Size Matters and How You Measure It: A Gram-Negative Antibacterial

Letter pubs.acs.org/journal/aidcbc

Cite This: ACS Infect. Dis. XXXX, XXX, XXX−XXX

Size Matters and How You Measure It: A Gram-Negative Antibacterial Example Exceeding Typical Molecular Weight Limits Fiorella Ruggiu, Shengtian Yang, Robert L. Simmons, Anthony Casarez, Adriana K. Jones, Cindy Li, Johanna M. Jansen, Heinz E. Moser, Charles R. Dean, Folkert Reck, and Mika Lindvall* Novartis Institutes for BioMedical Research, 5300 Chiron Way, Emeryville, California 94608, United States

Downloaded via UNIV OF GLASGOW on September 3, 2019 at 18:14:39 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: Monobactam antibiotic 1 is active against Gram-negative bacteria even though it has a higher molecular weight (MW) than the limit of 600 Da typically applied in designing such compounds. On the basis of 2D NMR data, the compound is able to adopt a compact conformation. The dimensions, projection area, and dipole moment derived from this conformation are compatible with porin permeation, as are locations of polar groups upon superimposition to the crystal structure of ampicillin bound to E. coli OmpF porin. Minimum inhibitory concentration (MIC) shifts in a porin knock-out strain are also consistent with 1 predominately permeating through porins. In conclusion, we describe a carefully characterized case of a molecule outside default design parameters where MW does not adequately represent the 3D shape more directly related to permeability. Leveraging 3D design criteria would open up additional chemical space currently underutilized due to limitations perceived in 2D. KEYWORDS: conformational flexibility, polar surface area, hydrophobic collapse, porins, permeability

N

by the internal electrostatics. They concluded that a molecule capable of adopting a conformation where the orientation permits steric passage through the bottleneck in combination with having a strong dipole moment component along the diffusion axis may be transported through the porin. The size of the steric barrier in the constriction zone has been evaluated previously. For instance, permeability of sugars of various sizes indicated that there is a cutoff for entry into Escherichia coli (E. coli) and Salmonella typhimurium porins at a molecular weight (MW) of 600 Da8,9 and that the pore diameter is ∼11 Å for E. coli porins.8 From the first crystal structure of E. coli OmpF, the eyelet was estimated to be 7 Å by 11 Å.3 In a recent study, Acosta-Gutiérrez et al.6 characterized several porins of Gram-negative pathogens and evaluated their minimal constriction zone radii between 2.5 and 3.5 Å (∼3 Å for E. coli OmpF) using crystal structures. In 2008, O’Shea and Moser10 analyzed the physicochemical property space for antibacterial compounds in comparison to other drugs and concluded that 95% of Gram-negative antiinfective drugs had a MW of 600 Da or less and that they were on average more polar in terms of cLogD and topological polar surface area than the Comprehensive Medicinal Chemistry (CMC) reference drugs. They also remarked that compounds should have the dual ability to be charged in order to diffuse through the porin and to be neutral in order to pass the inner

ew antibiotics are needed as resistance erodes efficacy of existing therapies, and multidrug resistant infections, such as carbapenem-resistant Enterobacteriaceae infections, become more common.1 Beta lactams such as carbapenems are considered last resort drugs for the treatment of resistant strains. Drug discovery of new scaffolds is impeded in part by the difficulty to penetrate the cell wall of Gram-negative pathogens, with challenges in both entry and efflux.2 Permeability into the Gram-negative pathogen is hindered by the asymmetric outer membrane, with an inner leaflet composed of lipids and an outer leaflet consisting of lipopolysaccharides (LPS). The outer membrane is particularly difficult for lipophilic molecules to diffuse through, probably due to the low fluidity of the LPS leaflet that is the barrier.3 Certain classes of antibiotics are known to pass through nonspecific hydrophilic porin channels,3 and many active compounds are assumed to enter via this mechanism.4,5 The outer membrane porin proteins (Omps) form a β-barrel filled with water and have an internal hourglass shape with a constriction zone comparable to an eyelet. For entry through Omps, small and polar compounds are needed to displace the water in the channel, interact with the polar amino acids, and cross the constriction zone. The constriction zone contains several negatively charged residues on one side and several positively charged residues on the opposite side, which create a strong transverse electric field.3,6 Bajaj et al.7 investigated the electrostatics of the diffusion of norfloxacin through the E. coli OmpF channel and observed that the mechanism is modulated © XXXX American Chemical Society

Received: July 14, 2019

A

DOI: 10.1021/acsinfecdis.9b00256 ACS Infect. Dis. XXXX, XXX, XXX−XXX

ACS Infectious Diseases

Letter

requirements are opposite: to pass through a porin lined with polar residues, it is desirable to mask hydrophobic groups and expose polar groups. Additionally, conformations need to be compact enough to pass through porins with limited width. Erythronolides and in particular azithromycin have been suggested to pass through porins due to their charge at pH 7.4 and flexibility to adopt compact conformations.14 Results and Discussion. In this communication, we have studied an analogue of compound 28, compound 115 (see Figure 1 and the Supporting Information for the synthesis), where the piperidine moiety is replaced by a pyrrolidine. Compound 1 behaves similarly to compound 28 and inhibits several Gram-negative pathogens, including carbapanemresistant strains, as shown by the minimum inhibitory concentration (MIC) against wild type strains (see Table 1). Our assumption is that the compound penetrates the outer membrane through porins to reach its target, the penicillinbinding proteins 3, in the periplasm. To determine the conformations of compound 1, a 2D 1 H−1H NOESY NMR was performed in D2O. In particular, two long-range restraints were identified between protons H1, H5, and H28 as well as protons H2, H4, and H38 (see Figure 2). Although the signal intensities for these restraints are low,

membrane and permeate mammalian cells. Similar observations were made by Brown et al.11 where they observed that small and polar compounds from the AstraZeneca screens accumulated better in Gram-negative pathogens. The size dependency of porin entry has commonly been translated to a cutoff in MW between 450 and 600 Da when selecting or designing compounds for Gram-negative antibiotic research. In this work, we present a molecule challenging the default design limit of 600 Da. In a previous report,12 monobactams, a known class of antibiotics, were further optimized to address carbapenem-resistant Enterobacteriaceae. Monocyclic β-lactams are not affected by metallo-β-lactamases, which degrade all other classes of β-lactams. During the design process, the team synthesized compound 28 (see Figure 1), which inhibits

Figure 1. Structures of a monobactam synthesized by Reck et al.12 (compound 28) and its analogue (compound 1).

growth of several Enterobacteriaceae despite a relatively high MW of 706 Da (see Table 1). We hypothesized that the flexibility of the compound enables it to adopt compact conformations able to pass through the Omp eyelet. The proposed mechanism is similar to how molecules such as cyclosporins and erythronolides form compact conformations, effectively masking their polar groups, in order to pass through nonpolar membranes.13 Yet in this case, the permeability

Figure 2. Long-range NMR restraints observed in 2D 1H−1H NOESY NMR for compound 1 and the protonation state used in modeling.

Table 1. MICs against Several Gram-Negative Pathogens for Compound 28 and Compound 1 MIC [μg/mL] strain name Escherichia coli

Pseudomonas aeruginosa

Klebsiella pneumoniae

Acinetobacter baumannii Enterobacter cloacae Serratia marcescens

description ATCC ATCC ATCC ATCC ATCC

25922 BAA-2469 BAA-2471 27853 BAA-47

ATCC 43816 ATCC 700603 ATCC BAA-2146

ATCC 19606 GC4142

cpd. 28

cpd. 1

0.19 0.71 8 2.3 4 8 8 4 0.2 0.71 0.4 0.4 0.79 0.42 4 6.7 0.71

0.25 0.125 8 3.7 4 8 8 4 0.25 0.25 0.25 0.125 0.5 0.125 4 8 0.5

lactamase expressed in strain New Delhi metallo-β-lactamase 1 (NDM-1)a NDM-1a

metallo-β-lactamase VIM-1a metallo-β-lactamase SPM-1a metallo-β-lactamase GIM-1a extended-spectrum β-lactamase (ESBL) SHV-18 NDM-1a serine carbapenemase KPC-2a serine carbapenemase KPC-3a ESBL CTX-M-14 AmpC β-lactamase metallo-β-lactamase IMP-1a

a

Confers resistance to carbapenems. B



Letter

Figure 3. (a) Structure of ampicillin. (b) Sectional view of ampicillin in complex with E. coli OmpF (PDB ID: 4gcp). (c) Top view of the same complex with an electrostatic surface.

Figure 4. (a) Conformer corresponding to 2D NMR long-range restraints superposed on ampicillin in E. coli OmpF (PDB ID: 4gcp) shows no clashes with protein and retains charge−charge interactions with arginine 167 and serine 125. (b) Sectional view of the superposed compound 1 in E. coli OmpF with the depiction of the dipole moment in black. Minimal dimensions (9.7 Å × 11.3 Å) and the dipole moment are almost ideally aligned (27°). (c) Top view of the same complex with the electrostatic surface.

We also examined a crystal structure of E. coli OmpF with ampicillin bound close to the constriction zone (PDB ID: 4gcp)20 (Figure 3). In order to permeate a porin, a drug should have favorable interactions with the protein but not be stabilizing enough for the compound to stay bound.20 Ampicillin is a penicillin antibiotic and contains an amide appended to a 4-membered lactam motif in common with compound 1. Conformers of compound 1 were superposed on this motif. One of the four conformers consistent with the 2D NMR restraints could fit the pocket without clashes (see Figure 4). The conformer is compact and exposes high polarity (3D-PSA = 417) with polar and ionizable groups placed at the perimeter of the molecule. The sulfonate of compound 1 overlays well with the carboxylate of ampicillin, which has ionic interactions with arginines 167 and 168. In the coordinate system of the overlay, additional reasonable locations for the functional groups of compound 1 can be identified. Its sulfonate may interact with serine 125. The positively charged amidine and the charged pyrrolidine are located in parts of the porin occupied by water molecules. The aminothiazole may interact with tyrosine 32 through hydrogen bonding or engage in π−π stacking with the aromatic ring. The triazole is within the hydrogen bonding distance of arginine 132 and serine 125. Conformations were further investigated to verify if they are in line with the crystal structure dimensions of the E. coli OmpF eyelet (7 Å × 11 Å)3 and the dipole moment criteria as

they are consistently present at different mixing times and thus are not noise (see the Supporting Information). On the basis of the reference restraint between protons H28 and H31 and the r−6 decay of the NOESY signal,16 the distances of the longrange restraints are estimated to be ∼5.4 Å, which indicates that the compound is mainly in compact conformations in water. NOESY NMR restraints are averaged distances over the population of conformers that the compound can adopt. In order to identify conformers generated by a force field that would be consistent with these restraints, we chose to apply a cutoff of 7 Å to the restraint distances in the conformers. The protonation state of compound 1 for conformational analysis was estimated using Moldiscovery moka 2.6.5 trained on the Novartis internal pKa data.17,18 The double zwitterion was predicted at pH 7.4, as depicted in Figure 2, with a 98% population (see the Supporting Information). Conformers were generated using CCG MOE 2018.0119 with and without the electrostatic term to increase their diversity. A maximum of 100 conformers was set, and otherwise, default parameters were used, which include the use of low mode molecular dynamics with the Amber10:EHT force field and an implicit solvent at a dielectric constant of 80. Proton distances corresponding to the long-range 2D NMR restraints were calculated via a python script (see the Supporting Information) to select conformers consistent with the experiment. Four lowenergy conformers were compliant with the 7 Å cutoff. C



Letter

observed by Bajaj et al.7 A python script (see the Supporting Information) was written to measure the minimal rectangle that could encompass a conformer to inquire if it could spatially fit through the constriction zone. This assessment is a simplified model to the porin size problem akin to a kid’s shape sorting box toy; however, it takes into account 3D aspects of molecules ignored by the MW descriptor. Similarly, AcostaGutiérrez et al.6 suggest from their modeling for one to use the average minimal projection area from conformers generated by molecular dynamics to characterize the size of molecules for porin entry. Minimal rectangle dimensions found by the script on the four stable conformers consistent with the NMR restraint vary between 8.7 and 9.7 Å and 11.3 and 12.1 Å with areas between 100 and 109 Å2 using van der Waals radii from Bondi21 (see Table S4). This indicates that compound 1 can adopt conformations with projected dimensions very close to the E. coli OmpF crystal structure eyelet size. This estimation does not take into account protein flexibility, which may result in the possibility of slightly bigger molecules permeating. In their 2018 study,22 Samanta et al. suggested by molecular dynamics that the narrow P. aeruginosa OccK8 porin could accommodate a larger substrate through a substrate induced expansion. The angle between the dipole moment and a theoretical diffusion axis, i.e., an axis perpendicular to the minimum rectangle, varied between 9° and 52°, which for 3 conformations is close to the ideal according to Bajaj et al.7 (0°). The dimensions for the conformer depicted in Figure 4 are 9.7 Å × 11.3 Å with an angle of 27°. The compact conformations as well as the good fit into the E. coli OmpF crystal structure gave us confidence that compound 1 was probably able to traverse E. coli OmpF and by extension potentially other porins. To test this hypothesis, we determined whether genetic deletion of multiple porins would decrease susceptibility to compound 1, which would be indicative of reduced compound entry into cells. Meuskens et al. recently designed a set of mutants derived from E. coli BL21(DE3)23 where four genes encoding abundant β-barrel proteins in the outer membrane (OmpA, OmpC, OmpF, and LamB) were deleted. Susceptibility to compound 1 was reduced 8-fold in this mutant relative to its porin replete parent strain (Table 2) consistent with compound entry through porins. In contrast, the activity of two siderophore-based inhibitors MC-124 and cefiderocol,25 which rely on TonBmediated siderophore transporters for cell entry, was unaffected by porin loss. The latter suggests that deletion of the porins was not making the cells generally more susceptible to inhibition of cell wall synthesis, consistent with the shift in activity of compound 1 reflecting reduced entry into cells. The activity of several other nonsiderophore-based β-lactam controls, i.e., aztreonam, cefepime, and ceftazidime, was also shifted (Table 2), consistent with the expectation that these drugs permeate into the periplasm of Gram-negative bacteria through porins.4,5 In addition, compound 1 does not contain a siderophore moiety, and knocking out TonB, the membrane proteins carrying iron chelates inside the bacteria, in K. pneumoniae does not significantly affect its activity. Compound 1 also had moderate activity against P. aeruginosa (Table 1), despite this organism having a notably less permeable outer membrane than E. coli, in part because porins in P. aeruginosa tend to be more selective.26 The specific routes of entry of 1 into P. aeruginosa were not determined in this study. The 3D design guideline discussed here is intended for improving permeation through general porins, but antibacte-

Table 2. MICs for Compound 1, the Siderophores MC-1 and Cefiderocol, and Reference Antibiotics in Parent BL21(DE3) and OmpA, OmpC, OmpF, and LamB KnockOut Strain of E. coli strain name Escherichia coli

description

parent BL21(DE3)

compound 1 MC-1a cefiderocola aztreonam cefepime ceftazidime chloramphenicol ciprofloxacin tigecycline

0.125 0.5 0.5 ≤0.06 ≤0.06 0.125 2 ≤0.004 0.125

Klebsiella pneumoniae

Omp knock-out mutant BL21(DE3) omp8 MIC [μg/mL] 1 1 0.5 1 0.5 1 2 0.015 0.125

clinical isolate from urine

tonB knockout mutant

1 0.5 0.5 >64 16 >64 >64 32 0.5

0.5 32 32 >64 16 >64 >64 32 0.5

a The dependence of MC-1 and cefiderocol on TonB mediated siderophore transporters is shown here by a shift in susceptibility of Klebsiella pneumoniae upon mutation of tonB. The tonB mutant was selected by plating Klebsiella pneumoniae on medium containing cefiderocol at 4× the MIC. The tonB gene has a deletion of nucleotides 197−210 resulting in a frameshift.

rials can have other routes of entry, with additional molecular design options. Conclusion. In summary, we have described a monobactam antibiotic, compound 1, which has a higher MW than the upper limit of 600 Da typically applied in designing such compounds, yet the compound is active against Gram-negative bacteria. The default molecular weight limit is intended to increase the likelihood of permeation through porins. Here, we demonstrate that, despite a MW of 692 Da, the compound is able to adopt a conformation, derived from 2D NMR data, where two main moieties of the molecule are folded onto one another. The resulting compact conformation, when superimposed to a related antibiotic, ampicillin, crystallized in a porin, is compatible with the dimensions and polarity of the narrowest part of the porin. MIC shifts in porin knock-out strains are also consistent with compound 1 predominantly permeating through porins. Therefore, in this case, the 2D size description by MW does not adequately describe 3D volume and shape, which relate more closely to permeability. In conclusion, we have contributed a carefully characterized case of a molecule breaking simple 2D design parameters. We challenge the medicinal and computational chemistry community to move more frequently to 3D design criteria, which would open up additional chemical space currently underutilized due to limitations perceived in 2D.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsinfecdis.9b00256. For compound 1: synthesis, 1D/2D NMR data, MIC results and details for pKa prediction, conformational analysis, overlays, size evaluation, and dipole moment (PDF) D



■

Letter

(15) Aulakh, V. S., Casarez, A., Lin, X., Lindvall, M., McEnroe, G., Moser, H. E., Reck, F., Tjandra, M., Simmons, R. L., Yifru, A., and Zhu, Q. (2015) Monobactam Organic Compounds for the Treatment of Bacterial Infections, Patent WO2015148379A1. (16) Evans, J. N. S., Ed. (1995) Biomolecular NMR Spectroscopy, Chapter 3, Oxford University Press. (17) Milletti, F., Storchi, L., Sforna, G., and Cruciani, G. (2007) New and Original PKa Prediction Method Using Grid Molecular Interaction Fields. J. Chem. Inf. Model. 47 (6), 2172−2181. (18) Gedeck, P., Lu, Y., Skolnik, S., Rodde, S., Dollinger, G., Jia, W., Berellini, G., Vianello, R., Faller, B., and Lombardo, F. (2015) Benefit of Retraining PKa Models Studied Using Internally Measured Data. J. Chem. Inf. Model. 55 (7), 1449−1459. (19) (2018) Molecular Operating Environment (MOE), 2018.01, Chemical Computing Group ULC, Montreal, QC, Canada. (20) Ziervogel, B. K., and Roux, B. (2013) The Binding of Antibiotics in OmpF Porin. Structure 21 (1), 76−87. (21) Bondi, A. (1964) Van Der Waals Volumes and Radii. J. Phys. Chem. 68 (3), 441−451. (22) Samanta, S., Bodrenko, I., Acosta-Gutiérrez, S., D’Agostino, T., Pathania, M., Ghai, I., Schleberger, C., Bumann, D., Wagner, R., Winterhalter, M., et al. (2018) Getting Drugs through Small Pores: Exploiting the Porins Pathway in Pseudomonas Aeruginosa. ACS Infect. Dis. 4 (10), 1519−1528. (23) Meuskens, I., Michalik, M., Chauhan, N., Linke, D., and Leo, J. C. (2017) A New Strain Collection for Improved Expression of Outer Membrane Proteins. Front. Cell. Infect. Microbiol. 7, 464. (24) McPherson, C. J., Aschenbrenner, L. M., Lacey, B. M., Fahnoe, K. C., Lemmon, M. M., Finegan, S. M., Tadakamalla, B., O’Donnell, J. P., Mueller, J. P., and Tomaras, A. P. (2012) Clinically Relevant Gram-Negative Resistance Mechanisms Have No Effect on the Efficacy of MC-1, a Novel Siderophore-Conjugated Monocarbam. Antimicrob. Agents Chemother. 56 (12), 6334−6342. (25) Ito, A., Nishikawa, T., Matsumoto, S., Yoshizawa, H., Sato, T., Nakamura, R., Tsuji, M., and Yamano, Y. (2016) Siderophore Cephalosporin Cefiderocol Utilizes Ferric Iron Transporter Systems for Antibacterial Activity against Pseudomonas Aeruginosa. Antimicrob. Agents Chemother. 60 (12), 7396−7401. (26) Chevalier, S., Bouffartigues, E., Bodilis, J., Maillot, O., Lesouhaitier, O., Feuilloley, M. G. J., Orange, N., Dufour, A., and Cornelis, P. (2017) Structure, Function and Regulation of Pseudomonas Aeruginosa Porins. FEMS Microbiol. Rev. 41 (5), 698−722.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Johanna M. Jansen: 0000-0003-3937-6243 Folkert Reck: 0000-0001-7250-2526 Mika Lindvall: 0000-0002-5637-8948 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS

■

REFERENCES

The authors are grateful to all scientists who were part of the Infectious Diseases research group at Novartis Institutes for BioMedical Research in Emeryville, CA, USA.

(1) Watkins, R. R., and Bonomo, R. A. (2016) Overview: Global and Local Impact of Antibiotic Resistance. Infect. Dis. Clin. North Am. 30 (2), 313−322. (2) Masi, M., Réfregiers, M., Pos, K. M., and Pagès, J.-M. (2017) Mechanisms of Envelope Permeability and Antibiotic Influx and Efflux in Gram-Negative Bacteria. Nat. Microbiol. 2, 17001. (3) Nikaido, H. (2003) Molecular Basis of Bacterial Outer Membrane Permeability Revisited. Microbiol. Mol. Biol. Rev. 67 (4), 593−656. (4) Hancock, R. E. W., and Bell, A. (1988) Antibiotic Uptake into Gram-Negative Bacteria. Eur. J. Clin. Microbiol. Infect. Dis. 7 (6), 713− 720. (5) Delcour, A. H. (2009) Outer Membrane Permeability and Antibiotic Resistance. Biochim. Biophys. Acta, Proteins Proteomics 1794 (5), 808−816. (6) Acosta-Gutiérrez, S., Ferrara, L., Pathania, M., Masi, M., Wang, J., Bodrenko, I., Zahn, M., Winterhalter, M., Stavenger, R. A., Pagès, J. M., et al. (2018) Getting Drugs into Gram-Negative Bacteria: Rational Rules for Permeation through General Porins. ACS Infect. Dis. 4 (10), 1487−1498. (7) Bajaj, H., Acosta Gutierrez, S., Bodrenko, I., Malloci, G., Scorciapino, M. A., Winterhalter, M., and Ceccarelli, M. (2017) Bacterial Outer Membrane Porins as Electrostatic Nanosieves: Exploring Transport Rules of Small Polar Molecules. ACS Nano 11 (6), 5465−5473. (8) Nikaido, H. (1994) Porins and Specific Diffusion Channels in Bacterial Outer Membranes. J. Biol. Chem. 269 (6), 3905−3908. (9) Koebnik, R., Locher, K. P., and Van Gelder, P. (2000) Structure and Function of Bacterial Outer Membrane Proteins: Barrels in a Nutshell. Mol. Microbiol. 37 (2), 239−253. (10) O’Shea, R., and Moser, H. E. (2008) Physicochemical Properties of Antibacterial Compounds: Implications for Drug Discovery. J. Med. Chem. 51 (10), 2871−2878. (11) Brown, D. G., May-Dracka, T. L., Gagnon, M. M., and Tommasi, R. (2014) Trends and Exceptions of Physical Properties on Antibacterial Activity for Gram-Positive and Gram-Negative Pathogens. J. Med. Chem. 57 (23), 10144−10161. (12) Reck, F., Bermingham, A., Blais, J., Capka, V., Cariaga, T., Casarez, A., Colvin, R., Dean, C. R., Fekete, A., Gong, W., et al. (2018) Optimization of Novel Monobactams with Activity against Carbapenem-Resistant Enterobacteriaceae − Identification of LYS228. Bioorg. Med. Chem. Lett. 28 (4), 748−755. (13) Matsson, P., Doak, B. C., Over, B., and Kihlberg, J. (2016) Cell Permeability beyond the Rule of 5. Adv. Drug Delivery Rev. 101, 42− 61. (14) Capobianco, J. O., and Goldman, R. C. (1994) Macrolide Transport in Escherichia Coli Strains Having Normal and Altered OmpC and/or OmpF Porins. Int. J. Antimicrob. Agents 4 (3), 183− 189. E


Size Matters and How You Measure It: A Gram-Negative Antibacterial

Recommend Documents