the Experimental section for a detailed analysis of the equations used to calculate these parameters). A similar analysis was also conducted for each PPTase using BpsA as a variable substrate at a fixed (25 μM) CoA concentration (Table 2).
Competition assay for estimation of PPTase kinetic parameters with alternative carrier protein substrates
The key limitation of the assay discussed above is that it only provides Michaelis–Menten parameters for a candidate PPTase with a single non-native carrier protein substrate (i.e. the PCP domain of BpsA). To generate more biologically relevant data, we developed a competition assay capable of providing a relative measure of kcat/Km for a given PPTase with any other carrier protein substrates capable of being purified. The assay consists of two phases. The first is the competition phase, in which BpsA and the subject PPTase are incubated with CoA and Mg2+ in the presence of different concentrations of a purified carrier protein. On a molar basis, 25% of CoA (250 nM) compared with BpsA (1 μM) was used, establishing a scenario in which BpsA and the added carrier protein competed for a limited pool of CoA. Thus, if the subject PPTase has a greater affinity for the added carrier protein than for BpsA, less BpsA will be converted into the active holo form. Following this incubation, the second (measurement) phase of the assay was initiated by addition of ATP and L-glutamine, and A590 was measured continuously in a microplate reader. Since indigoidine synthesis velocity is directly proportional to the concentration of holo-BpsA in a reaction, indigoidine synthesis velocities determined for a reaction during the measurement phase indicate the amount of holo-BpsA generated during the competition phase.
Four isolated carrier proteins were assessed in this assay: bPCP, pPCP, mPCP and npArCP. By establishing replicate competition reactions across a serial dilution series of carrier protein, it was possible to generate IC50 curves for individual carrier protein/PPTase combinations (Figure 3). IC50 values were derived for the four different carrier proteins with each of the three PPTases used in this study (Table 3). For a given PPTase/carrier protein combination, the derived IC50 value indicates the concentration of carrier protein required to compete for 50% of the available CoA. At this concentration, the average velocity of carrier protein modification is equal to the velocity of BpsA modification. The competition assay therefore provides an indirect but quantitative means for assessing velocity of carrier protein modification, a property which is dependent on kcat/Km for the PPTase–carrier protein interaction. In isolation, the derived IC50 values provide information about velocity of carrier protein modification relative to BpsA; however, these can be converted into absolute values using the previously determined kcat and Km values for each PPTase with BpsA as a variable substrate (Table 2). The derived kcat/Km for each PPTase–carrier protein combination is presented in Table 3 (see the Experimental section for a detailed analysis of the equations used to calculate these parameters).
Evaluation of PPTase inhibition by 6-NOBP
As noted previously, essential PPTase enzymes in pathogenic organisms are attractive targets for drug discovery. The ability of BpsA to act as a reporter for PPTase activity provides a convenient means to screen in vitro for novel inhibitors of any PPTase able to activate BpsA from apo to holo form. We demonstrated the potential of this screen using the previously identified Sfp inhibitor 6-NOBP . Each of our three PPTases was evaluated for inhibition across a 2-fold serial dilution series of 6-NOBP (250–0.12 μM) with CoA concentration fixed at either 10 or 2.5 μM. We found that all three PPTases were inhibited by this compound (Figure 4), with IC50 values ranging from 9.1 to 10.8 μM for the CoA concentration set at 10 μM; and 2.1 to 4.0 μM for the CoA concentration set at 2.5 μM (Table 4). We noted that pre-activated holo-BpsA was also somewhat inhibited by 6-NOBP, although inhibition of this NRPS in isolation was less than for the co-incubated PPTase/apo-BpsA reactions (IC50 value for holo-BpsA of 29.2 μM). As outlined in Table 4, conversion of IC50 values using the Cheng–Prusoff equation for competitive inhibition , and substituting the previously established Km values for each of the three PPTases with CoA, results in Ki values for 6-NOBP ranging from 0.4 to 5.2 μM.
The BpsA reporter assays outlined in the present study provide an array of biochemical tools for rapid, flexible and detailed investigation of PPTase enzymes. The underlying principle behind each of the assays is that the extent to which BpsA has been converted from the apo into the holo form by a candidate PPTase can be determined by measuring velocity of indigoidine synthesis, which we showed to be directly proportional to the concentration of holo-BpsA in a reaction (Figure 1C). The only requirement is that the candidate PPTase be capable of recognition and activation of the native PCP-domain of BpsA; this can easily be tested by simple co-expression of the PPTase gene with bpsA in E. coli.
As demonstrated using three PPTases as examples, our assays allowed for rapid and reproducible assessment of PPTase kinetic parameters with respect to both CoA and BpsA as variable substrates. The kcat and Km derived with BpsA as a variable substrate may then be used in conjunction with a CoA competition assay to derive estimates of kcat/Km for the same PPTase with different carrier protein substrates. In the present study, the Km values we measured for PcpS and Sfp with respect to CoA (3.79 and 0.62 μM respectively) were in reasonable agreement with values that have been previously determined using the traditional HPLC/radiolabelling method (1.1 and 0.7 μM respectively) [14,26]. However, the kcat values we determined (4.5 and 0.13 min−1 for PcpS and Sfp respectively) were substantially lower than those previously reported (169 and 102 min−1 respectively) with CoA as the variable substrate. This is most likely a consequence of our assays utilizing a non-native PCP rather than a native ACP substrate, at a much lower carrier protein concentration than that typically used in HPLC assays and at a pH that is not optimal for either Sfp or PcpS . It is worth noting that, when PcpS is examined with BpsA as a variable substrate in the presence of saturating CoA, the kcat derived for PcpS is considerably higher (41.2 min−1; Table 2).
The third enzyme characterized in this study, PP1183, has not previously been characterized in vitro. However, a previous heterologous expression study has shown that Ps. putida KT2440 possesses a broad-substrate specificity PPTase capable of activating myxobacterial carrier proteins, alongside bioinformatic analysis identifying PP1183 as the sole candidate PPTase in Ps. putida . We have now confirmed that PP1183 possesses PPTase activity both in vivo and in vitro. Under the conditions employed, PP1183 activated BpsA with a kcat/Km comparable with that of Sfp and approximately 5-fold lower than PcpS.
Although our measured kcat values for PcpS and Sfp were substantially lower than those obtained by HPLC/MS assays, this is not of great concern, as the turnover rates derived with BpsA as a substrate do not in isolation hold great significance for study of these enzymes. Importantly though, in the context of our carrier protein CoA competition assays, the kcat/Km derived for each PPTase with BpsA provides a powerful means for evaluating the relative efficacy of different PPTases with different carrier protein substrates. Our results from the present study indicate that each of the three PPTases tested have broad substrate specificity, with each being capable of recognition and activation of PCPs derived from Streptomyces lavendulae (bPCP), Ps. aeruginosa (pPCP) and M. aeruginosa (mPCP) NRPS enzymes as well as an aryl carrier protein from N. punctiforme (npArCP). The efficiency with which each PPTase activated the different carrier proteins, as indicated by the derived estimates of kcat/Km, varied widely. It was particularly interesting to note that, although generally more efficient than Sfp with PCP substrates, the two Pseudomonas PPTases (PcpS and PP1183) activated npArCP with comparatively low efficiency (kcat/Km of 0.65 and 0.19 min−1·μM−1 respectively), whereas Sfp activated this substrate with higher efficiency (kcat/Km of 3.51 min−1·μM−1) than any of the other carrier proteins with which it was tested. It was also interesting to note that the IC50 value derived for PP1183 with bPCP (0.30 μM) was markedly lower those that derived for Sfp and PcpS (~0.7 μM). As bPCP is the PCP of BpsA expressed and purified in isolation, this result suggests that PP1183 might activate isolated PCP domains more efficiently than those found in the context of a complete NRPS module.
We have also demonstrated the utility of our BpsA reporter for characterization of PPTase inhibitors, conducting a proof-of-principle study with the previously identified Sfp inhibitor 6-NOBP . Our assays enabled IC50 values for 6-NOBP-mediated inhibition of three candidate PPTases to be determined, with the previously established Km values for CoA allowing conversion of IC50 into Ki values. This flexible approach could readily be adapted to high-throughput screening of chemical libraries for inhibitors of a particular PPTase. PPTases are attractive targets for antibiotic development, owing to their central role in both primary and secondary bacterial metabolism [1,14], and studies aimed at uncovering inhibitors have already met with some success [20,21,31–33]. The methods we outline require only commonly available laboratory reagents, and standard His6-tagged affinity purification methods consistently yield between 40 and 60 mg of recombinant protein from a single 500 ml expression culture, enough enzyme for at least 20000 assay reactions in a standard 384-well microplate final volume of 50 μl. BpsA exhibits no loss of activity during 6 months storage at 4 °C and its activity is not impaired by DMSO concentrations up to 5%. These characteristics provide a basis for a robust and economical screening platform. Furthermore, the ability of the assays to rapidly measure PPTase activity at a variety of CoA concentrations means they could also be used to investigate efficacy (Ki) and mode of inhibition (competitive against non-competitive) for any lead compounds uncovered. BpsA assays could prove a useful adjunct to previously described FRET and fluorescence polarization high-throughput screening techniques [19,20,34] as they do not make use of artificial substrates, and therefore more closely mimic true physiological conditions. The fact that our assays do not depend on fluorophore–CoA conjugates means they could also be applied to the study of PPTases which do not tolerate such substrates in place of CoA. Our interest in this area is ongoing, and in other work we have demonstrated that PCP domain-swapping in conjunction with directed evolution can be used to generate recombinant BpsA constructs that are now recognised by PPTase enzymes which cannot phosphopantetheinylate the native PCP of BpsA (e.g. E. coli EntD; J.G. Owen, M.J. Calcott and D.F. Ackerley, unpublished work). Similar approaches could be used to generate reporter genes for chemical library screens targeting any PPTase enzyme, even those that do not recognize wild-type BpsA.
In addition to the applications we have demonstrated in the present study, we recognize that the CoA competition assay we outline may have other uses in developing PPTases for biotechnology. For example, it could be used for discovery of new tags for site-specific labelling of proteins. Existing technology allows for orthogonal labelling of two proteins with fluorescent CoA conjugates based on the differing efficiency of Sfp and AcpS catalysed modification of specific peptide or carrier protein fusion tags [35–38]. The techniques we have developed to rapidly assess PPTase specificity provide a facile means for uncovering new combinations of PPTase and carrier protein/peptide tag to expand the capacity for orthogonal labelling beyond two combinations. Our competition assay could also be used to identify PPTase enzymes that have a broader carrier protein specificity than Sfp, and these could be useful in a range of applications, including genome mining of natural product biosynthetic gene clusters by phage display as described by Yin et al. . The authors of this ground-breaking study demonstrate that Sfp can be used to biotinylate NRPS and PKS carrier protein domains expressed from randomly cloned genomic fragments fused to phage coat proteins, and that the genes encoding these domains can then be recovered by streptavidin binding. In a test study, 22 carrier protein clones were recovered from comprehensive libraries of the Myxococcus xanthus genome. This represents just under 20% of the 119 carrier protein domains that have been annotated in the genome sequence of this organism, and it is likely that alternative PPTases with broader carrier protein specificity than Sfp would enable more thorough mining of such libraries. However, a comprehensive evaluation of PPTase specificity for a wide range of carrier protein substrates has not been feasible using previous technologies.
Jeremy Owen co-designed the project, conducted all hands-on experimental work, and co-wrote the manuscript. Janine Copp assisted with project design, technical advice, carrier protein isolation and manuscript preparation. David Ackerley was the primary investigator and co-designed the project, offered technical advice, obtained all funding, and co-wrote the manuscript.
This work was supported by the Royal Society of New Zealand Marsden Fund [contract number VUW0901] and the Victoria University Research Fund.
We thank Jonathan Dunne and Dr Bill Jordan at the Victoria University of Wellington Centre for Biodiscovery for helpful advice regarding the MALDI-TOF analysis of BpsA; and Dr Alan Clark for stimulating discussion of the BpsA kinetic assays.
Abbreviations: ACP, acyl carrier protein; ArCP, aryl carrier protein; bPCP, BpsA peptidyl carrier protein domain; FAS, fatty acid synthetase; FRET, fluorescence resonance energy transfer; IPTG, isopropyl β-D-thiogalactopyranoside; LB, Luria–Bertani; mPCP, peptidyl carrier protein domain of Microcystis aeruginosa non-ribosomal peptide synthetase/polyketide sythetase MycG; 6-NOBP, 6-nitroso-1,2-benzopyrone; npArCP, ArCP from the Nostoc punctiforme protein HetM; NRPS, non-ribosomal peptide synthetase; 4′-PP, 4′-phosphopantetheine; PCP, peptidyl carrier protein; PKS, polyketide sythetase; pPCP, PCP domain of the first module of the Pseudomonas aeruginosa NRPS PvdD; PPTase, phosphopantetheinyl transferase
- © The Authors Journal compilation © 2011 Biochemical Society