Real-time monitoring of live mycobacteria with a microfluidic acoustic-Raman platform.

Tuberculosis (TB) remains a leading cause of death worldwide. Lipid rich, phenotypically antibiotic tolerant, bacteria are more resistant to antibiotics and may be responsible for relapse and the need for long-term TB treatment. We present a microfluidic system that acoustically traps live mycobacteria, M. smegmatis, a model organism for M. tuberculosis. We then perform optical analysis in the form of wavelength modulated Raman spectroscopy (WMRS) on the trapped M. smegmatis for up to eight hours, and also in the presence of isoniazid (INH). The Raman fingerprints of M. smegmatis exposed to INH change substantially in comparison to the unstressed condition. Our work provides a real-time assessment of the impact of INH on the increase of lipids in these mycobacteria, which could render the cells more tolerant to antibiotics. This microfluidic platform may be used to study any microorganism and to dynamically monitor its response to different conditions and stimuli.

x-axis represents the time in hours and the y-axis displays the normalized peak intensity. Normalized peak intensity  The experimental parameters were first optimised to obtain a stable trap, a good Raman signal compared to the noise level and also a controlled temperature. A plastic cover was placed on top of the system and permits to control the temperature inside. The temperature had to be close to 37°C for the experiments as it is the optimal growth temperature for M. smegmatis. First, we investigated the impact of the amplitude of the acoustic trapping on the temperature. The temperature was set at 35.5°C inside the system and measured by sensors located on top of the chamber. Supplementary The bacterial concentration also had to be high as more bacteria facilitate and reinforced the trapping. Stable trap was achieved when using 4 mL of 7-day-old M. smegmatis re-suspended, after centrifugation and removal of the supernatants, in 500 µl of medium (8 times concentrated).
The bacteria were first trapped using 7 Vpp for 2 to 3 minutes, generating large aggregates of bacteria then the amplitude was reduced to 3 Vpp to maintain the trap throughout the experiment and the Raman measures. This helped creating a stable trap as once the bacteria were aggregated a lower trapping force was needed to maintain the trap.
The laser power had to be optimised to not disturb the trapping. With a laser power of 100 mW targeting trapped cells as described above we observed that the laser force was not breaking the trap and permitted to acquire wavelength modulated Raman (WMR) spectroscopy measures.
The acquisition time was optimised using those conditions to obtain good quality Raman spectra S12 with good signal to noise ratio. 50 seconds per spectrum of acquisition time was validated and used for all the experiments presented. The laser itself induced a small temperature increase of about 0.5°C.

S3.2 Bacterial concentration in the experiments
The bacterial concentration, in the chamber, for all the experiment is show in Supplementary Table 1 and ranged from 3.1×10 8 CFU· ml −1 to 1.0×10 9 CFU· ml −1 . The CFU· ml −1 were calculated using the 7-day-old cultures. Then 4 ml were spun down at 20,000 ×g for 3 minutes, the supernatants were discarded, and the pellets were then re-suspended in 500 µl of medium leading to 8 times concentrated bacterial suspensions.

S3.3 Temperature and laser power during the experiments
The laser power was controlled throughout all the experiments to ensure that the power remain close the initial value set at T = 0h always close to 100 mW. 108.63 ± 8.07 (1 SD) mW and the temperature at 36.14 ± 0.69 (1 SD)°C. S13

S3.4 Comparison of all biological replicates averaged Raman spectra
In the no-stress experiments a suspension of 7-day-old M. smegmatis was re-suspended in fresh 7H9 broth and trapped acoustically in the chamber. In the INH-stress experiments a suspension of 7-day-old M. smegmatis was re-suspended in fresh 7H9 broth with isoniazid (INH) at 32 µg·ml −1 and acoustically trapped in the same chamber. The bacteria were then measured using WMR spectroscopy for up to 8 hours during which both the temperature and the laser power were controlled every hour. The analysis of spectra focused in the fingerprint region between 600 cm −1 and 1800 −1 . Averaged WMR spectra acquired from three replicates are shown in Supplementary Fig. 2

S3.5 Evolution of Raman peaks over time in no-stress and INH-stress conditions
Raman peaks at 635 cm −1 and 1606 cm −1 . Supplementary Fig. 3 presents the evolution over time of the intensity of the Raman peaks located at 635 cm −1 and 1606 cm −1 . The increase in those Raman peaks was associated with tyrosine (See Table 1). In all 6 experiments we observe an increase in intensity in those two Raman peaks. However, in the INH-stress condition the increase is less strong and delayed in time. In the second INH-stress experiment, ( Supplementary Fig. 3 (e)) the intensity of the Raman peaks stabilises after 5 hours of measurement. In the third INH-stress experiment, we noted that in Supplementary Fig. 3 (f) the intensity of the Raman peaks at 635 cm −1 is not showing a matching pattern with 1606 cm −1 which was the case in all other experiments.
Raman peaks at 1040 cm −1 and 1130 cm −1 . Supplementary Fig. 4 presents the evolution over time of the intensity of the Raman peaks at 1040 cm −1 and 1130 cm −1 . The increase in the these S14 two Raman peaks were mainly associated with carbohydrate; it is also possible that 1130 cm −1 shows some information related to lipids in the INH-stress condition (See Table 1). In all six experiments we observe an increase in both Raman peaks. However, in the INH-stress condition the increase is less strong and delayed. In the second INH-stress experiment, ( Supplementary   Fig. 4 (e)) the intensity of the Raman peaks reduces after five hours of measurement.  Table 1). In the three no-stress experiments we observe that the intensity of those two Raman peaks is rapidly decreasing over time. In the INH-stress experiments the intensity of the Raman peak at 1150 cm −1 and 1523 cm −1 , is also reducing over time. In the second INH-stress experiment, Supplementary Fig. 6 (e), the intensity of the Raman peaks begin to reduce after five hours of measurement.

S15
Raman peak at 1007 cm −1 . Supplementary Fig. 7 presents the evolution over time of the intensity of the Raman peak located at 1007 cm −1 associated with phenyalanine (See Table 1). In no-stress and INH-stress experiments, we observe that the intensity of this Raman peak is relatively stable over time; only in Supplementary Fig. 7 (d) a small increase is observed.
Raman peak at 783 cm −1 . Supplementary Fig. 8 presents the evolution over time of the intensity of the Raman peak at 783 cm −1 associated with nucleic acids (See Table 1). In the no-stress condition the intensity of the Raman peak at 783 cm −1 is slightly increasing over time. In INHstress condition, we observe that the intensity of the Raman peak at 783 cm −1 is relatively stable stock was then further diluted 8x serially and plated onto a Middlebrook 7H11 agar plate for cfu counting in the Miles and Misra fashion. This was to confirm the cfu estimate made previously.

S16
After 3-5 days, or until confluent growth was evident in the positive control wells and no growth was seen in the negative control wells, the 96-well plate was removed from the incubator. The plate was then analysed in a 96-well plate reader at 600nm absorbance. These steps were repeated 4 times to yield a total n number of 12 for each dilution of INH. Supplementary Fig. 9 shows the dose dependant relationship between INH and M. smegmatis. The MIC was found to be between 16 and 32 µg·ml −1 with 32 µg·ml −1 yielding the MIC90 and 16 µg·ml −1 yielding the MIC50.

S3.7 Discontinued measurement
In order to understand if the strong increase, observed in all previous experiments, of the Raman peaks at 635 cm −1 , 1040 cm −1 , 1130 cm −1 and 1606 cm −1 was associated with the trapping itself (either a reaction to the Acoustic force or a reaction to the cell aggregation) an experiment using discontinuous measurement was designed as shown in Supplementary Fig. 10.
The bacterial concentration, in the chamber, was 3×10 8 CFU·ml −1 for the first experiment and 2.8×10 8 CFU·ml −1 for the second. The temperature and laser power are well controlled during the two repeated experiments. No continuous increase of those Raman peaks is observed over time in both experiments (see Supplementary Table 2 and 3), suggesting that the increase in intensity over time observed in the experiment using continuous acoustic trapping is associated with the cell aggregation or is a reaction to the acoustic force. S17