5-Carboxytetramethylrhodamine-Ampicillin Fluorescence Anisotropy

Higher enzyme concentrations were not used because the initial rates were too ..... The Supporting Information is available free of charge on the ACS ...
0 downloads 0 Views 619KB Size
Subscriber access provided by UNIV OF TEXAS DALLAS

Article

5-Carboxytetramethylrhodamine-Ampicillin (5-TAMRAampicillin) Fluorescence Anisotropy-based Assay of Escherichia coli Penicillin-Binding Protein 2 Transpeptidase Inhibition Adam B. Shapiro, Janelle Comita-Prevoir, and Mark A Sylvester ACS Infect. Dis., Just Accepted Manuscript • DOI: 10.1021/acsinfecdis.8b00351 • Publication Date (Web): 08 Mar 2019 Downloaded from http://pubs.acs.org on March 12, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Infectious Diseases

5-Carboxytetramethylrhodamine-Ampicillin (5-TAMRA-ampicillin) Fluorescence Anisotropy-based Assay of Escherichia coli PenicillinBinding Protein 2 Transpeptidase Inhibition Adam B. Shapiro*, Janelle Comita-Prevoir, and Mark Sylvester Entasis Therapeutics, 35 Gatehouse Drive, Waltham, MA 02451 USA KEYWORDS: Penicillin-binding proteins, PBP2, transpeptidase, Escherichia coli, fluorescence anisotropy, fluorescence polarization The high-molecular mass penicillin-binding proteins (PBPs) are the essential targets of the -lactam classes of antibacterial drugs. In the Gram-negative pathogen Escherichia coli, these include PBP1a, PBP1b, PBP2, and PBP3. Techniques that enable facile measurement of the potency of inhibition of these targets are valuable for understanding structure-activity relationships in programs aimed at discovering new antibiotics to combat drug-resistant infections. Continuous, fluorescence anisotropy-based assays for inhibition of soluble constructs of PBP1a, PBP2 and PBP3 from the serious Gramnegative bacterial pathogens Pseudomonas aeruginosa and Acinetobacter baumannii, and PBP3 from E. coli using the fluorescent phenoxypenicillin analog BOCILLIN FL have been described previously, but this technique was not useful for PBP2 from E. coli due to a lack of change in fluorescence anisotropy or intensity upon reaction. Here we report that a fluorescent analog of ampicillin, 5-carboxytetramethylrhodamine-ampicillin (5-TAMRA-ampicillin), was useful as the indicator in a continuous fluorescence anisotropy-based kinetic assay for inhibition of a soluble construct of PBP2 from E. coli. The assay enables measurement of the bimolecular rate constant for inhibition kinact/Ki. This measurement was made for representative drugs from four classes of -lactams and for the diazabicyclooctenone ETX2514. 5-TAMRA-ampicillin was also useful in a fluorescence anisotropy-based assay for P. aeruginosa PBP2, and in fluorescence intensity-based assays with PBP1a and PBP3 from P. aeruginosa and A. baumannii, and PBP3 from E. coli.

There is an urgent need for new antibacterial drugs to counter the spread of multidrug-resistant pathogenic bacteria1,2 . Programs aimed at the discovery of new antibacterial drugs often focus on identifying potent inhibitors of molecular targets. In target-based antibacterial drug discovery projects, quantitative assays for inhibition of molecular targets enable the development of structure-activity relationships for potential antibacterial compounds, assisting in the design of potent analogs.

transpeptidases continue to be of interest as the targets for new antibacterial drugs. Therefore, projects aimed at discovering inhibitors of HMM PBPs require assays for measuring the potency of the compounds under investigation.

The high molecular mass (HMM) penicillin-binding proteins (PBPs) of bacteria are the molecular targets of the -lactam antibacterial drugs3,4, which include penicillins, cephalosporins, carbapenems, and monobactams5. These drugs are mechanism-based covalent inhibitors of the transpeptidase catalytic activity of the PBPs, which is responsible for forming the peptidic crosslinks of the peptidoglycan cell wall. In addition to the -lactams, some of the recently introduced diazabicyclooctanone and diazabicyclooctenone (DBO) class of mechanism-based serine -lactamase inhibitors exhibit intrinsic antibacterial activity due to HMM PBP transpeptidase inhibition6-10. As validated targets for antibacterial drugs, the HMM PBP

In a typical application of the gel assay, cell membranes are first reacted for a specific period of time with a range of concentrations of the PBP inhibitor to form covalent adducts. The remaining unreacted PBPs are then reacted with radioactive penicillin or BOCILLIN FL, a commercially available fluorescent penicillin derivative, in the continuing presence of the inhibitor. The membranes proteins are separated by SDS-PAGE, and the radioactivity or fluorescence of each of the PBP bands is measured. The inhibitor concentration producing 50% inhibition of labeling (IC50) is then calculated. The advantages of the gel method are (1) it allows inhibition of all the PBPs to be measured simultaneously, (2) it uses easily-prepared

There are two main biochemical techniques for quantifying the potency of PBP transpeptidase inhibition: (1) SDS-PAGE gel-based assays using native PBPcontaining cell membranes, and (2) assays using purified, soluble PBP constructs11.

ACS Paragon Plus Environment

ACS Infectious Diseases 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

bacterial cell membranes, and (3) it uses widely available equipment for SDS-PAGE. Some disadvantages are (1) it is laborious when several compounds are to be tested, (2) expensive equipment for quantitative measurement of band intensity is needed, (3) quantitation can be difficult when the PBP bands are very faint due to their low abundance, (4) any residual -lactamase activity in the membrane preparation can affect the results by degrading the inhibitor and/or the indicator compound, and (5) the IC50 calculation does not capture the time-dependent kinetics of covalent inhibition, expressed as the bimolecular rate constant kinact/Ki, or account for dissociation of the inhibitor. We previously reported methods for fluorescence anisotropy-based measurement of kinact/Ki for inhibition by -lactam drugs of the transpeptidase activities of purified, soluble constructs of the HMM PBPs PBP2 and PBP3 from the Gram negative pathogen Pseudomonas aeruginosa12,13. These assays measure, in a continuous microplate format, the reaction of the PBPs with BOCILLIN FL by means of the resulting increase in its fluorescence anisotropy. Inhibitors compete with BOCILLIN FL and thereby cause a decrease in the rate of the reaction. The value of kinact/Ki is obtained from a global fit of reaction progress curves obtained at several inhibitor concentrations to a kinetic model by numerical integration. The advantages of this method, given the availability of a fluorescence polarization-capable microplate reader, are (1) highthroughput, enabling the necessary data to be collected for several compounds or PBPs simultaneously in a short time (8 compounds in 40 minutes in a typical implementation, using a single PBP, or fewer compounds with multiple PBPs), and (2) its ability to supply the key kinetic parameter kinact/Ki. The inhibitor’s dissociation rate constant koff can also be measured in some cases. Assays for HMM PBP inhibition such as the BOCILLIN FL gel assay and the fluorescence anisotropy assay make use of reactions with -lactams rather than with the natural peptidoglycan substrate. Nevertheless, these assays are useful for investigating structure-activity relationships of HMM PBP inhibitors due to the fact that the assays directly measure the reactivity of inhibitors with the active site serine nucleophile of the PBPs. In the case of the fluorescence anisotropy assay, the measured bimolecular rate constant kinact/Ki combines information about both the reactivity (kinact) and the binding affinity Ki. The key challenges for using the fluorescence anisotropy assay to measure PBP inhibition kinetics are (1) obtaining a transpeptidase-active soluble construct of the PBP of interest, and (2) obtaining a sufficiently large change in anisotropy. We previously reported the preparation of soluble constructs of the HMM PBPs PBP1a, PBP2, and PBP3 from the Gram-negative pathogens Escherichia coli, P. aeruginosa, and Acinetobacter baumannii6,12,13. Unfortunately, there was no change in the fluorescence anisotropy of BOCILLIN FL when it reacted with E. coli PBP2, presumably because rapid rotation of the BODIPY

Page 2 of 14

fluorophore continued even when the BOCILLIN was bound to the PBP, nor was there a change in fluorescence intensity. The reason why BOCILLIN FL is an effective fluorescence anisotropy reagent for several HMM PBPs but not E. coli PBP2 is unknown. Presumably, some structural differences exist between the HMM PBPs that enable immobilization of the fluorophore upon reaction with the probe in some cases but not others. There is currently no published x-ray crystal structure of a PBP bound to BOCILLIN. It is important to measure the potency of PBP2 inhibition because PBP2 is essential for bacterial cell growth. Inhibition of PBP2 alone or in combination with other HMM PBPs, therefore, is responsible for or contributes to antimicrobial activity of -lactam drugs. Mecillinam is an example of a clinically useful -lactam antibiotic that selectively inhibits PBP2, and PBP2 is potently inhibited by carbapenems24. Moreover, the enhancer effect of recently reported DBO -lactamase inhibitors is attributed to inhibition of PBP26,9. In order to measure kinact/Ki for the DBO ETX2514 with E. coli PBP2, we sampled the reaction progress curves at various times by running portions of the reaction mixtures on SDS-PAGE to separate the purified PBP from the BOCILLIN FL, then measured the fluorescence intensities of the bands, an extremely laborious procedure6. To solve this problem, we prepared other fluorescent lactams, namely derivatives of ampicillin, to find one with a sufficient change in fluorescence anisotropy to use in the kinetic assay. This resulted in the identification of 5carboxytetramethylrhodamine-ampicillin (5-TAMRAampicillin) as suitable for this purpose. Here we describe results from the E. coli PBP2 fluorescence anisotropy-based assay using this new reagent, and also show that it can be used to measure acylation of several other HMM PBPs.

RESULTS AND DISCUSSION Fluorescence anisotropy assay with ampicillin derivatives. The -lactam antibiotic ampicillin (Fig. 1A) is a convenient starting point for the synthesis of fluorescent derivatives because of the presence of a primary amino group to which commercially available Nhydroxysuccinimide esters of fluorophores can be attached in a single-step reaction. Ampicillin derivatives attached to fluorescein and biotin have been reported14,15. The lack of a flexible 2-carbon linker between the fluorophore and the -lactam in the TAMRA-ampicillin derivative, as compared to BOCILLIN FL, may afford an improved performance in a fluorescence anisotropy assay by limiting rotational motion of the fluorophore in the bound state (“propeller effect”). Three fluorescent ampicillin derivatives were made using the fluorophores BODIPY FL (Fig. 1B), 5-TAMRA (Fig. 1C), and 6-TAMRA (Fig. 1D). In each case, the fluorescence anisotropy of the derivative increased upon reaction with E. coli PBP2 (Fig. 2 and Supporting Information Fig. S1A and S1B). The largest increase was observed with 5-TAMRA-ampicillin (Fig. 2). The

2 ACS Paragon Plus Environment

Page 3 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Infectious Diseases

fluorescence anisotropy of 5-TAMRA-ampicillin increased by 110 milli-units upon reaction with E. coli PBP2, and the rate of anisotropy change increased with increasing PBP2 concentration. There was almost no change in the total fluorescence intensity upon reaction (Supporting Information Fig. S1C), as desired for a fluorescence anisotropy assay. Figure 1. Chemical structures of ampicillin and fluorescent derivatives. (A) Ampicillin, (B) BODIPY-FL-ampicillin, (C) 5TAMRA-ampicillin, (D) 6-TAMRA-ampicillin.

pH optimization for E. coli PBP2 assay with 5-TAMRAampicillin. We observed previously with P. aeruginosa PBP2 that the pH optimum for the reaction with BOCILLIN FL was between 6.0 and 6.413. Similarly, with E. coli PBP2, the rate of the reaction with 5-TAMRAampicillin was highest at pH 6.0 in the pH 6.0-8.0 range, and decreased with increasing pH in sodium phosphate buffer (Fig. 3A). The same trend of decreasing reaction rate with increasing pH was observed with HEPES buffer between pH 7.0 and 8.0, with slightly lower activity in 50 mM HEPES buffer than in 0.1 M sodium phosphate buffer at equivalent pH (Supporting Information Fig. S2.). Further experiments with E. coli PBP2 were therefore performed in sodium phosphate buffer at pH 6.0. The initial rate of the anisotropy increase was directly proportional to the E. coli PBP2 concentration (Fig. 3B). The reaction of E. coli PBP2 with 6-TAMRA-ampicillin was also faster at pH 6 than at pH 7. The size of the anisotropy change of 6-TAMRA-ampicillin at pH 6 was smaller than that of 5-TAMRA-ampicillin at pH 6 (Supporting Information Fig. S3). Figure 2. Time-dependent fluorescence anisotropy increase of 30 nM 5-TAMRA-ampicillin upon reaction with E. coli PBP2 at pH 7.0. Progress curves from 4 replicate wells were averaged. The reaction volume was 6 l.

Measurements of kinact/Ki. To measure kinact/Ki for E. coli PBP2 transpeptidase inhibitors using the fluorescence anisotropy assay with 5-TAMRA-ampicillin, the same approach was employed as described previously for P. aeruginosa PBP2 and PBP3 using BOCILLIN FL6,12,13. Reaction progress curves measured at multiple inhibitor concentrations were globally fit by numerical integration to a simple kinetic model in which the inhibitor competes with 5-TAMRA-ampicillin for reaction with PBP2. The model does not deconvolute the initial equilibrium binding of the inhibitor to PBP2, characterized by the equilibrium dissociation constant Ki, and the subsequent formation of the covalent complex, characterized by the rate constant kinact. The reason for this is that both steps are presumed to produce the same fluorescence anisotropy increase, hence they cannot be distinguished. Therefore, only the composite second-order rate constant kinact/Ki was determined. Active site titration. When measuring kinact/Ki by the above method, we found that highly potent inhibitors could be used to perform an active site titration of the PBP. This was due to the fact that a high-quality fit to the data could not be obtained for very potent inhibitors if the nominal value of the enzyme concentration entered into the kinetic analysis program was higher than the effective active site concentration. To identify the effective active site concentration, the enzyme concentration setting was iteratively reduced until the value of goodness-of-fit parameter chi-squared was minimized. This procedure showed that the effective active site concentration of E. coli PBP2 was only 30% of the nominal concentration based on a protein assay, suggesting either that 70% of the protein was catalytically inactive, that the protein assay overestimated the concentration of the enzyme, the enzyme was impure, or a combination of the 3 issues. The PBP2 concentration obtained from the active site titration was used thereafter. For relatively weak inhibitors, an error in the enzyme concentration has a negligible effect on the value of kinact/Ki, but does have a substantial effect on the value of k+1, the rate constant for the reaction of PBP2 with 5-TAMRA-ampicillin. For the most potent inhibitors, an

3 ACS Paragon Plus Environment

ACS Infectious Diseases accurate effective enzyme concentration is necessary to obtain a close fit of the model to the data. Figure 3. (A) Effect of pH on the rate of the reaction of 30 nM 5-TAMRA-ampicillin with 120 nM E. coli PBP2. Each buffer consisted of 0.1 M sodium phosphate with 0.01% Triton X-100. Progress curves from 8 replicate wells were averaged. (B) Initial rate of anisotropy increase versus E. coli PBP2 concentration in the reaction with 5-TAMRA-ampicillin. The buffer was 0.1 M sodium phosphate pH 6.0 + 0.01% Triton X100. Progress curves from 8 replicate wells were averaged. Data from the first 100 seconds were used to calculate initial rates. Higher enzyme concentrations were not used because the initial rates were too fast to measure accurately.

160

Relative anisotropy x 1000

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

A

140 120

pH 6.0 pH 6.5 pH 7.0 pH 7.5 pH 8.0

100 80 60

Page 4 of 14

Inhibition by ETX2514 of PBP labeling in E. coli membranes is shown in Fig. 4B. The gel shows that ETX2514 selectively inhibits PBP2, with an IC50 of 250 nM. Based on the 40-min endpoint in the fluorescence anisotropy assay, the IC50 in that assay is 160 nM, similar to the result from the gel assay. It is noteworthy that the fluorescence anisotropy assay produced a similar potency measurement as the gel assay despite the different environments of the PBP2 in each case, i.e. isolated in solution versus in the native membrane. Figure 4. (A) Fit of the kinetic model (solid lines) to the data (points) for inhibition of the reaction of 100 nM E. coli PBP2 with 30 nM 5-TAMRA-ampicillin by various concentrations of ETX2514, as shown in the legend in M. Curves and data for ETX2514 concentrations between 2.56 and 20.48 M have been omitted for clarity but were included in the calculation. The values of k+1 and kinact/Ki were 78,000 M-1s-1 and 16,000 M-1s-1, respectively. There was no discernible dissociation of the E-I complex, i.e. k-2 5 x 10-5 s-1 existed, a value of 1 x 10-4 s-1 was entered into the fitting program initially. In most cases, the final value of koff fell below 5 x 10-5 s-1. The carbapenems, as a class, were the most potent inhibitors of E. coli PBP2, with kinact/Ki in the 104 – 105 M-1s-1 range. Among the other -lactams tested, only mecillinam (also called amdinocillin), a specific inhibitor of PBP216, had potency comparable to the carbapenems. A few of the penicillins and cephalosporins had kinact/Ki values in the 103 – 104 M-1s-1 range, but most exhibited weaker or undetectable inhibition. The monobactam aztreonam, a PBP1a- and PBP3-specific drug17, showed no detectable inhibition of E. coli PBP2, and the other monobactams also had weak or undetectable inhibition. Included in Table 1 are published E. coli membrane SDSPAGE assay IC50s for most of the -lactams tested in the E. coli PBP2 fluorescence anisotropy assay. Data were selected from papers in which several drugs were tested in the same way. Since the methods used by different authors were not identical, the IC50s should not be compared between papers. In addition to the data shown in Table 1, Bush et al18 reported PBP2 IC50s in the gel assay of >200 g/ml for both aztreonam and tigemonam, consistent with their undetectable acylation rate constants in the fluorescence anisotropy assay. It should be noted that the usual procedure for the gel assay is to preincubate the membranes with the inhibitor prior to the addition of labeled penicillin. In contrast, the fluorescence anisotropy measurements were made without inhibitor preincubation. Instead, the inhibitor and 5-TAMRAampicillin were mixed at the same time with PBP2. Preincubation has the effect of increasing the potency of the inhibitor compared with no preincubation. The potency of inhibition measured by the gel assay in concentration units cannot be compared directly with the kinact/Ki values obtained from the anisotropy assay in M-1s-1 units because of the difference in their units as well as because of the typical use of a preincubation in the gel assay but not in the anisotropy assay. Nevertheless, in general, there is strong qualitative agreement in terms of order of potency between the fluorescence assay and the gel assay in the sense that low IC50 corresponds to high kinact/Ki, although the agreement is imperfect in a few cases. A possible reason for the disagreement is the presence of trace levels of -lactamase activity in membrane preparations that degrade different -lactams with different levels of efficiency.

Collecting and analyzing the data for one measurement for each of the 36 compounds listed in Table 1 in order to measure kinact/Ki required approximately 2 days’ work for a single investigator and consumed 46 g of PBP2 and 1½ 384-well assay plates. In order to determine whether measurable values of k-2 in the fluorescence anisotropy assay are due to dissociation of the intact inhibitor from the E-I complex or dissociation of a hydrolyzed product, a separate experiment can be performed in which hydrolysis of the -lactam ring of the inhibitor is directly monitored by a change in ultraviolet absorbance. An example is shown in Supporting Information Fig. S4B, which shows that mecillinam (k-2 = 2.4 (±0.3) x 10-4 s-1) was hydrolyzed by E. coli PBP2. The second-order rate constant for hydrolysis can be measured by such an experiment, but the value of this constant cannot be directly compared to the first-order rate constant k-2 obtained from the fluorescence anisotropy experiment because the 2 constants have different units (M-1s-1 versus s-1, respectively). 5-TAMRA-ampicillin reactions with other PBPs. 5TAMRA-ampicillin was also useful in fluorescence-based assays for other HMM PBP soluble constructs. A substantial anisotropy increase was observed upon reaction with P. aeruginosa PBP2 (Fig. 5A). There was a slight decrease in fluorescence intensity of 12% over the course of the 40-min reaction with 330 nM PBP2. The anisotropy increase with A. baumannii PBP2 was very small at about 10 milli-units with 330 nM PBP2 in 40 min, which was also the case when BOCILLIN FL was used. This was accompanied by a 5% decrease in fluorescence intensity. To use the small fluorescence anisotropy change to measure acylation rate constants for A. baumannii PBP2, it is necessary to average the progress curves from multiple wells. No significant change in fluorescence anisotropy or intensity was observed with 200 nM E. coli PBP1a and 30 nM 5-TAMRA-ampicillin after 40 minutes When 5-TAMRA-ampicillin was reacted with PBP1a or PBP3 from P. aeruginosa and A. baumannii, or with PBP3 from E. coli, there were large decreases in fluorescence intensity as well as increases in fluorescence anisotropy. Since the large decrease in fluorescence intensity distorts the anisotropy-based progress curve, leading to an inaccurate value of kinact/Ki, the intensity decrease is a better signal to use for monitoring the reaction with these PBPs. The result for P. aeruginosa PBP1a is shown in Fig. 5B. Similar results were obtained with A. baumannii PBP1a and PBP3 and E. coli PBP3 (Supporting Information Fig. S5A-C). With P. aeruginosa PBP3, there was also a large 5TAMRA-ampicillin fluorescence intensity decrease upon reaction, but the reaction was so rapid that it went to completion too quickly to measure with 30 nM 5-TAMRAampicillin and the range of PBP concentrations used previously. Since the PBP concentration must be equal to or higher than the 5-TAMRA-ampicillin concentration in order to achieve the maximal signal, it was necessary to lower the concentrations of both reactants. The 5-TAMRA-

5 ACS Paragon Plus Environment

ACS Infectious Diseases

reducing both the 5-TAMRA ampicillin and PBP3 concentrations. Satisfactory 40-min progress curves were obtained with 10 nM 5- TAMRA-ampicillin and 20 nM E. coli PBP3 (Supporting Information Fig. S5C).

ampicillin concentration was reduced from 30 nM to 1 nM. The reaction was nearly complete after 40 min with 1.2 nM PBP3 (Fig. 5C). With E. coli PBP3, the reaction with 5TAMRA-ampicillin was also so strong that it required

Fig. 5. (A) Time-dependent fluorescence anisotropy increase of 30 nM 5-TAMRA-ampicillin upon reaction with P. aeruginosa PBP2 at pH 6.0. Progress curves from 4 replicate wells were averaged. The reaction volume was 6 l. (B) Timedependent fluorescence intensity decrease of 30 nM 5-TAMRA-ampicillin upon reaction with P. aeruginosa PBP1a at pH 7.0. Progress curves from 2 replicate wells were averaged. The reaction volume was 30 l. Progress curves for PBP1a concentrations between 210 and 330 nM have been omitted for clarity. (C) Time-dependent fluorescence intensity decrease of 1 nM 5-TAMRA-ampicillin upon reaction with P. aeruginosa PBP3 at pH 7.0. Progress curves from 8 replicate wells were averaged. The reaction volume was 30 l. Progress curves for PBP3 concentrations between 1.2 and 2.8 nM have been omitted for clarity. (D) Inhibition of A. baumannii PBP3 by aztreonam in the 5-TAMRA-ampicillin fluorescence intensity assay. The 30-l reaction volume contained 30 nM 5-TAMRA-ampicillin, 120 nM A. baumannii PBP1a, and various concentrations of aztreonam (shown in the legend in M) in 0.1 M sodium phosphate (pH 7.0) and 0.01% Triton X-100. The progress curves for 0.02 to 0.16 M aztreonam are omitted for clarity but were included in the calculation. The best-fit values of kinact/Ki and koff are shown in Table 2. The data are shown as points and the best-fit curves are shown as lines. RFU, relative fluorescence units.

280000

Fluorescence (RFU)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 14

240000

D

20.48 10.24 5.12

200000

2.56

160000

1.28

120000

0.64

80000

0.32 0

40000 0

The PBP1a/PBP3-selective inhibitor aztreonam and the PBP2-selective inhibitor mecillinam were used to test the ability of the 5-TAMRA-ampicillin fluorescence anisotropy and fluorescence intensity kinetic assays to measure kinact/Ki for PBP1a and PBP3 from A. baumannii and P.

400 800 1200 1600 2000 2400 Time (s)

aeruginosa, PBP3 from E. coli, and PBP2 from P. aeruginosa. As expected, mecillinam only inhibited PBP2 and aztreonam only inhibited PBP1a and PBP3 at the maximal tested concentration of 20 M. Values of kinact/Ki were calculated for each example of inhibition (Table 2).

6 ACS Paragon Plus Environment

Page 7 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Infectious Diseases

The results are closely comparable to those reported previously using the BOCILLIN FL fluorescence anisotropy assay19. An example of the data and fit to the kinetic mode in a 5-TAMRA-ampicillin fluorescence intensity assay for A. baumannii PBP1a inhibition by aztreonam is shown in Fig. 5D. 5-TAMRA-ampicillin extends the fluorescence anisotropy assay to E. coli PBP2, a PBP that was not accessible with BOCILLIN. It is possible that future developments of novel probes will further extend the utility of fluorescence anisotropy and/or fluorescence intensity assays to those PBPs for which neither BOCILLIN nor 5-TAMRAampicillin produces a sufficient signal.

CONCLUSION The 5-TAMRA-ampicillin fluorescence anisotropy assay provides the means quickly to collect quantitative inhibition kinetic parameters of a large number of E. coli PBP2 inhibitors with minimal expenditure of consumables and purified protein. 5-TAMRA-ampicillin is easily synthesized from commercially available reagents in a single step. 5-TAMRA-ampicillin can also be used in fluorescence anisotropy- or fluorescence intensity-based kinetic assays for inhibition of soluble constructs of other HMM PBPs. 5-TAMRA-ampicillin provide a readily accessible alternative to BOCILLIN FL for fluorescencebased PBP assays in cases where the latter provides an insufficient signal, or when a longer excitation wavelength than that of BOCILLIN FL is desired.

METHODS Chemicals. Ampicillin, cefepime, cefixime, cefoxitin, ceftazidime, cefuroxime, and imipenem were from US Pharmacopeia (Rockville, MD). Amoxicillin, aztreonam, nafcillin, and piperacillin were from MP Biomedicals (Santa Ana, CA). Carbenicillin, carumonam, cefazolin, cefoperazone, cefotaxime, cefsulodin, ceftibuten, ceftriaxone, cephalexin, cephalothin, cloxacillin, dicloxacillin, mecillinam, meropenem, methicillin, oxacillin, penicillin G, and ticarcillin were from MilliporeSigma (St. Louis, MO). Cefpodoxime was from Alfa Aesar (Haverhill, MA). Doripenem was from Sequioa Research Products (Pangbourne, UK). Ertapenem was from Thermo Fisher Scientific (Waltham, MA). Biapenem was from AvaChem Scientific (San Antonio, TX). Tebipenem was from BOC Sciences (Shirley, NY). Tigemonam was prepared by Pharmaron (Irvine, CA) according to published routes20. ETX2514 was prepared as described previously6. Bodipy FL N-hydroxysuccinimide (NHS) ester was from BroadPharm (San Diego, CA). 5TAMRA NHS ester and 6-TAMRA NHS ester were from MedchemExpress (Monmouth Junction, NJ). BOCILLIN FL was from Thermo Fisher Scientific. Preparation of PBPs. Soluble constructs of PBPs were prepared as described previously6,12,13.

Synthesis of fluorescent ampicillin derivatives The synthesis of fluorescently labeled ampicillin analogs was conducted using commercially available preactivated NHS fluorescent dyes and ampicillin sodium (MilliporeSigma). The final compounds were isolated from byproducts by reverse phase chromatography, and characterized by LC-MS and 1H NMR analysis. The BODIPY analog was prepared as follows: ampicillin sodium (9.5 mg, 0.027 mmol) and (2,5-dioxopyrrolidin-1yl) 3-(2,2-difluoro-10,12-dimethyl-1,3λ^5-diaza-2λ^4boratricyclo[7.3.0.0^3,7]dodeca-3,5,7,9,11-pentaen-4yl)propanoate (BODIPY FL NHS) (10.7 mg, 0.027 mmol) were combined and dissolved in dimethylsulfoxide (DMSO) (0.3 mL). The reaction mixture was stirred for 30 minutes at room temperature, then purified by reverse phase with a 5.5 g ISCO C18AQ column (Teledyne ISCO, Lincoln, NE) eluted with 0%-100% acetonitrile in water, with detection of absorbance at 214 nm. Two fractions containing (2S,5R,6R)-3,3-dimethyl-7-oxo-6-[[(2R)-2-[3(2,2-difluoro-10,12-dimethyl-1-aza-3-azonia-2λ^4boratricyclo[7.3.0.0^3,7]dodeca-3,5,7,9,11-pentaen-4yl)propanoylamino]-2-phenyl-acetyl]amino]-4-thia-1azabicyclo[3.2.0]heptane-2-carboxylic acid (BODIPY FLampicillin) were lyophilized to afford a bright orange solid, 6.2 mg (36% yield). 1H NMR (300 MHz, DMSO-d6,): δ 1.37 (s, 3H), 1.49 (s, 3H), 2.24 (s, 3H), 2.45 (s, 3H), 2.63 (t, 2H, J = 8 Hz), 3.07 (t, 2H, J = 7.5 Hz), 3.77 (s, 1H), 5.23 (d, 1H, J = 3.9 Hz), 5.32 (dd, 1H, J = 3.9, 8 Hz), 5.75 (d, 1H, J = 8.3 Hz), 6.27 (s, 1H), 6.35 (d, 1H, J = 4.1 Hz), 7.06 (d, 1H, J = 4.1 Hz), 7.28 (m, 3H), 7.41 (m, 2H), 7.66 (s, 1H), 8.60 (d, 1H, J = 8.3 Hz), 8.87 (d, 1H, J = 8.2 Hz). EIMS m/z: [M+H]+ 624, tr = 1.03 min., 90% pure. The 5-TAMRA analog was prepared by adding DMSO (0.1 mL) to solid ampicillin sodium (1.4 mg) (approximate weight) and 5-TAMRA NHS (~1 mg) in a conical vial. The reaction mixture was stirred for two hours and applied directly to a 6 g Interchim RPAQ C18 column (Montluçon, France) and eluted with 0%-100% acetonitrile in water, with detection of absorbance at 214 nm. The fractions containing the product peak were lyophilized to give the product 2-[3-(dimethylamino)-6-dimethyliminioxanthen-9-yl]-5-[[(1R)-2-oxo-1-phenyl-2-[[(2S,5R,6R)-2carboxy-3,3-dimethyl-7-oxo-4-thia-1azabicyclo[3.2.0]heptan-6yl]amino]ethyl]carbamoyl]benzoate (5-TAMRAampicillin) (1.6 mg, quantitative yield) as a purple solid. 1H NMR (300 MHz, DMSO-d6): δ 1.38 (s, 3H), 1.46 (s, 3H), 2.94 (s, 12H), 3.79 (s, 1H), 5.29 (d, 1H, J = 3.8Hz), 5.37 (dd, 1H, J = 3.8, 8.1Hz), 5.96 (d, 1H, J = 8.1 Hz), 6.51 (m, 6H), 6.61 (br s, 1H), 7.34 (m, 4H), 7.55 (d, 2H, J = 6.9Hz), 8.26 (dd, 1H, J = 1.6, 8.1 Hz), 8.55 (m, 1H), 8.80 (d, 1H, J = 8.1 Hz), 9.30 (d, 1H, J = 8.3 Hz). EIMS m/z: [M+H]+ 762, tr= 0.85min., 90% pure. The 6-TAMRA analog was prepared by adding DMSO (0.3 mL) to solid [(2S,5R,6R)-3,3-dimethyl-7-oxo-6-[[(2R)-2amino-2-phenyl-acetyl]amino]-4-thia-1-

7 ACS Paragon Plus Environment

ACS Infectious Diseases 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

azabicyclo[3.2.0]heptane-2-carbonyl]oxysodium (ampicillin) (7.39 mg, 0.02 mmol) and 2-[3(dimethylamino)-6-dimethyliminio-xanthen-9-yl]-4-(2,5dioxopyrrolidin-1-yl)ox ycarbonyl-benzoate (6-TAMRA NHS) (5 mg, 0.01 mmol)

8 ACS Paragon Plus Environment

Page 8 of 14

Page 9 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Infectious Diseases

Table 1. Measurements of kinact/Ki and koff (k-2) for inhibition of E. coli PBP2 by -lactam antibiotics measured with the 5-TAMRA-ampicillin fluorescence anisotropy assay. Averages ± standard deviations of 3 measurements made on separate occasions are shown. The lower limit of measurement of kinact/Ki and koff were 10 M-1s-1 and 5 x 10-5 s-1, respectively. When kinact/Ki values of 10 were measured on separate occasions, both results are shown. Published results from SDS-PAGE gel-based assays with E. coli membrane proteins are shown for comparison. Antibiotic

kinact/Ki (M-1s-1)

koff (s-1)

Gel assay IC50 (g/ml) Ref. 21

Penicillins

Ref. 22, 23

Ref. 24

amoxicillin

310 ± 50