Macrocycle peptides delineate locked-open inhibition mechanism for microorganism phosphoglycerate mutases

Glycolytic interconversion of phosphoglycerate isomers is catalysed in numerous pathogenic microorganisms by a cofactor-independent mutase (iPGM) structurally distinct from the mammalian cofactor-dependent (dPGM) isozyme. The iPGM active site dynamically assembles through substrate-triggered movement of phosphatase and transferase domains creating a solvent inaccessible cavity. Here we identify alternate ligand binding regions using nematode iPGM to select and enrich lariat-like ligands from an mRNA-display macrocyclic peptide library containing >1012 members. Functional analysis of the ligands, named ipglycermides, demonstrates sub-nanomolar inhibition of iPGM with complete selectivity over dPGM. The crystal structure of an iPGM macrocyclic peptide complex illuminated an allosteric, locked-open inhibition mechanism placing the cyclic peptide at the bi-domain interface. This binding mode aligns the pendant lariat cysteine thiolate for coordination with the iPGM transition metal ion cluster. The extended charged, hydrophilic binding surface interaction rationalizes the persistent challenges these enzymes have presented to small-molecule screening efforts highlighting the important roles of macrocyclic peptides in expanding chemical diversity for ligand discovery.

, error bars are defined as described in Supplementary Table 3.   , adapted here in 1536-well microtiter plate format for the PGM orthologs and isozymes used in this study.
Initial rate conditions determined from the continuous (kinetic) assay was used to calibrate an end-point bioluminescent assay format for a PGM profiling panel to evaluate the inhibitors developed in this project.  Fig. 3b). The initial rate (v i ) for each enzyme-substrate reaction was determined and plotted against molar substrate concentration to generate Michaelis-Menten curves, while the reciprocal of the rate and molar substrate concentrations were re-plotted as Lineweaver-Burk graphs in GraphPad Prism and apparent K M values were estimated for each respective enzyme.

1536-well format kinetic assay
1536-well format luminescence assay ATP generated from the pyruvate kinase (PK) catalyzed conversion of phosphoenolpyruvate (PEP) to pyruvate was utilized to configure a luminescence output for the PGM enzyme panel. Supplementary Table 1  The pyruvate kinase (PK) coupling enzyme was included as an additional specificity control in the enzyme panel.

Various concentrations of iPGM and dPGM enzymes listed in
PK concentrations of 0.3 units (~930 nM) or 0.15 units (~460 nM) were dispensed in a total volume of 4 µl of the above assay buffer into respective wells of 1536-well white/solid bottom plates as previously described. Two µl of PEP substrate solution prepared at an equivalent concentration to 3-PG substrate were added to the PK enzyme solution as described above, for a final assay concentration of 0.4 mM PEP and 3 mM ADP. The protocol for this assay profile panel is given in Supplementary Table 1.
SPPS cyclic peptides were tested under initial assay conditions described in Supplementary Table 1 designed to give robust and uniform signal to background across the 7 enzyme PGM panel. Concentration response curves (CRCs) were fit using a 4 (see Online methods) or 5 parameter Hill equation (below). Model selection was determined by an extra-sum-of-squares F test for each ortholog condition. The majority of cyclic peptides displayed hyperbolic responses; those with steep Hill slopes or requiring a 5 parameter fit 3 were revaluated at assay conditions employing lower iPGM concentrations.

parameter Hill equation:
Where S0 is the signal at zero concentration, Smax is the signal at infinite concentration, n is the Hill slope, LogEC 50 is the log of the concentration at half-maximal signal, X is the log of the concentration, and S is the asymmetry parameter. The capillary passed through a capacitively-coupled contactless conductivity detector (Tracedec) with the detection spot located approximately 2 cm from the sample reservoir. The buffer reservoir was filled with electrophoresis buffer (30 mM Tris-HCl pH 8.0, 20 mM MgCl 2 ), and enzyme reactions were mixed and run directly in the sample reservoir. The magnesium concentration in the electrophoresis buffer was chosen to optimize the resolution between the 2-PG and 3-PG signals 5 in the GEMBE electopherogram ( Fig. 1c and Supplementary Fig. 4).

C. elegans iPGM titration
The equilibrium ratio of 2-PG to 3-PG predicted from the standard free energy is approximately 1:7 at room temperature 6 . Consequently, with the GEMBE assay, the typical change in signal is approximately 11x larger for the reaction starting with 2-PG and converting to 3-PG than for the reaction starting with 3-PG and converting to 2-PG. For the cofactor independent enzymes (B. malayi iPGM, C. elegans iPGM, and E. coli iPGM), the mutase reaction was found to be reversible in the GEMBE assay. Therefore, reactions with those enzymes were run starting with pure 2-PG and monitoring the conversion of 2-PG to 3-PG to maximize signal. For the cofactor dependent enzyme, H. sapiens dPGM, the mutase reaction was found to be irreversible in the GEMBE assay, with much faster reaction rates found for the conversion of 3-PG to 2-PG. Reactions with that enzyme were therefore started with pure 3-PG and the conversion of 3-PG to 2-PG was monitored.
The analytical separation of the product and substrate was carried out as follows. The buffer reservoir pressure was maintained at 30 kPa between separations and during sample loading. Once a sample was loaded and the GEMBE separation initiated (as described below), the pressure was reduced to 20 kPa for 30 s with the high voltage off. The high voltage (+2 kV) was then turned on, and the pressure was further reduced to a starting pressure of between 750 Pa and 2500 Pa and held constant for approximately 14 s. Note that the results presented here were obtained over several months using different capillaries with nominally identical properties. Because of slight differences in the electroosmotic properties and inner diameter of the capillaries, the optimal starting pressure varied between sets of analyses. The pressure was then reduced at a rate of 12.5 Pa/s until both 2-PG and 3-PG had been detected (216 s to 240 s). The pressure was then increase to 20 kPa and held constant for 10 s. The high voltage was turned off, and the pressure was increased to 30 kPa for at least 30 s before the start of the next GEMBE separation. The GEMBE separation was Analysis of the GEMBE data and calculation of reaction rates was similar to that previously reported 7 . Briefly, the detector signal vs. time data for each electropherogram (see Supplementary Fig. 4a, for example) was fit to a functional form consisting of the sum of 3 complementary error functions and a quadratic baseline: The 3 error functions correspond to 2-PG, 3-PG, and an unknown species present in the enzyme stock solutions.
Calibration measurements indicated that the resulting best fit values for 2 and 3 were proportional to the concentration of the analytes, 2-PG and 3-PG, respectively. The percent conversion was then calculated from: The reaction rate was determined by the slope of a linear fit to the percent conversion vs. reaction time data for the first 4 GEMBE separations with each sample. The reaction rate was normalized by the rate of reaction from a no inhibitor control. Data was modeled to a four (see Online methods) or five parameter Hill equation for initial estimation of IC 50 s using Prism GraphPad (see above).
Size exclusion chromatography Samples were analyzed and fractionated on a Superdex 75 16/600 column using an AKTA Pure system. Samples, 500 µL, were eluted at 1 mL/min in buffer containing 30mM Tris, 150 mM NaCl and 2 mM  Table 4 and Fig. 17, left panel), molecular masses were verified by MALDI-TOF MS analysis (Supplementary Table 4 and Fig. 17, right panel), using a microflex or autoflex instrument (Bruker Daltonics).
Ring junction confirmed by MSMS spectrum and fragment analysis (Supplementary Fig. 18 (site1) and Asp 37, Ser 86, Asp 467 and His 468 (site2). When these sites were refined as Mg 2+ ions, residual positive electron density was observed suggesting that a larger metal ion is present. Modelling Mn 2+ ions at each site resulted in a B-factor of approximately 17 Å 2 at site1, but site2 contained some positive electron density and a B-factor of approximately 7 Å 2 . Additionally, the coordinating distances between the surrounding residues and the ion at site 2 were approximately 1.9 Å to 2.0 Å which are shorter than expected for a Mn 2+ ion (2.1 Å to 2.2 Å). Phased anomalous difference maps were calculated, using data collected at a wavelength of 1.0000 Å, which yielded peak heights of: site 1 (5.7σ) and site 2 (11.7σ) as shown in Supplementary Figure 12a. Zn 2+ ions may occupy site 2 based on the coordinating distances and anomalous signal, and homology of the iPGM phosphatase domain to that of E. coli alkaline phosphatase 8 .
The theoretical anomalous signal at a wavelength of 1.0000 Å, is 2.6 e and 1.4 e for Zn 2+ and Mn 2+ ions respectively.
Additionally, data were collected a wavelength of 1.9016 Å which is on the low energy side of the Mn 2+ absorption edge.
The theoretical anomalous signal at this wavelength is 1.1 e and 0.49 e for Zn 2+ and Mn 2+ ions respectively. A phased anomalous difference map was calculated using the 1.9016 Å data and produced no peaks at site 1 whereas peaks of approximately 5σ were observed at site 2. Subsequent refinement of Mn 2+ and Zn 2+ ions at site1 and site2 respectively yielded no residual positive electron density which further supported our assignments for these sites. The coordination of these metals by C. elegans iPGM is shown in Supplementary Figure 12b and the distances are listed in Supplementary   Table 6. It should be noted that an Mg 2+ binding site, from the crystallization solution, was also observed as shown in Supplementary Figure 12c.
C. elegans viability assay C. elegans strains N2 (wild type) was used for compound testing. Worms were handled using standard methods. They were cultivated on nematode growth medium (NGM) plates in a 20°C incubator and fed on living Escherichia coli strain OP50. The E. coli OP50 cells were precultured overnight at 37°C before spreading on the surface of NGM plates. Synchronous liquid culture of L1 C. elegans obtained through egg bleaching and individual L4 picked from NGM plates under the microscope were used for testing. Either 20 L1s or 1 L4 were placed in individual well of 96-well plate in 100uL of S medium supplied with dead E. coli HB101 as food source. Compounds were added into the well at different concentrations and incubated with worms at 20°C for up to 7 days. The effect of compounds on worm health was measured by food consumption, development and reproduction. Food consumption was measured by a decline in OD 600 nm readings monitored using a SpectraMax M5 microplate reader. Worm development and F1 progeny production were monitored visually under a microscope.