Droplet‐Based Single‐Cell Measurements of 16S rRNA Enable Integrated Bacteria Identification and Pheno‐Molecular Antimicrobial Susceptibility Testing from Clinical Samples in 30 min

Abstract Empiric broad‐spectrum antimicrobial treatments of urinary tract infections (UTIs) have contributed to widespread antimicrobial resistance. Clinical adoption of evidence‐based treatments necessitates rapid diagnostic methods for pathogen identification (ID) and antimicrobial susceptibility testing (AST) with minimal sample preparation. In response, a microfluidic droplet‐based platform is developed for achieving both ID and AST from urine samples within 30 min. In this platform, fluorogenic hybridization probes are utilized to detect 16S rRNA from single bacterial cells encapsulated in picoliter droplets, enabling molecular identification of uropathogenic bacteria directly from urine in as little as 16 min. Moreover, in‐droplet single‐bacterial measurements of 16S rRNA provide a surrogate for AST, shortening the exposure time to 10 min for gentamicin and ciprofloxacin. A fully integrated device and screening workflow were developed to test urine specimens for one of seven unique diagnostic outcomes including the presence/absence of Gram‐negative bacteria, molecular ID of the bacteriaas Escherichia coli, an Enterobacterales, or other organism, and assessment of bacterial susceptibility to ciprofloxacin. In a 50‐specimen clinical comparison study, the platform demonstrates excellent performance compared to clinical standard methods (areas‐under‐curves, AUCs >0.95), within a small fraction of the turnaround time, highlighting its clinical utility.


Droplet-based single-cell measurements of 16S rRNA enables integrated bacteria identification and pheno-molecular antimicrobial susceptibility testing from clinical samples in 30 min
Aniruddha M. Kaushik † , Kuangwen Hsieh † , Kathleen E. Mach, Shawna Lewis, Christopher M. Puleo, Karen C. Carroll, Joseph C. Liao, and Tza-Huei Wang * Summary of supplementary figures and tables:

Figures
Description Figure S1 Two-color laser induced fluorescence (LIF) detector and bulk reaction assessment device Figure S2 Bulk pheno-molecular AST of E. coli ATCC 25922 Figure S3 Modular droplet device for flexible assay characterization Figure S4 One-step sample pretreatment protocol Figure S5 Bulk sensitivity of the PNA probe assay Figure S6 Quantification of antibiotic effect on E. coli in droplets Figure S7 Schematic of the entire DropDx platform Figure S8 Thermal platform characterization Figure S9 DropDx device and droplet residence time Figure S10 Validation of pheno-molecular AST in the clinical comparison study Figure S11 Clinical comparison study workflow Figure S12 Pilot studies for power analysis and threshold determination Tables  Description  Table S1 Summary of designed PNA Probes Table S2 Flow rates for monodisperse droplet generation Table S3 Fluorescence signal of unquenched PNA probes over quenched background Table S4 Summary of urine samples tested and final results in clinical comparison study Table S5 Clinical performance (PPVs and NPVs) of DropDx ID and AST

E . c o li
A T C C 2 5 9 2 2

E . c o li
A T C C 2 5 9 2 2 + 8 µ g /m L G e n t a m i c i n Figure S2. Bulk pheno-molecular AST of E. coli 25922. PNA probe signal from E. coli cells increases for increasing culture durations, indicative of increased 16S rRNA production as bacteria replicate. In the presence of gentamicin, there is a relatively lower production of 16S rRNA over increasing incubation durations. In bulk, the inhibitory effect of 16S rRNA is noticeable after 90 min of culture. Figure S3. Modular droplet device for flexible assay characterization. A PDMS microfluidic device consisting of a droplet generation region and a normally disconnected droplet detection region served as the platform for assay characterization in droplets. Generated droplets enter into a Tygon tube which traverses over heaters that facilitate bacterial lysis (at 95 °C) and PNA probe hybridization (at 60 °C) before re-entering the device for detection. Hybridization duration of droplets was controlled by varying the length of Tygon tubing that rested on the hybridization heater. Droplet volume was controlled by controlling the height of the channels within in the device. As such, separate devices were used for generation of 1 pL, 4 pL, and 30 pL droplet volumes. Scale bars are ~100 μm. Figure S4. One-step sample pretreatment protocol. (A) Our one-step pretreatment protocol includes a single infusion of urine/MH solution through an appropriately sized syringe filter.
(i) Some urine samples can emit a high auto-fluorescence background. In such cases, (ii) 2fold or (iii) or 4-fold dilution of the sample in MH broth is necessary to improve signal to background ratio of positive droplets versus empty droplets, and recover the expected frequency of positive droplets based on the input concentration of bacteria (~10%, here). (B) Particulates in urine can impose a high limit of blank (i.e., frequency of empty droplets with high fluorescence intensities from blank/culture-negative urine samples) and hamper bacteria quantification. We tested 3 different syringe filters of varying pore sizes -GE Whatman (pore size 1.2 μm), GE Whatman (pore size: 2.0 μm), and iPOC-Dx Primecare (pore size gradient from 35 μm to 2.5 μm). One-step filtration of the urine/MH mixture through the iPOC-Dx filter reduces limit of blank by more than an order of magnitude, while ensuring up to 94% bacterial recovery for quantification. Figure S5. Bulk sensitivity of the PNA probe assay. The signal from bacteria spiked into urine samples were measured and compared to the fluorescence from no-bacteria controls in the same samples. For bulk reactions, at least 1.5 x 10^8 CFU mL -1 bacteria must be present in order to effectively measure signal over urine background.

Figure S6. Quantification of antibiotic effect on E. coli in droplets.
For E. coli suspended in MH broth exposed to gentamicin, the normalized positive droplet population decreases as the concentration of gentamicin increases. Figure S7. Schematic of the entire DropDx platform (not to scale). The experimental setup consists of a modular (not pictured here) or integrated microfluidic device that rests on individually controlled Peltier heaters. The detection region of the device is aligned to a 2color LIF detector. Syringe pumps are used to control the flow rates of urine samples, PNA probes, and droplet generation oil, and are finely tuned to generate stable droplets and propel the droplets through the device for the required incubation durations. The thermal rig is able to reliably deliver the correct temperature to the rig with minimal spatial and temporal temperature variation. Figure S11. Clinical comparison study workflow. Data collection was divided into 3 phases. Phase 1, "AST Pilot Study" included the first 15 urine samples tested. Pilot studies were used to establish a data-agnostic threshold for susceptibility/resistance calls and determine the minimum number of samples required for adequate statistical power. Phase 2, "ID Validation Study" included the next 16 samples tested and was used to set up thresholds for the ID classification categories used in this workflow. Phase 2 was also used to validate our measurements for susceptibility/resistance. The final set of samples was used to validate both ID classification categories as well as susceptibility/resistance. (B) (i) Power analysis was conducted using the pilot study data to determine that at least 20 samples must be interrogated in the validation phase to ensure a statistical power (1-β) of 90% and a confidence (α) of 95%. (ii) ROC analysis of the pilot study data was used to determine a data-agnostic threshold that maximizes the Youden's Index of the pilot dataset. This threshold was kept constant for all subsequent data analyzed. (C) Power analyses and ROC-based threshold determination was repeated in Phase 2, the ID Pilot Study, for each classification criteria utilized in DropDx.
Supplementary Tables   Table S1. Summary of designed PNA Probes. We designed and tested 4 unique PNA probes, specific to the species E. coli and P. mirabilis, the Enterobacterales order, and the bacterial kingdom (eubacteria). Custom DNA quenchers were designed tagged with an Iowa-Black quencher that spanned 10-11 complimentary bases to the PNA probes.