Discovery of a New Antibiotic Demethoxytetronasin Using a Dual-Sided Agar Plate Assay (DAPA)

Abstract

For over a century, researchers have cultured microorganisms together on solid support—typically agar—in order to observe growth inhibition via antibiotic production. These simple bioassays have been critical to both academic researchers that study antibiotic production in microorganisms and to the pharmaceutical industry’s global effort to discover drugs. Despite the utility of agar assays to researchers around the globe, several limitations have prevented their widespread adoption in advanced high-throughput compound discovery and dereplication campaigns. To address a list of specific shortcomings, we developed the dual-sided agar plate assay (DAPA), which exists in a 96-well plate format, allows microorganisms to compete through opposing sides of a solid support in individual wells, is amenable to high-throughput screening and automation, is reusable, and is low-cost. Herein, we validate the use of DAPA as a tool for drug discovery and show its utility to discover new antibiotic natural products. From the screening of 217 bacterial isolates on multiple nutrient media against 3 pathogens, 55 hits were observed, 9 known antibiotics were dereplicated directly from agar plugs, and a new antibiotic, demethoxytetronasin (1), was isolated from a Streptomyces sp. These results demonstrate that DAPA is an effective, accessible, and low-cost tool to screen, dereplicate, and prioritize bacteria directly from solid support in the front end of antibiotic discovery pipelines.

Introduction

Dating back to the early 1900s, some of the first clinical antibiotics such as penicillin and gramicidin were discovered using simple and inexpensive agar assays that revealed growth inhibition between competing microorganisms.^1,2 Throughout the next century, agar provided a foundation for microbial drug discovery efforts, serving as both a solid support for the isolation of microorganisms and as an accessible medium on which microbial growth inhibition could be detected. The use of agar resulted in the discovery of most of the major natural product (NP) antibiotic classes used in the clinic. Despite the positive attributes of agar-based bioassays (accessible, rapid, and low cost), generally speaking, academic research programs gradually replaced them by employing more advanced bioassay screening in the front end of the NP discovery pipeline.^3,4 But an opportunity remained to design a high-throughput assay to screen for growth inhibition directly from bacterial cell mass on agar. Such an early screening step would greatly reduce the need to build NP fraction libraries, the end result of a series of costly steps that are often a prerequisite to screen a library of bacterial isolates in more advanced bioassay screens. Furthermore, access to these advanced screens or to the collaborative expertise to implement them is often an obstacle for smaller laboratories in academia.

Traditional agar growth inhibition bioassays operate under a central concept: a solid support, typically agar, provides a surface from which the antibiotic activity of a compound, fraction, or opposing microorganism can be assessed.⁵ These assays have been effective because of their simplicity, efficiency, accessibility, and low cost, and they have been well summarized.⁶ Despite these advantages, several limitations exist. These include (1) unwanted same surface interactions between microorganisms since spreading of spores, overgrowth/swarming behavior, and chemical signaling between neighboring bacteria often interfere with a competition experiment; (2) difficulty observing or accurately quantifying growth inhibition, as many bacteria are transparent and not readily amenable to visual measurements; (3) multiple laborious steps involved in assay execution; and (4) compatibility with high-throughput automation since solid support assays often either occur on standard circular Petri dishes or require small, individual agar plugs to be cut/removed from one plate and placed onto another. To overcome some of these challenges, a few notable innovations have been made; these are discussed in detail in the Supporting Information (Supp. discussion).⁷⁻¹⁰ In order to overcome the aforementioned limitations, we developed the dual-sided agar plate assay (DAPA), which exists in a 96-well plate format, allows microorganisms to compete on opposing sides of a solid support in individual wells, is amenable to high-throughput screening and automation, and is low-cost. The plate was custom engineered out of non-toxic material, is autoclavable, and therefore reusable. Herein, we validate the use of DAPA as a tool for antibiotic discovery and demonstrate its capacity in the front end of this process, as it allows for compound dereplication and hit prioritization directly from bacteria grown in multiwell plates on solid support. This process is highlighted in the discovery of the new antibiotic demethoxytetronasin (1).

Results and Discussion

DAPA Plate Design

The dual-sided agar plate was 3D-printed from non-toxic resin, designed in 48- and 96-well formats, and made to be compatible with high-throughput automation (Figure 1i). Each well contains a custom-built channel designed to anchor the solid support plug—typically agar—in place. In order to suspend agar plugs in each well, a silicon base was created from a mold and inserted into one side of the plate. This provides a surface upon which molten agar is poured and will settle. After the silicon base is removed, agar plugs remain suspended in each well of the plate. Various combinations of environmental microorganisms, biological and chemical elicitors, and pathogens can be inoculated on either side of each agar plug, which allows for diffusion of metabolites and facile detection of growth inhibition, particularly when species contain an intrinsic or engineered fluorescent label. Fluorescence and optical density can be employed as orthogonal methods of visualizing growth inhibition (Figure 1iii). The well design allows for layering of multiple nutrient media types if one growth condition is not suitable for the two bacteria under investigation.

Figure 1.

Design of DAPA. (i) DAPA plate and the silicone insert. (ii) Sideview of a single DAPA well; T = top and B = bottom. A red silicone insert allows wells to be filled with agar, and custom-engineered interior channels allow the plug to be suspended in the middle of the well upon the removal of silicon. An environmental bacterium (green) is incubated and produces metabolites into the agar plug. (iii) DAPA plate is inverted, and a pathogen (blue) is added. After an appropriate incubation period, growth inhibition may be assessed using orthogonal methods of detection including fluorescence and optical density via a redox dye such as MTT.

Validation of DAPA

DAPA performance was evaluated according to a set of criteria that are essential for the design of robust in vitro NP antibiotic bioassays.¹¹ First, the ability of DAPA to measure growth inhibition of a pathogen was evaluated against antibiotic standards and antibiotic-producing ATCC strains. The following statistical parameters were characterized: screening window coefficient (Z′), signal to noise (S/N), signal to background (S/B), coefficient of signal variation (CV %) and assay variation (CVA %), and signal window (SW). A Z′-factor, known as the screening window coefficient, measures the dispersion of the assay signals. For a cell-based assay to be considered reliable, the Z′ should be greater than 0.4.⁹Z′ values in the range of 0.44–0.96 were derived from relative fluorescence units (RFU), and OD₆₀₀ data determined when three pathogens (Pseudomonas aeruginosa ATCC 10145GFP, Enterobacter aerogenes ATCC 13048 mCherry, and Staphylococcus aureus ATCC 29213) were screened against known antibiotics moxifloxacin and gentamicin (Table S1). Signal-to-noise ratios were well within the acceptable range, observed as 15.1 and 61.3 in RFUs and OD₆₀₀, respectively. A summary of full parameters that indicate that DAPA satisfies sufficient assay performance and sensitivity is listed in Table S2. Validation of the growth inhibition quantification process in DAPA can be found in the Supporting Information (Figure S1A).

Next, to provide evidence that DAPA plates may be used to detect growth inhibition from competition assays with environmental test microorganisms, a panel of antibiotic-producing strains (cycloserine producer Streptomyces lavendulae ATCC 11924, tetrazomine/capreomycin producer Saccharothrix mutabilis ATCC 23892, neomycin producer Streptomyces fradiae ATCC 10743, and bacilysin producer Bacillus subtilis PY79) was screened against the panel of three pathogens previously described. Working directly with live strains, NP antibiotic production levels are expected to vary as a function of nutrient media type, and this phenomenon was observed (Figure S1B). Generally speaking, S. fradiae showed strong inhibition against the pathogen panel across media types, as measured by RFUs and OD₆₀₀. S. mutabilis, S. fradiae, and B. subtilis showed varying levels of tetrazomine/capreomycin, neomycin, and bacilysin production, respectively, as expected. Overall, the observation of growth inhibition was achieved using live antibiotic-producing strains. From these DAPA results, a threshold of ≥90% pathogen inhibition was set in the following screen in order to prioritize strains that produce a sufficient concentration of antibiotics for isolation/structure elucidation experiments.

Screening of an Environmental Bacterial Library

A library of 217 environmental bacterial isolates was screened against the aforementioned panel of three pathogens using DAPA. This library consisted of 195 isolates derived from freshwater environments in Iceland; 13 isolates from Lake Michigan sediment; 2 isolates from marine sediment in Massachusetts; and 7 isolates derived from Great Lakes freshwater sponges. All isolates are in the phylum Actinomycetota (72, formerly Actinobacteria), Pseudomonadota (103, formerly Proteobacteria), or Bacillota (42, formerly Firmicutes). Each environmental isolate was incubated into four individual wells in a 96-well DAPA plate, with each well containing a different nutrient type (A1, M1, NZSG, Blood). After a 5 day incubation period, a thin layer of Muller Hinton agar was poured on the bottom side of each agar plug, and the pathogen was inoculated. This step is only necessary if the researcher chooses to grow test isolates and pathogens on different nutrient media. After an overnight incubation period, growth inhibition was assessed using a fluorescence readout for P. aeruginosa and E. aerogenes and optical density measurement with redox active MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) for S. aureus. Each pathogen was assessed separately for a total of 2604 individual experiments. A total of 55 environmental strains inhibited ≥90% growth of any one pathogen (22 against P. aeruginosa; 37 against E. aerogenes; and 9 against S. aureus). We then employed a hit prioritization scheme. Each of the 55 hits was grown in 100 mL of liquid culture, extracted, and subjected to a standard single dose microbroth inhibition assay. Based on antibiotic potency against their respective target pathogen, the top eight hits that exhibited ≥90% growth inhibition were advanced to compound dereplication efforts. Secondary confirmation bioassays, like the microbroth one described here, are useful to further prioritize hits according to the specific research infrastructure available, which in this case was the ability of an environmental bacterium to reproduce bioactivity when grown in liquid culture. However, this step is not required to successfully carry out the DAPA screening pipeline.

In regard to the detection of growth inhibition in the DAPA assay, it is possible to perform both qualitative and quantitative analyses. For qualitative analyses, MTT detection can be performed directly in DAPA wells. This is typically employed when speed and high-throughput are desired. Quantitative analyses can be achieved by using a liquid transfer of pathogen cell mass to a blank 96-well plate, adding MTT, and obtaining readout with a plate reader, although this adds an additional step and consumables to the pipeline (see Supporting Information for the protocol).

Dereplication of Known Antibiotics from DAPA Agar Plugs

One major advantage of DAPA is the ability to dereplicate known antibiotics directly from agar plugs. Select DAPA agar plugs were extracted, filtered, and analyzed using UPLC-qTOF MS/MS analysis. Known antibiotics were dereplicated upon the comparison of tandem MS spectra with entries in the Global Natural Products Social Molecular Networking (GNPS) platform.¹² In total, from the 8 prioritized hits in the antibiotic screen, 9 structures were dereplicated directly from acetone/methanol extracts of single DAPA agar plugs (Figures 2, S2). This provides a major advantage to groups engaging in the bioactive NP discovery from microorganisms. Researchers have the option to screen small bacterial cell mass in a high-throughput bioassay and perform compound dereplication directly from single DAPA wells. This is an inexpensive alternative to what is typically a laborious front-end endeavor that requires the creation of strain and subsequent NP fraction libraries, a process that can take months to years and cost tens to hundreds of thousands of dollars.

Figure 2.

(A) One screening plate of environmental bacteria against P. aeruginosa ATCC 10145GFP using fluorescence detection; this represents one of the 37 DAPA plates used in this screen. Extraction/UPLC-MS/MS/GNPS analysis of the duplicate wells highlighted in blue led to dereplication of the known antibiotic bacillomycin LC2 (Figure S2). (B) List of all known antibiotics dereplicated directly from DAPA wells in this screen; three-dimensional structures are putative since our MS analysis is not sufficient to determine the configuration at each stereocenter.

Discovery of New Antibiotic Demethoxytetronasin from Hit Strain K1063

Strain K1063 was prioritized for antibiotic discovery based on its growth inhibition activity of ≥90% against S. aureus ATCC 29213 in the DAPA assay and for the absence of known compounds observed after LCMS/MS/GNPS analysis. It was isolated from sediment collected using SCUBA in the small fissure between the North American and Eurasian continental plates at Nesgja, Iceland. It was identified by 16S rRNA analysis to be a Streptomyces sp. After fermentation in 34 L of NZSG nutrient media, the cells and broth were extracted and subjected to several successive normal- and reversed-phase chromatographic experiments using an antibiotic assay to guide the isolation of bioactive compounds. In the course of bioassay-guided fractionation, the known ionophore antibiotic tetronasin (2) and new antibiotic demethoxytetronasin (1) were isolated (Figures 3, S3). Tetronasin (2) was identified on the basis of structure elucidation via one- and two-dimensional NMR analysis (Table 1, Figures S4 and S5), HRMS analysis (Figure S6), comparison of ¹H and ¹³C NMR resonances/¹H coupling constants to those published in the literature,¹³⁻¹⁵ and comparison of optical rotation (OR) data to published values.^16,17 This provided a reliable foundation to elucidate the structure of 1, as the singular difference between the structures is the absence of a methoxy moiety at C-27 (Figure 3). It should be noted that minor resonance doubling was observed in the NMR spectra of 1 and 2 caused by the interconversion between free acid and salt forms. The molecular formula of the demethoxytetronasin (1) sodium adduct was C₃₅H₅₄O₈Na as determined by HRESIQTOFMS analysis and was 14 mass units less than that of 2 (Figure S7). The ¹H and ¹³C NMR data of 1 were nearly identical to those of 2 with the exception that no methoxy resonance was present (δ_C 57.8 in 2, C-35), and a characteristic shielding effect at C-27 was observed (δ_C 68.4 in 1, δ_C 78.8 in 2, Figures S4a, S8 and S9). COSY, HSQC, and HMBC NMR experiments allowed the construction of the planar structure of 1 (key COSY and HMBC correlations listed in Figures 3, S10 and S12).

Figure 3.

Antibiotic structures isolated from strain K1063 and key 2D NMR correlations of 1.

Table 1. ¹H NMR (600 MHz) and ¹³C NMR (125 MHz) Data for 1 in CDCl₃.

position	δH (J in Hz)	δC
1		177.5
2		96.9
3		193.5
4	4.28 d (15.0)	70.4
	4.36 d (15.1)
5		202.6
6	3.90 dq (7.1, 4.1)	43.2
7	1.85 td (10.7, 4.4)	47.7
8	1.38 m	33.9
9	1.11 m	35.3
	1.64 m
10	1.24 m	25.6
	1.59 m
11	1.01 m	35.3
	1.37 m
12	2.46 qd (10.7, 4.4)	36.0
13	5.14 d (10.4)	141.7
14		130.6
15	3.28 d (9.8)	90.1
16	1.33 m	32.6
17	1.21 m	31.9
	1.81 m (dq 13.0, 3.5)
18	1.44 qd (13.1, 3.8)	31.3
	1.66 m
19	3.85 m	78.3
20	5.53 dd (9.0, 15.2)	130.9
21	6.05 dd (9.9, 15.1)	141.3
22	2.22 m	39.4
23	3.68 dd (4.0, 10.3)	85.9
24	2.32 m	34.4
25	1.58 m	34.3
	1.89 td (11.7, 6.5)
26	4.20 ddd (10.7, 6.4, 2.3)	79.7
27	3.84 m	68.4
28	0.98 d (5.2)	16.3
29	0.97 d (6.3)	8.7
30	1.18 d (6.5)	19.7
31	3.78 d (11.1)	56.3
	4.27 d (11.2)
32	0.56 d (6.6)	18.1
33	0.94 d (6.2)	16.2
34	0.93 d (6.8)	13.6

The relative configuration of 1 was solved based on (a) comparison of ¹H and ¹³C NMR chemical shifts with experimentally isolated 2, and values reported in the literature for NP 2(13,15) and synthetic 2;¹⁷ (b) comparison of experimental ³J_HH coupling constants with literature values;^14,15 and (c) comparison of correlations from two-dimensional NOESY NMR experiments of 1 to the solution structure of tetronasin published by Martinek et al.¹⁸ Knowledge of the solution structure was critical to understanding long-range NOESY correlations. NOESY correlations between H₃-29 and H-12 and from H-8 to H-12 confirmed the relative configuration of stereocenters about the cyclohexane moiety. The configuration of the C-13–C-14 double bond was confirmed to be E based on a NOESY correlation of H₂-31 to H-20 and from H-13 to H-15; these correlations also supported the diaxial positioning of H-15 and H-19. Measured ³J_HH coupling constants of H-15, -16, and -19 supported their axial configurations in the methylated tetrahydropyran moiety and agreed well with those reported in the literature.^14,15 Furthermore, the observed ³J_HH coupling constant (J_20,21 = 15.1 Hz) is in accordance with a trans-oriented double bond. NOESY correlations from H₃-34 to H-22 and H-26 confirm the position of the methyl group and other stereocenters in the tetrahydrofuran ring and agree well with the solution model of tetronasin.¹⁸ Configurations in the furan ring were also confirmed through the calculation of ³J_HH coupling constants and the comparison of experimental J_{22,23; 23,24; 26,27} values with those reported in Grandjean and Laszlo.¹⁴ The proposed absolute configuration is based on a direct comparison of experimental ECD results of 1 with isolated 2, which were superimposable (Figure S14).

Antibiotic Activity of 1 and 2

Compound 1 is part of a rare class of linear polyketides, of which only three members have been reported to date.¹⁹ Tetronasin (2) was previously reported as an ionophore antibiotic that exhibits inhibition of ruminal microorganisms.^20,21 It is known to disrupt the membrane ion gradient, in particular through strong binding to Na⁺ and Ca⁺ ions. This transport and subsequent membrane depolarization result in cell death.^14,18,22 Compounds 1 and 2 were evaluated for in vitro antibacterial activity against a diverse panel of pathogenic bacteria (Tables 2, S3). Both 1 and 2 exhibited growth inhibitory activity against S. aureus and the causative agent of Lyme disease Borellia burgdorferi, although the observed activity of 2 was significantly more potent than that of 1. Neither of the compounds exhibited activity against our panel of mycobacteria (Table S3), and each showed IC₅₀ values of >85.0 μM (1) and 19.1 μM (2) when tested for cytotoxicity against Vero cells. Interestingly, the absence of the methoxy group resulted in a more than 7- and 40-fold change in activity against S. aureus and B. burgdorferi, respectively. Previous studies have shown that the C-35 methoxy oxygen in 2 is involved the coordination of tetronasin to a sodium ion.^18,23 It is plausible that the loss of the methoxy group diminishes the capacity of 1 to bind to sodium (or calcium) ions and disrupt the membrane ion gradient, resulting in weaker observed growth inhibition activity. Further studies would be required to confirm this.

Table 2. Antibacterial Activity of 1 and 2^a.

			MIC (μM)
	1	2	MXF	VCM	DXC
B. burgdorferi	1.44	<0.03		0.21	0.45
S. aureus	1.21	<0.16	<0.24		<0.22
P. aeruginosa	>40	>40	3.24		12.8
E. aerogenes	>40	>40	<0.24		3.02

^aMXF = moxifloxacin; VCM = vancomycin; DXC = doxycycline. B. burgdorferi (B31 ATCC 35210); S. aureus (ATCC 29213); P. aeruginosa (ATCC 10145GFP); E. aerogenes (ATCC 13048).

Conclusions

The use of a 3D-printed, dual-sided agar plug assay was demonstrated as an effective means to screen an environmental bacterial library, dereplicate known antibiotics from individual wells of the assay plate, and guide the identification of new antibiotics. After validating the DAPA assay according to published bioassay guidelines, it was used to screen 217 environmental bacteria against a panel of pathogens. From this screen, 9 antibiotics were dereplicated directly from agar plugs of “hits” via LCMS/MS analysis, which allowed us to prioritize strain K1063 for further analysis. The new ionophore antibiotic demethoxytetronasin (1) along with tetronasin (2) were identified from K1063 by spectroscopic analysis. Both were shown to inhibit B. burgdorferi and S. aureus in vitro. The unique DAPA plate design overcomes many challenges of previous solid support systems by (i) allowing for contactless bacterial competition assays; (ii) allowing for the use of different growth media for test bacteria and pathogens; (iii) facilitating the use of orthogonal methods of growth inhibition; and (iv) making the medium/high-throughput screening of environmental microorganisms accessible to laboratories on a smaller budget (DAPA is autoclave-resistant and thus reusable, reducing reliance on disposable plastics). It can be employed as a unique cost- and time-saving measure for researchers who wish to engage in high-throughput screening and rapid NP dereplication in the front end of a microbial antibiotic discovery pipeline since it does not necessitate the creation of environmental isolate and NP fraction libraries. With access to MS instrumentation compatible with the analysis of microbial cell mass (such as MALDI-TOF MS), one may incorporate bacterial strain dereplication on a list of bioactive “hits”. As the technology driving MS analysis of bacterial colonies advances, it should also be possible to perform extraction-free dereplication of antibacterial hits through integration of MALDI-TOF, DESI MS, or similar techniques. These two improvements to the pipeline would extend the information obtained from small amounts of bacterial cell mass in the front end of antibiotic discovery.^24,25 Beyond this study, DAPA is well suited for use in other analyses involving polymicrobial systems. The plate can be utilized to study multi-media metabolic profiling of bacterial colonies, to assess the effects of one (or more) microbial species on NP production, or to study cell-to-cell communication. Regardless of use, it can be conveniently coupled to several MS-based bioinformatic applications that inform front-end NP discovery and facilitate rapid strain/compound dereplication and sample prioritization.

Methods

General Experimental Procedures

Optical rotation was measured on a PerkinElmer 241 polarimeter using a 100 mm cell. UV spectra were recorded on a Varian Cary 5000 spectrophotometer. ECD spectra were recorded on a JASCO J-815 spectrometer. NMR spectra were obtained on a Bruker 600 (125) MHz AVANCE III NMR spectrometer equipped with a 5 mm TCI cryogenic inverse probe with z-axis pfg and TopSpin version 3.2 operating software at the University of Illinois at Chicago Center for Structural Biology. Chemical shifts (δ) are given in ppm, and coupling constants (J) are reported in Hz. ¹H and ¹³C NMR chemical shifts were referenced to the CDCl₃ (δ_H 7.26 ppm and δ_C 77.0 ppm). High-resolution mass spectra were obtained on a Bruker COMPACT ESIQTOF mass spectrometer at the University of Illinois at Chicago. High-performance liquid chromatography (HPLC-DAD) data were obtained using a Hewlett-Packard series 1100 system controller and pumps with a model G1315A diode array detector (DAD) (Hewlett-Packard, Palo Alto, CA, USA) equipped with a reversed-phase Discovery amide C16 column (250 × 4.6 mm, 5 μm) at a flow rate of 0.9 mL·min^–1 (Figure S3). Preparative HPLC was performed using a Waters Prep LC4000 system equipped with a Water486 UV/Vis detector and a Isco Combiflash (Teledyne) equipped with a silica column (12 g) using ELSD and UV detection at a flow rate of 30 mL/min. All solvents were spectroscopic grade.

DAPA Plate Design

The dual-sided agar plate is made of Dental SG resin (Formlabs, Somerville, MA, USA) using a Form2 3D printer (Formlabs) with 123.0 × 81.0 × 16.0 mm dimensions. Each well contains a custom-built channel designed to anchor the solid support plug in place. A silicon base was created from a mold and is designed to support molten agar as it is poured. Free access to the full set of detailed code and instructions to 3D print the DAPA plate can be obtained by contacting the corresponding author (for academic use only).

DAPA Growth Inhibition Assay

Two layers of solid nutrient media were utilized in each well of the DAPA plate. To support the growth of the test bacterium, the first layer (200 μL of nutrient medium, 1.5% agar) was poured in on the top side of the DAPA plate, while the removable silicon base was inserted. Pouring agar into wells was performed by hand, although it can be readily integrated with automated liquid handling systems. After the agar was solidified, the silicon base was removed. Bacterial isolates were inoculated on the nutrient medium layer using a sterile cotton swab from their respective isolation plate. The DAPA plate was then covered by sterile universal microplate lids (Nunc). The plates were placed in an incubator at 30 °C for between 3 and 5 days, depending on the optimal incubation period for the test bacteria. Note: incubation times and conditions vary widely based on target bacteria; the user should determine the most suitable conditions. After the incubation period, the second agar layer (100 μL of MHB with 0.1 M potassium phosphate, 1.5% agar) was poured in on the bottom side of the DAPA plate to support the growth of the pathogen. After the agar solidified, 25 μL of pathogen (6.0 × 10⁶ CFU/mL P. aeruginosa; 4.1 × 10⁷ CFU/mL E. aerogenes; 25 μL of OD₆₀₀ = 0.1 S. aureus) in sterile PBS solution was inoculated from cryostock on the MHB layer. After an overnight incubation period, growth inhibition of the pathogen was measured.

Visualization of Growth Inhibition in DAPA

Fluorescence and optical density were employed in order to visualize the growth inhibition of three microbial pathogens used in this study. Fluorescent signals were detected using a plate reader (CLARIOstar), and fluorescent images were obtained using a fluorescence imaging system (IVIS). P. aeruginosa ATCC 10145GFP was purchased, as it contains a multicopy vector encoding the green fluorescent protein (from Aequorea victoria), GFPmut3. E. aerogenes ATCC 13048 was transformed by inserting an mCherry-containing plasmid (Addgene, Watertown, MA, USA). S. aureus ATCC 29213 was tested without fluorescent labeling. P. aeruginosa ATCC 10145GFP and E. aerogenes ATCC 13048mCherry growth in DAPA was visualized using either a plate reader or a fluorescent imager. S. aureus was visualized using a 3-(4,5- dimethylthiazole-2-yl)-2,5-diphenyltetrazolium bromide (MTT) indicator dye. The extent of growth was determined by calculating RFUs for fluorescence and OD₆₀₀ for MTT. For MTT measurement, 100 μL of PBS was added to the surface of the pathogen (S. aureus). After pipetting to mix pathogen cells with PBS, 50 μL of solution was transferred to 100 μL of 5 mg/mL of MTT solution (in PBS) in an empty 96-well plate. After 30 min, OD₆₀₀ was measured. The latter step is only necessary if quantitative analyses are required; qualitative growth inhibition analyses can be performed directly in the DAPA wells with MTT. For full protocol, see the Supporting Information.

Extraction of Agar Plugs from the DAPA Plate

After the antibiotic assay was performed, agar plugs that exhibited growth inhibition were removed with a spatula and placed into individual flasks. 50 mL of methanol was added to each flask, and the solution was left overnight to kill any potential viable pathogenic bacteria. Plugs were then individually ground with a mortar and pestle and placed back into their flasks (with original methanol solution) and then on a shaker overnight to allow for extraction. Methanol was then filtered and evaporated. The resulting extract was subjected to LCMS/MS analysis on a Bruker COMPACT ESIQTOF MS, and data were uploaded into and analyzed using GNPS molecular networking.¹²

Bacterial Transformation

E. aerogenes ATCC 13048 was transformed by an mCherry-containing plasmid, pCherry10 (Addgene no.24664). The strain DH5alpha containing the plasmid was grown at 37 °C in Luria broth (LB). The plasmid was extracted using a QIAGEN Plasmid Midi Kit no.12143. The preparation and transformation of competent E. aerogenes cells followed standard procedures.²⁶ Successful transformants were selected by plating on LB agar containing 200 μg/mL of hygromycin.

Preparation of Cryostocks of Pathogen Strains

P. aeruginosa ATCC 10143 GFP, E. aerogenes ATCC 13048mCherry, and S. aureus ATCC 29213 were inoculated on MHA medium containing 100 μg/mL of ampicillin at 37 °C overnight. A colony from each inoculum was transferred to 10 mL of MHB in a 50 mL tube and incubated at 37 °C overnight.

In Vitro Broth Microdilution S. aureus (ATCC 29213), P. aeruginosa (ATCC 27853), and E. aerogenes (ATCC 13048) Assays

Bacterial stocks with known cell counts (CFU/mL) were diluted in Mueller Hinton II Broth (Cation-Adjusted) (CAMH) to achieve inoculums of 2 × 10⁵ CFU/mL. 100 microliters of this inoculum was added to individual wells in 96 well plates containing an equal volume of CAMH broth. Each of the three pathogens was incubated with test compounds overnight at 37 °C after which absorbance was read at 570 nm. Moxifloxacin and doxycycline were employed as positive controls. A microdilution procedure was followed to determine the MIC values, as described by the European Committee on Antimicrobial Susceptibility Testing (EUCAST).²⁷ MIC was defined as the lowest concentration of the compound that prevented detectable growth. All MICs were determined in duplicates (due to a low sample amount).

In Vitro Broth Microdilution B. burgdorferi Assay

B. burgdorferi ML23 pBBE22luc originally derived from strain B31-ATCC35210, which harbors a plasmid encoding optimized luciferase (luc) under a strong borrelia promoter (Pflab-luc), was used for the assay. The culture was grown in Barbour Stoenner-Kelly (BSK-H) medium containing 300 μg/mL of kanamycin at 33 °C for 3 days until a logarithmic phase was achieved. After incubation, cell counts were made with a Petroff-Hauser cell-counting chamber under dark-field microscopy and adjusted to 1 × 10⁶ CFU/mL with fresh BSK-H medium. B. burgdorferi culture, and test compounds were incubated at 33 °C for 4 days. 50 microliters of 2 mM D-luciferin was added to detect the luminescence signal, and the MIC was calculated. The MIC was defined as the lowest concentration of the compound required to observe ≥90% inhibition of pathogen growth.

Cytotoxicity Assay

Green monkey kidney cells (Vero, ATCC CRL-81) were used to assess the cytotoxicity, as previously described.^28,29 Vero cells were cultured in Eagle’s minimum essential medium (MEM) containing 10% fetal bovine serum (FBS) supplemented with a penicillin and streptomycin antibiotic mix. The cells were microscopically quantified into a cell density of 2 × 10⁵ cells/mL. A 100 microliter portion of cell suspension was added to the test compounds and incubated at 5% CO₂ and 37 °C for 3 days. At the end of the incubation period, 20 μL of 0.6 mM resazurin was added, and the fluorescence readout was measured with excitation/emission wavelengths of 530/590 nm after 4 h of incubation to calculate IC₅₀ values. Rifampin and bedaquiline were used as positive controls.

Collection and Identification of Strain K1063

Strain K1063 was isolated from the sediment collected using SCUBA at 4.3 m depth in 2.2 °C freshwater in the small fissure between the North American and Eurasian continental plates at Nesgja, Iceland. The amplified 16S rRNA gene fragment (1250 bp, GenBank accession number OQ709800.1) displayed 100% BLAST nucleotide identity to that of Streptomyces sp. strain GGS53 (GenBank accession number OK037580.1).

Isolation of Compound 1

Strain K1063 was grown in 34 L in NZSG media at 21 °C while shaking at 220 rpm. On day 7, the extracellular secondary metabolites were absorbed from the fermentation broth using 1:1 Amberlite XAD-16 and 7HP polymeric resins, followed by extraction with 1:1 acetone/methanol and partitioning between water and ethyl acetate. The organic layer was dried under vacuum to afford 7.3 g of the extract. It was then fractionated using silica gel flash column chromatography (100 g of silica). The 8:2 ethyl acetate/hexane fraction exhibited bioactivity and was separated using automated normal phase silica (12 g) chromatography at 10–40% ethyl acetate in hexanes over 12 min, and 40–100% to 15 min. Fractions 4–8 were pooled and separated over semi-preparative reversed-phase Discovery Amide-C16 HPLC (4.6 × 250 mm, 0.9 mL/min) using a gradient of 7:3 A/B to 100% B over 30 min (A = 10% acetonitrile in 5 mM ammonium formate, pH 9; B = 90% acetonitrile in 5 mM ammonium formate, pH 9) to yield compound 1 (t_R 13.5 min, 0.9 mg, Figure S3). An ammonium formate buffered system was employed in order to elute compounds 1 and 2 during reversed-phase HPLC separation. Under basic conditions (pH 9.0–10.0), the tetronic acid moiety can be converted to a tetronate moiety, which carries a negative charge and subsequently increases the polarity of tetronacins during chromatographic separation (note: compounds 1 and 2 were not eluted from C18, C8, or amide columns under isocratic 100% MeOH conditions without an ammonium formate buffered system).

Demethoxytetronasin (1)

Compound 1 was isolated as an amorphous solid (0.9 mg). [α]_D²⁵ −49 (c 0.1, MeOH). ¹H NMR (600 MHz, CDCl₃) and ¹³C NMR (125 MHz, CDCl₃), see Table 1, Figures S4a, S8, and S9. For complete two-dimensional NMR data, see Figures S10–S13. All raw NMR data files were deposited in the open access Natural Products Magnetic Resonance Database (www.np-mrd.org).³⁰ HRESI-TOF MS m/z 589.3722 [M + H]⁺ (calcd for C₃₄H₅₃O₈: 589.3696; Figure S7).

Tetronasin (2)

Compound 2 was isolated as an amorphous solid (5.0 mg). [α]_D²⁵ −47 (c 0.1, MeOH). For complete one- and two-dimensional NMR, HR-qTOF-MS/MS, and CD spectra, see Figures S4–S6, S14.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsinfecdis.3c00171.

• Spectroscopic data consisting of one- and two-dimensional NMR data of 1 and 2, MS and CD spectra of 1 and 2, bioassay data, and other materials (PDF)

The authors declare the following competing financial interest(s): Drs. Lee and Murphy are in the process of filing a patent for this invention for purposes of commercialization. Free access to the full set of detailed code and instructions to 3D print the DAPA plate can be obtained by contacting the corresponding author (free for all academics).