Rapid Antimicrobial Susceptibility Testing by Stimulated Raman Scattering Imaging of Deuterium Incorporation in a Single Bacterium

Abstract

To slow and prevent the spread of the antimicrobial resistant infenctions, rapid antimicrobial susceptibility testing (AST) is in urgent need to quantitatively determine the antimicrobial effects on pathogens. It typically takes days to complete the AST by conventional methods based long-time culture, and they do not work directly for clinical samples. Here, a rapid AST method enabled by stimulated Raman scattering (SRS) imaging of deuterium oxide (D₂O) metabolic incorporation is reported. Metabolic incorporation of D₂O into biomass and the metabolic activity inhibition upon exposure to antibiotics at the single bacterium level are monitored by SRS imaging. The single-cell metabolism inactivation concentration (SC-MIC) of bacteria upon exposure to antibiotics can be obtained after total 2.5 h of sample preparation and detection. Furthermore, this rapid AST method is directly applicable to bacterial samples in complex biological environments, such as urine or whole blood. SRS metabolic imaging of deuterium incorporation is transformative for rapid single-cell phenotypic AST in clinic.

SUMMARY:

This protocol presents rapid antimcrobial susceptibility testing (AST) within 2.5 h by single-cell stimulated Raman scattering imaging of D₂O metabolism. This method is applicable to bacteria in urine or whole blood environment, which is transformative for rapid single-cell phenotypic AST in clinic.

INTRODUCTION:

Antimicrobial resistance (AMR) is a growing global threat to the effective treatment of infectious disease.¹ It is predicted that AMR will cause an additional 10 million deaths per year and $100 trillion global GDP loss by 2050, if no action for combating antibiotic-resistant bacteria is taken.^1,2 It stressed the urgent need of rapid and innovative diagnostic methods for antibiotic susceptibility testing (AST) of infectious bacteria to slow down the emergence of antibiotic-resistant bacteria and therefore reduce the related mortality rate.³ To ensure the best possible clinical outcome, it is crucial to introduce effective therapy within 24 hour. However, the current gold standard method, like diffusion or broth dilution method, usually takes at least 24 h for the preincubation procedure for clinical samples and additional 16–24 h to obtain the minimal inhibitory concentration (MIC) results. Overall, these methods are too time-consuming to guide immediate decision for infectious disease treatment in clinic, which leads to the emergence and spread of antimicrobial resistance.⁴

Genotypic AST methods, such as polymerase chain reaction (PCR)-based techniques,⁵ have been developed for rapid detection. Such techniques measure the specific resistance genetic sequences in order to provide rapid AST results. They don’t rely on time-consuming cell culture, however, only specific known genetic sequences with resistance are tested. Its application is limited to various bacterial species or different mechanisms of resistance. Also, they cannot provide MIC results for therapy decisions.^6,7 Besides, novel phenotypic methods for rapid AST are under development to overcome these limitations,⁸ including microfluidic devices,^9–13 optical devices,^14–16 phenotypic AST quantifying the nucleic acids copy number,^17,18 and Raman spectroscopic methods.^19–24 These methods reduce time to guide AST results, however, most of them are only applicable to bacterial isolates, not direct clinical specimen, and still require long-time preincubation.

In this work, we present a method for rapid determination of the susceptibility of bacteria in urine and whole blood via monitoring cellular metabolic activity by SRS imaging. Water (H₂O) takes part in the vast majority of essential biomolecular synthesis process in living cells. As the isotopologue of water, through enzyme-catalyzed H/D exchange reaction between the redox-active hydrogen atom in NADPH and the D atom in D₂O, deuterium can be incorporated into biomass inside a cell.^25,26 Deuterated fatty acid synthesis reaction is mediated by the deuterium labeled NADPH. And the D₂O incorporation into reactions of amino acids results in the deuterated proteins production.²⁶ (Figure 1) In this way, the newly synthesized C–D bond-containing biomolecules in single microbial cells can be employed as a general metabolic activity marker to be detected. To read out de novo synthesized C-D bonds, Raman spectroscopy, a versatile analytical tool providing specific and quantitative chemical information of biomolecules, is widely used to determine antimicrobial susceptibility and can significantly reduce the testing time to a few hours.^27–30 However, due to the inherent low efficiency of the Raman scattering process, the spontaneous Raman spectroscopy is of low detection sensitivity. Thus, it is challenging to obtain real-time image results using spontaneous Raman spectroscopy. Coherent Raman scattering (CRS), including coherent anti-Stokes Raman scattering (CARS) and stimulated Raman scattering (SRS), has reached high detection sensitivity because of the coherent light field to generate orders of magnitude larger than that of spontaneous Raman spectroscopy, thereby rendering high-speed, specific, and quantitative chemical imaging at the single cell level.^31–39

Figure 1. Scheme for D₂O incorporation into deuterated lipid and protein.

Deuterium can be incorporated into biomass inside a cell through enzyme-catalyzed H/D exchange reaction between the redox-active hydrogen atom in NADPH and the D atom in D₂O. Deuterated fatty acid synthesis reaction is mediated by the deuterium labeled NADPH. The D₂O incorporation into reactions of amino acids results in the deuterated proteins production. Adapted from ref⁴⁰.

Here, based on our most recent work,⁴⁰ we present a protocol for rapid determination of the metabolic activity and antimicrobial susceptibility by femtosecond SRS C-D imaging of D₂O incorporation of bacteria in normal medium, urine and whole blood environment at the single cell level. Femtosecond SRS imaging facilitates monitoring metabolic response in the presence of antibiotics and deuterium at the single bacterium level in as fast as 10 min. Single-cell metabolism inactivation concentration (SC-MIC) against antibiotics is obtained within 2.5 h. The SC-MIC results are validated by standard MIC test via broth microdilution. Our method is applicable to determine antimicrobial susceptibility of bacteria urinary tract infection (UTI) and bloodstream infection (BSI) pathogens with a much reduced assay time in comparison with the conventional method, which opens the opportunity for rapid phenotypic AST in clinic at the single-cell level.

PROTOCOL:

1. Preparation of bacteria and antibiotics stock solution

   • 1.1
     Prepare the antibiotics (gentamicin sulfate or amoxicillin) stock solution at a concentration of 10 mg/mL dissolved in 1x phosphate-buffered saline (PBS) or dimethyl sulfoxide (DMSO) solvent in 1.5 mL micro tubes. Gentamicin sulfate is dissolved in sterile PBS solution. Amoxicillin is dissolved in sterile DMSO solvent. Thereafter, store the antibiotic solution at 2–8 °C as suggested.
   • 1.2
     To make D₂O containing cation-adjusted Mueller-Hinton Broth (MHB) media, add 220 mg of MHB broth base to 10 mL of D₂O to make 100% D₂O containing medium. Sterilize the media solution by filtering with filters in 200 nm pore size.
     NOTE: This protocol is always used for making and sterilizing medium solutions in further steps.
   • 1.3
     To prepare bacterial samples for SRS imaging, add 2 mL of normal MHB media, which does not contain deuterium, to a sterile round-bottom tube and then take it to prewarm at 37 °C.
   • 1.4
     Use a sterile loop to select one bacterial (Escherichia coli BW25113 or Pseudomonas aeruginosa ATCC 47085) colony from the fresh culture on a triptic soy agar plate, and then suspend it in the prewarmed culture media by gently vortex to prepare the bacteria suspension.
   • 1.5
     Incubate bacteria at 37 °C in a shaker until the logarithmic phase is reached.
2. D₂O incorporation treatment in the presence of antibiotics (Figure 2a)

   • 2.1
     Check the bacterial concentration by measuring the optical density with a photometer at a wavelength of 600 nm.
   • 2.2
     Dilute the bacteria solution to the normal MHB medium, which does not contain deuterium, to reach a final cell concentration of 8X10⁵ CFU/mL. Vortext gently to mix the bacterial cells.
   • 2.3
     Seperate the bacteria solution to 300 μL aliquots in ten 1.5 mL micro tubes.
   • 2.4
     Prepare two-fold serial dilutions of the test antibiotics, gentamicin or amoxicillin, using the aliquots of bacteria solution as a dilution medium. For both of gentamicin and amoxicillin, the serial concentrations cover from 0.25 to 8 μg/mL.
   • 2.5
     Incubate 300 μL of bacteria with certain antibiotic (gentamicin or amoxicillin) containing MHB medium for 1 h.
   • 2.6
     Incubate 300 μL of bacteria with no antibiotics as well. This will be a negative control to inspect the bacterial metabolic activity without antibiotics treatment.
   • 2.7
     In the meantime of incubation, prepare the serial diluted concentration of antibiotics in the 100% D₂O containing medium with the same concentration gradient of antibiotics. For both of gentamicin and amoxicillin, the serial concentrations cover from 0.25 to 8 μg/mL.
   • 2.8
     After 1 h antibiotic treatment, add 700 μL of serial diluted antibiotic and 100% D₂O-containing MHB medium to the antibiotic-pretreated bacteria (in steps 2.5, 2.6) respectively, and then incubate at 37 °C in an incubation shaker at 200 revolutions per minute (rpm) for additional 30 min. Add 700 μL of antibiotic-free and D₂O-free MHB medium to bacteria as a negative control. In this step, the final concentration of D₂O in the medium for the test is 70%.
   • 2.9
     First centrifuge the 1 mL of antibiotic and D₂O-treated bactria sample at 8000 rpm for 5 min at 4 °C, then wash samples twice with purified water. Fix samples by 10% (w/v) formalin solution and store them at 4 °C.
3. Preparation of bacteria in urine environment (Figure 2b)

   • 3.1
     To prepare Escherichia coli BW25113 at the logarithmic phase, follow the steps at 1.
   • 3.2
     Check the bacterial concentration by measuring the optical density (OD).
   • 3.3
     To mimic the clinical UTI samples,^14,18,41 spike the E. coli solution into 10 mL of deidentified urine to reach a final cell concentration of 10⁶ CFU/mL.
   • 3.4
     Filter the E. coli spiked urine using a 5-μm filter, and then separate bacterial solution to 300 μL aliquots into ten 1.5 mL micro tubes 10 eppendorf tubes.
   • 3.5
     The next steps of D₂O incorporation treatment in the presence of antibiotics are the same as the previous steps shown from 2.4 to 2.9.
4. Preparation of bacteria in blood environment (Figure 2c)

   • 4.1
     To prepare Pseudomonas aeruginosa ATCC 47085 at the logarithmic phase, follow the steps at 1.
   • 4.2
     To mimic the clinical bloodstream infections samples,^42,43 spike P. aeruginosa in 1 mL of deidentified human blood to reach a final concentration of 10⁶ CFU/mL.
   • 4.3
     Add 9 mL of sterile purified water to lyse the blood.
   • 4.4
     Filter the P. aeruginosa spiked blood using 5-μm filter, and then harvest bacteria to 1 mL volume by centrafugation at 8000 rpm for 5 min at 4 °C.
   • 4.5
     Separate the spiked blood to 300 μL aliquots into ten 1.5 mL micro tubes.
   • 4.6
     The next steps of D₂O incorporation treatment in the presence of antibiotics are the same as the previous steps shown from 2.4 to 2.9.
5. SRS imaging of D₂O metabolic incorporation in a single bacterium

   • 5.1
     Wash 1 mL of fixed bacteria solution with purified water and then centrifuge at 8000 rpm for 5 min at 4 °C. Remove the supernatant. Enrich the bacteria solution to about 20 μL.
   • 5.2
     Deposit the bacteria solution on a poly-L-lysine coated coverglass. Sandwich and seal the sample for SRS imaging.
   • 5.3
     In the SRS microscope, a tunable femtosecond laser with an 80-MHz repetition rate provides the pump (680 to 1300 nm) and Stokes (1045 nm) excitation lasers. To image bacteria at the C-D vibrational frequency at 2167 cm⁻¹, input and tune the pump wavelength to 852 nm in InSight graphical user interface control software on a computer. Measure the laser power using a power meter. Set the power of pump laser at the sample to ~8 mW and the power of Stokes laser at the sample to ~40 mW by adjusting the ND filter in front of the laser output.
   • 5.4
     By adjusting the screws of the reflection mirrors, spatially align the pump and Stokes beams and direct the two beams into an upright microscope equipped with 2D galvo mirror system for laser scanning. Use a 60× water immersion objective to focus the pump and Stokes lasers on the sample. Use an oil condenser to collect the signals from the sample in forward direction. Use a bandpass filter to filter out the Stokes laser before directing into a photodiode. Extract the stimulated Raman signal by a lock-in amplifier and detect the signals by the photodiode.
   • 5.5
     Set each SRS image to contain 200 × 200 pixels and the pixel dwell time for 30 μs in the LabView control panel. The total acquisition time for one image is ~1.2 s. Set the step size for 150 nm, so the image size is about 30X30 μm². Image at least three field of views for each sample.
6. Image processing and data analysis (Figure 3)

   • 6.1
     To obtain the average C-D signal intensity, open and process SRS images with ImageJ software.
   • 6.2
     First convert SRS images into 8-bit type images with inverted color by clicking “Image”-“Type”-“8-bit”, and then “Edit”- “Invert” buttons in the ImageJ software.
   • 6.3
     Then, filter the images with Gaussian blur by clicking “Process”-“Filters”-“Gaussian blur” buttons and set the “Sigma (Radius)” to 1.
   • 6.4
     Use image threshold adjustment to select bacterial area. Click “Image”-“Adjust”-“Threshold” to ensure the selected bacterial sizes match those in the original SRS images. Eliminate small particles by adjusting the size threshold to determine the particles. Click “Apply”.
   • 6.5
     Apply “Analyze”-“Particles analysis” buttons to label and determine the area of bacteria.
   • 6.6
     By click the “show all” button in the “ROI manager” to the original unprocessed SRS image, label the same area of bacteria, determine the average intensity of each data point by click “measure” button in the “ROI manager”.
   • 6.7
     Circle the background area in the SRS image and measure the average intensity of background. The average C-D intensities of each bacterium is obtained by deducting the background signal intensity.
7. Quantitation of antimicrobial susceptibility via SC-MIC

   • 7.1
     The cut-off value at 0.60 to determine the SC-MIC is established according to the statistical analysis of the SRS C-D intensities of the metabolism-active and metabolism-inhibited conditions for bacteria upon various concentrations of drug exposure.⁴⁰ The C-D intensities for the antibiotic-susceptible and antibiotic-resistant groups were fitted with normal distribution. Plot the receiver operating characteristic (ROC) curve and the cut-off threshold is evaluated at 0.60. Based on this cut-off value, the SC-MIC as an indicator of the efficacy of antibiotics can be defined to determine the metabolically inactive and metabolically active group.
   • 7.2
     To quantitatively analyze the SRS imaging data, plot the histograms of C-D signal intensities for each bacteria group treated with the serial dituted antibiotic concentration in the Origin software. The colored data points stand for different individual bacterium.
   • 7.3
     Normalize the C-D intensities of antibiotic-treated group to the mean intensity of the control group without antibiotic treatment. Determine the SC-MIC results of different bactera and antibiotic combinations by quantifying the SRS signal intensities at C-D region versus various concentration of antibiotics using the cut-off value at 0.60.
   • 7.4
     Validate and compare the SC-MIC readout with the MIC determined using conventional broth microdilution assay.
   • 7.5
     According to the Clinical and Laboratory Standards Institute (CLSI), the susceptibility category based on the SRS metabolic imaging results for each tested bacterial strain is interpreted as “susceptible”, “resistant”, or “intermediate”.

Figure 2. Workflow of rapid AST by SRS metabolic imaging of deuterium incorporation.

(a) D₂O incorporation treatment in the presence of antibiotics in MHB medium, and the following SRS imaging procedures. (b) Preparation of bacteria in urine environment. (c) Preparation of bacteria in blood environment. Adapted from ref⁴⁰.

Figure 3. Automated image processing and data interpretation.

(a) Raw SRS image. (b) Image after intensity threshold adjustment to determine the area of bacterial cells. (c) Data points selected after particle analyze step. (d) The corresponding data points selected in the raw image. (e) Results of the corresponding data points in the raw image. (f) Statistical results of the average intensity of the data points after subtraction of background. Scale bar: 10 μm. Adapted from ref⁴⁰.

REPRESENTATIVE RESULTS:

The effect of incubation time on deuterium incorporation is measured by spontaneous Raman microspectroscopy at the C-D (2070 to 2250 cm⁻¹) and C-H (2,800 to 3,100 cm⁻¹) region (Figure 4a). The time-lapse single-cell Raman spectra of P. aeruginosa cultured in 70% D₂O containing medium show increasing CD/CH intensity over incubation time from 0 to 3 h. (Figure 4b) The increasing C-D abundance in single microbial cells reveals that D₂O is incorporated into deuterated biomolecules inside the cell.

Figure 4. Effect of incubation time on deuterium incorporation into bacteria.

(a) Time-lapse measurement at C-D (2070 to 2250 cm⁻¹) and C-H (2,800 to 3,100 cm⁻¹) region by spontaneous Raman microspectroscopy (N=20). (b) Histogram plot of CD/CH intensity ratio plot over D₂O incubation time for bacteria in (a). Each colored point stands for a measurement from a single bacterium. Error bars represent the standard error of the mean (SEM).

Following the protocol, P. aeruginosa was incubated with serially diluted gentamicin for 1 h and then 70% D₂O for additional 30 min. SRS metabolic imaging at ~2162 cm⁻¹ (Figure 5a) was conducted. The C-D intensities upon antibiotic treatment are divided by the mean value of the control group which is without antibiotic treatment. The quantitative statistical analysis (Figure 5b) showed that C-D signals of P. aeruginosa were significantly lower at 2 μg/ml or higher gentamicin concentration than that without gentamicin treatment (0 μg/ml). Using the cut-off threshold at 0.60, the P. aeruginosa was matabolically inhibited at 2 μg/mL and higher concentrations of gentamicin. The dotted line is showing the defined cut-off value at 0.60 by in Figure 5b. In this way, the SC-MIC for P. aeruginosa against gentamicin in normal MHB medium was determined to be 2 μg/ml. This SC-MIC value is varified to be within the one-fold difference range with the MIC (4 μg/ml) determined by the broth microdilution method (Figure 5c). Taken together, SC-MIC determined by our technology enables quantification of antimicrobial susceptibility.

Figure 5. SC-MIC determination using SRS imaging of bacterial metabolic incorporation of D₂O againt antibiotics in normal medium, urine and blood environment.

(a) SRS imaging at C-D vibration (2168 cm⁻¹) and the corresponding transmission images of P. aeruginosa in the presence of D₂O with the addition of serially diluted gentamicin in normal MHB medium. (b) Quantitative analysis of SRS C-D intensity of P. aeruginosa in (a). The colored data points in the histogram stand for different individual bacterium. The dotted line indicates the cutoff value at 0.60. (c) The comparison of the SC-MIC readout with the MIC by broth microdilution method, and susceptibility category for P. aeruginosa according to the CLSI. (d) SRS imaging at C-D vibration (2168 cm⁻¹) and corresponding transmission images of E. coli in urine after incubation in D₂O with the serially diluted amoxicillin. (e) Quantitative analysis of SRS C-D intensity in (d). (f) The comparison of the SC-MIC readout and susceptibility category for E. coli in normal MHB and in urine. (g) SRS imaging at C-D vibration (2168 cm⁻¹) and corresponding transmission images of P. aeruginosa in blood after incubation in D₂O with the serially diluted gentamicin. (h) Quantitative analysis of SRS C-D intensity in (g). (i) The comparison of the SC-MIC readout and susceptibility category for P. aeruginosa in normal MHB and in blood. S: sensitive. Number of cells N ≥ 10 per group. Error bars represent the standard error of the mean (SEM). Scale bar: 10 μm. Adapted from ref⁴⁰.

To explore the potential of rapid AST by SRS imaging of deuterium metabolic incorporation for clinical applications, especially for the most prevalent infection which is UTI, we tested bacteria-spiked urine sample using E. coli, which is the most common pathogen to cause UTI infection.⁴⁴ To mimic the clinical UTI samples at a relavent bacterial concentration, E. coli is added to the deidentified urine to a final concentration of 10⁶ CFU/ml. After sample purification, the urine sample were incubated with amoxicillin and D₂O. The clean background in the SRS images showed that the sample preparation protocol was applicable for rapid AST measurement (Figure 5d). The SC-MIC for the E. coli-spiked urine sample against amoxicillin was determined to be 4 μg/ml (Figure 5e), which has the same susceptibility readout with the MIC (8 μg/ml) by conventional broth dilution method for pure E. coli in normal MHB medium (Figure 5f). These results collectively showed that rapid AST by SRS imaging of deuterium metabolic incorporation is of great potential for clinical diagnosis to UTI infectious pathogens.

As compared with UTI infection, rapid AST for BSI pathogens is much more challenging for in situ study of bacterial metabolic activity, as a lot of blood cells presenting in blood. To investigate the applicability of rapid AST by SRS imaging of D₂O metabolic incorporation for clinical BSI samples, P. aeruginosa spiked in deidentified human blood was detected. As shown in Figure 5g, the C-D intensity of SRS image at 2168 cm⁻¹ was dominated by bacterial signals. Since the red blood cells do not have metabolic activity to uptake D₂O for further biosynthesis, the C-D signals were originated from the metabolic deuterium incorporation of live bacteria. The cross-phase modulation or photothermal signal of debris or red blood cells species contributed to the weak background signals, without affecting the quantitative analysis of the SC-MICs. The SC-MIC result for P. aeruginosa in blood was determined to be 2 μg/ml (Figure 5h), which agreed well with the conventional standard MIC result for P. aeruginosa in normal growth medium (Figure 5i). Taken together, these results showed that SRS metabolic imaging of deuterium metabolic incorporation can be a rapid AST method to determine the SC-MIC for bacteria in BIS infections.

Conclusion

Rapid AST can be obtained by assessing the response of metabolic activity of bacteria to antibiotic treatment using single-cell SRS metabolic imaging within 2.5 hours from sample to SC-MIC results. Bacteria metabolic activity response and the antimicrobial susceptibility can be detected by monitoring the metabolic incorporation of D₂O, which are incoporated for biomolecule synthesis, by SRS imaging of C–D bonds. Since water are ubiquitously used in living cells, SRS metabolic imaging provides a universal method for rapid AST. The rapid AST method is applicable to detect bacteria in complex biological environments, such as urine or whole blood at single bacterium level. The SC-MIC can be determined after 1 h culture of bacteria in urine and blood, which is considered transformative to shift the paradigm of UTI and BSI diagnosis from a time-consuming culture-dependent procedure to a culture-independent in situ approach. Therefore, it means tremendous reduction in diagnosis time as compared with the conventional broth microdilution method, which paves the way towards clinical translation allowing for on-time identification of appropriate antimicrobial agents for precise treatment.

DISCUSSION:

• aCritical steps in the protocol

The protocols of the antibiotics treatment described here are following the guidelines of the CLSI, in which the MHB medium is a suggested medium that can be generally used for cultivation of a wide variety of microorganisms. A key parameter is that the bacterial cell number used for antimicrobial susceptibility testing is keep at about 5X10⁵ CFU/mL as recommended in CLSI. This is of critical importance for obtaining accurate and reproducible results. In antibiotics treatment experiments the bacteria concentration is set at 8X10⁵ CFU/mL. Higher bacteria concentration can lead to an increase in the MIC results. Once the bacterial suspension is adjusted, it must be used within 30 min to avoid changes in the bacterial cell concentration.

The cation-adjusted MHB medium contains 20–25 mg/L of Ca²⁺. Therefore, as the recommended calcium concentration for daptomycin susceptibility testing is 50 mg/L, ensure additional Ca²⁺ (solubilized in water and filter sterilized) in the concentration of 30μg/mL is further supplemented.

Another critical step in the presented method is the incubation time of bacteria upon antibiotics exposure and D₂O incorporation. Because the generation time of bacterial life cycle is roughly 30 to 60 min, it is important to influence bacterial metabolic activity upon antibiotic exposure for a certain time. This test has been evaluated for a variety of bacteria/antibiotic combinations to antibiotic exposure for 1 h prior to following 0.5-h D₂O treatment. The first 1 h antibiotic treatment step was essential to influence the bacterial metabolic activity. Next, bacteria were incubated with D₂O-containing and antibiotic-containing medium for additional 30 min. The final antibiotic concentration are maintained at the same level, and the final concentration of D₂O was adjusted to 70%. Overall, after 1 h antibiotic preculture and following 0.5 h of D₂O and antibiotic incorporation, the SC-MIC results were then determined by SRS metabolic imaging of the bacterial metabolic activity. This design minimized the impact of D₂O influence of antimicrobial activity to bacteria, and also leads to a comparable readout of SC-MIC results with the MICs by conventional method.

• bModifications and troubleshooting of the method

Table 1. Troubleshooting table.

Table 1.

Troubleshooting table.

Problem	Possible reason	Solution
Little or no bacterial cell number in imaging field of view	Bacterial density in the solution is too low	Centrifuge for longer time to further enrich bacteria
Photodamage of the bacterial cells during SRS imaging	The laser power used is too high	Tune down the laser power to an appropriate value
No SRS signal is detectable	The spatial and temporal overlap of the pump and Stokes beam is not optimized	Align the pump and Stokes beams using a standard sample deuterated dimethyl sulfoxide

• cLimitations of the method

In the SC-MIC measurements, we prepared in parallel 40 samples including 5 different antibiotics, each with 8 concentrations at the same time. However, because there are a lot of mannual operation procedures, the total assay time for detecting the AST for 5 different bacteria-antibiotics combinations is longer than 2.5 h. In our method, each SRS image, containing ~20 bacteria, was obtained within ~1.0 second in one single shot at 30 μs pixel dwell time. The total MIC assay time to study 10 antibiotics for one strain would be less than 2 h from sample to MIC readout, which has tremendous possibility to perform high throughput measurement. In future work, to further improve the throughput, automated sample preparation and imaging data acquisition method will be employed.

• dThe significance of the method with respect to existing/alternative methods

In conventional culture-based AST methods, to obtain bacterial isolates for further measurements, it is necessary to pre-incubate clinical specimens for hours. Advanced AST methods for clinical UTI sample, such as Raman spectroscopy,²⁹ nanoliter array,⁶ and digital nucleic acid quantification,¹⁸ have been developed to get rid of long-time preincubation. Compared with UTI infection, the BSI or sepsis is much more life-threatening,^18,45 where rapid AST is urgently needed for precise diagnostics in clinic. A microscopic imaging method to measure bacterial colony formation from positive blood cultures has been reported to provide MIC results.⁴⁶ However, it takes at least 6 h to grow bacteria to conduct the AST assay. Furthermore, commercial automatic systems⁴⁷ and mass spectrometry^48,49 strategies can provide AST readout from positive blood cultures. However, the MIC results for clinical decision cannot be provided. The AST results and the MIC readout are significant to avoid excess dosage of antibiotics to patients to cause potential side effects in clinics, to slow and prevent the spread of the antimicrobial resistant infenctions.^50,51 Compared with the existing spontaneous Raman microscopy-based AST methods, our technology tremendously reduced data acquisition time (ca. 600 times less) due to orders-of-magnitude signal enhancement. In this work, we demonstrate rapid AST by SRS imaging of deuterium metabolism in single bacteria at a clinically relevant bacterial concentration (10⁵~10⁶ CFU/ml) in either urine or whole blood environment. As shown in previous results, the MIC results can be determined after 1-h antibiotic treatment and 30-min mixture of D₂O and antibiotics incubation into bacteria in urine and blood. Our method can provide MICs and susceptibility classification for each strain-antibiotic combination within 2.5 h, and therefore opens a new avenue to clinical translation. To summarize, without the need of preculturing and bacterial division, our method has an enormous potential in the field of rapid and high throughput AST in infectious diseases.

• eImportance and potential applications of the method in specific research areas

Our SC-MIC method by SRS metabolic imaging is applicable to detect MICs and provide susceptibility classification for infectious pathogens when dealing with abundant varieties of strain-antibiotic combinations for clinical use. The SC-MIC is determined after 30-min of D₂O incorporation into bacteria in urine and blood, which means a tremendous reduction in diagnosis time compared with the conventional broth dilution method costing 16 to 24 h for preincubation. To deliver pathogen identification information for clinical decision-making, SRS metabolic imaging technology can be further integrated with diagnostic platforms that are capable of rapid pathogen identification, for example, matrix-assisted laser desorption ionization-time-of-flight mass spectrometry.^49,52,53, Combining in situ pathogen identification and rapid AST diagnosis, it could be of great potential for translation into clinic that allows for on-time identification of appropriate antimicrobial agents for precise treatment.

TABLE OF MATERIALS:

Name	Company	Catalog Number	Comments
Acousto-optic modulation	Gooch&Housego	R15180-1.06-LTD	Modulating stokes laser beam
Amoxicillin	Sigma Aldrich	A8523-5G
Bandpass filter	Chroma	HQ825/150m	Block the stokes laser beam before the photodiode
Calcium chloride	Sigma Aldrich	C1016-100G	Cation adjustment
Cation-adjusted Mueller-Hinton Broth	Fisher Scientific	B12322	Antimicrobial susceptibility testing of microorganisms by broth dilution methods
Centrifuge	Thermo Scientific	75002542
Cover Glasses	VWR	16004-318
Culture tube with snap cap	Fisher brand	149569B
Daptomycin	Acros	A0386346
Deuterium oxide		151882	Organic solvent to dissolve antibiotics
Deuterium oxide-d6	Sigma Aldrich	156914	Organic solvent as a standard to calibrate SRS imaging system
Escherichia coli BW 25113	The Coli Genetic Stock Center	7636
Eppendorf polypropylene microcentrifuge tubes 1.5 mL	Fisher brand	05-408-129
Gentamicin sulfate	Sigma Aldrich	G4918
Hydrophilic Polyvinylidene Fluoride filters	Millipore-Sigma	SLSV025NB	pore size 5 μm
ImageJ software	NIH	Version: 2.0.0-rc-69/1.52t	Image processing and analysis
Incubating orbital shaker set at 37 °C	VWR	97009-890
Inoculation loop	Sigma	BR452201-1000EA
InSight DeepSee femtosecond pulsed laser	Spectra-Physics	Model: insight X3	Tunable laser source and fixed laser source at 1045 nm for SRS imaging
Lock-in amplifier	Zurich Instrument	HF2LI	Demodulate the SRS signals
Oil condenser	Olympus	U-AAC	NA 1.4
Pseudomonas aeruginosa ATCC 47085 (PAO1)	American Type Culture Collection (ATCC)	ATCC 47085
Photodiode	Hamamatsu	S3994-01	Detector
Polypropylene conical tube 15 mL	Falcon	14-959-53A
Polypropylene filters	Thermo Scientific	726-2520	pore size 0.2 μm
Sterile petri dishes	Corning	07-202-031
Syringe 10 mL	Fisher brand	14955459
UV/Vis Spectrophotometer	Beckman Coulter	Model: DU 530	Measuring optical density at wavelength of 600 nm
Vortex mixer	VWR	97043-562
Water objective	Olympus	UPLANAPO/IR	60×, NA 1.2