Small molecule inhibitors of disulfide bond formation

by the bacterial DsbA-DsbB dual enzyme system

Maria A. Halili1*, Prabhakar Bachu1*, Fredrik Lindahl1*, Chérine Bechara2, Biswaranjan Mohanty3, Robert C. Reid1, Martin J Scanlon3, Carol Robinson2, David P. Fairlie1#, Jennifer L. Martin1#

1Division of Chemistry and Structural Biology, Institute for Molecular Bioscience, The University of

Queensland, Brisbane, Queensland 4072, Australia. 2Department of Chemistry, South Parks Road,

Oxford OX1 3QZ, United Kingdom 3Faculty of Pharmacy and Pharmaceutical Sciences, Monash

Institute of Pharmaceutical Sciences, Monash University (Parkville Campus) 381 Royal Parade,

Parkville Victoria 3052, Australia

* contributed equally

#co-corresponding authors: [email protected], [email protected]

1 ABSTRACT (~150 words)

The bacterial enzymes DsbA and DsbB together mediate introduction of stabilizing disulfide bonds into bacterial virulence factors, and may be novel targets for antivirulence drugs. Here, we report the discovery of synthetic analogues of ubiquinone (dimedone derivatives) that inhibit disulfide bond

formation (IC50 ~1 μM) catalysed by E. coli DsbA:DsbB, a dual enzyme redox system. The inhibition mechanism involves covalent modification of a free cysteine in each enzyme, revealed by mass spectra for each complex and supported by NMR spectra for reduced E. coli DsbA. A vinologous anhydride in each inhibitor is cleaved by the thiol, which becomes covalently modified to a thioamide by a propionyl substituent. Inhibition of bacterial DsbA:DsbB catalyzed disulfide bond formation, together with its importance in regulating virulence, supports this enzyme system as a viable target for antimicrobial agents. Inhibiting bacterial virulence rather than growth may slow development of resistance to antibacterial drugs.

I think TOC should be modified to show the key finding, that the thiol attacks the carbonyl. Maybe just show an arrow from the SH to the carbonyl.

2 Introduction

The development of bacterial resistance to antibiotics is a global concern, and has been described as a looming public health crisis (1, 2). Multi-drug resistant strains of Gram-negative bacteria, such as

Escherichia coli, Klebsiella pneumoniae and Pseudomonas aeruginosa, as well as Gram-positive bacteria such as Staphylococcus aureus, have proven to be difficult to treat with current antibiotics (3).

New approaches involving different mechanisms of drug action, that can bypass bacterial resistance mechanisms used against existing drugs, are needed to combat bacterial infections. One strategy is to target bacterial pathogenicity and virulence, rather than interfering with bacterial survival. This strategy may reduce the selective pressure for resistance development (4, 5).

Many bacterial virulence factors require structural bracing by disulfide bonds for their stability and function (6, 7). Disulfide bonds in bacteria are introduced in proteins by the DiSulfide Bond (DSB) system that acts on a wide range of virulence factors including protein toxins, adhesins, and secretion systems (7, 8).

The archetypal E. coli K-12 DSB system (REFERENCE showing this nomenclature which wont be understood by anyone including me) comprises the periplasmic oxidoreductase E. coli DsbA (EcDsbA) and the membrane-bound protein E. coli DsbB (EcDsbB). EcDsbA is highly oxidizing, with a redox potential of –120 mV. The active-site sequence Cys30-Pro-His-Cys33 has unusual properties. First,

Cys30 has a highly acidic pKa of 3–3.5. Second, the reduced dithiol form is more stable than the oxidized disulfide form of the enzyme (8, 9). EcDsbB is a membrane protein with two reactive cysteine pairs (Cys41-Cys44 and Cys104-Cys130) located on its two periplasmic loops. These act in concert

3 with ubiquinone (UQ8) or menaquinone (MK8) to maintain EcDsbA in the active oxidized form (10-

13).

The DSB system has been described as a master regulator of virulence in bacteria (7). Deletion or mutation results in pleiotropic effects and attenuated virulence with no effect on bacterial growth (7, 8).

Perhaps the most compelling evidence for its importance in virulence comes from studies on

Burkholderia pseudomallei, a Gram-negative bacterium that is highly pathogenic, naturally drug resistant and for which infections have a high mortality rate. All mice infected with wild type B. pseudomallei died, whereas all mice infected with the strain missing DsbA (ΔdsbA) survived even after

42 days, with little or no phenotypic evidence of infection (14). Similarly, deletion of dsbA in

Haemophilus influenzae had significantly lower bacteremia counts in animal models (15). DSB proteins are therefore implicated as possible targets for development of antivirulence drugs.

We set out to discover novel inhibitors of the membrane protein EcDsbB. During NMR screening of membrane proteins, other had identified several small molecules from a drug-like fragment library that bound very weakly to EcDsbB (16). We selected one of these compounds, shown to compete with ubiquinone for binding to EcDsbB, as a starting point for inhibitor development. Here, we report our systematically explored structure-activity relationship (SAR) studies, focusing on a 5,5-dimethyl-1,3- cyclohexanedione (dimedone) scaffold.

4 Results and Discussion

To assess DSB inhibition, a rapid screening assay was installed to monitor effects on the formation of disulfide bonds in a synthetic peptide substrate. The redox sensitive fluorescence substrate allows measurement in real-time of the oxidoreductase activity of the DSB system. The assay has been used previously to characterize the oxidase activity of several DSB proteins (17, 18). Here, it was used as a screening tool to rapidly identify potential inhibitors. However the assay does not discriminate between inhibitors of EcDsbA or EcDsbB. Representative data from the assay are shown in Figure 1A, with decreasing fluorescence correlating with increasing inhibitor concentration, and indicating inhibition of

DSB activity. A plot of the initial rates of reaction versus the log of the concentration of inhibitor gave a concentration-response curve for calculating inhibitor potency (Figure 1B)

Inhibitor design initially focused on compound 1 (Figure 1D), reported to compete weakly with ubiquinone at its binding site in EcDsbB (16). To our knowledge no follow-up structure activity

relationships have been reported. Its reported potency (IC50 100 µM) referred to inhibition of ubiquinone reduction (19). In the peptide oxidation assay used here, 1 proved to be an extremely poor

inhibitor (IC50 2-5 mM). The assays differed in a number of respects. First, the peptide oxidation assay utilized a membrane preparation of EcDsbB and either the lipid components or the different detergents used may have interfered with the inhibitor accessing its site of action. Moreover, the EcDsbB we used was likely to be bound to ubiquinone, which the inhibitor may have had to dislodge in order to inhibit activity. This may not have been the case in the previously study (19), where the cyanoborohydride used to attach EcDsbB to nanodiscs for assays may have reduced ubiquinone to ubiquinol.

Nevertheless, some inhibition was detectable in our assay for compound 1 (see Figure 1), so we

5 decided to derive other potential inhibitors from it. A summary of the SAR data is given in Table 1 for a range of small organic molecules derived from 1.

We focused on the core 5,5-dimethyl-1,3-cyclohexanedione (dimedone) scaffold. To probe the enzyme active site further, we stripped off the 2-propionoyl, 4-phenyl and the two 5-methyl groups from the dimedone ring in 1 and added a phenylbutanoyl chain to create the novel phenylbutanoate compound 2.

This structure mimics to some extent the endogenous ubiquinone cofactor of EcDsbB, with an aliphatic

tail and bulky headgroup (Figure 1 C) and was a much more potent inhibitor (IC50 24 µM) than 1.

When the 5,5-dimethyl substituents were reinserted (3) there was little improvement (IC50 19 µM).

Incorporating the ester onto the dimedone scaffold (at the C2 position via lactone formation) inactivated the compound (4). To probe the binding space further, one of the 5-methyl groups of 3 was

replaced by a phenyl group and this improved inhibitor potency 2-fold (5, IC50 9.5µM).

Next, we investigated whether replacing the 4-phenylbutanoate with a simple propionate (6) would retain activity. This modification slightly improved potency compared to 5. We therefore concluded that the bulky chain extending from the ester group was not required for potency, however the ester moiety was essential for inhibition as its removal in 7 abolished activity. To further probe the space around the phenyl moiety, a meta-chlorine substituent was added to the phenyl group of 6, giving 8 and

a significant improvement in inhibitor potency (IC50 1.3 µM).

Using compound 8 as a core structure, we then designed derivatives to probe the ester functionality

further, but these gave little or no improvement in IC50 (data not shown). Replacing the propionyl group

6 with a chlorobutanoyl (9) did not improve inhibitory activity. Removing the chlorine and maintaining a

four-carbon chain also did not improve inhibitory activity (IC50 2.6 µM vs 1.3 µM for 10 and 8, respectively) and replacing the propionyl with a bulky pivaloyl group (11) decreased potency by >20-

fold (IC50 43 µM).

The importance of the ester group was further assessed by modifying it into an ether linkage or changing the ketone group into a ketoxime (12 and 13). However both changes rendered the molecule inactive. Returning to the scaffold of 9, replacing the C5-methyl group with a benzyl substituent (14)

almost doubled the IC50 to 2.2µM (15 vs 9). An additional chlorine, ortho- or para- in the phenyl ring

(15 and 16 respectively), did not improve activity significantly. Substitution of the m-chlorine with a

phenyl ring (17) increased IC50 >2-fold (17 vs 8). Replacing m-chlorine with other halogens showed

that fluorine (18) reduced the inhibitory activity by 3-fold giving IC50 3.6 µM (18 vs 8). However, the

larger bromine (19) improved the inhibitory activity a little (IC50 1.1 µM). Compound 19 was the most active inhibitor from this study.

To investigate interactions formed between the inhibitor and protein, mass spectrometry techniques were applied. First, mass spectra of the purified EcDsbB (solubilized in detergent micelles) were obtained following nanoflow electrospray (nanoES) using a gas phase activation strategy described previously (20). Two isoforms of the protein were detected (Figure 2A), the major form at m/e 21,200

KDa. In the presence of three different inhibitors (9, 19 and 19 should this be 8, 18, 19 as below?) at varying concentrations, successive binding of 56 Da entities to DsbB was observed. These entities were not detached upon higher activation conditions, implying a covalent interaction.

7

To further demonstrate covalent binding, an equimolar mixture of EcDsbB and 8 was subjected to denaturing LC-MS/MS. Four residues modified by a 56 Da entity were detected: three of these were lysines (at position 167, 169 and 170 in the C-terminal peptide 167KAKKRDLF174), and the fourth was a cysteine at position 130 in the fragment 125VASGDCAERQW135 (Figure 2B). To gain insight into the nature of the covalent modification, solutions of 8, 18 and 19 were subjected to electrospray by direct infusion to test the stability of the inhibitors upon nanoES in the gas phase. All intact protonated inhibitors were detected, the major forms being with one DMSO adduct. However, a fragment ion resulting in the neutral loss of 56 Da was also present. When isolating the parent ion of the intact inhibitor, the major peak corresponded to the neutral loss of 56 Da even at the minimum collision energy. These results demonstrate that the inhibitors are not stable upon ESI or in the gas phase, and have a tendency to fragment losing a propionate.

Addition of the propionate to the C-terminal Lys residues of EcDsbB is unlikely to affect the activity of the protein. However, covalent attachment of the propionate to EcDsbB Cys130 will adversely affect the ability of EcDsbB to re-oxidize DsbA, because Cys130 is essential for this process (11). Although

Cys130 is relatively buried in the structures of the EcDsbA-EcDsbB complex, there is a degree of flexibility in the region of Cys130 (21, 22) that may expose this residue to the solvent, allowing covalent attachment of the compounds.

Since the peptide oxidation assay does not discriminate between inhibition of EcDsbA and EcDsbB, we also investigated the effect of the compounds on EcDsbA by nanoES. Mass spectra showed that Cys33 in the fragment [27SFFCPHCY34] was also covalently modified by a 56 Da adduct (data not shown).

Further investigation of the binding of inhibitors to EcDsbA was undertaken using 1H,15N-HSQC

8 NMR. Compound 19 was added to a sample of both oxidized and reduced forms of 15N-labeled

EcDsbA causing localized perturbations around the active site (Figure 2C). Subsequently, each complex was re-purified by size exclusion chromatography (SEC) and a 1H,15N-HSQC re-acquired. In the case of oxidized EcDsbA, SEC removed compound 19 and the HSQC spectrum was consistent with apo-oxidized EcDsbA. In the case of reduced EcDsbA, SEC had no effect on the HSQC spectrum obtained for the complex. These results suggest that 19 forms a covalent complex with reduced

EcDsbA, but binds non-covalently to oxidized EcDsbA adjacent to the active site, supporting the results obtained from mass spectrometry where covalent attachment of the adduct to only one Cys in each of the EcDsbA and EcDsbB enzymes suggests a degree of specificity and the formation of a stable enzyme-inhibitor complex before covalent attachment occurs (Figure 3A). Compounds 1-19 are more correctly described as “vinylogous anhydrides” than esters and covalently modify the nucleophilic side chains of cysteine and lysine by efficiently transferring the propionyl group, thus giving rise to the 56

Da adducts of the protein observed in the mass spectra (Figure 3B). Insert reference: J. Phys. Org.

Chem. 2008, 21 1022–1028 DOI 10.1002/poc.1435 The alternative pathway of Michael addition followed by elimination of propionate (Figure 3B) was not observed by mass spectroscopy, consistent with calculations that predict this pathway to be of minor significance. Same Reference

Our initial starting point was a compound (1) identified to weakly bind and inhibit the activity of the membrane protein EcDsbB. We generated a library of analogues, including compounds that over 1,000 times more potent than 1 in inhibiting DSB-catalysed formation of disulfide bonds, the most potent inhibitor 19 binding covalently to a single cysteine in EcDsbB and EcDsbA. Both enzymes are candidate targets for the development of new antimicrobials. In summary, the findings validate the

DSB system as a potential new target for small molecule inhibitors. Since this system is important for bacterial virulence, new inhibitors developed for the DSB system may be antivirulence drugs, and the 9 dimedone inhibitors have been instrumental tools here to validate the DSB system as a viable target for small molecule inhibitors. Further efforts to examine more druggable compounds against this target system may lead to a new class of antivirulence drugs to thwart current bacterial resistance mechanisms.

10 METHODS

Protein purification

EcDsbA in a pMCSG7 vector was expressed as a cytoplasmic protein with an N-terminal His6-tag and a TEV protease-recognition site. The plasmid was inserted into BL21(DE3)pLysS cells and transformed cells were used for protein expression by auto-induction with ampicillin. TALON resin

(Clontech, Australia) was pre-equilibrated with buffer in a gravity column and supernatant passed

through resin to bind His6-tagged EcDsbA. The His6-tag was removed by TEV protease digest. Cleaved protein was collected by running sample over pre-equilibrated TALON and collecting the flow- through. Protein was further purified by running over a size-exclusion column (GE Superdex 75 16/60) and peak fractions pooled and assessed for purity by SDS-PAGE gel.

The dsbB gene in a pQE-70 vector (JCB851) with a C-terminal His6-tag was obtained from our collaborators at Monash University and Professor James Bardwell (University of Michigan). EcDsbB contains the mutations Cys8Ala and Cys49Val to aid in purification. EcDsbB was expressed and purified as described previously (18) except for buffer, which was PBS. Membrane-bound EcDsbB was quantified using SDS-PAGE 12% gels, and used in the peptide oxidation assay. For mass spectrometry,

EcDsbB was purified from the membrane by re-suspending membrane in PBS buffer and solubilized with detergent n-dodecyl-b-D-maltopyranoside (DDM) 1% w/v. Membrane debris was removed by ultracentrifugation at 38,000 rpm for 1 hr at 4°C. Supernatant was collected and PrepEase (USB

Corporation, Cleveland, OH) used to bind protein. Further purification was done using size-exclusion chromatography (Superose 6 10/300 GL) with PBS containing 0.03% w/v DDM. Fractions were monitored by SDS-PAGE gels to check purity.

11 Peptide oxidation assay

The peptide substrate CQQGFDGTQNSCK with a 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) chelate on the N-terminus and a methylcoumarin amide (MCA) group coupled to the ε- amino group of the lysine was used. Lyophilized peptide (Anaspec, USA) was reconstituted in buffer

(100mM imidazole, pH 6) at 2 mM concentration. Europium trifluoromethansulfonate (Sigma Aldrich,

Australia) was added to the peptide solution at a molar ratio of 2:1 peptide to europium salt. The solution was left for 5 min at room temperature, then aliquoted into eppendorf tubes and immediately snap frozen in liquid nitrogen.

Inhibitor screening was performed in 384-well white plates (Perkin Elmer, Australia). Protein was added to the final concentrations of 80 nM EcDsbA and 1.6 µM EcDsbB membrane solution, in 50 mM

MES, 50 mM NaCl, 2 mM EDTA buffer, pH 5.5. Peptide substrate was added last to a final concentration of 8 µM in a total 50 µL volume. Assay was run on a Synergy H1 Multimode Plate

Reader (Biotek, USA) with excitation 340 nm and emission 615nm, a delay of 100 m s before read and

100 m s read time. Compounds were analysed in duplicates per plate, and tested at least three times on separate occasions. Data was plotted and analyzed using GraphPad Prism version 5.0 (GraphPad

Software, Inc. USA). For calculating the IC50 values, the initial rates of the reaction were calculated over the linear portion of the curve (the first 10 or 20 min of each reaction) and normalized as a percentage relative to no inhibitor present.

Synthesis of DsbB inhibitors

A detailed description of the synthesis of compounds is given in the supporting information.

12

Mass spectrometry determination of EcDsbB

Aliquots of EcDsbB were buffer exchanged against 200 mM ammonium acetate, 0.02% DDM, and

EcDsbA against 200 mM ammonium acetate, using Micro Bio-spin 6 columns (BioRad). Inhibitor stock solutions were freshly prepared in DMSO at 10 mM concentration. Inhibitors were added to the protein at different concentrations, final DMSO was at 5 % to be consistent with the screening assays.

For nanoflow electrospray mass spectrometry, spectra were acquired on a modified Q-ToF II mass spectrometer (Waters) (26) using in-house prepared gold-coated glass capillaries (27). To preserve non- covalent interactions, the MS parameters were: capillary voltage 1.5 kV, cone voltage 80 V, extractor 5

V, source backing pressure 6-8 mbar and collision cell pressure of 5 psi. Collision cell energy was between 60 and 200 V. Inhibitors were analysed in a Q Exactive hybrid quadrupole Orbitrap mass spectrometer (Thermo Scientific). The covalent binding sites were determined by denaturing LC-

MS/MS: EcDsbB or EcDsbA (either in ammonium acetate buffer or in 50 mM HEPES, 50 mM NaCl, pH 6.8, supplemented with 0.02 % DDM for EcDsbB), were mixed with inhibitor at equimolar ratios, the mixture was then ethanol precipitated and digested with chymotrypsin. Peptides were separated by reversed-phase nanoLC (DionexUltiMate 3000 RSLC nano System, Thermo Scientific) and directly analysed in a LTQ-Orbitrap XL hybrid mass spectrometer (Thermo Scientific). Potential binding sites were identified using the MassMatrix Database Search Engine (28, 29) then validated manually. Search parameters were as follows: Carbamidomethylation of cysteines and oxidation of methionine residues were allowed as variable modifications; the mass accuracy filter was 10 p.p.m. for precursor ions and

0.8 Da for fragment ions; minimum pp and pp2 values were 5.0 and minimum pptag was 1.3.

NMR spectroscopy

13 Interactions of PB-1-20 with oxidized and reduced EcDsbA were investigated by [15N,1H]-HSQC solution NMR spectroscopy. First, a [u-13C,15N]-labeled oxidized sample was prepared with protein concentration 100 µM and 50 mM HEPES (pH 6.8), 50 mM NaCl, 2% (v/v) d6-DMSO, 5% (v/v) D2O.

The oxidized sample (protein concentration ~ 180 µM; 50 mM HEPES (pH 6.8), 50 mM NaCl) was further reduced by adding 100 fold excess of DTT, incubating for 1h at room temperature and then buffer exchanged using a PD10 column to remove DTT. A 200 mM stock solution of PB-1-20 was prepared in d6-DMSO and diluted to 20 mM and 10 mM. Finally, for each state, three samples were prepared with EcDsbA: PB-1-20 ratio 1:0.5, 1:1 and 1:10 with a fixed concentration of 2% d6-DMSO,

100 µM protein and 5% (v/v) D2O. All data was collected using approximately 160 µL sample volume in a 3 mm thick walled NMR tube. All NMR experiments were acquired on a 600 MHz spectrometer equipped with a CryoProbe at 298 K. The data were processed by Topspin 3.2 and analyzed by CARA.

SUPPORTING INFORMATION:

Synthetic and analytical methods for the compounds described here.

This material is available free of charge via the Internet at http://pubs.acs.org.

ACKNOWLEDGMENTS

We thank Stephanie Tay and Wilko Duprez for technical support. JLM is an ARC Australian Laureate Fellow (FL0992138). DPF is an NHMRC Senior Principal Research Fellow (1027369). MAH, PB and FL were supported by the ARC Laureate Award. JLM is also an Honorary NHMRC Fellow (455829). DPF and RCR acknowledge support from ARC (DP130100629) and from the ARC Centre of Excellence in Advanced Molecular Imaging (CE140100011). CB and CR acknowledge support from ERC IMPRESS. MJS and BM acknowledge support from the NHMRC (1009785).

14 REFERENCES

1. Hogberg, L. D., Heddini, A., and Cars, O. (2010) The global need for effective antibiotics: challenges and recent advances, Trends Pharmacol. Sci. 31, 509-515. 2. Hojgard, S. (2012) Antibiotic resistance - why is the problem so difficult to solve?, Infect. Ecol. Epidemiol. 2, 18165- 18171. 3. Boucher, H. W., Talbot, G. H., Bradley, J. S., Edwards, J. E., Gilbert, D., Rice, L. B., Scheld, M., Spellberg, B., and Bartlett, J. (2009) Bad bugs, no drugs: no ESKAPE! An update from the Infectious Diseases Society of America, Clin. Infect. Dis. 48, 1-12. 4. Cegelski, L., Marshall, G. R., Eldridge, G. R., and Hultgren, S. J. (2008) The biology and future prospects of antivirulence therapies, Nat. Rev. Microbiol. 6, 17-27. 5. Rasko, D. A., and Sperandio, V. (2010) Anti-virulence strategies to combat bacteria-mediated disease, Nat Rev Drug Discov 9, 117-128. 6. Depuydt, M., Messens, J., and Collet, J. F. (2011) How proteins form disulfide bonds, Antioxid. Redox Signaling 15, 49-66. 7. Heras, B., Shouldice, S. R., Totsika, M., Scanlon, M. J., Schembri, M. A., and Martin, J. L. (2009) DSB proteins and bacterial pathogenicity, Nat. Rev. Microbiol. 7, 215-225. 8. Shouldice, S. R., Heras, B., Walden, P. M., Totsika, M., Schembri, M. A., and Martin, J. L. (2011) Structure and function of DsbA, a key bacterial oxidative folding catalyst, Antioxid. Redox Signaling 14, 1729-1760. 9. Ito, K., and Inaba, K. (2008) The disulfide bond formation (Dsb) system, Curr. Opin. Struct. Biol. 18, 450-458. 10. Inaba, K., Takahashi, Y. H., Fujieda, N., Kano, K., Miyoshi, H., and Ito, K. (2004) DsbB elicits a red-shift of bound ubiquinone during the catalysis of DsbA oxidation, J. Biol. Chem. 279, 6761-6768. 11. Kadokura, H., and Beckwith, J. (2002) Four cysteines of the membrane protein DsbB act in concert to oxidize its substrate DsbA, EMBO J. 21, 2354-2363. 12. Stenson, T. H., and Weiss, A. A. (2002) DsbA and DsbC are required for secretion of pertussis toxin by Bordetella pertussis, Infect. Immun. 70, 2297-2303. 13. Takahashi, Y. H., Inaba, K., and Ito, K. (2006) Role of the cytosolic loop of DsbB in catalytic turnover of the ubiquinone-DsbB complex, Antioxid Redox Signal 8, 743-752. 14. Ireland, P. M., McMahon, R. M., Marshall, L. E., Halili, M., Furlong, E., Tay, S., Martin, J. L., and Sarkar-Tyson, M. (2014) Disarming Burkholderia pseudomallei: structural and functional characterization of a disulfide oxidoreductase (DsbA) required for virulence in vivo, Antioxid. Redox Signal 20, 606-617. 15. Rosadini, C. V., Wong, S. M., and Akerley, B. J. (2008) The periplasmic disulfide oxidoreductase DsbA contributes to Haemophilus influenzae pathogenesis, Infect. Immun. 76, 1498-1508. 16. Fruh, V., Zhou, Y., Chen, D., Loch, C., Ab, E., Grinkova, Y. N., Verheij, H., Sligar, S. G., Bushweller, J. H., and Siegal, G. (2010) Application of fragment-based drug discovery to membrane proteins: identification of ligands of the integral membrane enzyme DsbB, Chem. Biol. 17, 881-891. 17. Vivian, J. P., Scoullar, J., Rimmer, K., Bushell, S. R., Beddoe, T., Wilce, M. C., Byres, E., Boyle, T. P., Doak, B., Simpson, J. S., Graham, B., Heras, B., Kahler, C. M., Rossjohn, J., and Scanlon, M. J. (2009) Structure and function of the oxidoreductase DsbA1 from Neisseria meningitidis, J. Mol. Biol. 394, 931-943. 18. Walden, P. M., Heras, B., Chen, K. E., Halili, M. A., Rimmer, K., Sharma, P., Scanlon, M. J., and Martin, J. L. (2012) The 1.2 A resolution crystal structure of TcpG, the Vibrio cholerae 15 DsbA disulfide-forming protein required for pilus and cholera-toxin production, Acta Cryst., Sect D: Biol. Crystallogr. 68, 1290-1302. 19. Bader, M., Muse, W., Zander, T., and Bardwell, J. (1998) Reconstitution of a protein disulfide catalytic system, J. Biol. Chem. 273, 10302-10307. 20. Barrera, N. P., Di Bartolo, N., Booth, P. J., and Robinson, C. V. (2008) Micelles protect membrane complexes from solution to vacuum, Science 321, 243-246. 21. Inaba, K., Murakami, S., Suzuki, M., Nakagawa, A., Yamashita, E., Okada, K., and Ito, K. (2006) Crystal structure of the DsbB-DsbA complex reveals a mechanism of disulfide bond generation, Cell 127, 789-801. 22. Sperling, L. J., Tang, M., Berthold, D. A., Nesbitt, A. E., Gennis, R. B., and Rienstra, C. M. (2013) Solid-state NMR study of a 41 kDa membrane protein complex DsbA/DsbB, J. Phys. Chem. B 117, 6052-6060. 23. Nelson, J. W., and Creighton, T. E. (1994) Reactivity and ionization of the active site cysteine residues of DsbA, a protein required for disulfide bond formation in vivo, Biochemistry 33, 5974-5983. 24. Ondo-Mbele, E., Vives, C., Kone, A., and Serre, L. (2005) Intriguing conformation changes associated with the trans/cis isomerization of a prolyl residue in the active site of the DsbA C33A mutant, J. Mol. Biol. 347, 555-563. 25. Vivian, J. P., Scoullar, J., Robertson, A. L., Bottomley, S. P., Horne, J., Chin, Y., Wielens, J., Thompson, P. E., Velkov, T., Piek, S., Byres, E., Beddoe, T., Wilce, M. C., Kahler, C. M., Rossjohn, J., and Scanlon, M. J. (2008) Structural and biochemical characterization of the oxidoreductase NmDsbA3 from Neisseria meningitidis, J. Biol. Chem. 283, 32452-32461. 26. Sobott, F., Hernandez, H., G. McCammon, M. G., Tito, M. A., and Robinson, C. V. (2002) A Tandem Mass Spectrometer for Improved Transmission and Analysis of Large Macromolecular Assemblies, Anal. Chem. 74, 1402-1407. 27. Hernández, H., and Robinson, C. V. (2007) Determining the stoichiometry and interactions of macromolecular assemblies from mass spectrometry, Nat. Protoc. 2, 715-726. 28. Xu, H., and Freitas, M. A. (2007) A mass accuracy sensitive probability based scoring algorithm for database searching of tandem mass spectrometry data, BMC Bioinformatics 3, 133-142. 29. Xu, H., Yang, L., and Freitas, M. A. (2008) A robust linear regression based algorithm for automated evaluation of peptide identifications from shotgun proteomics by use of reversed- phase liquid chromatography retention time, BMC Bioinformatics 9, 347-361.

16 Compound IC50 ± SEM (µM) n 1 2000-5000 (100 # ) 3

2 24 ± 1.6 3

3 19 ± 1 4

4 NA 3

5 9.4 ± 0.5 3

6 6.9 ± 0.3 4

7 NA 3

8 1.3 ± 0.03 6

9 2.2 ± 0.08 4

10 2.6 ± 0.1 3

11 43 ± 3.1 3

12 NA 3

13 NA 3

14 2.2 ± 0.08 4

15 1.9 ± 0.05 4

16 1.5 ± 0.09 3

17 2.9 ± 0.1 3

18 3.6 ± 0.2 3

19 1.1 ± 0.02 6

NA = no activity # value from Fruh et al (16) Table 1. Structure-activity relationship results of compounds screened in the peptide oxidation assay.

All compounds were tested in duplicate on at least three separate occasions. IC50 values were calculated from the initial rates of the reaction and the means and standard errors of mean are reported.

17 A50000 B 120 Compound 19 e s 63µM 100 pon 16µM s e r

40000 80

4µM e c U

F 1µM 60 en R

0.25µM sc e 40

30000 r 0µM luo

F 20

20000 % 0 0 1000 2000 3000 4000 5000 -8 -7 -6 -5 -4 -3 time s Conc M

O C O

O CH3 O O O Ubiquinone-8 O O Ubiquinone-1

D O O O O O O O O OH O O O 1 2 3 4

O O O O O OH O

O O O O 5 6 7 8 Cl O O O Cl O O O O

O O O O

9 10 11 12 Cl Cl Cl Cl O O O O O O O Cl OH O N O O

13 Cl 15 Cl 16 Cl 14 Cl Cl

O O O O O O

O O O 17 18 19 F Br

Figure 1: Representative graphs of inhibitory activity using the peptide oxidation assay. (A) Raw data of fluorescence emission. As the concentration of inhibitor 19 increases, less fluorescence emission is observed. Data shown are from duplicates ± SD (B) A dose-response curve generated from the initial

rates of the reaction, used to calculate the IC50 of 19. Data were from n=6 experiments and the mean ±

SEM presented. (C) Structures of ubiquinone-8 and ubiquinone-1 (D) Structures of small molecule inhibitors used in this study. 18 A 100 B 639.78

VASGDC(S-CH2CONH2)AERQW

640.28

100μM

Relative Intensity (%) 640.78

641.28 641.78 75μM m/z 100 639.28

VASGDC(S-COC2H5)AERQW Relative Intensity (%)

25μM 639.78 Relative Intensity (%)

640.28

0μM 640.82

m/z m/z 1626 1634 1642 1650 C 0.10

0.08

0.06

0.04 CSP / ppm CSP

0.02

0 50 100 150 Sequence

0.6

0.5

0.4

0.3

CSP / ppm CSP 0.2

0.1

0 50 100 150 Sequence 19

Figure 2: Binding of inhibitors to EcDsbB. A) nanoES spectra of the 13+ charge state of 75µM

EcDsbB apo (lowest graph) or in the presence of 19 at different concentrations as indicated. Apo- protein labeled in black, binding to the first, second and third inhibitor molecules are labeled (green, orange and purple, respectively). B) CID MS spectrum of the doubly charged peptide ion corresponding to the sequence [125VASGDCAERQW135] with the Cys at position 130 modified either by a carbamidomethyl group (top) or by the covalently bound inhibitor (bottom). Both sequences were verified by MS/MS. C) Chemical shift perturbations observed in the 15N-HSQC spectrum of (top) oxidised EcDsbA and (bottom) reduced EcDsbA upon addition of 19 (1 mM). In each case the perturbations are clustered around the active site of EcDsbA. In each case the largest perturbations are mapped onto the structure of EcDsbA (1FVK and 1A2L, respectively) with perturbed residues coloured red.

20 A DsbA O S (C33) (C30)S

O S S S S (C104) (C130) (C41) (C44)

DsbB UQ

nuchleophiles B Inhibitor (EcDsbB) O O cysteine cysteine OH O O R O S HS H R R S O or Ester hydrolysis or O O lysine R H2N Cl O lysine Cl Cl R HN 56Da adducts observed O O R R R H O HS cysteine O S S O or Michael addition HO O R O O H2N lysine

Cl Cl Cl Product not seen

Figure 3: Schematic of EcDsbA:EcDsbB complex with the propionate adduct. A) Covalent attachment of the propionate to EcDsbB Cys130 prevents the protein from reoxidizing via a disulfide-dithiol exchange with the Cys41-Cys44 pair. At the same time, covalent attachment of the propionate to

EcDsbA Cys33 prevents nucleophilic attack on the Cys30 and EcDsbA cannot be re-oxidized. B) The mechanism by which the inhibitors modify the enzymes is direct acylation (top) rather than the alternative Michael addition/elimination sequence (bottom).

21