2. Current Antimicrobial Susceptibility Testing (AST) Methods

Fighting the threat of multidrug-resistant pathogens requires a multi-disciplinary approach in which rapid AST plays a critical role. The classical method to determine antibiotic susceptibility is the disk diffusion method ^{[3][9][10][11]}. This well-established method requires a growth period before the actual disk test is performed, which also is based on further growth during 16 to 20 h. Since some pathogenic bacteria are non-culturable, other methods have to be used. Therefore, new methods that also allow one to perform AST on non-culturable microorganisms in a short time frame ^[8] are needed. Current AST methods can be divided into phenotypic and molecular tests ^[12][13][14].

Table 1). Since these methods mostly rely on microbial culturing and thus replication, the performance of these tests takes 1 to 3 days ^[3][4]. Agar dilution assays, i.e., disk diffusion and E-test methods, are flexible and simple methods that are commonly used in clinical microbiology labs (

Table 1). They allow one to determine the minimal inhibitory concentration (MIC). A MIC test can also be used using broth dilution assays, where the MIC corresponds to the lowest concentration of antibiotic that completely inhibits bacterial growth and lacks visible turbidity ^[16]. Broth macrodilution assays have been miniaturized and automated ^[3]. Several commercial semi-automated or fully automated instruments have been developed, such as the MicroScan WalkAway, Vitek-2, BD Phoenix, Wider System and Sensititre system ^{[3][4][7][17][18][19][20][21][22][23][24][25][26][27]}. The time–kill test is a tool for obtaining information on the dynamic interaction between the antimicrobial and the microbial strain ^[14]. The time–kill curve reveals a time- or concentration-dependent antimicrobial effect and can be used to determine synergism or antagonism between drugs in combinations ^{[28][29][30][31][32]}. Optical-based AST methods have been developed to measure the growth rate, such as the “multiplexed automated digital microscopy (MADM)” method ^[33][34][35] and the oCelloscope ^[36], as well as to measure morphological changes of single cells upon antibiotic treatment ^[37] (

Table 1

Table 1

^{[57][58][59][60][61][62][63][64]}. The use of MALDI-TOF MS for AST lies in the combination of MALDI TOF MS identification with an established AST method, such as the combination with Vitek-2 ^[65] or the BD Phoenix system ^[66][67]. MALDI-TOF MS has also been combined with stable isotope labeling by amino acid in cell culture (SILAC). This MS method can identify the metabolically inactive microorganisms due to the action of the antibiotic ^[68]. ATP bioluminence assays can provide a fast antibacterial ^[69][70][71], antimycobacterial ^[72][73] and antifungal testing ^[74][75] where the growth is determined based on the ATP quantification. Another molecular marker for growth that has been used is 16S rRNA ^[76].

Molecular techniques rely on the determination of a particular fingerprint associated with the resistance to a specific antibiotic ^[15][48][49] (

Table 2). Real-time PCR techniques and specifically constructed DNA microarrays have been developed to detect a spectrum of genes that could be related to resistance to different antibiotics ^[15][48][50]. Some of these techniques (e.g., the Xpert MTB/RIF assay ^[51][52][53]) have been commercialized and are characterized by a very high reliability and speed of execution ^[13]. In the last 10 years, various methods have been developed that are based on matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) ^[54]. MALDI-TOF MS allows for the fast identification of the microbial species ^[55][56]

In this study, we will essentially focus on a novel way to characterize the susceptibility of microorganisms to antibiotics. The technique relies on the detection of the nanometric scale oscillations that characterize living cells. Several years ago, our team demonstrated that all living organisms oscillate at a nanometric scale and that these oscillations end as soon the organism dies ^[77]. Highlighting such minute movements on a single microorganism requires highly sensitive devices such as atomic force microscopes (AFMs). These instruments are particularly adapted to such challenges, since they can detect displacements in the range of 0.1 Å with a temporal resolution in the range of microseconds. As an illustration, the typical distance between two carbon atoms in an organic molecule is about 2 Å. The very first and straightforward application of such a life monitor is rapid AST. The aim of this article is to describe the working principle of these novel devices, to review their contributions to the field of AST and to discuss their future applications.

3. AFM Nanomotion Setup and Measurement

The technique is relatively simple to set up. A detailed procedure describing the preparation measurement and the data processing steps can be found in Venturelli et al. ^[78]. Briefly, the first step consists of functionalizing a relatively soft (0.06 N/m) AFM cantilever with a cross linking molecule such as glutaraldehyde, paraformaldehyde, APTES ((3-aminopropyl)triethoxysilane) or fibronectin. To ensure a stronger binding, we recommend a suspension of the microorganism in a phosphate-buffered saline (PBS) solution first. Cell membrane parts, various peptides or amino acids present in traditional culture media can hide the attachment spots on the cross-linking molecules. To ensure the attachment, the cantilever is immersed in a droplet containing the bacteria for about 15 min. The sensor is eventually inserted in the analysis chamber of the AFM to start the measurement. Biologically oriented instruments are preferable since they are designed to operate in liquids and permit one to exchange the “imaging” medium during the measurement. Custom built devices such the one depicted in 

The technique is relatively simple to set up. A detailed procedure describing the preparation measurement and the data processing steps can be found in Venturelli et al. [150]. Briefly, the first step consists of functionalizing a relatively soft (0.06 N/m) AFM cantilever with a cross linking molecule such as glutaraldehyde, paraformaldehyde, APTES ((3-aminopropyl)triethoxysilane) or fibronectin. To ensure a stronger binding, we recommend a suspension of the microorganism in a phosphate-buffered saline (PBS) solution first. Cell membrane parts, various peptides or amino acids present in traditional culture media can hide the attachment spots on the cross-linking molecules. To ensure the attachment, the cantilever is immersed in a droplet containing the bacteria for about 15 min. The sensor is eventually inserted in the analysis chamber of the AFM to start the measurement. Biologically oriented instruments are preferable since they are designed to operate in liquids and permit one to exchange the “imaging” medium during the measurement. Custom built devices such the one depicted in 

Figure 1 can also be used.

3 can also be used.

Figure 13.

 Dedicated nanomotion detection apparatus. 1. analysis chamber with the AFM cantilever, 2. inlet, 3. outlet, 4. laser inlet, 5. prisms to orient the laser beam onto the cantilever and the photodiodes, 6. photodiodes.

The analysis chamber is then filled with the culture medium and the laser beam as well as the two- or four-segment photodiodes are adjusted to achieve the highest possible sensitivity. The measurement usually starts 5–10 min after the insertion of the cantilever into the analysis chamber. This delay permits the liquid medium and the cantilever to reach a thermal steady state. Typical measurements are carried out with a sampling rate of about 20 kHz. Such a high rate is preferred, since it permits one to capture the resonant frequency of the cantilever and to assess the correct position of the laser beam. The oscillations of the lever are recorded for about 15 min in the culture medium before the addition of the antibiotic. Usually, after 10–15 min exposure of bacteria to the drug, the oscillation amplitude drops if the bacteria are sensitive and remains stable or even increases if they are resistant. The experiment can be stopped at this stage; however, we usually inject an additional killing agent such as glutaraldehyde or paraformaldehyde to ensure that the oscillation amplitude drops to zero once all the organisms present on the cantilever are dead.

4. AFM Nanomotion Data Processing

The data processing step consists of a high pass filtering of the original data set to get rid of the thermal drift of the cantilever. The resulting data are eventually processed to extract the variance in a temporal window of 10 s. The variance of the signal is up to now the most sensitive parameter we found to distinguish between living and dead cells. A trial-and-error process in which we attempted to maximize the difference between signals recorded on living and dead samples determined the size of this 10 s window. It is important to mention that the amplitude of the variance signal directly correlates to the nutrient concentrations in the analysis chamber. This observation significantly extends the technique application domains.

Interestingly, frequency domain analysis did not reveal up to now any preferential peak (i.e., frequency) that we could attribute to the specific bacterial species or a metabolic state. However, we noticed that on the fast Fourier transforms (FFT) of the signal, the largest difference between living and dead cells is located between 0.2 and 100 Hz. This frequency window is very stable among all the living organisms that we, and other groups, explored up to now ^[79].

Interestingly, frequency domain analysis did not reveal up to now any preferential peak (i.e., frequency) that we could attribute to the specific bacterial species or a metabolic state. However, we noticed that on the fast Fourier transforms (FFT) of the signal, the largest difference between living and dead cells is located between 0.2 and 100 Hz. This frequency window is very stable among all the living organisms that we, and other groups, explored up to now [142].