Bacterial Cytological Profiling (BCP) as a Rapid and Accurate Antimicrobial Susceptibility Testing Method for Staphylococcus aureus

Abstract

Successful treatment of bacterial infections requires the timely administration of appropriate antimicrobial therapy. The failure to initiate the correct therapy in a timely fashion results in poor clinical outcomes, longer hospital stays, and higher medical costs. Current approaches to antibiotic susceptibility testing of cultured pathogens have key limitations ranging from long run times to dependence on prior knowledge of genetic mechanisms of resistance. We have developed a rapid antimicrobial susceptibility assay for Staphylococcus aureus based on bacterial cytological profiling (BCP), which uses quantitative fluorescence microscopy to measure antibiotic induced changes in cellular architecture. BCP discriminated between methicillin-susceptible (MSSA) and -resistant (MRSA) clinical isolates of S. aureus (n = 71) within 1-2 h with 100% accuracy. Similarly, BCP correctly distinguished daptomycin susceptible (DS) from daptomycin non-susceptible (DNS) S. aureus strains (n = 20) within 30 min. Among MRSA isolates, BCP further identified two classes of strains that differ in their susceptibility to specific combinations of beta-lactam antibiotics. BCP provides a rapid and flexible alternative to gene-based susceptibility testing methods for S. aureus, and should be readily adaptable to different antibiotics and bacterial species as new mechanisms of resistance or multidrug-resistant pathogens evolve and appear in mainstream clinical practice.

Keywords: Antibiotic resistance; Multidrug resistant bacteria; Staphylococcus aureus; Susceptibility tests.

Fig. 1

BCP methodology and analysis. Each antibiotic was added to exponentially growing cells and samples were collected for imaging hourly. Changes in cytological parameters were measured using CellProfiler. For each strain, three sets of profiles (each represented by three images), were obtained over three different days.

Fig. 2

Discrimination of MRSA from MSSA with BCP. (a) Images of MRSA (ATCC 33591) and MSSA (ATCC 29213) after 2 h treatment either with no antibiotics, 10 μg/mL of oxacillin, or 20 μg/mL of cephalexin. In all the images, cell walls were stained with WGA-Cy5 (red), DNA was stained with DAPI (blue) and SYTOX Green (green). SYTOX Green is membrane impermeable and only stains permeabilized cells. (b) Cartoon showing the observed cellular phenotypes. (c–e) Linear discriminant analysis (LDA) plots were calculated from a cytological profile library consisting of 22 parameters measured for 71 isolates, 37 MRSA strains (red) and 36 MSSA strains (blue), treated with each antibiotic measured in triplicates. The mean of BCP profiles of each of the 71 strains (diamonds) is plotted after 2 h of treatment with (c) 10 μg/mL of oxacillin, (d) 20 μg/mL of cephalexin, (e) from the combined data from the measurements in (c) and (d). USA100 (N315; black square) and USA300 (TCH1516; black x) profiles were projected onto the MRSA/MSSA LDA plot. (f) LDA plots calculated from BCP profiles measured after 1 h treatment with 20 μg/mL of cephalexin. Blind test profiles (filled circles; n = 30) are overlaid on LDA plots of the training set

Fig. 3

BCP determination of daptomycin susceptibility. (a) Daptomycin non-susceptible (DNS; gs5) and daptomycin susceptible (DS; ATCC29213) cell walls were stained with WGA-Cy5 (red) while the DNA was stained with DAPI (blue) and SYTOX Green (green). Images were taken after treatment with no antibiotics or 5 μg/mL of daptomycin for 30 min. (b) LDA plots were calculated from cytological profiles after 30 min of treatment with daptomycin where parameters were the mean of triplicates. The LDA plot consists of the mean cytological profiles 8 DNS strains (red) and 12 DS strains (blue), all of which forms the training set. Blind test profiles (filled circles; n = 12) are overlaid on the training set.

Fig. 4

Two classes of MRSA strains. (a) Principal component analysis (PCA) plot of training set data comprise of 37 MRSA (red) and 36 MSSA (blue) strains. (b) LDA plot showing the two classes of MRSA strains. The transformation matrix for the LDA plot was calculated from the cytological profiles of strains in the top and bottom quartiles of PC2 and used to plot 12 MRSA^HL (green diamonds) and 24 MRSA^LL (purple diamonds) after 2 h of treatment with 10 μg/mL of oxacillin. Each data point represents the mean of triplicate cytological profiles for a different clinical isolate. The BCP profiles for the blind tests (filled circles), USA300 (TCH1516; black x) and USA100 (N315; black square) were overlaid on the training set (diamonds). (c) Time-lapse microscopy of MSSA (ATTC29213), MRSA^LL (ATCC33591), MRSA^HL (wr7) was taken while strains were grown on agarose pads with 5 μg/mL oxacillin. (d–e) Comparison of the net growth rates of these three groups on agarose pads (d) without antibiotics and (e) with oxacillin. (f) Cumulative lysis of different strain types on agarose pad with oxacillin.

Fig. 5

Genetic characterization and susceptibility to dual-beta lactam treatment for the two MRSA subgroups. (a) Dendrogram calculated using unweighted paired-group method with arithmetic mean (UPGMA) for the multilocus sequence typing (MLST) allelic profiles of MRSA^LL (purple) and MRSA^HL (green). (b–c) Synergy assay curves. The position of each curve represents the combination of antibiotic concentrations that inhibits growth. Each curve represents the average of three independent synergy checkerboard assay for a single strain. There are 12 MRSA (green) strains that were not viable when treated with 1/4 Cmax of both antibiotics in combination; each of these was MRSA^HL strains. There are 24 MRSA strains (purple) that were viable; each corresponds to MRSA^LL strains. The dotted purple curves represent two strains (wr12; cb15) that were viable at 1/4 Cmax of both antibiotics in at least one of the three independent synergy checkerboard assays.