Rapid ultrasensitive detection platform for antimicrobial susceptibility testing

Abstract

Rapid detection and phenotyping of pathogenic microbes is critical for administration of effective antibiotic therapies and for impeding the spread of antibiotic resistance. Here, we present a novel platform, rapid ultrasensitive detector (RUSD), that utilizes the high reflectance coefficient at high incidence angles when light travels from low- to high-refractive-index media. RUSD leverages a principle that does not require complex manufacturing, labeling, or processing steps. Utilizing RUSD, we can detect extremely low cell densities (optical density [OD] ≥ 5 × 10-7) that correspond to approximately 20 bacterial cells or a single fungal cell in the detection volume, which is nearly 4 orders of magnitude more sensitive than standard OD methods. RUSD can measure minimum inhibitory concentrations (MICs) of commonly used antibiotics against gram-negative and gram-positive bacteria, including Staphylococcus aureus, Pseudomonas aeruginosa, and Escherichia coli, within 2 to 4 h. Here, we demonstrate that antibiotic susceptibility tests for several pathogens can rapidly be performed with RUSD using both small inoculum sizes (500 cells/mL) and larger inoculum sizes (5 × 105 cells/mL) used in standard antibiotic susceptibility tests. We anticipate that the RUSD system will be particularly useful for the cases in which antibiotic susceptibility tests have to be done with a limited number of bacterial cells that are available. Its compatibility with standard antibiotic susceptibility tests, simplicity, and low cost can make RUSD a viable and rapidly deployed diagnostic tool.

Fig 1. RUSD.

(a) Optical schematics of the RUSD apparatus, which utilizes the high reflectance coefficient of light at the interface between thin and dense media (water and glass) at high incidence angles. (b) The focal distance of the laser is set such that the incidence angle (θ) of the laser at the fiber wall remains > 89°. At θ > 89°, the reflectance coefficient of laser light is R > 0.82 for both s and p polarizations of light, allowing the laser beam to travel in liquid and reflect multiple times within the fiber. (c) During travel, any photon encountering a cell or a particle is either absorbed or scattered, losing the angular condition to be guided by the fiber (i.e., θ > 89^o), and any cell on the light path reduces the total intensity proportional to its cross-sectional area and creates a 2D shadow-projection image on the optical detector. The current from the reverse-biased optical detector is passed through a resistor in which the potential difference is converted to a signal for data acquisition. (d) The voltage signal as a function of cell density. Simulation data based on the mathematical model described here and actual RUSD exhibit a high degree of concordance (R² = 0.97) up to OD > 10⁻³ and (R² = 0.93) for the whole data range used here. (e) Signal strength versus number of cells and cell dimensions in RUSD: As the cell size increases, signal strength is multiplied proportionally to the cell cross-sectional area. The cell shape is taken as a square pattern with the specified dimensions. The region marked with red asterisks indicates the boundary of detection at which the signal is above the 1 mV limit. RUSD signal rapidly saturates over approximately OD 0.1 as cells optically block the fiber cross section. OD, optical density; PDMS, polydimethylsiloxane; RUSD, rapid ultrasensitive detector.

Fig 2. RUSD detects extremely low bacterial and fungal cell counts.

(a) Top panel: Counts for C. glabrata at approximately OD 10⁻⁶ displays a Poisson distribution with counts of 0, 1, 2, and 3 with a mean value of about 1. Bottom panel: Increasing the yeast cell density shifts distribution toward a normal distribution with a mean value of about 5.3 cells. (b) S. cerevisiae at an OD nearly 10⁻⁶ (top panel) results in a Poisson distribution with counts of 0, 1, and 2 and a mean value of approximately 0.48; (lower panel) a slight increase of cell density moves distribution toward a normal distribution with a mean value of approximately 4 cells. (c) Counts for E. coli culture at (top panel) an OD of approximately 5 × 10⁻⁷, yields about 25 cells; (lower panel) increasing the bacterial cell density slightly shifts distribution toward a normal distribution with a similar mean value of about 24 cells with smaller standard deviation, which shows the sensitivity limit of the device. ODs of bacterial cultures grown in LB (started from OD roughly 5 × 10⁻⁷) and recorded by RUSD: (d) E. coli (MG1655), (e) S. aureus (RN4220), (f) P. aeruginosa (PAO1), and (g) an E. coli clinical isolate (ET-CI28). (The data for Fig 2 can be found in S1 Data). LB, lysogeny broth; OD, optical density; RUSD, rapid ultrasensitive detector.

Fig 3. iFAST yields antibiotic resistance profile of pathogenic bacteria in 1 to 3 h.

Bacterial cultures with initial OD of approximately 10⁻⁶ were grown in different concentrations of antibiotics, and their densities were periodically monitored in RUSD. Heat map plots show growth of bacteria as a function of time and antibiotic dosage. Colored bars indicate ODs of bacterial cultures. Red diamonds indicate MIC values measured at different time points at which bacterial growth is statistically significant (OD > LOQ). Below each heat map plot, a dose-response curve is provided for the first statistically significant MIC measurement. A calibration curve is generated for every strain before every measurement, and LOQ values are shown by using dashed magenta lines. (a) Time-resolved dose-response measurements for laboratory strains of P. aeruginosa (PAO1), E. coli (MG1655), and S. aureus (RN4220) in levofloxacin, a DNA gyrase inhibitor. (b) Time-resolved dose-response measurements for a clinical E. coli isolate (ET-CI28) in levofloxacin, P-T (a beta-lactam cell-wall synthesis inhibitor and beta-lactamase inhibitor), and amikacin (an aminoglycoside). (The data for Fig 3 can be found in S2 Data). iFAST, in Fiber Antibiotics Susceptibility Testing; LOQ, limit of quantification; MIC, minimum inhibitory concentration; OD, optical density; P-T, piperacillin-tazobactam; RUSD, rapid ultrasensitivity detector.

Fig 4. iFAST yields reliable results in shorter times in comparison to standard commercial clinical antibiotic susceptibility testing methods.

(a) Utilizing the high sensitivity of the RUSD platform, iFAST can track growth of clinical isolates starting from the initial OD = 2 × 10⁻³ in antibiotic susceptibility test panels (NM43, Beckman Coulter). Cell densities of a clinical E. coli isolate (CIET-001) growing in different doses of 29 antibiotic compounds were periodically recorded at hourly intervals. Cell densities exceeding the growth threshold (4 × 10⁻³, horizontal red dashed lines, one doubling) are shown in green filled circles. Cell densities lower than the threshold are shown in gray filled circles. (b) Growth information was converted to a binary resistance map for the test panel, in which blue and white indicate growth and no growth in a given well, respectively. The Matthews correlation coefficient is calculated using the true–false matrix between observed and expected resistance data. (c) Correlation coefficient between iFAST measurements and clinical data changes as a function of growth threshold and hovers to acceptable levels as incubation times increase. For consistency and reliability purposes, we chose a 4 h incubation time and threshold of one population size doubling for the rest of the clinical E. coli strains. (Right panel) Histogram showing the correlations between iFAST measurements and clinical reports demonstrate the accuracy and power of the iFAST assay. (The data for Fig 4 can be found in S3 Data). Ak, amikacin; Am, ampicillin; A/S, ampicillin/sulbactam; Aug, amoxicillin/clavulanic acid; Azt, aztreonam; CA, clavulanic acid; Cax, ceftriaxone; Caz, ceftazidime; Cf, cephalothin; Cft, cefotaxime; Cfz, cefazolin; Cp, ciprofloxacin; Cpe, cefepime; Crm, cefuroxime; Etp, ertapenem; Fd, nitrofurantoin; FN, false negative; FP, false positive; Gm, gentamicin; iFAST, in Fiber Antibiotics Susceptibility Testing; Imp, imipenem; LOQ, limit of quantification; Lvx, levofloxacin; Mer, meropenem; Mxf, moxifloxacin; Neg Contr., negative control; OD, optical density; Pi, piperacillin; P/T, piperacillin/tazobactam; RUSD, rapid ultrasensitivity detector; Te, tetracycline; Tgc, tigecycline; TN, true negative; To, tobramycin; TP, true positive; T/S, trimethoprim/sulfamethoxazole.