Ultra-fast parallelized microfluidic platform for antimicrobial

Subscriber access provided by UNIV OF LOUISIANA

Article

Ultra-fast parallelized microfluidic platform for antimicrobial susceptibility testing of gram positive and negative bacteria Wenjing Kang, Saheli Sarkar, Zhi Shen Lin, Seamus McKenney, and Tania (Tali ) Konry Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.9b00939 • Publication Date (Web): 02 Apr 2019 Downloaded from http://pubs.acs.org on April 3, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Ultra-fast parallelized microfluidic platform for antimicrobial susceptibility testing of gram positive and negative bacteria

Wenjing Kang1, Saheli Sarkar1, Zhi Shen Lin1, Seamus McKenney1, and Tania Konry1*

1Department

of Pharmaceutical Sciences, School of Pharmacy, Bouve College of Health

Sciences, Northeastern University, 140 The Fenway, Boston, MA, 02115

*Address

Correspondence to:

Dr. Tania Konry Department of Pharmaceutical Sciences Northeastern University 140 The Fenway, R 441, Lab 446 Boston, MA 02115

Tel.: (617) 898-7840 Email: [email protected]

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Abstract Antimicrobial susceptibility testing (AST) is an essential diagnostic procedure to determine the correct course of treatment for various types of pathogen infections. Patients are treated with broad spectrum antibiotics until AST results become available, which has contributed to the emergence of multi-drug resistant bacteria worldwide. Conventional AST methods require 16-24 hours to assess sensitivity of the bacteria to a given drug and establish its Minimum Inhibitory Concentration (MIC). A rapid AST assay can assist clinicians in making an informed choice of targeted therapy and avoid unnecessary over-prescription. Here, we have developed a highly parallelized droplet microfluidic platform that can screen four antibiotics/pathogens simultaneously and assess antibiotic sensitivity in 15−30 min. The device consists of four integrated microdroplet arrays, each hosting over 8000 docking sites, which can be operated individually or jointly for greater flexibility of operation. Small numbers (1−4) of bacterial cells were entrapped in droplets of 110 pL volume and monitored dynamically over 2 hours. This imaging-based AST approach was used to determine the growth rates of four types of clinically relevant bacteria known to cause urinary tract infection (UTI) in millions of patients. We quantified doubling times of both gram positive (Staphylococcus aureus, Enterococcus faecalis) and gram negative bacteria (e.g., Escherichia coli, Klebsiella pneumoniae) with varying levels of antibiotic resistance. Six concentrations of bactericidal and bacteriostatic antibiotics (oxacillin and tetracycline respectively) were tested to determine the MIC of the strains as well as the heterogeneity in growth profiles of bacteria at single cell resolution. The MIC determined from phenotypic analysis in droplets matched the MIC obtained from broth microdilution method for all strains. The advantages of the proposed droplet-based AST, including rapid drug sensitivity response, morphological analysis and heterogeneity in antibiotic-resistance profiles, make it an


Page 2 of 26

Page 3 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


excellent alternative to standard phenotypic AST with potential applications in clinical diagnostics and point of care testing.

Keywords: Droplet microfluidics, single cell study, high throughput screening (HTS), antimicrobial susceptibility testing (AST), minimum inhibitory concentration (MIC)



1. Introduction Antimicrobial susceptibility tests (AST) are commonly performed by clinical microbiology laboratories all over the world to determine sensitivity of microorganisms such as bacteria, parasites and viruses to specific antimicrobial agents. While some bacteria continue to show susceptibility to known antibiotics (e.g., Streptococcus pyogenes for penicillin), several types of bacteria can potentially acquire resistance to key life-saving agents.1 This includes isolates from bacterial strains such as Enterobacteriaceae, Pseudomonas, Staphylococcus, Enterococcus, and Streptococcus pneumoniae. In fact, the World Health Organization (WHO) has reported that resistance to first-line drugs is now globally prevalent for Klebsiella pneumoniae (K. pneumoniae) and Escherichia coli (E. coli), which results in treatment failure for 50% of all patients.2 E. coli and K. pneumoniae also cause urinary tract infections (UTI) in approximately 7 million patients annually in US.3-6 The evolution of antimicrobial resistance (AMR), not only for UTI but also for various diseases, are being closely monitored by WHO7, 8 and various international institutions.9 Bacteria with AMR result in increased patient morbidity and mortality.9, 10 Apart from the healthrelated concerns, the common occurrences of multi-drug resistant bacteria in healthcare facilities cause an immense burden to economies all over the world, increasing billions in expenditure every year.8 Thus there is an urgent need for identifying the appropriate antibiotic required for the bacterial strain, as well as estimating the range of dose most likely to eliminate bacterial growth, before beginning treatment to minimize AMR. Currently, there are two types of AST: genotypic and phenotypic. Genotypic ASTs rely on detection of known genetic markers (e.g., mutations) linked with resistance mechanisms. Techniques such as polymerase chain reaction (PCR), padlock probe-dependent rolling circle amplification and whole-genome sequencing have been successfully employed for AST.11-13

1


Page 4 of 26

Page 5 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


While these approaches are typically fast (about 3.5 hours) and sensitive, they are predicated on available information about resistance pathways for the species under consideration. Very few genes have been clearly associated with phenotypic resistance, which leads to limited effectiveness of these tests.1 Additionally, the expression of a genetic marker associated with resistance may not automatically result in phenotypic resistance.14, 15 Phenotypic ASTs serve as the gold standard method for routine clinical evaluation as they allow physicians to determine bacteria growth in the presence of antibiotics. Conventional phenotypic AST includes liquid media-based approaches such as broth microdilution assay and solid media-based ASTs such as disk diﬀusion assay or EtestTM. Both types of assays determine the overall behavior of a large heterogeneous cell population.1,

16

They require prolonged

measurement time (16−24 h1 or up to 4 days17) based on absorbance or visual determination of turbidity caused by multiple cycles of bacteria division18. This translates to a delay of at least one day in initiating targeted therapy with narrow-spectrum agents. This would, in turn, selectively generate or maintain strains with broader spectrum of AMR and potentially worsen the AMR situation. Furthermore, it has been argued that the turbidity of a bacterial culture can result in false positives due to alteration in the morphological characteristics of the bacteria when exposed to antibiotics.19 The development of novel phenotypic ASTs that track proliferation of small subsets of bacteria and reduce detection times significantly to help make clinicians a more informed choice is thus an active field of investigation. A number of microfluidics-based ASTs have been reported recently, with advantages such as rapid analysis, high throughput, and low cost.15, 19-24 Droplet microfluidics has been utilized previously to encapsulate single bacteria in picoliter-volume droplets to enable morphological analysis and response to drug treatment by imaging over time.23 Continuous profiling of individual

2



bacteria at single cell resolution by imaging AST can provide details about various phenotypic features such as cell division, filament formation, shape and motility, all of which may be used to determine antibiotic resistance. This approach also permits detection of minimum inhibitory concentration (MIC), which is defined as the lowest concentration of an antimicrobial drug that prevents visible growth of a pathogen.25 Microfluidic ASTs report outcomes in 15 min to 3.5 hours, depending on the doubling time of the bacteria and the analytical assay employed.15, 22, 24 This is a significant improvement in response time compared to the 16-24 hours needed by conventional phenotypic AST. Furthermore, by encapsulating small numbers of bacteria in a droplet, it is possible to test heteroresistance, defined as the variability in resistance or susceptibility exhibited by subpopulations of isogenic bacteria to a given antibiotic.26 However, the throughput of AST by the previous generation of droplet microfluidic platforms were not sufficient to allow multiplexed analysis of more than one bacteria species or antibiotic concentrations.23 In this study, we developed a large-scale high throughput droplet microfluidics AST platform that consists of four high throughput arrays, each composed of >8000 droplet docking sites. The integrated design is flexible enough to allow independent operation of one unit for samples of limited volume, or joint operation of all units for simultaneous analysis of four combinations of bacteria and/or antimicrobial agents. This accelerates the drug screening process considerably compared to the previous generation of droplet (also known as ScanDrop) platform.21 We selected four clinically relevant strains of UTI-associated bacteria, all of which survived robustly in droplets at low initial loading density of 1−4 bacterial cells per droplet. The MICs of two common antibiotics known to be active against gram positive strains (oxacillin) and gram negative strains (tetracycline)27, 28 were determined in the droplet AST and subsequently confirmed by broth microdilution AST in 96-well plates. Significant differences in proliferation were

3


Page 6 of 26

Page 7 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


observed for all bacteria strains within 30 min in the droplets, particularly at higher doses of antibiotics. The droplet microfluidic AST system thus provides fast, reliable analysis of antibiotic susceptibility for clinical isolates and may be utilized in future as a diagnostic approach to minimize usage of broad-spectrum antimicrobials in large doses.

2. Materials and Methods Bacteria culture and sample preparation. We prepared the bacteria samples based on the protocols from ATCC. One day before the experiment, we started culturing a frozen stock of bacteria at 37 °C for overnight with constant agitation at ~200 rpm. On the day of the experiment, we diluted the overnight bacteria culture for 50× in fresh medium, and let it shake for 2 h at 37°C. We prepared dilutions of oxacillin and tetracycline reagents and stored them at 4 °C until needed. Further details are described in Supplementary Materials and Methods in supporting information (SI). Droplet microfluidic AST. We loaded bacteria suspension and four selected concentrations of antimicrobial reagents into individual 1 mL syringes as the aqueous phase, and flowed them with the oil carrier fluid. We monitored the bacterial growth in at least 100 droplets using the time-lapse imaging function for a total duration of 2 h at 15-min intervals. Details about the operation of the microfluidic system are described in SI.

3. Results and Discussion 3.1 Design and optimization of microfluidic droplet-based AST platform. Here we sought to design a high throughput droplet microfluidic platform that permits susceptibility testing of multiple bacterial strains simultaneously. The integrated quadruplex droplet device consists of four droplet generation and docking arrays that can be controlled concurrently or independently by three on-chip manifolds. To ensure robust monodisperse droplet generation in the device, we

4



first optimized channel dimensions and flow rates by simulating droplet generation in the critical flow-focusing junction (Fig. 1a). The simulations were based on modified Navier-Stokes equation that predicts fluid velocity and pressure in the designed microchannels,29 and were performed with COMSOL Multiphysics® Modeling Software and Microfluidics Module (as detailed in Supplementary Results). The dimensions of the droplets and the corresponding droplet docking sites were selected to ensure imaging of one array in 4±1 minutes, thereby facilitating rapid data acquisition from the entire platform. The diameter of the droplets was thus set to 60µm. We tested several combinations of channel geometries, including a range of channel height (15−200 µm) and channel width (30 µm, 40 µm, and 100 µm). The simulation results showed that a channel of 60 µm height and 30 µm width generated droplets of 49−54 µm diameter (Fig. 1a). The diameter of the droplets could be further adjusted to 60µm by changing the ratio of oil to aqueous phase flow rates from 200:100 µL/h to 800:200 µL/h. Based on the results of the simulation, each unit of the quadruplex droplet microfluidic platform was designed to hold 8010 (90 × 89) droplets of ≤ 60µm diameter (Fig. 1b), which allowed us to significantly reduce the overall screening time as four concentrations of antibiotics were tested in one round of experiment. The four droplet generation and docking arrays are connected by three on-chip manifolds (as indicated in Fig. 1b-c) that permit combining and/or splitting input fluids such as cell suspension and drug solution both from and to all four units. The entire device has two sets of aqueous inlets: a common inlet (inlet 1) shared by all four units, and independent inlets (inlet 2) on the side of each unit (Fig. 1bc). In addition, each unit also has its own independent oil inlet, waste outlet and array outlet which can be controlled separately or together if they are combined onto the three on-chip cross-shaped manifolds. This highly multiplexed design can be used to screen four combinations of bacteria and/or antibiotics

5


Page 8 of 26

Page 9 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


simultaneously in an accelerated manner, thus opening up possibilities of large scale pharmaceutical screening. The width of the serpentine channels and the flow-focusing junction in each microfluidic unit was set at 30 µm as optimized by the simulation. The flow rates of the oil to aqueous phase was maintained at 4:1 ratio. As predicted, the diameter of the droplets obtained was found to be 59.4±2.5 µm, resulting in droplet volumes of 109.6±0.008 pL. The device performed robustly, generating water-in-oil droplets that filled the four parallel arrays and remained stably trapped for up to 24 hours (Fig. 1d-f). We further assessed the frequency of bacterial encapsulation at various initial concentration of bacteria suspension. Our experimental results match estimates predicted by Poisson probability distribution in Fig S1 (SI). As shown for all screened strains, the small volume of media per droplet (~110 pL) did not inhibit growth. This optimized platform was employed to quantify bacterial growth kinetics in the presence of antibiotics. 3.2 Quantification of MIC and drug sensitivity in droplet microarray. We tested the performance of the system using four clinically relevant uropathogenic bacterial strains that have been used as references in AST.30, 31 Two of the chosen strains are gram positive − S. aureus 29213 and E. faecalis 29212,32,33 and two are gram negative − E. coli 25922 and K. pneumoniae 700603.34,35 The response of these bacteria to two well-known drugs oxacillin and tetracycline was determined as both antibiotics have been applied previously to treat various bacterial infections.27 These drugs have been shown to be effective against specific bacterial strains while other strains exhibit tolerance and/or resistance. This allowed us to explore the possibility of quantifying phenotypic resistance at single cell resolution. The critical features of the four selected bacteria strains, including their growth conditions, expected phenotypes (sensitive versus resistant) and the type and concentrations of antibiotics tested are listed in Table S1 (SI). The doubling time and the

6



MIC for each bacteria/drug combination was quantified from the growth trends in droplets. As defined previously, the lowest drug concentration that restricted bacteria growth in the droplets was determined to be the MIC.25 Additionally, MIC was also measured from standard AST of the same bacteria culture on the same day in 96-well plate-based broth microdilution so as to confirm the findings of the droplet-based AST.18 3.2.1 Susceptibility of gram positive bacteria to oxacillin. We first screened the grampositive strains for oxacillin susceptibility (Fig. 2). Oxacillin, like other beta-lactam antibiotics, promotes lysis of bacteria by interfering with specific penicillin-binding proteins (PBPs) present in the cell wall. We selected two gram-positive bacteria, S. aureus 29213 which is known to be inhibited by oxacillin (i.e., sensitive)36, 37 and E. faecalis 29212 which is known to be resistant to this drug.28, 37 Accordingly, we selected lower doses of oxacillin for the sensitive strain (0.03−0.5 µg/mL), and higher doses for the resistant strain (4−32 µg/mL). S. aureus 29213 depicted a doubling time of 36 min and natural spherical morphology in the droplets in the absence of antibiotics (Fig. 2a-b). This doubling time for S. aureus is similar to that reported by other groups (46±10 min).38 The average normalized growth curves of S. aureus at different concentrations of oxacillin are demonstrated in Fig. 2b. Increasing the concentration of oxacillin slowed growth, extending doubling time progressively to 56−224 min (Fig. 2b, Table S1). Even the lowest concentration of oxacillin (0.03 µg/mL) significantly inhibited growth (p < 0.0002) compared to the control condition (no drug) as early as 15 min. As expected, all higher concentrations (0.06−0.5 µg/mL) also depicted significant growth inhibition over 15−60 min in droplets. Some evidence of bacterial replication was detected in the presence of 0.03−0.06 µg/mL, but not 0.12 µg/mL, at 1.5 and 2 hours (Fig. 2b). Based on these findings, the MIC of oxacillin for S. aureus 29213 in droplets was determined to be 0.12 µg/mL. The response of the strain in the

7


Page 10 of 26

Page 11 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


independent standard broth microdilution AST (Fig. 2c) also indicate that the absorbance of the wells with ≥ 0.12 µg/mL oxacillin are significantly low (p < 0.00006 for 0.12 µg/mL). The detected MIC for S. aureus 29213 agreed with the range of MIC (≤ 0.5 µg/mL) reported in literature.36, 37 The doubling time of E. faecalis 29212 was found to be 38 min in droplets, which falls within the previously reported values of ~26 min 39 and 48.6±3.7 min40 in other platforms (Table S1). Addition of 4 µg/mL of oxacillin resulted in significant reduction in bacteria numbers (p value 0.03) by 30 min in droplets. All concentrations of the antibiotic inhibited E. faecalis growth significantly (p < 0.01) at 60 min. E. faecalis has been known to exhibit resistance to oxacillin and it proliferated at lower concentrations of the antibiotic (fold change in bacteria numbers at 60min: 1.9−2.1 for 4−8 µg/mL). However, its growth was slowed (doubling times: 47 and 59 min respectively) compared to the control condition (no antibiotic: 38min) (Table S1). Further increase in oxacillin concentration to 16µg/mL inhibited growth markedly (Fig. 3a). The MIC of E. faecalis 29212, calculated in a similar manner as described above, was thus determined to be 16 µg/mL. The standard AST also revealed significant lack of proliferation at this concentration (p < 0.00004, Fig. 3b). The high value of MIC in the droplets confirms that this strain is relatively resistant to oxacillin as also reported in literature.28, 37 It is evident from these findings that the droplet-based AST matched the findings of the standard AST for both gram-positive strains (Fig. 3b). 3.2.2 Susceptibility of gram negative bacteria to tetracycline. We next screened the growth of two gram-negative strains − E. coli 25922 and K. pneumoniae 700603 upon addition of tetracycline. Similar to the gram-positive bacteria experiments, we selected two strains of gramnegative bacteria that are inhibited by relatively low (sensitive) or high (resistant) concentrations of tetracycline.

8



E. coli 25922 was considered an example of tetracycline-susceptible bacteria, with reported MICs < 4 µg/mL.41 Thus we evaluated this strain at 0.2, 0.5, 1, 4, and 8 µg/mL of the antibiotic (Fig. 4, Table S1). The average doubling time for E. coli 25922 bacteria alone was assessed to be 39 min, comparable to the published ranged of 20−68 min.42, 43 The doubling time was lengthened to 98−330 min at 0.2−8 µg/mL tetracycline in droplets (Table S1). The earliest inhibitory effect of the antibiotic was detected at 30 min (p value: 0.04 and 0.019 for 0.2 and 0.5 µg/mL respectively). This strain was sensitive to the growth-inhibitory effect of low levels of tetracycline (fold change in bacteria at 0.2 and 0.5 µg/mL: 2.2 and 1.8 respectively, compared to 4.1 fold in the absence of the antibiotic) (Fig. 4a). Based on these quantifications, MIC for E. coli was estimated to be 0.5 µg/mL. The standard AST experiments performed in 96-well plate confirmed the results obtained from the droplets (Fig. 4b). K. pneumoniae 700603, the tetracycline resistant strain,35 was treated with 1, 2, 4, 8, and 16 µg/mL. The bacteria grew robustly in the absence of tetracycline with a doubling time of 30 min (Table S1), which is slightly lower than 38−40 min44 time reported by other groups. Low doses of tetracycline such as 1 µg/mL downregulated growth profiles as early as 15 min (p value 0.00003 compared to control). Although 1 µg/mL of the drug diminished proliferation significantly (p

Ultra-fast parallelized microfluidic platform for antimicrobial

Recommend Documents