Biosensors and Bioelectronics 142 (2019) 111552

Contents lists available at ScienceDirect

Biosensors and Bioelectronics

journal homepage: www.elsevier.com/locate/bios

Emerging technologies for antibiotic susceptibility testing T

a a,b a a
Bhagaban Behera , Anil Vishnu G.K. , Suman Chatterjee , V.S.N. Sitaramgupta V ,
Niranjana Sreekumara, Apoorva Nagabhushana, Nirmala Rajendranc, Prathik B.H.d,
Hardik J. Pandyaa,∗
a
Biomedical and Electronic (10-6-10-9) Engineering Systems Laboratory, Department of Electronic Systems Engineering, Indian Institute of Science, Bangalore, India
b
Center for BioSystems Science and Engineering, Indian Institute of Science, Bangalore, India
c
IISc Medical Center, Indian Institute of Science, Bangalore, India
d
Indira Gandhi Institute of Child Health, Bangalore, India

A R T I C LE I N FO A B S T R A C T

Keywords: Superbugs such as infectious bacteria pose a great threat to humanity due to an increase in bacterial mortality
Bacteria leading to clinical treatment failure, lengthy hospital stay, intravenous therapy and accretion of bacteraemia.
Antibiotics These disease-causing bacteria gain resistance to drugs over time which further complicates the treatment.
AST Monitoring of antibiotic resistance is therefore necessary so that bacterial infectious diseases can be diagnosed
Emerging technologies
rapidly. Antimicrobial susceptibility testing (AST) provides valuable information on the efficacy of antibiotic
Microfabrication
agents and their dosages for treatment against bacterial infections. In clinical laboratories, most widely used AST
methods are disk diffusion, gradient diffusion, broth dilution, or commercially available semi-automated sys-
tems. Though these methods are cost-effective and accurate, they are time-consuming, labour-intensive, and
require skilled manpower. Recently much attention has been on developing rapid AST techniques to avoid
misuse of antibiotics and provide effective treatment. In this review, we have discussed emerging engineering
AST techniques with special emphasis on phenotypic AST. These techniques include fluorescence imaging along
with computational image processing, surface plasmon resonance, Raman spectra, and laser tweezer as well as
micro/nanotechnology-based device such as microfluidics, microdroplets, and microchamber. The mechanical
and electrical behaviour of single bacterial cell and bacterial suspension for the study of AST is also discussed.

1. Introduction global bacterial infections. There has been a significant rise in re-
sistance to most commercially accepted antibiotics. A detailed under-
Antibiotic resistance by pathogenic bacterial strains is a major standing of bacterial resistance mechanisms will help for developing
prevalent health crisis in the world today. Antibiotics are drugs that can novel antibiotic testing platforms for rapid treatment of bacterial in-
effectively kill (bactericidal antibiotics) or inhibit (bacteriostatic anti- fections (Heesemann, 1993). It is known that the bacteria acquire re-
biotics) the growth of bacteria causing infections in both humans and sistance to antibiotics by producing extended spectrum β-lactamase
animals (Nemeth et al., 2015). Alexander Fleming, a Scottish physician, enzyme (Soltani et al., 2014). The β-lactamase gene inactivates β-
is widely credited for the discovery of the world's first antibiotic (Pe- lactam antibiotics like penicillin and cephalosporins by hydrolyzing the
nicillin G) from fungus Penicillium notatum in 1928. This discovery has β-lactam ring that plays a crucial role in inhibiting bacterial cell wall
saved the lives of millions of people across the globe from bacterial biosynthesis (Saravanan et al., 2018). Bacteria can mask the target drug
infections (Tan and Tatsumura, 2015). β-lactam antibiotics, a class of and reprocess DNA mechanisms, develop mutations, and express al-
broad-spectrum antibiotics including I, II and III generation cephalos- ternative proteins. They can also protect their target sites from anti-
porins, monobactams, carbapenems, β-lactamase inhibitors, and peni- biotic action by modifying them. Certain enzymes produced by bacteria
cillin are widely used in treating bacterial infections (Kong et al., 2010). are capable of altering the antibiotic molecule by inducing chemical
Cephalosporins are widely used as intravenous antibiotics and fluor- changes and thereby, destroying the molecule. In the case of gram-ne-
oquinolones are major oral antibacterial drugs used in the treatment of gative bacteria, drug uptake is prevented from permeating the inner
bacterial infections. Antibiotics are a widely used medication in treating cytoplasmic membrane (Munita and Arias, 2016). Antibiotic resistance

∗
Corresponding author.
E-mail address: hjpandya@iisc.ac.in (H.J. Pandya).

https://doi.org/10.1016/j.bios.2019.111552
Received 22 April 2019; Received in revised form 27 July 2019; Accepted 29 July 2019
Available online 09 August 2019
0956-5663/ © 2019 Elsevier B.V. All rights reserved.
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

can also be acquired by either horizontal transfer of resistant genes by grow around the antibiotic disk or kills the bacteria around the disk
conjugation, transformation, and transduction or vertical transfer by forming a clear, visible ring around the disk. The diameter of the
mutation among different bacterial strains (Read and Woods, 2014). In growth area is a direct measure of the susceptibility of the bacteria to
some cases, antibiotics are expelled out from the cells before it can the antibiotic. The advantages of disk diffusion process are its simpli-
reach the target location. city, cost-effectiveness, and easy interpretation. However, the main
The life-threatening condition of a patient can be avoided by iden- disadvantage of this process is that it needs manual processing and
tifying correct bacteria causing the condition and the most effective lacks automation. Further, this process provides only qualitative and
antibiotic against this bacterium considering its resistance against an- not quantitative results. Additionally, disk diffusion has been employed
tibiotics. Biosensors being highly sensitive, specific, repeatable, rapid, in different ways to study AST. Khatoon et al. used silver nanoparticles
label-free, inexpensive, and compact are a preferred choice in devel- (AgNPs) in place of conventional drugs to study AST using disk diffu-
oping point-of-care platforms for AST. A biosensor detects bacteria sion (Khatoon et al., 2017). Similarly, the antibacterial activity of al-
through a two-step process viz. i) biological recognition element to ginate films incorporated with graphene oxide was studied using disk
recognize pathogen of interest which may include antibodies, enzymes, diffusion method (Frígols et al., 2019). AST using agar and broth di-
whole cells, nucleic acids, receptors, aptamers, peptides or lipids, and lution is one of the oldest methods used to determine the minimum
ii) transducers to convert the biological signal generated by recognition inhibitory concentration (MIC). In the agar dilution method, a known
element into a measurable electrical signal (Bhalla et al., 2016; number of bacterial cells are marked as a spot on an agar plate con-
Mehrotra, 2016; Turner, 2013). Currently, microbiological techniques taining different concentrations of antibiotic (Balouiri et al., 2016a;
like bioassays, screening methods including disk diffusion, anti- Reller et al., 2009c). After incubation, the spots are examined for
microbial gradient diffusion, broth dilution, and cell culture are some of bacterial growth. The agar dilution method is capable of testing one
the common methods adopted by health care facilities for measuring antibiotic at a time whereas more than one antibiotic can be tested at
the effectiveness of antibiotics (Puttaswamy et al., 2018; Reller et al., once in broth dilution. In the broth dilution method, antibiotics are
2009a). The rapid detection and identification of pathogenic bacteria diluted two folds in a liquid growth medium and incubated after in-
that are typically present in low concentrations in patient samples is the oculation with a standard bacterial suspension. Post incubation samples
limitation of these methods. Such typical bacterial growth and sus- are examined to find out MIC. This method is also used to measure
ceptibility testing require more than 24 h to obtain results. In the cur- time-kill i.e. rate of bacterial death for different concentration of drugs
rent scenario, accurate, evidence-based, and targeted antibiotic pre- over a period of time up to 24 h (Gupta et al., 2015). The advantages of
scribing by clinicians are problematic and hence they are forced to these dilution processes are reproducibility and cost-effectiveness.
prescribe broad-spectrum antibiotics. Thus, there is a pressing need for However, it is labour-intensive and expensive in nature. A nearly si-
highly sensitive, specific, repeatable, rapid, label-free, inexpensive, and milar technique called microbroth dilution is now employed to reduce
compact point-of-care platforms for antimicrobial susceptibility testing reagents and processing time. The antimicrobial gradient method also
(AST). AST profiling are performed in two ways: genotypic and phe- called E-test combines the principles of both dilution and diffusion
notypic based methods. Genotypic methods, though rapid and accurate, methods to determine the minimum inhibitory concentration (MIC) of
are however limited by their cost, labour intensiveness, and difficulty in the drug (Belkum and Dunne, 2013; Citron et al., 1991; Puttaswamy
translating into point-of-care platforms in many cases (Nseir and Povoa, et al., 2018; Reller et al., 2009b, 2009a). This method provides direct
2015; Schofield, 2012). Phenotypic methods, through their funda- quantification of AST by dilution and diffusion of antibiotic in an agar
mental interactions with the physical environment, provide more ave- plate. The process relies on creating a predefined, continuous con-
nues for emerging engineering technologies to be applied to them to centration gradient of the antibiotic to be tested in an agar medium to
improve their efficiency and versatility. In this review, we provide an determine susceptibility. In this procedure, an antibiotic-soaked strip is
overview of different approaches for phenotypic AST developed over taken with an increasing concentration gradient from one end to the
the years after a brief discussion on conventional AST techniques. We other and is deposited on the agar surface. The intersection of the
have primarily focused on devices/tools such as, microcalorimetry, growth inhibition ellipse and the strip is determined as the MIC value of
flow cytometry, microfluidics, microcantilever technology, high-fre- the drug (Balouiri et al., 2016a; Reller et al., 2009c). This process is
quency electromagnet sensor, pH sensor, mass spectrometry, capaci- simple to implement and thus, is regularly used. However, E-test is not
tance sensor, nuclear magnetic resonance, microsound detection, cost effective for large-scale AST measurement of multiple drugs. A
Raman spectroscopy, semiconductor quantum well, fluorescence de- good correlation between the MIC values obtained through broth di-
tection, and impedance methods to measure AST profile in terms of lution or agar dilution method and E-test has been reported (Baker
simplicity, cost-effectiveness, and low processing time. The key tech- et al., 1991; Berghaus et al., 2015; Gupta et al., 2015). E-test can also be
niques discussed have been graphically portrayed in Fig. 1 and their used to investigate the antibacterial effect when multiple drugs are used
highlights have been summarized in Table 1. The publication statistics (White et al., 1996). To profile AST of two antibiotics on a sample si-
for articles on antibiotic or antimicrobial susceptibility testing is shown multaneously, an E-test strip soaked with the first antibiotic is placed on
in Electronic Supplementary Information (ESI) along with a figure de- an agar plate coated with the bacteria under study. Another strip
picting past, present and future trends in AST. soaked with the next antibiotic is used to replace the first strip after 1 h.
In the commercial version of E-test, 5–6 strips can be placed radially on
2. Classical and commercial AST the surface of an agar plate coated with bacteria. After overnight in-
cubation, a growth inhibition area, in the shape of an ellipse, at the
Conventional ASTs are based on bacterial growth with and without immediate vicinity of the test strip can be observed and MIC can be
antibiotics on a solid agar plate or in a liquid growth medium. The test determined from this observation (Jorgensen and Ferraro, 2009). E-test
is carried out on agar plate by identifying inhibition zones after culture is also used to test critical samples like blood and cerebrospinal fluid
whereas, in liquid-based methods, change in optical density is mea- (CSF) with both gram-positive and gram-negative bacteria. The MIC
sured. Several successful results have been achieved using this method value in μg/mL range can be observed at the ellipse edge that intersects
while testing on different bacteria and antibiotics (Reller et al., 2009b, the strip. The advantages of this method are easy test procedures, able
2009a; Syal et al., 2017a). In the disk diffusion method, a disk im- to determine contamination, and ability to test fastidious organisms
pregnated with antibiotic drug is placed on an agar plate which is which are not reliable in disk diffusion methods. However, E-test is
coated with the test bacterium (Reller et al., 2009c). Then the plate is relatively expensive compared to disk diffusion and dilution methods.
incubated overnight to allow the antibiotic to diffuse from the filter In comparison, liquid suspension-based methods are faster than those
paper into the agar. If an antibiotic is effective, it stops the bacteria to based on solid-state medium owing to increased growth rate

2
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

Fig. 1. Highlights of different techniques for AST applications.

capabilities. But liquid suspension-based methods are more laborious 3. Engineering technologies for AST
than solid state ones. To address this problem, several instruments like
Phoenix (made by Becton-Dickinson, Franklin Lakes, NJ), Microscan 3.1. Flow cytometry AST
(made by Beckman Coulter, Brea, CA), Sensititre (made by Thermo
Scientific Waltham, Massachusetts, USA) and Vitek-1/Vitek-2 (made by Flow cytometry is a technique where optical and fluorescence
bioMérieux, Marcy-L'Etoile, France) have been developed for high characteristics are employed in cell counting, cell sorting, chromosome
throughput AST. BD Phoenix is based on turbidity reading and colori- preparation, and biomarker detection by suspending them in a stream
metric change. This system can use up to 99 panels (cassettes) and of fluid and passing them in front of a detector (Adan et al., 2017). Cell
provide results in ∼6–16 h. Similarly, Microscan uses 96-well micro- morphology and cell counts are measured by considering light scat-
dilution tray and is capable of handling more than 40 trays at a time. In tering and excitation/emission spectra from different fluorescent ma-
this system, AST is examined by recording bacterial turbidity using a terials such as FRET dyes, fluorophores or fluorescent proteins. In the
specialized photodetector and typical process time is ∼4.5–18 h. Sen- context of AST it is known that antibiotics induce changes in different
sititre uses 96-well plates and is capable of handling 64 plates. In this morpho-functional and physiological characteristics like membrane
instrument, bacterial growth is monitored by measuring fluorescence potential, cell size, cytoplasmic membrane integrity, and amount of
intensity and the typical process time is over 18 h. The Vitek AST in- DNA (Adan et al., 2017; Chaturvedi et al., 2004; Jepras et al., 1997;
strument uses 64 well plate and can handle up to 240 plates by pro- Tracy et al., 2010). This is leveraged in the methods that use flow cy-
viding results over 4 h. The major limiting factor of these commercial tometry to perform AST. This enables faster processing as compared to
systems is the need to culture large concentrations of bacteria for a conventional methods as changes in physiological parameters caused
sufficiently long time to faithfully detect antibiotic effects on bacterial by antibiotics is faster than growth inhibition processes (1–2 h vs
growth. 16–24 h). A comparative study of AST using flow cytometry-based and
However, these systems strictly follow the Clinical and Laboratory conventional techniques by Gauthier et al. showed 93.9% accuracy
Standards Institute (CLSI) and European Committee on Antimicrobial (Gauthier et al., 2002). This method has been applied to several bac-
Susceptibility Testing (EUCAST) guidelines for antimicrobial suscept- terial species and combinations of antibiotics by using various dye/
ibility testing which comes across as a big advantage. Though these fixation combinations (Chaturvedi et al., 2004; Faria-Ramos et al.,
instruments are compatible with a wide spectrum of bacteria and an- 2013; Gauthier et al., 2002; Huang et al., 2015; Jepras et al., 1997).
tibiotics, they cannot handle polymicrobial bacteria and their corre- Although flow cytometry can produce AST results within few hours, it
sponding antibiotics. The turbidity method has the uncertainty that it has not been used significantly due to inefficiency for complex patient
assumes bacterial growth and absorbance plot as linear. Also, these samples, staining inefficiency of dyes, autofluorescence of certain bac-
instruments are prohibitively expensive to use in a low-resource setting. terial creating noise, and unable to differentiate the effect of bacter-
icidal or bacteriostatic antibiotics. Additionally, verification and vali-
dation of the clinical database require an enormous amount of work to

3
 Table 1
Summary of conventional and emerging engineering AST technologies.
Methods Antimicrobial testing Test principle Automatic/ Test Bacteria tested Cost MIC Features Ref.
B. Behera, et al.

technology Manual time (h) detected

Traditional Disk diffusion Filter paper containing antibiotic Manual 16–24 Almost all kind of Low Yes (not Simple interpretation, approved (Bauer et al., 1966, 1959; Reller
of desired concentrations are bacteria and yeast consistent) by CLSI, labor-intensive, high et al., 2009a)
placed on agar surface which is detection limit, and MIC
coated with bacteria determination is not consistent
Agar dilution Different concentrations of Manual 16–24 Almost all kind of Low Yes Approved by CLSI, and simple (Baker et al., 1991; Balouiri et al.,
antibiotics added to agar plates bacteria interpretation, labor-intensive, 2016a, 2016b; Berghaus et al., 2015;
and bacteria coated on the slow and high detection limit Gupta et al., 2015)
surface
Broth dilution or Broth Different concentrations of Manual 16–24 Almost all kind of Low Yes Appropriately determines MIC, (Baker et al., 1991; Balouiri et al.,
microdilution antibiotics are added to bacterial bacteria approved by CLSI, labor-intensive 2016b; Berghaus et al., 2015;
culture media and requirement of high reagent Jorgensen and Ferraro, 2009; Reller
volume et al., 2009a)
E-test A strip soaked with gradient Manual 12–20 Almost all kind of Medium Yes MIC determination, easy (Baker et al., 1991; Berghaus et al.,
concentrations of antibiotic kept bacteria interpretation, approved by CLSI, 2015; Mittman et al., 2009; White
on agar surface which is coated labor-intensive, expensive as et al., 1996)
with bacteria compared to other classical
techniques
Commercial Flow cytometry Cells viability are counted with Semi- 2–6 Almost all kind of High Yes semi-Automated, and relatively (Faria-Ramos et al., 2013; Gauthier
instruments the help of dyes automatic bacteria simple, low detection limit, and et al., 2002; Huang et al., 2015;
expensive Ramani et al., 1997; Saint-Ruf et al.,
2016)
Isothermal Measures heat flowrate (which is Automatic ∼24 Almost all kind of Medium- Yes Safe to operate even to high-risk (Bonkat et al., 2013; Braissant et al.,
microcalorimetry proportional to physical or bacteria High bacteria, time-consuming and 2009; Howell et al., 2012; Lewis and
chemical change) of bacterial can't determine the exact nature Daniels, 2003; Nawattanapaiboon

4
suspension in the presence or of physical or chemical change et al., 2015; Tafin et al., 2012; von
absence of antibiotics Ah et al., 2009)
Sensititre Fluorescence measurement of Automatic 18–24 Almost all kind of Medium- Yes Sensitive, relatively inexpensive, (Alexander et al., 2007; Cuenca-
enzymatic activities of bacterial bacteria and yeast High approved by CLSI, time- Estrella et al., 2010; Espinel-Ingroff
suspension in the presence of consuming et al., 1999)
antibiotics
Vitek-1/Vitek-2 Measure bacterial growth in the Automatic 6–24 Almost all kind of Medium- Yes Sensitive, relatively inexpensive, (Cuenca-Estrella et al., 2010; Eigner
presence of antibiotics by bacteria High and approved by CLSI, time- et al., 2005; Jiang et al., 2016;
recording light attenuation using consuming and not suitable for Mittman et al., 2009)
a photometer POC applications
BD Phoenix Recording of bacterial Automatic 4–16 Almost all kind of Medium- Yes Sensitive, and relatively (Eigner et al., 2005; Mittman et al.,
suspension turbidity and bacteria High inexpensive, approved by CLSI, 2009; Snyder et al., 2008)
colorimetric changes in the time-consuming
presence of antibiotics
Mechanical Asynchronous Measurements of magnetic beads Semi- 1–6 E. coli High Yes Highly sensitive, single bacteria (Kinnunen et al., 2012, 2011; Sinn
magnetic bead rotational frequency automatic observation, not tested with other et al., 2012, 2011)
rotation sensor bacteria, and system is complex
Cantilever Measurement of cantilever Semi- ∼2 E. coli, S. aureus High Yes Highly sensitive, complex (Etayash et al., 2016; Gfeller et al.,
fluctuations by placing bacteria automatic fabrication process and complex 2005; Lissandrello et al., 2014; Longo
cells on it measurement et al., 2013)
Crystal resonator Measurement of change in Semi- 15 min- E. coli, E. faecalis High No Highly sensitive, and complex (Cermak et al., 2016; France et al.,
resonant frequency automatic 2h measurement 2016; Johnson et al., 2017)
Optical Surface plasmon Measurement of refractive index Automatic ∼2 E. coli, High Yes Sensitive, fast, complex (Chiang et al., 2009; Tawil et al.,
resonance S. epidermidis, MRSA, 2013)
MSSA
SERS Measurement of Raman scattered Automatic ∼2 E. coli, K. pneu, S. sapro, High Yes Sensitive, fast, moderately (Chabot et al., 2013; Galvan and Yu,
light signals with the help of E. faecalis complex, expensive, Complex 2018; Liu et al., 2016, 2009;
metallic nanoparticles algorithm needed Martinez-Perdiguero et al., 2013;
Premasiri et al., 2017, 2017)
(continued on next page)
Biosensors and Bioelectronics 142 (2019) 111552
 Table 1 (continued)

Methods Antimicrobial testing Test principle Automatic/ Test Bacteria tested Cost MIC Features Ref.
technology Manual time (h) detected
B. Behera, et al.

Laser tweezers Raman spectra study of optically Automatic ∼4 E. coli High Yes Sensitive, Fast, characterize single (Moritz et al., 2010a; Pilát et al.,
trapped bacterial cell's surface cell, moderately complex, 2018; Samadi et al., 2015)
Expensive, Complex algorithm
needed
Phase shift Optical (reflectance) study of Si Automatic 2–3 E. coli High Yes Sensitive, fast, real-time, Leonard et al. (2017)
Spectroscopy micropillar topologies and complex, expensive, complex data
colonized bacteria in suspension accusation
Electrical Impedance Impedance measurement of Semi- 0.5–1.5 E. coli, MRSA Medium Yes Simple, rapid, real-time, (Puttaswamy et al., 2012; Safavieh
bacterial suspension automatic inexpensive, miniaturized et al., 2017)
Capacitance Capacitance measurement of Manual 2.5 E. coli, S. thyphi, P. Medium Yes Simple, real-time, inexpensive (Jo et al., 2018; Niyomdecha et al.,
bacterial suspension aeruginosa, S. epidermidis, miniaturized 2017)
S. aureus, B. subtilis
Electrochemical Measurement of change in Manual 1 E. coli Low- Yes Rapid, Inexpensive, Moderately (Ertl et al., 2000; Mann and
current due to electrochemical Medium complex, chemical wastage Mikkelsen, 2008; Onishi et al., 2018)
reactions
Engineering Gradient MFD Concentration gradient formed Semi- 2.5–4 P. aeruginosa, E. coli, S. High Yes High throughput, multiple (Derzsi et al., 2016; Hou et al.,
approaches on the chip by various design automatic aureus, Salmonella antibiotics with different 2014b; Kim et al., 2015; Li et al.,
typhimurium concentration can be tested in one 2014)
shot. complex design
pH Sensor Measurement of effective optical Semi- 2 E. coli High Yes Rapid, no long-term pre- Tang et al. (2013)
thickness of porous silicon automatic incubation required, small sample
embedded pH sensitive material volume, expensive equipment
requirement,
Microfluidic agarose Bacteria immobilized in agarose Semi- 4–10 206 bacteria tested by Medium Yes Small sample volume, Real-time, (Choi et al., 2017, 2016, 2013)
channel and monitored using microscope automatic (Choi et al., 2017) relatively slow, inexpensive

5
equipment requirement,
Droplet microfluidics Monitoring of growth of bacterial Automatic 3 Different bacteria Medium Yes Rapid, automated, Single bacteria (Chung et al., 2016; Jiang et al.,
by confining them in a nanolitre can be studied, small sample 2016; Kaushik et al., 2017;
volume volume, high throughput, Sabhachandani et al., 2017; Sinn
multiplexing possible, expensive et al., 2011)
equipment requirement,
complicated chip design
Dielectrophoresis Monitoring dielectrophoretic Semi- 1.5–4 P. aeruginosa, E. coli, K. Medium Yes Fast, high throughput, (Braff et al., 2013; Chung et al.,
induced AST behaviour of elongated bacteria automatic pneumoniae multiplexing possible 2012, 2011; Peitz and van Leeuwen,
2010; Su et al., 2017)
Stress-Induced AST High flow rates on-chip stress Semi- 1–2 S. aureus, E. cloacae, E. High Yes Effective, rapid, high throughput, (Kalashnikov et al., 2018, 2017,
immobilized bacteria in the automatic coli, K. pneumoniae, P. expensive, chances of leakage 2014, 2012)
presence and absence of aeruginosa during testing, high sample
antibiotic; number of dead volume required
bacteria tracked via microscopy
Biosensors and Bioelectronics 142 (2019) 111552
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

be certified by CLSI (van Belkum and Dunne, 2013). their characterization in the presence of antibiotics (Galvan and Yu,
2018). Liu et al. have demonstrated SERS based AST of E. coli and S.
3.2. Isothermal microcalorimetry (IMC) aureus using an array of Ag-nanoparticles imbedded in nanochannels of
anodic aluminium oxide (Liu et al., 2009). In this case, the character-
Isothermal microcalorimetry (IMC) enables heat flow measurement, istic peaks of S. aureus and E. coli cell wall at 730 cm−1 and 1330 cm−1
which is directly related to physical or chemical processes (based on the respectively were studied in the presence of lysostaphin and lysozyme.
metabolic activity of growing bacteria). In IMC, a microorganism con- The destruction of cell walls due to these antibiotics shifts the positions
taining glass ampoule is placed in a measuring channel and heat flow is of these peaks significantly. The same team has extended this work to
continuously measured. These continuous signals of heat flow are quantify MIC of oxacillin and imipenem when MSSA and wild-type E.
evaluated and calibrated with the amount of heat being produced in the coli were treated with these antibiotics respectively by measuring the
ampule. IMC experiment is designed in such a way that observed heat intensities of specific biomarker peaks (Liu et al., 2016). The char-
flow is directly related to the process of interest by considering exo- acteristic peak intensities at 730 cm−1 of S. aureus and 654 cm−1 of E.
thermic and endothermic reactions in the ampoule (Lewis and Daniels, coli without (taken as reference) and with antibiotics were calibrated to
2003). For example, growth of Escherichia coli (E. coli) in M9 medium find out MIC. In 2016, Hadjigeorgiou et al. performed AST of Klebsiella
containing glucose and lactose as carbon sources induces metabolic pneumoniae, 16 Proteus, and E. coli using SERS followed by the machine
activities such as glucose respiration, glucose fermentation, and lactose learning method known as Linear Discriminant Analysis (LDA) and
fermentation. Heat production from each metabolic process is calcu- Leave-One-Out Cross Validation (LOOCV) PC transformation
lated and evaluated by measurement data (Braissant et al., 2009). In- (Hadjigeorgiou et al., 2016). This method could differentiate between
stead of large volume ampule, researchers have developed chip mi- resistant, intermediate, and sensitive bacteria with 94% correct classi-
crocalorimeters which use only a few micro/nanolitre of biological fication. Similarly, Premasiri et al. studied 2 bacterial strains using
samples (Johannessen et al., 2002; Lee et al., 2009). This chip calori- SERS and applied the post partial least squares-discriminant analysis
meter has a heat power sensor developed using Micro Electro- machine learning technique (Premasiri et al., 2017). This method
mechanical Systems (MEMS) technology where all essential compo- provided resistant and susceptible bacteria with > 95% sensitivity
nents like conductance sensor, heat sink, temperature sensors, heating and > 99% specificity. SERS-based AST is a label-free and non-invasive
resistors, and sample containers are integrated on a single chip (van method.
Herwaarden, 2005). IMC has several advantages such as: (i) testing in
sealed ampules, assuring safety towards high-risk bacteria, (ii) no 3.5. Laser tweezers
manual manipulation of ampoules is needed, and (iii) antimicrobial
activity can be calculated during the lag phase which cuts down on the The study of biological samples using Raman spectra was extended
wait time of measurement. Some of the limitations of this method are: through a method called Laser Tweezers Raman spectroscopy (LTRS).
(i) it requires reference sample at the desired temperature, (ii) AST In LTRS, individual cells are trapped in a laser focal volume using a
measurement is carried out in a closed ampoule where environmental Gaussian laser beam. LTRS provides unique advantages as it allows to
parameters such as oxygen concentration available may change over characterize single cells in terms of bacterial identification and their
time, and (iii) non-specific heat flow may create problems. Overall, IMC metabolic activity. Moritz et al. have developed an inverted microscope
shows fast and quantitative AST performance at an affordable price. having a laser source of wavelength 785 nm for LTRS study of E. coli
(Moritz et al., 2010a). In this study, at a given time an average of five
3.3. Optical density (OD) measurement bacterial cells were captured. The Raman spectra intensities in range
710–1280 cm−1 of the trapped bacterial cells were studied over in-
Optical density measures the concentration of bacteria in a sus- cubation time. Principal component analysis (PCA) was used as a ma-
pension. When a light beam passes through a bacterial suspension, the chine learning technique to differentiate bacteria under normal and
scattered light detected by a spectrophotometer is correlated with antibiotic influenced growth. Recently, Pilát et al. have used an opto-
bacteria density in colony-forming units (CFU). OD measurement of a fluidic platform to manipulate single cell using LTRS after transferring
bacterial suspension with antibiotics added can delineate between re- individual bacteria cell into the microchambers (Pilát et al., 2018). In
sistant and sensitive strains (Friedman et al., 2011). This method pro- this LTRS system, individual cells were placed in micro-chambers with
vides AST results in a few hours. However, it can neither calculate MIC the help of optical tweezer and antibiotics induced morphological
nor it is suitable for low concentrations of bacterial suspension. changes were evaluated using Raman spectra and post PCA computa-
tion. In another study, Samadi et al. used a 1064 nm laser beam to trap
3.4. Raman spectroscopy E. coli and studied the motility of the flagella in the presence of ethyl
alcohol (Samadi et al., 2015). The killing time of E. coli was calculated
Raman spectroscopy is employed to measure molecular vibrations using statistical analysis of power spectral density and their auto-cor-
and thus is a useful technique to study biomolecules. In Raman spec- relation function. LTRS is generally applied in cases where the bacterial
troscopy intensity and wave number of in-elastic scattered light is cell immobilizes upon antibacterial treatment owing to reduced avail-
measured. However, Raman spectra signals from biological samples are ability of cellular energy for flagellar movement in comparison to
weak which limits its application. To work around this shortcoming, normal healthy cells. This difference in motility is detected using this
several research groups have combined Raman spectra and machine technique.
learning approaches (discriminant analysis) to differentiate between
live and dead bacteria (Athamneh et al., 2014; Kastanos et al., 2010; 3.6. Surface Plasmon Resonance (SPR)
Moritz et al., 2010b; Walter et al., 2011). Alternately, the weak Raman
signals are amplified by using metallic nanoparticles (plasmon resonant Surface Plasmon Resonance (SPR) is a physical process in which
particles such as silver (Ag), gold (Au)) through the process called electron attached to the metal surface oscillates (at the metal-dielectric
Surface-Enhanced Raman Spectroscopy (SERS). The efficiency and interface) under the influence of plane-polarized light. Since the oscil-
sensitivity of SERS has also been further improved through the use of lations of electrons are extremely sensitive to the refractive index of the
covalently bounded fragmented antibodies conjugated with the nano- dielectric, the change in the refractive index is measured when bio-
particles(Baniukevic et al., 2013). The use of antibodies in AST im- molecules are attached to the metal surface. SPR biosensors are label-
proves the capture efficiency of bacteria. SERS has been applied in free, ultra-sensitive, and allows for real-time detection (Lee et al., 2015;
several bioanalytical processes including bacterial identification and Martinez-Perdiguero et al., 2013; Nath and Chilkoti, 2004). In most

6
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

known applications, SPR is used to detect and identify cells, bacteria, or free-standing structure made of layers of SiO2, Si3N4, and Au. The
viruses based on affinity (Abdulhalim et al., 2008; Bai et al., 2012; growth of antibody immobilized E. coli on the gold surface of free-
Chabot et al., 2013; Firdous et al., 2018; Piliarik et al., 2009; Yanase standing structures was monitored using extraordinary optical trans-
et al., 2014). In 2009, Chiang et al. used SPR platform to test E. coli with mission (EOT). The bacterial growth in the presence of ampicillin and
ampicillin and S. epidermidis with tetracycline qualitatively (Chiang tetracycline changes the refractive index of free-standing structures
et al., 2009). In this study, one side of a prism was coated with a thin resulting in shift of EOT peak. Though SPR based AST techniques are
gold layer to increase the phase of the propagating optical wave. Am- rapid and consume a small quantity of analytes (down to single cell),
picillin strongly affects the synthesis of the cell wall of E. coli JM109 the system is costly, bulky and complex.
and to detect this effect, the refractive index of cells near the cell-gold
interface is measured. A significant SPR angle shift of susceptible strain 3.7. Optical path difference
and resistant strain of E. coli was observed. Also, significant SPR angle
shift was observed when the susceptible and resistant strain of S. epi- In another case, Leonard et al. developed a biofunctionalized closely
dermidis was treated with tetracycline (a protein synthesis inhibitor). spaced silicon micropillar array platform as a phase grating to study
Similarly, Tawil et al. have studied on methicillin-resistant, methicillin- AST (Leonard et al., 2017). These pillar arrays provided a platform to
susceptible and borderline oxacillin-resistant S. aureus (Tawil et al., colonize bacterial cells and act as an optical transducing element when
2013). In this study, genomic DNA was extracted from S. aureus and optical phase-shift is measured with the help of reflectometric inter-
amplified using the PCR technique and further evaluated using SPR by ference spectroscopy as shown in Fig. 2 (II). Periodic Si micropillars
detecting specific penicillin-binding protein. Similarly, Nawattanapai- were functionalized with Wheat Germ Agglutinin (WGA) to bind bac-
boon et al. have studied methicillin-resistant S. aureus by employing teria. The optical path difference (OPD) between reflected light from
SPR imaging as a DNA biosensor (Nawattanapaiboon et al., 2015). On the top surface of the pillar array and the bottom surface was measured.
the other hand, Syal et al. have studied a unique phenotype AST The OPD was strongly dependent on the refractive index of liquid-filled
method using plasmonic imaging and tracking (PIT) of E. coli O157:H7 medium and height of pillar. Post-colonization of bacteria, the re-
as shown in Fig. 2(I) (Syal et al., 2017b, 2016). This plasmonic imaging fractive index of the medium changes and this was measured as a
setup used a high numerical aperture objective (NA 1.49) and an in- change in OPD. The refractive index decreases when antibiotic above
verted microscope for contrast imaging of 47 nm thick gold sensor chip. MIC was used as it inhibits the colonization, while untreated coloni-
The motion of individual bacterial cells was tracked with nanometer zation over time increases the refractive index. This AST method pro-
precision by considering the contrast change of plasmonic image. The vided result in less than 3 h in real-time and can also provide MIC.
action of antibiotic over period slows down nanoscale motions of the
bacterial cell, which was tracked by the plasmonic imaging technique. 3.8. Photoluminescence
In another scenario, Kee et al. used a plasmonic nanohole-based bio-
sensor for monitoring growth of E. coli and evaluating AST performance In another optical detection approach, Nazemi et al. developed an
(Kee et al., 2013). Plasmonic nanoholes were fabricated using standard innovative AST based on Photoluminescence (PL) emission of photo-
microfabrication processes involving thermal oxidation, low-pressure corroded GaAs/AlGaAs quantum well (QW) (Nazemi et al., 2017). The
chemical vapor deposition (LPCVD), lithography, sputtering, and QW was formed by depositing 30 pairs of 6 nm thick GaAs and 10 nm
etching. A photonic crystal-like structure with different through-hole thick Al0.35Ga0.65 layers sequentially on a semi-insulating GaAs sub-
diameters ranging from 200 nm to 350 nm was fabricated through a strate. The QW biochip was functionalized and kept in a bacterial

Fig. 2. (I). (a) Schematic setup of plasmonic imaging and tracking of bacterial cells. Surface plasmons were formed when p-polarized light is subjected to gold coated
glass sensor chip and reflected light is detected using CCD, (b) Bacterial cell's movement with and without antibiotic treatment, (c) Image of bacterial movement with
respect to varying plasmonic image contrast. (d) The plot showing the vertical distance between the plasmon surface and live bacterial cell. (e)The plot showing the
vertical distance between the plasmon surface and dead bacterial cell. Reprinted with permission from (Syal et al., 2016). Copyright (2015) American Chemical
Society. Fig. 2(II). (a) Schematic representation of phase-shift reflectometric interference spectroscopy AST using photonic chip of pillar-type gratings, (b) Typical
reflectance spectra from the grating collected by a spectrometer, (c) seeding stage: bacterial suspension was allowed to settle inside Si pillar chip, (d) Schematic
representation of biochip below and above MIC values, (e) Schematic representation of increase in refractive index below MIC, (f) Schematic representation of
decrease in refractive index above MIC. Reprinted with permission from (Leonard et al., 2017). Copyright (2017) American Chemical Society.

7
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

suspension flow cell so that the electrostatic interaction of bacterial droplet microfluidic for capturing and AST monitoring of E. coli
cells with the biochip surface can be monitored by analyzing the in- (Sabhachandani et al., 2017). In this case, E. coli suspension and anti-
tensity and maxima of the PL spectrum. Under the influence of anti- microbial agent are encapsulated in droplet form by using droplet mi-
biotics, the electrostatic interaction was observed to be decreasing crofluidics (MFD). In the encapsulated droplet, E. coli from the patient's
(owing to bacterial death). This caused changes in the PL maxima po- sample were easily captured by beads conjugated with antibody. Both
sition over time. This method could provide AST results in less than 3 h antibiotic-induced bacterial growth and phenotype filament formations
for two strains of E. coli. However, it used a complex and expensive was studied via fluorescence imaging. This technique had the capability
components and results of the evaluation with other bacterial species to detect a single cell and produce AST results in 30 min. All these beads
have not been reported. assisted methods have a low throughput in screening for an antibiotic
or bacteria (in the case of detection) and are limited by an inability to
3.9. Bead-based AST be multiplexed for different assays. He et al. integrated an im-
munomicrobead-based microfluidic platform for bacterial detection
A group of researchers at the University of Michigan, have worked and AST (He et al., 2014). In this study, four PDMS microfluidic units
on asynchronous magnetic bead rotation (AMBR) sensor to measure the (one for identification and three for AST with different antibiotics) with
growth and drug susceptibility of individual bacterial cells (Kinnunen common output were fabricated on a single chip. In one-unit, bacterial
et al., 2012, 2011; Sinn et al., 2012, 2011). In their sensors, magnetic suspension mixed with antibody-coated microbeads were examined
beads were introduced in bacterial suspension and placed within an using via fluorescence imaging for identification. Remaining three MFD
external rotating magnetic field to control the rotation of the beads. made of linear concentration gradient generator was used for AST. This
Over time the bacterial cells attach to the beads forming bead-bacteria device could identify bacteria (E. coli) in the range of 101–105 CFU/μL
complexes. The changes in shape, volume, magnetic moment, and en- in 30 minutes and provide AST results in 4 h.
vironment (viscosity) of the bead-bacteria complex was measured by
considering the rotational frequency. In 2011, Kinnunen et al. used a 3.10. Nanomechanical sensor
single ball-shaped magnet upon which a single E. coli bacterial cell can
be attached when the magnet is functionalized with anti-E. coli In the field of phenotypic AST, nanomechanical sensor-based AST
(Kinnunen et al., 2011). During the growth of attached bacteria, the cell has been an emerging approach in the last decade. A nanomechanical
length changes due to cell division which leads to change in rotational sensor is an ultrasensitive oscillator (cantilevers in most cases) which
dynamics. The affected growth dynamics creates a unique detectable measures the change in mass induced by the adsorption or selective
signature in the rotational dynamics after addition of antibiotics. By chemical binding of specific target substance onto the substrate or
measuring these torque variations, change in length of bacteria as low specimen under study. Atomic Force Microscopy (AFM) based canti-
as 80 nm was sensed. In the improvement of this technique, they had levers are a versatile tool for biochemical sensing applications because
used a self-assembled clustered magnetic microparticles to detect of its operational capabilities in different environments such as gases,
streptomycin and gentamicin MIC against E.coli (Kinnunen et al., 2012). liquids, or even in vacuum (Lang and Gerber, 2008). It is well known
This method was rapid (1.5 h) as small changes in bacterial growth was that drugs induce changes in the morphology of bacterial cells i.e. de-
clearly measurable over such a small timeframe. However, this process pending on the concentration of antibiotics, varying effects from the
requires delicately suspended cultures and was not effective for higher creation of pores to complete lysis of the cell wall. The change in
concentrations of bacteria. To solve this problem, a novel 48-sensor morphology of susceptible cells due to antibiotic treatment has been
prototype using standard microwell plates stacked between an aligned successfully measured using AFM tip by several researchers (Du et al.,
light source and photodiodes was used (Kinnunen et al., 2014). In this 2008; Eaton et al., 2008; Fernandes et al., 2009). However, this tech-
case, the rotational frequency was temporally tuned to observe the nique has several limitations such as the complexity of sample pre-
growth of a wide spectrum of bacterial concentrations ranging from 108 paration, the requirement of an AFM expert to perform the study, the
to 5 × 103 CFU/mL AST of S. aureus. In the AMBR sensor discussed complexity of data interpretation, and an amount of time consumed
above, particle-to-surface stiction effects affect the efficiency and sen- (Longo and Kasas, 2014). Additionally, it does not provide real-time
sitivity of the system leading to limited time for reproducing the results. AST results. To overcome these problems, Longo et al. have developed a
To alleviate this shortcoming Sinn et al. have used a water-in-oil droplet microcantilever coupled with an AFM tip to study the micro-motions of
microfluidics-based AMBR sensor to rapidly measure bacterial growth, cells (Aghayee et al., 2013; Longo et al., 2013). Bacterial cells were
susceptibility, and MIC as shown in Fig. 3 (Sinn et al., 2011). In this attached to the surface of the cantilever and its dynamic fluctuations
device, individual AMBR sensors were enclosed in the microfluidic were measured as a function of time. Nanometer-scale motions of E.
droplets and growth of the individual as well as small bacterial popu- coli, and S. aureus bacteria and their metabolic activities in the presence
lations was monitored. In 2012, the same group developed a similar of antibiotics was studied by measuring the amplitude of fluctuations. A
droplet microfluidics platform using AMBR sensor-based viscometer to schematic representation of bacteria detection and AST profiling is
measure AST. The change in the metabolism, motility, and concentra- shown in Fig. 4(I)(Longo et al., 2013). This work was further extended
tion of bacteria changes the viscosity of the medium, which is leveraged by Lissandrello et al. to measure the nature of force applied by bacterial
(Sinn et al., 2012). These studies are limited to E. coli and thus, more cells on the microcantilever (Lissandrello et al., 2014). It was observed
studies are needed to be performed to make it more reliable for rapid that the amplitude of the fluctuations strongly depends on the oscilla-
AST in clinics. tion frequency as well as the number of bacterial cells attached to the
In another study, Chung et al. used carboxylate-/amine-modified surface of the cantilever. Also, the nanoscale motions of prokaryotic
polystyrene particles and optical diffusometry principle to quantify the and eukaryotic cells and their viability (at a metabolic level) were de-
Brownian motion of particles (Chung et al., 2016). The Brownian mo- tected using this technique (Kasas et al., 2015). Thus, it is a promising
tion of particles was recorded using a high-speed camera at 10 Hz frame technology to study and characterize the real-time interactions of
rate for each measurement. When live bacteria (P. aeruginosa) are at- bacterial cells with antibiotics. Recently, Johnson et al. studied bac-
tached to the surface of particles, they move more vigorously due to terial vibrations by placing them on a resonant crystal (France et al.,
gaining additional energy from bacteria. However, dead bacteria cause 2016; Johnson et al., 2017). In this case, bacterial-induced phase noise
more dragging and thus diffusivity decreases. In this AST method, the i.e. perturbations of resonant crystal boundary conditions and resonant
temporal diffusivity of the particles is analyzed and it showed excellent frequency was measured over time to study AST. On exposure of
results for P. aeruginosa against gentamicin within 2 h. Recently, Sab- polymyxin B and ampicillin to E. coli, the change in phase noise was
hachandani et al. have used bead-based biosensors combined with observed in 7 and 15 min respectively.

8
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

Fig. 3. (a) Schematic representation of microfluidic channel fabrication using standard lithography process, (b) Image of the microfluidic device, (c) An image of the
microfluidic device placed inside the electromagnet to generate rotating magnetic field, (d) Magnified image of magnetic bead rotating in an external rotating
magnetic field, (e) Image of droplet formation containing magnetic beads. Reprinted with permission from (Sinn et al., 2011). Copyright (2011) The Royal Society of
Chemistry.

Another efficient mechanical sensing method is based on quartz to provide AST results for multiple antibiotics is also challenging. In this
crystal microbalance (QCM). In 2015, Ma et al. utilized electrochemical regards, Kara et al. had used fine MFD filled with PBS solution to
QCM-based lectin biosensor for AST profiling (Ma et al., 2015). Anti- monitor ultra-movement of bacteria via measuring electrical resistance
biotics induce changes in the carbohydrate and lectin adhesin structures through the channel (Kara et al., 2018). When bacteria cells were
of bacterial cell walls, which leads to alterations of lipopolysaccharide present inside the fine microfluidic channel, the effective diameter of
(LPS) chain lengths and lectin expressions. Thus, the binding strength of the channel was reduced resulting in a voltage drop due to resistance
bacteria cells on QCM surface via LPS was crippled. The effect of ci- faced by ions. These small voltage fluctuations were recorded with the
profloxacin, ceftriaxone, and tetracycline on E. coli suspension showed help of lock-in-amplifiers over a period of time to measure the effects of
at least 10% change in resonant frequency after 1 h. the antibiotic on the E. coli. Though this method is simple, rapid, and
requires small to no sample preparation, it will be not useful for non-
motile bacteria (e.g. S. aureus). In another study, resonator frequencies
3.11. Opto-mechanical techniques depend AST was carried out by placing multiple mass sensors in a
suspended microfluidic channel (Cermak et al., 2016). Here, these re-
In another approach, buoyant mass and instantaneous growth rates sonant mass sensors were placed along the channel but separated by
of individual bacterial cells were measured with the help of MFD em- ‘delay: to give sufficient time to grow’ to measure growth rate i.e. in-
bedded on resonant cantilever (Gfeller et al., 2005; Godin et al., 2010; creased mass in the successive sensor. This serial suspended micro-
Knudsen et al., 2009). The position of individual bacterial cells in an fluidic resonator was tested for various cells including Saccharomyces
MFD was determined by measuring the changes in the resonant fre- cerevisiae, E. coli, and E. faecalis bacteria cells. For AST profiling, the
quency. This method could detect different bacterial cells and their normal and bacterial suspension mixed with antibiotics was made to
changes in volume, mass, and density. The difference between anti- flow through the channel and their growth was measured at 0.02 pg/h
biotic-resistant and antibiotic-susceptible bacteria was also verified. A resolution.
similar approach was developed by Etayash et al. for AST measurement
by integrating photothermal infrared spectroscopy with a bi-material
microchannel cantilever (Etayash et al., 2016). This device was de- 3.12. Microfluidics technologies
signed to measure the adsorbed mass, adsorption stress, and mid-in-
frared spectroscopy of the adsorbates (in this case, bacterial cells) In microfluidics technology, an extremely small amount of fluid can
which improved the sensitivity and selectivity. This scheme is shown in be accommodated and manipulated to yield interesting biophysio-
Fig. 4(II). In this case, antibodies captured specific bacteria cells on the chemical changes and effects (Whitesides, 2006). The size of these
cantilever. These immobilized bacterial cells upon absorption of heat devices can vary from the nanometer to centimeter scales. With the help
from specific frequency ranges of the incident IR light caused changes of advanced nanotechnology and microfabrication techniques, re-
in the resonant frequency of the cantilever. The live and antibiotic-in- searchers have fabricated miniaturized chambers (micro-liter to pico-
duced dead bacteria was studied by measuring changes in the oscilla- liter), channels as well as other structures like porous membranes, slits,
tion frequency of the cantilever. These nanomechanical sensors for AST etc. (Leester-Schädel et al., 2016; Whitesides, 2006). Poly-
application are rapid as compared to other methods. However, it re- dimethylsiloxane (PDMS), glass, silicon, silicone elastomer, or plastic
quires a pure isolated bacterial culture and may not be useful for mixed are the most preferred material in microfluidics device fabrication. In
bacterial population or non-bacterial cells. Implementing multiplexing biological applications, the prime utility of ‘lab-on-chip’ platform is to

9
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

Fig. 4. (I) (a) Schematic and SEM image of cantilever functionalized with bacteria, (b) In an optical image (bottom), several attached bacteria can be visualized, (b)
Schematic representation of acquisition chamber (A.C.) with help from laser beam, L and detector, D, (c) Schematic representation of cantilever fluctuations during
bacterial presence, (d) Bacterial presence fluctuations measured in presence of antibiotics. Reprinted with permission from (Longo et al., 2013). Copyright (2013)
Nature Publishing Group Fig. 4(II) (a) Schematic representation of microfluidic channel on a bi-material cantilever (BMC) for AST application, (b) Cross-sectional
Scanning Electron Microscopy (SEM) image of microfluidic channel, (c) Cross-sectional schematic of the microfluidic channel filled with bacterial suspension and
antibiotics, (d) Green fluorescent image of microfluidic channel filled with bacterial suspension, (e) SEM image of microfluidic channel tip, Time evolution of (f)
cantilever deflection, (g) Resonance frequency of cantilever beam, as more and more bacteria gets immobilized on the cantilever using mAb, and (h) Frequency range
over which bacteria absorbs IR light detected through nanomechanical deflection. Reprinted under creative commons licence from (Etayash et al., 2016) Nature
Publishing Group.

create a controlled physical or chemical environment via geometric have several advantages like allowing for multifunctional im-
manipulations with the capability to manipulate liquids and study the plementation (e.g. amplification, cell lysis, and hybridization), ensuring
changes in the biochemical behaviour. Diverse microfluidic archi- high sensitivity, reducing the workload, and making the testing less
tectures have been used for different applications such as gradient expensive.
generation, droplet-formation, microchamber array formation, etc. The In the last twenty years, several researchers have studied bacterial
changes in physical or chemical behaviour of biological samples in culture using microfluidics and have found that bacterial division in
microfluidics were monitored through various modalities such as op- small volume is faster, which may potentially reduce AST time. Connell
tical, magnetic, and electrical. Overall, these lab-on-chip platforms et al. showed that bacterial behavior was strongly dependent on the

10
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

against three different concentrations of antibiotics against P. aerugi-

nosa. AST was microscopically evaluated by counting cell numbers and
morphological observation.

3.13. Gradient microfluidics

In a 96-well plate, a concentration gradient antibiotic can be gen-

erated and tested against bacteria. However, this process is laborious
and consumes large volumes of reagents. With the help of microfluidic
engineering technology, some research groups have developed on-chip
gradient generator (Hou et al., 2014a; Li et al., 2014). These gradient
generators were formed in discrete or continuous versions with the help
of different channel configurations, microfluidic traps, or hydrogel
slabs. Kim et al. have fabricated a fivefold gradient generator to
monitor biofilm disruption under the influence of several antibiotics
(Pil Kim et al., 2010). Similarly, Hou et al. have used a commercially-
available microfluidic device having a 6 μL chamber to test a range of
Fig. 5. Schematic of the two-layer disc for gradient generation. Gradient of antibiotic concentrations against E. coli, S. aureus, and Salmonella Ty-
antibiotic was generated by rotating the disc. Reprinted with permission from phimurium (Hou et al., 2014b). The bacterial response over a linear
(Tang et al., 2018). Copyright (2018) The Royal Society of Chemistry. concentration gradient of antibiotics was observed using a phase con-
trast microscope to determine growth over time and against various
geometry of microfluidics (Connell et al., 2013). In 2010, Chen et al. concentrations to obtain the MIC value. A similar approach analogous
demonstrated that growth of bacterial cell strongly depends on the to classical E-test was used to determine MIC corresponding to an op-
volume of the culture medium (Chen et al., 2010). They proved that the timal bacterial concentration (Malmberg et al., 2016). Recently, Tang
enhanced surface area-to-volume ratio of growth chamber like in mi- et al. developed an automated linear concentration gradient generator
crofluidics increases the growth rate of cells. Researchers are interested based on centrifugal microfluidics for AST applications (Tang et al.,
in single bacterial cell separation and its AST behavior over time. For 2018). This device consists of a multilayer microfluidics platform in the
this, several techniques such as droplet microfluidics, microchambers, shape of stacked discs where the individual layer was fed with DI water
bacterial capturing and their confinement, bacterial traps, and chan- (layer-1), ampicillin (layer-2) and E. coli suspension (layer-3) as shown
nels/tracks have been developed (Choi et al., 2014, 2013; Dai et al., in Fig. 5. After spinning the disc at various speed, 16 linear gradient
2013; Long et al., 2013; Lu et al., 2013; Peitz and van Leeuwen, 2010; antibiotic concentrations were generated in the metering chamber and
Rowat et al., 2009; Sinn et al., 2011). Baltekin et al. used a narrow they were further mixed with an equal amount of E. coli suspension
microfluidic trap to study AST and were able to provide results in less after passing through capillary valves. Once each fluid reaches the
than 30 min (Baltekin et al., 2017). The growth of trapped E. coli in the mixing chamber, the disc was incubated, and absorbance of each
presence of antibiotics was studied using automated phase contrast chamber was measured sequentially. This device provided AST results
microscopy and post image processing. in 3 h. Kim et al. have developed a miniaturized broth microdilution
With emerging engineering technologies, there is a wealth of work platform based on concentration gradient microfluidics for AST appli-
dealing with microfluidic for AST applications as reviewed by Campbell cation(Kim et al., 2015). In this study, pressure driven microfluidics
et al. and Dai et al. (Campbell et al., 2016; Dai et al., 2016). One such was used to generate different concentrations of antibiotic which were
study is a multiplexed microfluidics device developed by Mohan et al. mixed with bacterial culture (5 × 105 CFU/mL) in 30-nL chambers.
for AST application (Mohan et al., 2013). In this work, an array (4 × 6)- Bacterial behaviour was monitored using standard phase contrast mi-
of microwells, with each microwell straddled by fluid lines on either croscopy over time to determine the effect of antibiotics. In another
side for loading antibiotics and bacterial suspension was fabricated for study, Derzsi et al. developed a gradient microfluidic based on geo-
the study of bacteria-antibiotic interaction. The effects of an individual metry controlled hydrodynamic traps that passively meter, merge,
as well as multiple antibiotics with variable concentrations were stu- route, and retain nanoliter samples via capillary action (Derzsi et al.,
died on E. coli. This multiplexed microfluidics device provided AST 2016). This device was operated through five pipetting steps, three for
results in less than 4 h and the total volume of reagents used for this test loading the aqueous samples namely bacterial culture, antibiotic and
was ∼6 μl. In their next work, a similar kind of multiplexed micro- pure broth as a diluent and two steps for maintaining the flow in the
fluidic device was used for polymicrobial suspension (Mohan et al., device. The metering requirements in the device are for creating equal
2015). In majority cases in the literature, AST has been reported to be volumes of bacterial culture using pure broth and concentration gra-
performed on monomicrobial cultures whereas polymicrobial AST is dients of the antibiotics. The metering as well as the flow in the chip
complicated due to inadequate antimicrobial dosing regimen of dif- creates the required droplets for analysis. This chip gave a dilution ratio
ferent bacteria. This device had 48 wells for diffusion mixing and, half of 1.68 (although it was designed for 2) in 11-fold. It was successfully
the part of each microwell was covered by polymicrobial suspension tested with E. coli strain against ampicillin and provided AST results in
and the other half by antimicrobial solutions. The analysis of the kill 4 h after pipette loading.
curve after 16 h of incubation provided information about AST results. In most cases, microfluidic chips are operated by means of pumps.
In another study, Dai et al. have worked on multiplexed nanoliter re- On the other hand, Ran et al. have developed a pump-free micro-device
actors (1 nL) that allows for different cell cultures in parallel (Dai et al., for AST analysis (Ran et al., 2015). In this study, a circular well (5 mm
2013). The device was designed in such a way that flow channels and in diameter) and a V-shaped microfluidic channel (200 μm wide and
culture reactors will maintain uniform bacteria numbers in variable 1000 μm length on each side) were fabricated in such a way that the
environments in a single run. This device was integrated with a fluor- intersecting portion of the V-shaped channels would terminate by just
escence imaging system to study bacterial growth in the presence of touching the well wall. The channel and the well wall were separated
various concentrations of antibiotics. Recently, Matsumoto et al. have from each other by a set of parallel micropillars as a diffusion barrier.
used four channel based microfluidics device with one shared inlet for Upon loading antibiotic in the well, they start diffusing towards the
AST application (Matsumoto et al., 2016). This device allows for testing channels which were loaded with bacteria mixed agarose gel. The an-
tibiotic diffuses from the well towards the channels and as a result, a

11
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

Fig. 6. Schematic of two-dimensional linear gradient

AST: (a) Design to generate gradients with two dif-
ferent drugs. This device can't use full range of an-
tibiotics, (b) Modified design to generate full range
(0–100%) of antibiotics gradients. This design en-
ables to use multiple number of antibiotics and their
corresponding gradients. Reprinted with permission
from (Liu et al., 2017). Copyright (2017) Wiley On-
line Library.

concentration gradient is generated along the channels. The bacterial sensitive and resistant strains of S. aureus from human blood plasma. In
behavior in the presence of gradient antibiotic concentration was stu- this case, separate devices were used for droplet generation, droplet
died using fluorescence imaging. Though this method was simple and incubation, and droplet detection. To reduce AST time, Kaushik et al.
was able to perform AST within 4 h it was limited by the requirement developed a droplet-based fluorescent AST device with inbuilt incuba-
for large amounts of reagents. The gradient profiling is time dependent tion zone as shown in Fig. 7. (Kaushik et al., 2017). This device provides
and thus one has to wait till the diffusion process completes to start the AST results in 1 h and is one of the fastest phenotypic AST devices.
testing. Similarly, Baraban et al. used a novel millifluidic droplet analyzer
Instead of conventional microfluidics, Sun et al. have developed (MDA) to determine the growth rate of E.coli and calculate its MIC
hydrogel microfluidics by combining conventional plate culture to against specific antibiotics (Baraban et al., 2011). Digital screening of
study AST (Sun et al., 2016). A liquid mix of agar and sodium alginate multiple antibiotics or combined antibiotics on E. coli using an auto-
gel was cast on PDMS mold to solidify after which solid gel slices were mated microfluidic platform was carried out by Churski et al. (2012). In
bonded with help of CaCl2 solution. To generate a linear gradient of this case, resazurin (a fluorescent dye) was used to track bacterial be-
antibiotics across the square shaped solid porous gel, four channels haviour and AST was performed within 3 h. This device was pro-
were designed to surround this zone from four sides to flow antibiotics grammed for automatic droplet formation and sequential detection
and LB broth as shown in Fig. 6(a). Bacteria were cultured on above this using optical imaging. The major advantages of droplet microfluidics
square region and their interaction with antibiotics were monitored via were the ability to fine-tune droplet size, bacterial density, antibiotic
fluorescence imaging. This device was capable of testing against two concentrations, and reproducibility. In 2006, Jiang et al. demonstrated
gradients and the same group went on to improve this technique to test a unique droplet-based millifluidic technique to study AST accurately
with multiple antibiotics across a range of concentrations (Liu et al., (Jiang et al., 2016). This millifludic system was analyzed with a Mil-
2017). However, this design restricted the generation of complete liDrop Analyzer and results were in good agreement with classical as
concentration gradient range of 0–100%. In their improvised version, well as VITEK®2 technologies. However, droplet microfluidics has not
three layers of solid gel slices were bonded so that channels can bypass yet been tested with multiple bacteria or combinations of bacteria with
each other as shown in Fig. 6(b). By designing parallel channels in one different antibiotics and thus, more studies need to be carried out in this
layer and orthogonal parallel channels in another layer, several square field.
zones were formed. This device was able to generate multiple gradient
zones of individual or combination of antibiotics and also allowing for
the entire concentration range of 0–100% to be covered which was a 3.15. Stress-induced AST
limitation in the previous design. The test results of E. coli against
ampicillin, gentamicin and their combination were produced in less By applying mechanical stress and antibiotics, Kalashnikov et al.
than 5 h. However, the durability of this device was not tested and thus, have developed a microfluidics AST technique to accelerate the death of
more studies are needed. susceptible strains (shown in Fig. 8) (Kalashnikov et al., 2018, 2017,
2014, 2012). Bacterial cells were immobilized on the channel floor by
epoxy coating to hold them in place under high flow rates. Mechanical
3.14. Droplet microfluidics in AST and enzymatic stress was applied on these immobilized bacteria by
flowing growth media containing Sytox green, which is a fluorescent
In droplet microfluidics, droplets are formed either by continuous stain confirming the loss of cell viability. An enzyme Lysostaphin that
flow of the emulsion or by electrowetting. In AST applications, bacterial cleaves linkages in the bacterial cell wall, bacterial culture media, and
cells and the antibiotics to be tested against were encapsulated in the antibiotics at high speed by inducing enzymatic stress. The antibiotics
form of droplets in a favourable cellular environment (Kaminski et al., inhibit the cell wall repair pathways that get triggered by the stress-
2016). In 2008, Boedicker et al. used droplet microfluidics to accelerate inducing enzyme. The effect of antibiotics was examined via phase
S. aureus growth and measure AST against different antibiotics (Q. contrast and fluorescence microscopy through visualization of the Sytox
Boedicker et al., 2008). This device provided results after 7 h of in- Green dye. This stress-induced AST could distinguish between methi-
cubation and was limited only by the formation and measurement of cillin-resistant S. aureus and methicillin-sensitive S. aureus in less than
droplet volume. This device was also capable of delineating methicillin 30 min (Kalashnikov et al., 2014, 2012). In their next work, they

12
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

Fig. 7. (a) Schematic representation of droplet for-

mations embedded of single bacteria cell and anti-
biotics. These encapsulated droplets are incubated
(by keeping them in an incubation zone) for bacteria
growth and study droplets were analyzed using
fluorescence imaging, (b) Image of droplet genera-
tion during continuous flow, (c) Schematic re-
presentation of the droplet analysis by fluorescence
imaging using a laser excitation source and an ava-
lanche photodiode detector. Reprinted with permis-
sion from (Kaushik et al., 2017). Copyright (2017)
Elsevier.

3.16. Microchamber-based AST

In droplet microfluidics, droplets are motile and thus, manipulating,

handling, and tracking them is relatively complex. Recently, few re-
searchers have reported the development of microchamber (static)
based AST. In 2014, Weibull et al. fabricated nanoliter wells (336 wells
per device) which offers simple loading, multiplexing, and imaging for
AST applications (Weibull et al., 2014). The growth of E. coli in nano-
liter chamber against ciprofloxacin, cefotaxime, and ampicillin was
observed in real-time using optical density measurements. They have
developed a mathematical algorithm to calculate the lag phase of
bacterial growth which allowed for determining MIC within 4 h. Con-
sidering the design and structure of these microwells, it was difficult for
these wells to be completely isolated from each other during the testing
process, thereby casting apprehensions on the utility of this platform to
be used for multiplexing. Recently, Avesar et al. have developed che-
mically isolated nanoliter wells (8 nL) for rapid AST (Avesar et al.,
2017). Here, the multiplexing was carried out by creating rows of na-
nowells with a common delivery channel. The mixed bacterial culture
with antibiotics was used to fill the wells via a common delivery
channel. The wells were chemically isolated by flowing oil through the
Fig. 8. (a–b) Stress induced AST using microfluidics device. Cross-sectional same delivery path as shown in Fig. 9. Further, the biochemical activity
view of a single microfluidic channel AST action. Bacterial behavior was of the bacterial cells within each well was monitored using conven-
monitored using phase contrast and fluorescence microscope. Reprinted under tional fluorescence imaging. Also, an algorithm was developed for the
creative commons licence (Kalashnikov et al., 2017) Nature Publishing Group automated analysis of susceptible/resistant strains. In another study,
and reprinted with permission from (Kalashnikov et al., 2012) Copyright (2012) eight individual microchambers (volume-360nL) with two flow chan-
The Royal Society of Chemistry. nels were fabricated to determine the MIC for microbes (Takagi et al.,
2013). Channel 1 was used to load bacterial suspension and channel 2
successfully tested against several gram-negative bacteria namely, En- was used for suction of suspension into the chamber. Prior to making
terobacter cloacae, E. coli, Klebsiella pneumoniae, Pseudomonas aeruginosa the device via bonding of PDMS with the glass substrate, antibiotics
and gram-positive bacteria S. aureus(Kalashnikov et al., 2017). Most solution with 0/5 wt% BSA was loaded onto the PDMS chambers and
recently as a next step, the temperature, mechanical stress, chemical freeze-dried to form a matrix. Variable concentrations of different
stress, and media compositions were varied to optimize the AST per- freeze-dried matrices were used to study the growth of E. coli.
formance (Kalashnikov et al., 2018). Despite uniqueness and robustness
of this measurement modality, there are several concerns that needs to
be addressed: such as requirement of large volumes of reagents, pos- 3.17. Agarose based cell tracking
sibilities of leakage owing to high flow rate of liquid in the channels and
change in morphology of resistant bacteria after the experiment which In most cases, cells in MFD are in constant motion and thus tracking
could be considered as a collateral damage of the experiment itself. of individual cells is a difficult task. Also, shear flow affects the mor-
phology of cells. Few research groups have alternately used agarose in

13
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

Fig. 9. (a) Schematic representation of

multiplexed microfluidic agarose channel
for AST. This device is loaded with agar-
ose–bacteria mixture in centered inlet and
antibiotics at periphery inlet, (b) (1)
Channel without any loading, (2) agar-
ose–bacteria mixture loading, (3) Antibiotic
loading. Here diffusion of antibiotics hap-
pens through capillary valve, (4) Bacterial
behavior is recorded using a microscope
and a time lapse method. Reprinted with
permission from (Choi et al., 2013) Copy-
right (2013) The Royal Society of Chem-
istry.

MFD to freeze individual cells and study their AST profiles. Choi et al. cultured in glucose medium, was tested against ciprofloxacin, tetra-
have worked on microfluidic agarose channel based systems for cycline, azithromycin, and amikacin. It was found that the EOT is in-
tracking single bacterial cells and applied for rapid AST studies (Choi versely related to the pH i.e. lower EOT results from bacterial inhibition
et al., 2014, 2013). In their first system, liquid agarose with bacteria and higher EOT results from bacterial growth. This device provided AST
and culture media containing antibiotics was flown in two channels result within 2 h. However, after Tang et al. no other researcher has
separated by a capillary valve (shown in Fig. 9) (Choi et al., 2013). In explored this technique further.
the next step, bacteria cells were immobilized by solidifying the In general microfluidics devices require valves and actuators to
agarose. The culture media containing antibiotic was allowed to diffuse operate. However, Cira et al. developed a portable microfluidic device
towards the agarose frozen channel. Bacterial growth in the presence of that has dead-end chambers where bacterial samples were loaded and
different antibiotics with variable concentration was then monitored for isolated from others (J. Cira et al., 2012). In this case, a set of PDMS
AST. The same group extended this work using microfluidic agarose chambers were fabricated and dried antibiotics were stored in them.
channel integrated with a 96-well platform i.e. individual well and This PDMS layer containing chambers was bonded with a second PDMS
microfluidics channel were placed side-by-side and separated by a ca- layer having microfluidic channels to connect the chambers. The
pillary valve (Choi et al., 2014). In this study, the morphology of in- chambers, as well as the channel, were degassed via vacuum as PDMS is
dividual cells was analyzed in the presence of antibiotics with the help permeable to the flow of gases helping the chambers and channels to be
of time-lapse microscopy. An algorithm was developed to match with degassed. The bacterial suspension is then sucked into the device during
gold standard broth microdilution technique (91.5% agreement). In the loading process because of the vacuum inside. In this degassing
another study, the same device was used to carry out direct and rapid driven device, the interaction of antibiotics and bacteria in the chamber
antimicrobial susceptibility testing (dRAST) of bacterial samples from a was visualized with a pH indicator (colorimetric change) i.e. as bacteria
positive blood culture bottle and provided AST results (91.11% success) grow, they produce organic acids by degrading glucose and thus pH
in 6 h. This is of particular interest as most studies are only on clinical decreases. This is a self-loading device which needs only a few micro-
isolates (Choi et al., 2017). Recently, the same platform was used to test liters of the bacterial suspension. This AST device was tested with
on Mycobacterium tuberculosis. The drug susceptibility test results were various concentrations of vancomycin, tetracycline, and kanamycin
obtained in 9 days, whereas conventional drug susceptibility tests for against Enterococcus faecalis, Proteus mirabilisHI4320, Klebsiella pneu-
Mycobacterium tuberculosis takes around 4–6 weeks (Choi et al., 2016). moniae, and E. coli.
In 2014, Li et al. worked on a unique AST device having a thin agarose
gel membrane coated with a monolayer of bacterial suspension which 3.19. Dielectrophoresis
was sandwiched between glass and PDMS layer (Li et al., 2014). Two
parallel PDMS channels (namely source and drain) were allowed to Dielectrophoresis is a nonlinear electrokinetic phenomenon in
flow amoxicillin and plain medium so that a concentration gradient of which a force gets exerted on dielectric particles such as bacterial cells
amoxicillin formed between the source and drain through agarose. when a non-uniform electric field is applied. The particle movement is
Morphological dynamics of sandwiched bacterial cells under the in- strongly dependent on the strength of the local electric field, medium,
fluence of gradient of amoxicillin were recorded for AST. and the permittivity of the particles. Based on this principle, Johari
et al. have studied the viability of bacterial cells by applying an electric
3.18. Microfluidic pH sensor field of frequency 100–200 kHz (Johari et al., 2003). The movement of
the cells towards high and low electric fields was used to differentiate
During bacterial growth, the glucose-containing medium becomes between viable and non-viable cells. Similarly, Braff et al. discriminated
more acidic due to the accumulation of organic acids. Based on this bacterial species at the strain level for P. aeruginosa and S. mitis by
idea, Funfak et al. have used pH-dependent fluorescent polymer parti- immobilizing them with the electric field in a microfluidic channel
cles in microfluidic channels to study the metabolic activity of E. coli (Braff et al., 2013). The immobilized bacterial cells images were pro-
cells (Funfak et al., 2009). In 2013, Tang et al. developed a microfluidic cessed to study bacterial behaviour. Chung et al. have worked on die-
platform to measure changes in pH for AST applications (Tang et al., lectrophoresis-based AST by measuring dielectrophoresis behaviour
2013). In this device, a microfluidic channel was fabricated on the and elongation rates of E. coli (Chung et al., 2011). They designed a
porous silicon surface by integrating pH-sensitive chitosan hydrogel. quadruple electrode to create a non-uniform electric field as shown in
Bacterial cells were confined in a nanoliter size channel which results in Fig. 11(I) (a-c). Under the influence of β-lactam antibiotics, the cross-
rapid changes in metabolic activity and thus pH changes rapidly. Based over frequency (transition frequency as particles moves from high
on the pH of the medium, the chitosan hydrogel either shrinks or swells electric field region to low electric field region) of the bacterial sus-
and this was recorded with Fourier transform reflective interferometric pension and cell elongation was measured. This test provided AST re-
spectroscopy (FTRIFS) which measure the effective optical thickness sults in less than 2 h. Subsequently, they tested with cephalosporin-re-
(EOT) of chitosan hydrogel as shown in Fig. 10. E. coli suspension, sistant E. coli and cephalosporin-resistant K. pneumoniae against

14
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

Fig. 10. Schematic representation of microfluidics pH sensor. The effective optical thickness of porous silicon and chitosan hydrogel was measured by considering
antibiotics induced pH change. Reprinted with permission from (Tang et al., 2013) Copyright (2013) American Chemical Society.

cefazolin(CEZ) successfully (Chung et al., 2012). According to these bacterial suspension (Chotinantakul et al., 2014; Ertl et al., 2003, 2000;
results, CEZ treatment caused reduction of cross-over frequency of Mann and Mikkelsen, 2008). The measured current in amperometric
bacterial suspension to hundreds of kHz and cell length increased more electrochemical AST strongly depends on Fe(II) oxidation which is
than 10 μm (which is usually the average cell length of healthy cells). proportional to the number of cells available in the suspension. Under
This work was further extended to study bacterial behaviour in a re- the influence of antibiotics above MIC, the current decreases as com-
sistant and susceptible mode in the presence of different antibiotics pared to resistant bacterial suspension or a bacterial suspension with no
with variable concentrations (Su et al., 2017). Cell elongation of anti- antibiotics added. Amperometric/voltammetric AST systems are rela-
biotics induced susceptible bacteria, cell swelling, or lysis, as well as the tively faster than conventional as well as few engineered AST tech-
behaviour of resistant bacteria (whether the numbers are unchanged, or nologies. Besant et al. have studied a phenotypic electrochemical ap-
they increased), was monitored using this technique Fig. 11(I) (d). proach for AST applications (Besant et al., 2015). In this case, bacterial
In another approach, electrokinetic AST was used to capture bac- cells were captured in confined nanoliter wells integrated with
terial cells at the edge of microelectrodes by dielectrophoresis method working, counter and reference electrodes. The loaded bacterial cells
and they were analyzed with an automated image processing to single- were cultured with antibiotics and an electrochemically redox-active
cell resolution in the presence of antibiotics (Peitz and van Leeuwen, molecule like resazurin. Resazurin gets reduced actively when live re-
2010). In this study, an additional DC field was applied between in- sistant bacterial cells are present in the suspension leading to changes in
terdigitated electrodes and a redox reagent was flown in the micro- current profiling. If majority susceptible cells are there the extent of
fluidic channel to capture and immobilize bacteria on the electrode reduction reaction is lesser leading to negligible changes in the current
surface. AST of E. coli K12 against Polymyxin B was successfully mon- profile. This method provided AST results in one hour and is one of the
itored and determined MIC as well as IC50 values. A similar approach most efficient and low-cost techniques currently available.
by Lu et al. for AST was carried out by narrowing the channel width to
the range of bacterial cell diameter (Lu et al., 2013). In this study, E. coli
3.21. Impedance spectroscopy
cells were loaded onto the confined microchannels (0.5–10 μm width)
by the electrokinetic method and their growth was monitored at the
Impedance measurement is one of the simplest techniques used for
single cell level. The position of the bacterial cells inside the channel
AST, where an alternating current (AC) signals of a wide range of fre-
was tailored by controlling the electric field and subsequently, the
quencies are applied to a suspension and its impedance response is
growth rate was observed using phase contrast microscopy. The growth
measured. Under the influence of AC field, live bacterial cells in sus-
rates with and without ciprofloxacin antibiotic were easily differ-
pension act as polarized particles leading to charge build-up across their
entiated in less than 1 h. All these dielectrophoresis-based AST studies
cell membrane which acts like capacitors. The total capacitance
were performed on few bacterial species and strains and thus, may not
strongly depends on the number of bacteria cells available in the sus-
be generalized for an exhaustive antibiotic-bacteria combination yet.
pension. In most cases, antibiotics above MIC value causes lysis of cells
which leads to the increase in conductive species like carbonate and
3.20. Electrochemical sensing lactate in the suspension as it gets out of the cytoplasm of the cell into
the suspension medium. Total resistance, as well as capacitance change
Electrical biosensing is considered a robust, miniature, label-free, of the suspension, is measured to find out the effect of antibiotics.
low-cost, rapid, and easy-to-handle technique. These devices are highly Recently, Jo et al. has performed AST experiment by measuring the
sensitive and are considered as the most practical assessment tool for changes in capacitance of E. coli and S. aureus suspension in the pre-
point-of-care diagnosis (Bhalla et al., 2016; Luo and J. Davis, 2013; sence of gentamicin, ampicillin, and tetracycline (Jo et al., 2018). In
Mehrotra, 2016; Turner, 2013; Vigneshvar et al., 2016). In most cases, this study, an array of interdigitated electrodes (IDE) was fabricated on
biological targets are made to bind with the target antibody and the a glass substrate and each electrode area was confined by acrylic wells.
perturbations in the current or voltage signal are measured. Several The sensor surface between the electrodes was functionalized by DNA
biosensors like amperometric/voltammetric biosensors (current mea- aptamers to immobilize bacterial cells. The frequency dependent ca-
surement), potentiometric (voltage/charge measurement), conducto- pacitance of bacterial suspension with and without antibiotics was
metric (medium conductance measurement), impedance biosensors measured in real-time. It was found that capacitance of the suspension
(electrical impedance measurement with varying input potential and decreases over time when antibiotics above MIC was added and in-
frequency) and field-effect transistor (FET) biosensors (current or po- creases when this concentration was below MIC, indicating bacterial
tential measurement across a semiconductor by tuning gate potential) growth over time in the absence of effective antibiotics. In another AST
are widely used. Electrical biosensors have been frequently used to study, 3-aminophenylboronic acid (3-APBA) was used to bind to poly-
detect bacteria and their growth over time. In most amperometric saccharides on the bacterial cell walls to immobilize them and measure
electrochemical systems, ferri-/ferrocyanide is used to monitor the changes in capacitance with and without the antibiotic drug

15
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

Fig. 11. (I) (a) Typical electric field distribution of a quadruple electrode array used in dielectrophoresis method-based AST application, (b) Schematic representation
of dielectrophoretic AST device, (c) Typical behavior of suspension while doing AST, (d) Image of bacteria in resistant and susceptible mode. Reprinted with
permission from (Su et al., 2017) Copyright (2017) American Chemical Society. Fig. 11(II) (a) Schematic representation of impedance biosensor with interdigitated
electrodes, heater, and PMMA microfluidic channel for AST application, (b) Image of the device, (c) AST measurement using label-free electrical sensing on a
biosensor without surface modification, (d) AST measurement using same biosensor with surface modification. Reprinted with permission from (Safavieh et al., 2017)
Copyright (2017) American Chemical Society.
16
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

(Niyomdecha et al., 2017). This 3-APBA on a polytyramine (Pty) coated significantly when a combination of optical, mechanical, and electrical
gold electrode enables reusability of the device and thus multiple modalities was integrated with microfluidics. Different microfluidic
bacteria were tested with the same device after washing with an acidic technologies such as gradient, droplet, and multiplexing enables testing
buffer. Under the influence of antibiotics binding of the number of with a wide variety of bacterial species and antibiotics in minimum
bacteria to the 3-APBA was affected thereby changing the total capa- time. These observations point to a trend that a multi-parametric sen-
citance measured. This reusable sensing device produced AST results in sing modality combining existing and emerging technologies will help
2.5 h and was reused for 35 times successfully thus making it a potential improve the robustness of the systems. Emerging approaches such as
candidate for developing a future diagnostic tool. In 2012, Puttaswamy data mining and machine learning combined with critical automation
et al. have developed a microfluidic channel with electrodes for AST will hold the key for the next generation of AST systems. Considering
applications (Puttaswamy et al., 2012). By measuring the capacitance the increasing trend of susceptible bacteria gaining resistance to con-
of the suspension, the viability of bacterial cells under the influence of ventional treatment, there will be added value in technologies that can
antibiotics was tested. In this case, the impedance of the bacterial clearly delineate between resistant, susceptible and persistent bacterial
suspension was measured at 500 different frequencies to determine sample. This will pave the way for a more personalized treatment which
charge-polarization of living and dead cell membranes. The highlight of will provide the best prognostic outcome.
this work was that it could determine MIC in 4 h and provide in- However, from a translational point of view, the demonstration of
formation about the antibiotic influenced bactericidal or bacteriostatic the clinical value of these technologies is a critical aspect that is lacking
activity. One major limitation of this method was that the loading of in the literature. Except for the classical AST techniques and semi-au-
suspension for each measurement makes it labour-intensive. Recently, tomated systems, none of the emerging technologies discussed have
Safavieh et al. have used microfluidics-based impedance analyser to test gone through clearances by international bodies such as CLSI and
AST on E. coli and MRSA against gentamicin, methicillin, ampicillin, EUCAST. Only a handful of studies have reported measurements from
ciprofloxacin, erythromycin, and daptomycin antibiotics (Safavieh clinical patient samples or direct improvements in patient outcomes
et al., 2017). In this study, silver electrodes for impedance measure- using their reported technology. While the scientific potential of these
ment and carbon microheater for temperature stabilization were fab- emerging technologies is immense and open up new and innovative
ricated on a low-cost plastic substrate as shown in Fig. 11(II). A closed ways of handling an important healthcare challenge, their adaptability
microchannel was formed by attaching an engraved poly(methyl me- into the existing clinical workflow and cost-effectiveness in comparison
thacrylate) (PMMA) sheets on the plastic substrate that carried the with gold standard techniques needs to be explored. The trend towards
channel design. This microchip was functionalized with Anti-MRSA developing low-cost devices and point-of-care platforms is a positive
antibody to capture MRSA from the blood sample. The electrical im- step in this direction.
pedance of the bacterial suspension was monitored at frequency
1 kHz at 1V in the presence and absence of antibiotics in both functio-
Conflict of interest
nalized as well as non-functionalized conditions to test the efficacy of
the chip for affinity-based and antibody-capture based sensing. This
No conflict of interest exists.
device provided AST results in less than 90 min thus making it a po-
tential point-of-care technology for diagnosing patients suffering from
UTI. Recently, Zhang et al. have developed a unique AST technique Funding
based on a multichannel capacitively coupled contactless conductivity
detector (C4D) which the team termed as electrical bacterial growth No funding was received for this work.
sensor (EBGS) (Zhang et al., 2018). In C4D, the contactless conductivity
of the suspension was measured, which was proportional to the con- CRediT authorship contribution statement
centration and mobility of the ionic charge carriers in the suspension.
This EBGS system measured the conductivity of E. coli suspensions at Bhagaban Behera: Data curation, Methodology, Writing - original
the excitation frequency of 2.0 MHz at 16 V amplitude. The con- draft. G.K. Anil Vishnu: Writing - original draft. Suman Chatterjee:
ductivity of bacterial solutions in the presence of chloramphenicol, Writing - original draft. V.S.N. Sitaramgupta V: Writing - original
penicillin G, and malachite green was monitored over incubation time. draft. Niranjana Sreekumar: Writing - original draft. Apoorva
This sensor was also monitored online. In addition to the inherent ad- Nagabhushan: Writing - original draft. Nirmala Rajendran: Writing -
vantages of electrochemical sensing techniques, this method provided original draft. B.H. Prathik: Writing - original draft. Hardik J.
more utilities such as freedom from polarization and elimination of Pandya: Data curation, Methodology, Writing - original draft.
passivation risks.

Acknowledgements
4. Conclusion

Bhagaban Behera acknowledges the Science and Engineering

The engineering technologies for AST discussed in this article will
Research Board of Government of India for providing financial assis-
help in reducing sample volumes, testing duration, improve sensitivity
tance (National Postdoctoral Fellowship, File number: PDF/2017/
up to the level of a single bacterial cell, and potentially improve the
000644). Hardik J. Pandya acknowledges Indian Institute of Science,
portability of the overall system. Some of the emerging technologies
Bangalore for the start-up grant to establish the research and compu-
such as mechanical, magnetic, and optical sensing-based methods are
tational facilities at the Department of Electronic Systems Engineering.
highly sensitive but are limited by the complexity and cost of the
Hardik J. Pandya also acknowledges SERB (ECR/2017/001043) and
system. With advanced optical imaging techniques, researchers have
Rajiv Gandhi University of Health Sciences (RGHUS)-Indian Institute of
studied AST on single bacterial cells. These bacteria cells were captured
Science (IISc) collaborative project (RGUH0013) for providing financial
by physical confinement, dielectrophoresis, and/or agarose gel. On the
assistance.
other hand, different mechanical methods such as asynchronous mag-
netic bead rotation and cantilever fluctuation have been used to study
AST on bacterial colonies as well as single cells. The microfluidics-based Appendix A. Supplementary data
technologies provide several advantages as it require extremely low
sample volumes and provide high throughput enabling parallel testing Supplementary data to this article can be found online at https://
against multiple antibiotics. The efficiency of AST profiling improved doi.org/10.1016/j.bios.2019.111552.

17
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

References Ghannoum, M.A., 1999. J. Clin. Microbiol. 37, 591–595.

Etayash, H., Khan, M.F., Kaur, K., Thundat, T., 2016. Nat. Commun. 7, 12947.
Faria-Ramos, I., Espinar, M.J., Rocha, R., Santos-Antunes, J., Rodrigues, A.G., Cantón, R.,
Abdulhalim, I., Zourob, M., Lakhtakia, A., 2008. Electromagnetics 28, 214–242. Pina-Vaz, C., 2013. Clin. Microbiol. Infect. Off. Publ. Eur. Soc. Clin. Microbiol. Infect.
Adan, A., Alizada, G., Kiraz, Y., Baran, Y., Nalbant, A., 2017. Crit. Rev. Biotechnol. 37, Dis. 19, E8–E15.
163–176. Fernandes, J.C., Eaton, P., Gomes, A.M., Pintado, M.E., Xavier Malcata, F., 2009.
Aghayee, S., Benadiba, C., Notz, J., Kasas, S., Dietler, G., Longo, G., 2013. J. Mol. Ultramicroscopy 109, 854–860.
Recognit. JMR 26, 590–595. Firdous, S., Anwar, S., Rafya, R., 2018. Laser Phys. Lett. 15, 065602.
Alexander, B.D., Byrne, T.C., Smith, K.L., Hanson, K.E., Anstrom, K.J., Perfect, J.R., France, D., Johnson, W., Cordell, W., Walls, F., 2016. Biophys. J. 110 200a.
Reller, L.B., 2007. J. Clin. Microbiol. 45, 698–706. Friedman, A., Blecher, K., Sanchez, D., Tuckman-Vernon, C., Gialanella, P., Friedman,
Athamneh, A.I.M., Alajlouni, R.A., Wallace, R.S., Seleem, M.N., Senger, R.S., 2014. J.M., Martinez, L.R., Nosanchuk, J.D., 2011. Virulence 2, 217–221.
Antimicrob. Agents Chemother. 58, 1302–1314. Frígols, B., Martí, M., Salesa, B., Hernández-Oliver, C., Aarstad, O., Ulset, A.-S.T., Sӕtrom,
Avesar, J., Rosenfeld, D., Truman-Rosentsvit, M., Ben-Arye, T., Geffen, Y., Bercovici, M., G.I., Aachmann, F.L., Serrano-Aroca, Á., 2019. PLoS One 14, e0212819.
Levenberg, S., 2017. Proc. Natl. Acad. Sci. 114, E5787–E5795. Funfak, A., Cao, J., Wolfbeis, O.S., Martin, K., Köhler, J.M., 2009. Microchim. Acta 164,
Bai, H., Wang, R., Hargis, B., Lu, H., Li, Y., 2012. Sensors 12, 12506–12518. 279–286.
Baker, C.N., Stocker, S.A., Culver, D.H., Thornsberry, C., 1991. J. Clin. Microbiol. 29, Galvan, D.D., Yu, Q., 2018. Adv. Healthc. Mater. 7, 1701335.
533–538. Gauthier, C., St-Pierre, Y., Villemur, R., 2002. J. Med. Microbiol. 51, 192–200.
Balouiri, M., Sadiki, M., Ibnsouda, S.K., 2016a. J. Pharm. Anal. 6, 71–79. Gfeller, K.Y., Nugaeva, N., Hegner, M., 2005. Appl. Environ. Microbiol. 71, 2626–2631.
Balouiri, M., Sadiki, M., Ibnsouda, S.K., 2016b. J. Pharm. Anal. 6, 71–79. Godin, M., Delgado, F.F., Son, S., Grover, W.H., Bryan, A.K., Tzur, A., Jorgensen, P.,
Baltekin, Ö., Boucharin, A., Tano, E., Andersson, D.I., Elf, J., 2017. Proc. Natl. Acad. Sci. Payer, K., Grossman, A.D., Kirschner, M.W., Manalis, S.R., 2010. Nat. Methods 7,
114, 9170–9175. 387–390.
Baniukevic, J., Hakki Boyaci, I., Goktug Bozkurt, A., Tamer, U., Ramanavicius, A., Gupta, P., Khare, V., Kumar, D., Ahmad, A., Banerjee, G., Singh, M., 2015. J. Clin. Diagn.
Ramanaviciene, A., 2013. Biosens. Bioelectron. 43, 281–288. Res. JCDR 9, DC04–DC07.
Baraban, L., Bertholle, F., Salverda, M.L.M., Bremond, N., Panizza, P., Baudry, J., de Hadjigeorgiou, K., Kastanos, E., Pitris, C., 2016. IEEE Healthcare Innovation Point-Of-
Visser, J.A.G.M., Bibette, J., 2011. Lab Chip 11, 4057–4062. Care Technologies Conference. HI-POCT, pp. 158–161.
Bauer, A.W., Kirby, W.M., Sherris, J.C., Turck, M., 1966. Am. J. Clin. Pathol. 45, 493–496. He, J., Mu, X., Guo, Z., Hao, H., Zhang, C., Zhao, Z., Wang, Q., 2014. Eur. J. Clin.
Bauer, A.W., Perry, D.M., Kirby, W.M.M., 1959. AMA Arch. Intern. Med. 104, 208–216. Microbiol. Infect. Dis. 33, 2223–2230.
Belkum, A. van, Dunne, W.M., 2013. J. Clin. Microbiol. 51, 2018–2024. Heesemann, J., 1993. Infection 21 (Suppl. 1), S4–S9.
Berghaus, L.J., Giguère, S., Guldbech, K., Warner, E., Ugorji, U., Berghaus, R.D., 2015. J. Hou, Z., An, Y., Hjort, Karin, Hjort, Klas, Sandegren, L., Wu, Z., 2014a. Lab Chip 14,
Clin. Microbiol. 53, 314–318. 3409–3418.
Besant, J.D., Sargent, E.H., Kelley, S.O., 2015. Lab Chip 15, 2799–2807. Hou, Z., An, Y., Hjort, Karin, Hjort, Klas, Sandegren, L., Wu, Z., 2014b. Lab Chip 14,
Bhalla, N., Jolly, P., Formisano, N., Estrela, P., 2016. 60, 1-8. 3409–3418.
Bonkat, G., Solokhina, A., Rieken, M., Müller, G., Wyler, S., Gasser, T., Bachmann, A., Howell, M., Wirz, D., Daniels, A.U., Braissant, O., 2012. J. Clin. Microbiol. 50, 16–20.
2013. Eur. Urol. Suppl. 12, e167. Huang, T.-H., Ning, X., Wang, X., Murthy, N., Tzeng, Y.-L., Dickson, R.M., 2015. Anal.
Braff, W.A., Willner, D., Hugenholtz, P., Rabaey, K., Buie, C.R., 2013. PLoS One 8, Chem. 87, 1941–1949.
e76751. J. Cira, N., Y. Ho, J., E. Dueck, M., B. Weibel, D., 2012. Lab Chip 12, 1052–1059.
Braissant, O., Wirz, D., Göpfert, B., Daniels, A.U., 2009. FEMS Microbiol. Lett. 303, 1–8. Jepras, R.I., Paul, F.E., Pearson, S.C., Wilkinson, M.J., 1997. Antimicrob. Agents
Campbell, J., McBeth, C., Kalashnikov, M., Boardman, A.K., Sharon, A., Sauer-Budge, Chemother. 41, 2001–2005.
A.F., 2016. Biomed. Microdevices 18, 103. Jiang, L., Boitard, L., Broyer, P., Chareire, A.-C., Bourne-Branchu, P., Mahé, P., Tournoud,
Cermak, N., Olcum, S., Delgado, F.F., Wasserman, S.C., Payer, K.R., Murakami, M.A., M., Franceschi, C., Zambardi, G., Baudry, J., Bibette, J., 2016. Eur. J. Clin. Microbiol.
Knudsen, S.M., Kimmerling, R.J., Stevens, M.M., Kikuchi, Y., Sandikci, A., Ogawa, M., Infect. Dis. 35, 415–422.
Agache, V., Baléras, F., Weinstock, D.M., Manalis, S.R., 2016. Nat. Biotechnol. 34, Jo, N., Kim, B., Lee, S.-M., Oh, J., Park, I.H., Jin Lim, K., Shin, J.-S., Yoo, K.-H., 2018.
1052–1059. Biosens. Bioelectron. 102, 164–170.
Chabot, V., Miron, Y., Charette, P.G., Grandbois, M., 2013. Biosens. Bioelectron. 50, Johannessen, E.A., Weaver, J.M.R., Cobbold, P.H., Cooper, J.M., 2002. IEEE Trans.
125–131. NanoBioscience 1, 29–36.
Chaturvedi, V., Ramani, R., Pfaller, M.A., 2004. J. Clin. Microbiol. 42, 2249–2251. Johari, J., Hübner, Y., Hull, J.C., Dale, J.W., Hughes, M.P., 2003. Phys. Med. Biol. 48
Chen, C.H., Lu, Y., Sin, M.L.Y., Mach, K.E., Zhang, D.D., Gau, V., Liao, J.C., Wong, P.K., N193.
2010. Anal. Chem. 82, 1012. Johnson, W.L., France, D.C., Rentz, N.S., Cordell, W.T., Walls, F.L., 2017. Sci. Rep. 7,
Chiang, Y.-L., Lin, C.-H., Yen, M.-Y., Su, Y.-D., Chen, S.-J., Chen, H.-F., 2009. Biosens. 12138.
Bioelectron. 24, 1905–1910. Jorgensen, J.H., Ferraro, M.J., 2009. Clin. Infect. Dis. Off. Publ. Infect. Dis. Soc. Am. 49,
Choi, J., Jeong, H.Y., Lee, G.Y., Han, Sangkwon, Han, Shinhun, Jin, B., Lim, T., Kim, S., 1749–1755.
Kim, D.Y., Kim, H.C., Kim, E.-C., Song, S.H., Kim, T.S., Kwon, S., 2017. Sci. Rep. 7, Kalashnikov, M., Campbell, J., Lee, J.C., Sharon, A., Sauer-Budge, A.F., 2014. JoVE J. Vis.
1148. Exp, e50828.
Choi, J., Jung, Y.-G., Kim, J., Kim, S., Jung, Y., Na, H., Kwon, S., 2013. Lab Chip 13, Kalashnikov, M., Lee, J.C., Campbell, J., Sharon, A., Sauer-Budge, A.F., 2012. Lab Chip
280–287. 12, 4523–4532.
Choi, J., Yoo, J., Kim, K., Kim, E.-G., Park, K.O., Kim, Hyejin, Kim, Haeun, Jung, H., Kim, Kalashnikov, M., Lee, J.C., Sauer-Budge, A.F., 2018. Diagnostics 8, 24.
T., Choi, M., Kim, H.C., Ryoo, S., Jung, Y.-G., Kwon, S., 2016. Appl. Microbiol. Kalashnikov, M., Mueller, M., McBeth, C., Lee, J.C., Campbell, J., Sharon, A., Sauer-
Biotechnol. 100, 2355–2365. Budge, A.F., 2017. Sci. Rep. 7, 8031.
Choi, J., Yoo, J., Lee, M., Kim, E.-G., Lee, J.S., Lee, S., Joo, S., Song, S.H., Kim, E.-C., Lee, Kaminski, T.S., Scheler, O., Garstecki, P., 2016. Lab Chip 16, 2168–2187.
J.C., Kim, H.C., Jung, Y.-G., Kwon, S., 2014. Sci. Transl. Med. 6 267ra174-267ra174. Kara, V., Duan, C., Gupta, K., Kurosawa, S., J. Stearns-Kurosawa, D., L. Ekinci, K., 2018.
Chotinantakul, K., Suginta, W., Schulte, A., 2014. Anal. Chem. 86, 10315–10322. Lab Chip 18, 743–753.
Chung, C.-C., Cheng, I.-F., Chen, H.-M., Kan, H.-C., Yang, W.-H., Chang, H.-C., 2012. Anal. Kasas, S., Ruggeri, F.S., Benadiba, C., Maillard, C., Stupar, P., Tournu, H., Dietler, G.,
Chem. 84, 3347–3354. Longo, G., 2015. Proc. Natl. Acad. Sci. U. S. A. 112, 378–381.
Chung, C.-C., Cheng, I.-F., Yang, W.-H., Chang, H.-C., 2011. Biomicrofluidics 5, 021102. Kastanos, E.K., Kyriakides, A., Hadjigeorgiou, K., Pitris, C., 2010. J. Raman Spectrosc. 41,
Chung, C.-Y., Wang, J.-C., Chuang, H.-S., 2016. PLoS One 11. 958–963.
Churski, K., S. Kaminski, T., Jakiela, S., Kamysz, W., Baranska-Rybak, W., B. Weibel, D., Kaushik, A.M., Hsieh, K., Chen, L., Shin, D.J., Liao, J.C., Wang, T.-H., 2017. Biosens.
Garstecki, P., 2012. Lab Chip 12, 1629–1637. Bioelectron. 97, 260–266.
Citron, D.M., Ostovari, M.I., Karlsson, A., Goldstein, E.J., 1991. J. Clin. Microbiol. 29, Kee, J.S., Lim, S.Y., Perera, A.P., Zhang, Y., Park, M.K., 2013. Sens. Actuators B Chem.
2197–2203. 182, 576–583.
Connell, J.L., Ritschdorff, E.T., Whiteley, M., Shear, J.B., 2013. Proc. Natl. Acad. Sci. 110, Khatoon, U.T., Nageswara Rao, G.V.S., Mohan, K.M., Ramanaviciene, A., Ramanavicius,
18380–18385. A., 2017. Vacuum 146, 259–265.
Cuenca-Estrella, M., Gomez-Lopez, A., Alastruey-Izquierdo, A., Bernal-Martinez, L., Kim, S.C., Cestellos-Blanco, S., Inoue, K., Zare, R.N., 2015. Antibiotics 4, 455–466.
Cuesta, I., Buitrago, M.J., Rodriguez-Tudela, J.L., 2010. J. Clin. Microbiol. 48, Kinnunen, P., Carey, M.E., Craig, E., Brahmasandra, S.N., McNaughton, B.H., 2014. Sens.
1782–1786. Actuators B Chem. 190, 265–269.
Dai, J., Hamon, M., Jambovane, S., 2016. Bioengineering 3, 1–13. Kinnunen, P., McNaughton, B.H., Albertson, T., Sinn, I., Mofakham, S., Elbez, R., Newton,
Dai, J., Yoon, S.H., Sim, H.Y., Yang, Y.S., Oh, T.K., Kim, J.F., Hong, J.W., 2013. Anal. D.W., Hunt, A., Kopelman, R., 2012. Small 8, 2477–2482.
Chem. 85, 5892–5899. Kinnunen, P., Sinn, I., McNaughton, B.H., Newton, D.W., Burns, M.A., Kopelman, R.,
Derzsi, L., S. Kaminski, T., Garstecki, P., 2016. Lab Chip 16, 893–901. 2011. Biosens. Bioelectron. 26, 2751–2755.
Du, W.-L., Xu, Y.-L., Xu, Z.-R., Fan, C.-L., 2008. Nanotechnology 19, 085707. Knudsen, S.M., von Muhlen, M.G., Schauer, D.B., Manalis, S.R., 2009. Anal. Chem. 81,
Eaton, P., Fernandes, J.C., Pereira, E., Pintado, M.E., Xavier Malcata, F., 2008. 7087–7090.
Ultramicroscopy 108, 1128–1134. Kong, K.-F., Schneper, L., Mathee, K., 2010. APMIS Acta Pathol. Microbiol. Immunol.
Eigner, U., Schmid, A., Wild, U., Bertsch, D., Fahr, A.-M., 2005. J. Clin. Microbiol. 43, Scand. 118, 1–36.
3829–3834. Lang, H.P., Gerber, C., 2008. Microcantilever sensors. In: STM and AFM Studies on (Bio)
Ertl, P., Robello, E., Battaglini, F., Mikkelsen, S.R., 2000. Anal. Chem. 72, 4957–4964. Molecular Systems: Unravelling the Nanoworld, Topics in Current Chemistry.
Ertl, P., Wagner, M., Corton, E., Mikkelsen, S.R., 2003. Biosens. Bioelectron. 18, 907–916. Springer, Berlin, Heidelberg, pp. 1–27.
Espinel-Ingroff, A., Pfaller, M., Messer, S.A., Knapp, C.C., Killian, S., Norris, H.A., Lee, W., Fon, W., Axelrod, B.W., Roukes, M.L., 2009. Proc. Natl. Acad. Sci. 106,

18
B. Behera, et al. Biosensors and Bioelectronics 142 (2019) 111552

15225–15230. Bioelectron. 4, 1–6.

Lee, W., You, M., Wei, P., Chung, P., Lee, K., Tian, W., 2015. IEEE Sensors 1–4. Q. Boedicker, J., Li, L., R. Kline, T., F. Ismagilov, R., 2008. Lab Chip 8, 1265–1272.
Leester-Schädel, M., Lorenz, T., Jürgens, F., Richter, C., 2016. Springer, Cham, pp. 23–57. Ramani, R., Ramani, A., Wong, S.J., 1997. J. Clin. Microbiol. 35, 2320–2324.
Leonard, H., Halachmi, S., Ben-Dov, N., Nativ, O., Segal, E., 2017. ACS Nano 11, Ran, M., Wang, Y., Wang, S., Luo, C., 2015. Biomed. Microdevices 17, 67.
6167–6177. Read, A.F., Woods, R.J., 2014. Evol. Med. Public Health 147.
Lewis, G., Daniels, A.U. (Dan), 2003. J. Biomed. Mater. Res. B Appl. Biomater. 66B, Reller, L.B., Weinstein, M., Jorgensen, J.H., Ferraro, M.J., 2009a. Clin. Infect. Dis. 49,
487–501. 1749–1755.
Li, B., Qiu, Y., Glidle, A., McIlvenna, D., Luo, Q., Cooper, J., Shi, H.-C., Yin, H., 2014. Reller, L.B., Weinstein, M., Jorgensen, J.H., Ferraro, M.J., 2009b. Clin. Infect. Dis. 49,
Anal. Chem. 86, 3131–3137. 1749–1755.
Lissandrello, C., Inci, F., Francom, M., Paul, M.R., Demirci, U., Ekinci, K.L., 2014. Appl. Reller, L.B., Weinstein, M., Jorgensen, J.H., Ferraro, M.J., 2009c. Clin. Infect. Dis. 49,
Phys. Lett. 105. 1749–1755.
Liu, C.-Y., Han, Y.-Y., Shih, P.-H., Lian, W.-N., Wang, H.-H., Lin, C.-H., Hsueh, P.-R., Rowat, A.C., Bird, J.C., Agresti, J.J., Rando, O.J., Weitz, D.A., 2009. Proc. Natl. Acad. Sci.
Wang, J.-K., Wang, Y.-L., 2016. Sci. Rep. 6, 23375. U. S. A. 106, 18149–18154.
Liu, T.-T., Lin, Y.-H., Hung, C.-S., Liu, T.-J., Chen, Y., Huang, Y.-C., Tsai, T.-H., Wang, H.- Sabhachandani, P., Sarkar, S., Zucchi, P.C., Whitfield, B.A., Kirby, J.E., Hirsch, E.B.,
H., Wang, D.-W., Wang, J.-K., Wang, Y.-L., Lin, C.-H., 2009. PLoS One 4, e5470. Konry, T., 2017. Microchim. Acta 184, 4619–4628.
Liu, Z., Sun, H., Ren, K., 2017. ChemPlusChem 82, 792–801. Safavieh, M., Pandya, H.J., Venkataraman, M., Thirumalaraju, P., Kanakasabapathy,
Long, Z., Nugent, E., Javer, A., Cicuta, P., Sclavi, B., Cosentino Lagomarsino, M., M.K., Singh, A., Prabhakar, D., Chug, M.K., Shafiee, H., 2017. ACS Appl. Mater.
Dorfman, K.D., 2013. Lab Chip 13, 947–954. Interfaces 9, 12832–12840.
Longo, G., Alonso-Sarduy, L., Rio, L.M., Bizzini, A., Trampuz, A., Notz, J., Dietler, G., Saint-Ruf, C., Crussard, S., Franceschi, C., Orenga, S., Ouattara, J., Ramjeet, M., Surre, J.,
Kasas, S., 2013. Nat. Nanotechnol. 8, 522–526. Matic, I., 2016. Front. Microbiol. 7.
Longo, G., Kasas, S., 2014. Wiley Interdiscip. Rev. Nanomed. Nanobiotechnol. 6, Samadi, A., Zhang, C., Chen, J., Reihani, S.N.S., Chen, Z., 2015. Biomed. Opt. Express 6,
230–244. 112–117.
Lu, Y., Gao, J., Zhang, D.D., Gau, V., Liao, J.C., Wong, P.K., 2013. Anal. Chem. 85, Saravanan, M., Ramachandran, B., Barabadi, H., 2018. Microb. Pathog. 114, 180–192.
3971–3976. Schofield, C.B., 2012. Am. Soc. Clin. Lab. Sci. 25, 233–239.
Luo, X., J. Davis, J., 2013. Chem. Soc. Rev. 42, 5944–5962. Sinn, I., Albertson, T., Kinnunen, P., Breslauer, D.N., McNaughton, B.H., Burns, M.A.,
Ma, F., Rehman, A., Sims, M., Zeng, X., 2015. Anal. Chem. 87, 4385–4393. Kopelman, R., 2012. Anal. Chem. 84, 5250–5256.
Malmberg, C., Yuen, P., Spaak, J., Cars, O., Tängdén, T., Lagerbäck, P., 2016. PLoS One Sinn, I., Kinnunen, P., Albertson, T., McNaughton, B.H., Newton, D.W., Burns, M.A.,
11, e0167356. Kopelman, R., 2011. Lab Chip 11, 2604–2611.
Mann, T.S., Mikkelsen, S.R., 2008. Anal. Chem. 80, 843–848. Snyder, J.W., Munier, G.K., Johnson, C.L., 2008. J. Clin. Microbiol. 46, 2327–2333.
Martinez-Perdiguero, J., Retolaza, A., Otaduy, D., Juarros, A., Merino, S., 2013. Sensors Soltani, R., Ehsanpoor, M., Khorvash, F., Shokri, D., 2014. J. Res. Pharm. Pract. 3, 6.
13, 13960–13968. Su, I.-H., Ko, W.-C., Shih, C.-H., Yeh, F.-H., Sun, Y.-N., Chen, J.-C., Chen, P.-L., Chang, H.-
Matsumoto, Y., Sakakihara, S., Grushnikov, A., Kikuchi, K., Noji, H., Yamaguchi, A., Iino, C., 2017. Anal. Chem. 89, 4635–4641.
R., Yagi, Y., Nishino, K., 2016. PLoS One 11. Sun, H., Liu, Z., Hu, C., Ren, K., 2016. Lab Chip 16, 3130–3138.
Mehrotra, P., 2016. J. Oral Biol. Craniofacial Res. 6, 153–159. Syal, K., Iriya, R., Yang, Y., Yu, H., Wang, S., Haydel, S.E., Chen, H.-Y., Tao, N., 2016. ACS
Mittman, S.A., Huard, R.C., Della-Latta, P., Whittier, S., 2009. J. Clin. Microbiol. 47, Nano 10, 845–852.
3557–3561. Syal, K., Mo, M., Yu, H., Iriya, R., Jing, W., Guodong, S., Wang, S., Grys, T.E., Haydel, S.E.,
Mohan, R., Mukherjee, A., Sevgen, S.E., Sanpitakseree, C., Lee, J., Schroeder, C.M., Kenis, Tao, N., 2017a. Theranostics 7, 1795–1805.
P.J.A., 2013. Biosens. Bioelectron. 49, 118–125. Syal, K., Shen, S., Yang, Y., Wang, S., Haydel, S.E., Tao, N., 2017b. ACS Sens. 2,
Mohan, R., Sanpitakseree, C., V. Desai, A., E. Sevgen, S., M. Schroeder, C., A. Kenis, P.J., 1231–1239.
2015. RSC Adv. 5, 35211–35223. Tafin, U.F., Clauss, M., Hauser, P.M., Bille, J., Meis, J.F., Trampuz, A., 2012. Clin.
Moritz, T.J., Polage, C.R., Taylor, D.S., Krol, D.M., Lane, S.M., Chan, J.W., 2010a. J. Clin. Microbiol. Infect. 18, E241–E245.
Microbiol. 48, 4287–4290. Takagi, R., Fukuda, J., Nagata, K., Yawata, Y., Nomura, N., Suzuki, H., 2013. Analyst 138,
Moritz, T.J., Taylor, D.S., Polage, C.R., Krol, D.M., Lane, S.M., Chan, J.W., 2010b. Anal. 1000–1003.
Chem. 82, 2703–2710. Tan, S., Tatsumura, Y., 2015. Alexander Fleming (1881–1955): discoverer of penicillin.
Munita, J.M., Arias, C.A., 2016. Micro. Spectr. 4, 1–24. Singap. Med. J. 56, 366–367.
Nath, N., Chilkoti, A., 2004. Anal. Chem. 76, 5370–5378. Tang, M., Huang, X., Chu, Q., Ning, X., Wang, Y., Kong, S.-K., Zhang, X., Wang, G., Ho, H.-
Nawattanapaiboon, K., Kiatpathomchai, W., Santanirand, P., Vongsakulyanon, A., P., 2018. Lab Chip 18, 1452–1460.
Amarit, R., Somboonkaew, A., Sutapun, B., Srikhirin, T., 2015. Biosens. Bioelectron. Tang, Y., Zhen, L., Liu, J., Wu, J., 2013. Anal. Chem. 85, 2787–2794.
74, 335–340. Tawil, N., Mouawad, F., Lévesque, S., Sacher, E., Mandeville, R., Meunier, M., 2013.
Nazemi, E., Hassen, W.M., Frost, E.H., Dubowski, J.J., 2017. Biosens. Bioelectron. 93, Biosens. Bioelectron. 49, 334–340.
234–240. Tracy, B.P., Gaida, S.M., Papoutsakis, E.T., 2010. Curr. Opin. Biotechnol. Anal.
Nemeth, J., Oesch, G., Kuster, S.P., 2015. J. Antimicrob. Chemother. 70, 382–395. Biotechnol. 21, 85–99.
Niyomdecha, S., Limbut, W., Numnuam, A., Asawatreratanakul, P., Kanatharana, P., Turner, A.P.F., 2013. Chem. Soc. Rev. 42, 3184–3196.
Thavarungkul, P., 2017. Biosens. Bioelectron. 96, 84–88. van Belkum, A., Dunne, W.M., 2013. J. Clin. Microbiol. 51, 2018–2024.
Nseir, S., Povoa, P., 2015. Intensive Care Med. 41, 531–533. van Herwaarden, A.W., 2005. In: 8th Lähnwitzseminar on Calorimetry, vol 432. pp.
Onishi, K., Enomoto, J., Araki, T., Takagi, R., Suzuki, H., Fukuda, J., 2018. Analyst 143, 192–201.
396–399. Vigneshvar, S., Sudhakumari, C.C., Senthilkumaran, B., Prakash, H., 2016. Bioeng.
Peitz, I., van Leeuwen, R., 2010. Lab Chip 10, 2944–2951. Biotechnol. 4.
Pilát, Z., Bernatová, S., Ježek, J., Kirchhoff, J., Tannert, A., Neugebauer, U., Samek, O., von Ah, U., Wirz, D., Daniels, A.U., 2009. BMC Microbiol. 9, 106.
Zemánek, P., 2018. Sensors 18. Walter, A., Reinicke, M., Bocklitz, T., Schumacher, W., Rösch, P., Kothe, E., Popp, J.,
Piliarik, M., Vaisocherová, H., Homola, J., Rasooly, A., Herold, K.E. (Eds.), 2009. 2011. Anal. Bioanal. Chem. 400, 2763–2773.
Biosensors and Biodetection, Methods in Molecular BiologyTM. Humana Press, Weibull, E., Antypas, H., Kjäll, P., Brauner, A., Andersson-Svahn, H., Richter-Dahlfors, A.,
Totowa, NJ, pp. 65–88. 2014. J. Clin. Microbiol. 52, 3310–3317.
Pil Kim, K., Kim, Y.-G., Choi, C.-H., Kim, H.-E., Lee, S.-H., Chang, W.-S., Lee, C.-S., 2010. White, R.L., Burgess, D.S., Manduru, M., Bosso, J.A., 1996. Antimicrob. Agents
Lab Chip 10, 3296–3299. Chemother. 40, 1914–1918.
Premasiri, W.R., Chen, Y., Williamson, P.M., Bandarage, D.C., Pyles, C., Ziegler, L.D., Whitesides, G.M., 2006. Nature 442, 368–373.
2017. Anal. Bioanal. Chem. 409, 3043–3054. Yanase, Y., Hiragun, T., Ishii, K., Kawaguchi, T., Yanase, T., Kawai, M., Sakamoto, K.,
Puttaswamy, S., Gupta, S.K., Regunath, H., Smith, L.P., Sengupta, S., 2018. Arch. Clin. Hide, M., 2014. Sensors 14, 4948–4959.
Microbiol. 9. Zhang, X., Jiang, X., Yang, Q., Wang, X., Zhang, Y., Zhao, J., Qu, K., Zhao, C., 2018. Anal.
Puttaswamy, S., Lee, B.-D., Amighi, B., Chakraborty, S., Sengupta, S., 2012. J. Biosens. Chem. 90, 6006–6011.

19