A high-resolution screen identifies a preexisting beta-lactam that specifically treats Lyme disease in mice

Abstract

Lyme disease, caused by Borrelia burgdorferi in the United States, is an escalating human health problem that can cause severe disease if not properly treated. Doxycycline is the primary treatment option for Lyme disease; however, several concerns are associated with high-dose doxycycline treatment. For example, doxycycline is a broad-spectrum antibiotic and kills beneficial bacteria. Doxycycline is also known to produce unwanted off-target effects in eukaryotic cells. Some at-risk populations such as young children cannot be prescribed doxycycline, and in addition to these shortcomings, the treatment appears to fail in 10 to 20% of cases. We reasoned that safe, alternative therapies may currently exist but have not yet been found because of the challenges associated with drug screening approaches. We screened nearly 500 US Food and Drug Administration-approved compounds using an array of physiological, cellular, and molecular techniques. Top-performing candidates were counter screened to identify compounds that did not affect other bacterial phyla. Piperacillin emerged as a compound that eradicated B. burgdorferi at low-nanomolar concentrations by specifically interfering with the unusual, multizonal peptidoglycan synthesis pattern common to the Borrelia clade. Mechanistic in vitro studies identified the cellular target of piperacillin in B. burgdorferi and produced key insights that may explain both the specificity and efficacy of the compound. Further, in vivo studies using an experimental mouse infection model demonstrated that piperacillin treated animals at a 100-fold lower dose than the effective dose of doxycycline without affecting the murine microbiome. Our findings suggest that piperacillin may offer clinicians another therapeutic option for Lyme disease.

Figure 1.. In vitro drug library screen to identify new Lyme disease therapeutics.

(A) Microtiter colorimetric assay screened 466 FDA approved compounds whereby 10⁵ cells/mL were incubated with 100 nM of each compound for 5 days. Growth was determined by the media color change using absorbance values (562 nm) attained, relative to absorbance values attained by culture media alone (dotted red line). Each compound was assayed in duplicate and the mean shown after being grouped by class. Z factor for initial trial = 0.5128. (B) After repeating the above, and additional refinement, a sub-set of top-performing compounds (12%) were incubated with B. burgdorferi (100 nM final concentration) prior to co-incubation with fluorescent D-Alanine analogue HADA. Each condition was imaged by phase-contrast and epifluorescence microscopy, pixel intensity for each cell was determined, averaged by volume, and plotted relative to untreated HADA-labeled cells (dotted red line). Each treatment included > 300 cells. Compounds that are currently used to treat Lyme disease (*). (C) The same procedure described in B was performed, but after HADA labeling, peptidoglycan (PG sacculi) was purified and signal values attained were determined by microplate fluorescence. The inverse of the signals attained (1/HADA) were plotted relative to the inverse of the whole cell signal values attained for those that were significantly below the untreated HADA control cells. (D) The same studies performed in C, but the data was binned based on the compounds that produced whole cell values above the untreated, HADA-labeled control cells in B.

Figure 2.. Counter screen of select compounds and validation of top performing antibiotics

(A) The top- performing compounds previously identified were counter-screened on Escherichia coli, B. subtilis, and Staphylococcus aureus (100 nM final concentration). Each species was cultured to an optical density (OD, 600 nm) of 0.15, transferred to a microtiter plate containing 100 nM of each compound, and growth curves were performed at 37°C in duplicate. The growth rate was calculated for all treatments (denominator) and normalized by the untreated control values (numerator), attained for each species and the ratio shown as a heatmap. Compounds that are currently used to treat Lyme disease (*). (B) Piperacillin minimum inhibitory concentration (MIC) was calculated on replicate B. burgdorferi cultures in BSK-II complete liquid media and found to be 35 nM (+/− 5.66). (C/D) The bactericidal activity of piperacillin was determined by colony forming units (CFU), relative to doxycycline. Liquid cultures were treated with either 80 μM or 800 nM (final concentration) for 5 days and plated in semi-solid agar lacking treatment.

Figure 3.. Piperacillin specifically inhibits zonal PG synthesis downstream of establishing the divisome site.

(A) Phase contrast and epifluorescence micrographs of HADA labeled cells treated with 100 nM of piperacillin. Scale bar 5 μm. (B) Population-level demograph analysis of HADA signal in piperacillin treated cells. n > 300 cells. Heatmap is the normalized fluorescent signal intensity in a.u. (C) B. burgdorferi strain genetically engineered with a 2^nd copy of ftsA fused to monomeric, super-folder green fluorescent protein (msf-GFP) that was incubated with 0.2 mM HADA. Phase-contrast (upper left), FtsA-msfGFP (upper right), HADA (lower left), and merge (lower right) are shown. Scale bar 5 μm. (D) Population-level analysis of data collected in C. Cells were binned based on the number HADA zones present, the number of FtsA foci was determined for each, and the percentage in each category were calculated. (E) FtsA-msfGFP kymograph of relative signal intensity position over time (minutes). (F) Demograph analysis of FtsA-msfGFP signal in untreated (above) or piperacillin treated (below, 100 nM) cells. Each data set n > 300 cells. Heatmap is the normalized fluorescent signal intensity in a.u.

Figure 4.. Piperacillin specifically interacts with BB0136.

(A) BB0136 was identified in silico and probed for piperacillin affinity in competition assays. A recombinant BB0136 was produced, purified, and 1 ug was pre- incubated with a titration of piperacillin (left) or penicillin G (right), prior to incubation with Bocillin-FL. A fraction of each reaction was separated by SDS PAGE, imaged using fluorescence excitation and emission spectra for fluorescein (upper) and then stained with Coomassie (lower) to evaluate protein loading in each reaction. (B) Data collected in A from three independent experiments were used to assess the affinity of rBB0136 for piperacillin or penicillin G using Bocillin FL signal. (C) BB0136 structure (gray) modeled on the solved crystal structure of Streptococcus pneumoniae Pbp2x (green, reference 39). RMSD = 1.28. (D) Catalytic site and key residues in BB0136 (gray) or S. pneumoniae Pbp2x (green, reference 39).

Figure 5.. Low doses of piperacillin treat murine Lyme disease without impacting the microbiome.

(A) C3H mice were needle inoculated with 10⁴ B. burgdorferi cells. Twenty-one days later, animals were treated with effective doses (see table 2) of either piperacillin, piperacillin-tazobactam, doxycycline, ceftriaxone, or diluent control for seven days. At different stages during or post treatment, fecal samples were collected, genomic DNA purified, and 16S deep sequencing was performed to determine the phylogenic abundance and complexity driven by each treatment. Dosages are shown in milligrams of drug per kilogram of animal weight per day (m/k/d). Samples were collected pre- and post-treatment (p.t.). (B) Shannon index values of results attained in A.