Article pubs.acs.org/ac
Rapid Antibiotic Susceptibility Testing in a Microfluidic pH Sensor Yanyan Tang,† Li Zhen,† Jingqing Liu,‡ and Jianmin Wu*,† Institute of Microanalytical System, †Department of Chemistry, and ‡Industry Technology Research Institute, Zhejiang University, Hangzhou, 310058, China S Supporting Information *
ABSTRACT: For appropriate selection of antibiotics in the treatment of pathogen infection, rapid antibiotic susceptibility testing (AST) is urgently needed in clinical practice. This study reports the utilization of a microfluidic pH sensor for monitoring bacterial growth rate in culture media spiked with different kinds of antibiotics. The microfluidic pH sensor was fabricated by integration of pH-sensitive chitosan hydrogel with poly(dimethylsiloxane) (PDMS) microfluidic channels. For facilitating the reflectometric interference spectroscopic measurements, the chitosan hydrogel was coated on an electrochemically etched porous silicon chip, which was used as the substrate of the microfluidic channel. Real-time observation of the pH change in the microchannel can be realized by Fourier transform reflectometric interference spectroscopy (FT-RIFS), in which the effective optical thickness (EOT) was selected as the optical signal for indicating the reversible swelling process of chitosan hydrogel stimulated by pH change. With this microfluidic pH sensor, we demonstrate that confinement of bacterial cells in a nanoliter size channel allows rapid accumulation of metabolic products and eliminates the need for long-time preincubation, thus reducing the whole detection time. On the basis of this technology, the whole bacterial growth curve can be obtained in less than 2 h, and consequently rapid AST can be realized. Compared with conventional methods, the AST data acquired from the bacterial growth curve can provide more detailed information for studying the antimicrobial behavior of antibiotics during different stages. Furthermore, the new technology also provides a convenient method for rapid minimal inhibition concentration (MIC) determination of individual antibiotics or the combinations of antibiotics against human pathogens that will find application in clinical and point-of-care medicine.
P
infecting bacterial species and lack the ability to directly test the live bacteria’s function, such as susceptibility to particular antibiotics. Usually, AST data can be more accurately determined by a functional-based assay, especially for bacterial strains with unknown resistance mechanisms. Microfluidics is an attractive platform for working with cell culture10−12 and recently has become an attractive tool in microbiology study partly because of its ability to control the microbial culture in a very small volume.13−18 In the meantime, it has been reported that the growth of microbes in a microfluidic channel is obviously faster than that in bulk solution.17,19 Therefore, the determination of bacteria growth rate in a microfluidic chip is expected to be faster than the conventional methods. Among various approaches for bacteria detection, monitoring the metabolic products is one of the appropriate methods that could be applied in the bacterial functional assay.19,20 For example, it has been well-known that the growth medium becomes acidified owing to the accumulation of organic acids, which are the metabolic products of glucose or sugar during bacteria culture. Therefore, monitoring the pH change in culture medium has become one of the commonly used methods for bacterial functional assay. In
athogenic bacteria always cause infections such as tetanus, typhoid fever, diphtheria, syphilis, and leprosy.1,2 Although bacterial infections may be treated with antibiotics, these infections still cause a major health problem in a developing country, leading to a large amount of deaths annually around the world.3 In addition, research has shown that antibiotic resistance, in general, leads to additional cost, length of stay, morbidity, and mortality, presumably as a result of inappropriate therapy.4,5 Antimicrobial susceptibility testing (AST) is often performed to determine the antibiotic sensitivity of bacterial pathogens in clinical samples such as urine, blood, sputum, or wound swabs.6 However, the current AST method in clinical practice usually requires growth of bacteria into colonies in culture media spiked with antibiotic for more than 1 day, followed by a lengthy identification process involving morphological and biochemical testing or counting. The whole procedure will usually take about 2−4 days after the sample collection, resulting in the delay of patient treatment and high risk of severe infection. Therefore, rapid determination of antimicrobial susceptibility is urgently needed toward judicious management of infectious diseases in emergency situations. A fast AST process would be very helpful to significantly decrease the mortality rate and reduce the cost of treating patients with aggressive bacterial infections.7 Although polymerase chain reaction (PCR)-based methods enable detection within a few hours,8,9 these methods only provide a genetic profile of the © 2013 American Chemical Society
Received: November 15, 2012 Accepted: January 29, 2013 Published: January 29, 2013 2787
dx.doi.org/10.1021/ac303282j | Anal. Chem. 2013, 85, 2787−2794
Analytical Chemistry
Article
a bulk solution, the pH value can be easily measured by a conventional pH meter. Nevertheless, real-time pH measurement in a microfluidic channel with only micro/nanoliter volume is still a difficult task, mainly due to the lack of a pH probe in micro/nanometer scale. Previously, we have reported that swelling behavior of chitosan hydrogel film is very sensitive to the pH change in a near-neutral pH range.21 The hydrogel film was prepared by covalently cross-linking the chitosan with glycidoxypropyltrimethoxysilane (GPTMS) on the surface of a porous silicon (pSi) chip. The reflectance spectrum of the gelcoated pSi displays Fabry−Pérot interference fringes characteristic of a double layer. Monitoring the Fourier transform reflective interferometric spectroscopy (FT-RIFS) peak corresponding to the hydrogel layer allows direct, real-time observation of the reversible change in optical thickness upon cycling the pH of the tested solution.21 In this work, we integrated the pH sensing material with microfluidic channels in order to real-time monitor the pH change in a micro/ nanoliter liquid sample. It has been observed that the metabolic process of bacteria in the high surface-to-volume ratio of the microchannel was greatly accelerated probably due to the enhanced oxygen supply that occurred in the confined nanoliter-sized channel.17 Such confinement effect has been used for single-cell analysis in microfluidic devices.10−13,22 Consequently, a rapid accumulation of metabolic products will lead to a fast pH change in the growth media confined in microfluidic channel. With the help of FT-RIFS, the chemical signal can then be instantaneously transformed to an optical signal by the pH-sensitive hydrogel integrated in the microfluidic channel. We designate this system as a microfluidic pH sensor. Our results show that the system displays a good ability to reduce the time for monitoring the bacterial metabolic rate and consequently the growth rate. Furthermore, the microfluidics-based device also enables simultaneous execution of numerous assays in the same experiment, which is especially useful for rapid AST. We applied this technology to test the antibiotics susceptibility of model bacteria, Escherichia coli (E. coli), by measuring the bacterial inhibitory activity of several antibiotics. As a proof of concept, the minimal inhibition concentration (MIC) of tetracycline against E. coli was also evaluated by this technology. The technology established in this work offers two advantages over traditional bacterial detection and drug screening methods: (1) confinement of bacteria cells in a nanoliter size channel allows rapid accumulation of metabolic products and eliminates the need for long-time preincubation, thus reducing the detection time; (2) the whole bacterial growth curve can be obtained in less than 2 h, thus will provide more detailed information for studying the antimicrobial behavior of antibiotics. On the basis of this technology, rapid selection of appropriate antibiotics for the treatment of bacterial infection can be potentially realized. The technology will also provide a simple method for determining MIC of an individual antibiotic and combination of antibiotics against pathogens. That will find applications in clinical and point-of-care medicine.
Scheme 1. Schematic Diagrams Depicting the Assembly of the Microfluidic pH Sensor and Macrofluidic Flow Cella
a
(a) Procedure for assembly of microfluidic pH sensor. The test zone of etched pSi chip was defined by a PDMS film patterned with microchannels (length, ∼1.2 cm; width, ∼1 mm; depth, ∼250 μm) and microwells (diameter, ∼2 mm; depth, ∼250 μm). The total channel volume is calculated to be ∼0.8 μL. The chitosan pre-gel was quantitatively pipetted into each microwell and dried in room temperature. The microfluidic channels were finally sealed by a second PDMS film. (b) The schematic cross-sectional view of the microfluidic pH sensor. (c) The schematic view of the macrofluidic flow cell with chamber radius of 5.6 mm and depth of 2 mm.
(dimethylsiloxane) (PDMS) was chosen as the material for constructing the microchannels structure, since it has unique advantages, such as being transparent, gas-permeable, good biocompatibility, and easy to be sealed between two different substrates. The fabrication of the channel-patterned PDMS film is similar to commonly used methods,24 as described in the Supporting Information. For facilitating the FT-RIFS measurements, the porous silicon chip was used as the substrate of the microfluidic chip, and chitosan hydrogel for pH sensing was coated on the porous silicon. The detailed procedure for assembling the PDMS microfluidic channel with the optical sensing material is also provided in the Supporting Information. The cross-sectional view of the microfluidic sensing chip is depicted in Scheme 1b. The total volume of the microfluidic channel is around 0.8 μL. To compare the effect of channel size on the pH response, a macrofluidic flow cell with channel volume of ∼200 μL was also constructed as depicted in Scheme 1c. The porous SiO2 chip coated with hydrogel was placed in the flow cell. The inlet and outlet of the flow cell was connected to the chamber. Preparation of Bacterial Cell Culture and Suspension. E. coli was used as the model bacteria in this study. It was obtained by scraping the colony from an overnight-cultured sample in an LB agar plate. The E. coli suspension was prepared by diluting the bacterial cell with fresh Mueller−Hinton broth (MHB) with 1% glucose (w/v), and the cell density was adjusted to match the turbidity of a 0.5 McFarland standard (108 CFU/mL) using MHB (with 1% glucose). Then the suspension was further quantitatively diluted to reach an appropriate cell density with MHB (with 1% glucose).The bacterial suspension with different cell densities was used in subsequent experiment. Optical Measurement. The experimental setup for the optical measurements on the microfluidic-based sensing device is illustrated in Scheme 2a. The reflectance spectra of each test zone on the chip were acquired by an Ocean Optics USB2000+ spectrometer which was fitted to a microscopic lens via fiber optics. White light from a tungsten lamp (Ocean Optics) was
■
EXPERIMENTAL SECTION Fabrication of the Microfluidic pH Sensor. The porous silicon chip was prepared by an electrochemical etching method,21 which is briefly described in the Supporting Information. The whole procedure for the fabrication of the microfluidic pH sensor is illustrated in Scheme 1a. Poly2788
dx.doi.org/10.1021/ac303282j | Anal. Chem. 2013, 85, 2787−2794
Analytical Chemistry
Article
mechanism and representative result obtained by the microfluidic pH sensor are also illustrated in Scheme 2, parts b and c, respectively. The methods for measurement of pH response and bacterial metabolic rate in the microfluidic pH sensor are provided in the Supporting Information. Antibiotic Susceptibility Test with the Microfluidic pH Sensor. Each E. coli sample with initial cell density of 1 × 106 CFU/mL was cultured in glucose solution containing 1 μg/mL antibiotic. The bacterial suspensions spiked with different types of antibiotics (ciprofloxacin, tetracycline, azithromycin, amikacin) were introduced into the microfluidic pH sensor with a syringe pump. The temperature of the sensing device was kept at 37 °C. Nutrient media with and without E. coli were taken as positive and negative controls, respectively. The reflectance spectra were recorded at every 1 min, and the time course of the EOT value was obtained with IGOR software. In addition, the MIC of antibiotics was also measured by the same method as described above. Antibiotic solutions with different concentrations were added into E. coli suspensions with an initial cell density of 1 × 106 CFU/mL, respectively. Each sample was loaded into the microfluidic pH sensor, and the reflectance spectra were monitored in real time. For comparing the MIC data obtained in the microfluidic pH sensor with a conventional method, off-chip MIC measurements were performed as described in the Supporting Information.
Scheme 2. (a) Schematic Instrumental Setup for the Sample Detection; (b) Mechanism for Monitoring Bacterial Metabolism and Growth (the Growth of Bacteria Accelerates the Decrease of pH in the Microfluidic Channel and Increases the Swelling Ratio of pH-Sensitive Hydrogel); (c) Representative Fourier Transform Reflectometric Interference Spectra of the Chitosan−pSi Double-Layer Chipa
a
The highlighted peak in part c indicates an EOT shift before and after bacterial incubating.
■
fed through one arm of a bifurcated fiber-optic cable and focused through a lens onto the sensor chip at normal incidence. Reflected light was collected through the same optics, and data was analyzed with Spectrasuit software (Ocean Optics Inc.) as described previously.21,23 The EOT (effective optical thickness) was calculated with the Fabry−Pérot relationship:
RESULTS AND DISCUSSION Principle of the Optical pH Sensor. Reflectometric interference spectroscopy provides a convenient method to monitor the swelling behavior of the hydrogel in real time.21,25 Up to now, there are two types of hydrogel/porous silicon materials that have been used for pH sensing based on FTRIFS. One is the composite structure, in which the hydrogel was filled into the porous Si layer.25 The optical response of the composite structure to pH stimuli is attributed to the reversible swelling of hydrogel filled in pores of the porous Si that causes the change in the averaged refractive index of the porous layer. Another is termed a double-layer structure, in which the hydrogel layer was coated on the porous silicon (pSi) layer.21 In the present work, the double-layer structure was used for pH sensing. The hydrogel layer was formed by placing a drop of chitosan/GPTMS solution (containing ∼3% acetic acid) on the pSi chip and drying at room temperature for ∼16 h. During this
2nL = mλ
where λ is the wavelength of maximum constructive interference for a spectral fringe of order m, n is the average refractive index of the porous layer and its contents, and L is the thickness of the porous layer, hydrogel layer, or their combination. In our study, the quantity of 2nL, or the EOT, was determined by Fourier transformation of the reflectance spectrum calculated with the IGOR program (Wavemetrics, Inc.). Fourier transformation resulted in a sharp peak whose position was equal to the product value of 2nL. The sensing
Figure 1. (a) Reflectometric interference spectrum of pSi coated with chitosan hydrogel and (b) its corresponding FT-RIFS spectrum. The first major peak in FT-RIFS is assigned to the porous Si layer, and the second major peak is corresponding to the combined layer comprising the hydrogel and pSi layer. A red shift will be observed in the second peak upon the swelling of the chitosan layer in response to pH change. 2789
dx.doi.org/10.1021/ac303282j | Anal. Chem. 2013, 85, 2787−2794
Analytical Chemistry
Article
time, condensation of the silanol groups and cross-linking of the chitosan network occurs. GPTMS is an effective crosslinker for chitosan; it also has the ability to react with the silanol groups on the oxidized pSi surface through the formation of Si−O−Si bonds. The ability of GPTMS to form covalent bonds with both the hydrogel and the pSi contributes enhanced stability to the composite. The fundamental theories on the sensing mechanism of the hydrogel−pSi double-layer structure have been discussed in our previous work.21 The reflectometric spectrum of the hydrogel−pSi double layer represents a superposition of Fabry−Pérot interference spectra that originate in each of the two layers and their combination as shown in the Figure 1a. The Fourier transformation of this spectrum is expected to display three peaks, corresponding to the hydrogel layer, porous SiO2 layer, and their combined layer, respectively. The position of each peak along the x-axis gives the value of 2nL for that particular layer, and the amplitude of the peak represents the index contrast at the two interfaces defining the layer. However, as shown in Figure 1b, the FT spectrum only displays two major peaks. The position of the first major peak assigned to the porous Si layer is almost equal to the FT peak position of the corresponding porous silicon sample, indicating that the gel layer does not significantly infiltrate the porous layer. This layer can act as a reference layer that compensates the signal variation merely caused by the infiltration of different solutions with slightly different refractive indexes. The position of the second major peak is corresponding to the combined layer comprising the hydrogel and pSi layer. The disappearance of the FT peak corresponding to the hydrogel layer should be owing to the low refractive index contrast at the interface of liquid sample and hydrogel. The optical thickness of the hydrogel layer can be calculated by subtracting the 2nL of the second major peak with that of the first major peak. Accordingly, the thickness of the hydrogel layer is estimated to be around 3.2 μm, which highly agreed with the data obtained from scanning electron microscopy (SEM) measurements (Supporting Information, Figure S1). The volume phase transition of the chitosan hydrogel layer induced by the pH change will produce changes in its thickness (L) and refractive index (n), both of which can contribute to the shift in the 2nL value of the combined layer, referring to the position of the second peak in Figure 1b. Thus, the real-time FT-RIFS will provide the dynamic information on hydrogel swelling or shrinking in response to pH change. By calibrating the relationship between the EOT and pH value, optically measuring the pH of the sample solution can be realized. In order to measure the pH in a solution with a volume scale of microliters or nanoliters, we constructed a microfluidic pH sensor by integrating the hydrogel−pSi chip into a microfluidic channel as described in Experimental Section. pH-Responsive Behavior of the Chip-Based Optical Sensor. The pH response of the chitosan hydrogel derives primarily from the large number of amino groups on the polymeric chains of chitosan molecules. Since the pKa of chitosan is around 6.3, the hydrogel has a larger response in near-neutral pH range.26,27 This property makes the chitosan hydrogel a good candidate for sensing the pH variation in biological conditions. Figure 2 shows the real-time pH response of the hydrogel−pSi sensing chip. A stepwise optical response can be found when the pH of the solution decreased from 7.4 to 6.2 step by step at an interval of 0.4 pH unit. The EOT value of the double layer increases steadily with the decrease of pH,
Figure 2. Optical response of the hydrogel-based sensor integrated in the microfluidic chip (red trace) and macrofluidic flow cell (black trace). The volumes of the microfluidic chip and macrofluidic flow cell are around 0.8 and 200 μL, respectively. The EOT value increases step by step when the pH of the solution decreases stepwise from 7.4 to 6.2. The inset figure shows that the optical signal (ΔEOT) displays a nonlinear relationship within the pH range (6.2−7.4).
indicating an overall increase in optical thickness (nL value) of the hydrogel layer. As shown in the inset in Figure 2, the optical signal (ΔEOT) displays a nonlinear relationship in the pH range from 6.2 to 7.4. Nevertheless, an approximate linear relationship can be found in a small pH range. Therefore, the optical pH sensor based on the swelling behavior of the chitosan hydrogel will provide a new alternative for real-time monitoring of the pH change in biofluids, because in most situations the pH shift in biofluids is not so significant. The sensitivity of the hydrogel pH sensor mainly depends on the thickness and cross-linking ratio of the hydrogel film, which dominate the swelling behavior and, consequently, the pHresponsive behavior of the hydrogel. Compared with the one layer of hydrogel−porous silicon composite,25 the sensitivity of the double-layer structure has been greatly enhanced, due to that the change in the hydrogel thickness (L) is far greater than that of the refractive index (n). As show in this work, the EOT shift produced by the double-layer structure can be up to thousands of nanometers when pH shifted from 7.4 to 6.2. In a microfluidic sensor, blockage of the microchannel induced by hydrogel swelling should be considered. In the present chip design, the change of hydrogel thickness during the swelling process is around 0.8 μm calculated from the maximal shift of EOT value. Compared with the depth of the channel (∼250 μm), the swelling of the hydrogel sensing film will not impose a significant blockage effect on the sample loading. The fast signal response is another advantage of the microfluidic sensing platform, because it has very small dead volume and special fluid dynamics in the microscale channel. Herein, the influences of chip volume on the speed of optical response were investigated. Figure 2 shows that the optical response in the microfluidic chip (channel volume ∼0.8 μL) is much faster than that in a macrofluidic flow cell (channel volume ∼200 μL). The time for producing the initial response and maximal response was compared between the microfluidic and macrofluidic sensing device (Supporting Information, Figure S2). The results indicate that microfluidic chip not only produces fast initial response but also takes less time to reach a maximal response. Usually, the fast initial response 2790
dx.doi.org/10.1021/ac303282j | Anal. Chem. 2013, 85, 2787−2794
Analytical Chemistry
Article
results from the small dead volume of the microfluidic channel, whereas the rapid response from the initial to maximal should be ascribed to the fast swelling rate of the hydrogel. It has been well-known that the osmotic pressure created by the free counterions inside the hydrogel is a main driving force for the diffusion of solvent from solution into the hydrogel network.28,29 Consequently, the speed of solvent and ion flux diffusion decides the swelling rate of the hydrogel. A fast swelling rate is observed in microfluidic sensors probably owing to the small distance for solvent and ion flux diffusion. The pH response of the hydrogel is reversible. As indicated in the Supporting Information (Figure S3), the time trace of EOT undergoes a reversible change when pH cycled back to 7.4, since the charge density of chitosan decreased with the increasing of pH. Accordingly, the sensing chip can be reused. Monitoring Bacterial Metabolic Rate on the Microfluidic pH Sensor. It has been well-known that the growth medium becomes acidified as glucose was metabolized by bacteria, leading to the accumulation of organic acids and consequently the decrease of pH in growth medium. Usually, the rate of pH change can reflect the metabolic rate of a bacterial sample,19,20,30 and the metabolic rate is assumed to be related with the growth rate of bacteria. In a large chamber for biochemical reaction, the pH change caused by the microbial metabolism can be readily measured by a conventional pH meter. However, when the sample volume is reduced to microliter or nanoliter scale, it is almost impossible for a conventional pH meter to measure the pH. Microfluidic chips have proven to be a good platform for bacteria culture and antibiotic drug screening.12,13,15,19 Therefore, approaches for real-time monitoring of the pH change of nutrient medium in a microfluidic chip are urgently needed. Recently, pH indicator has been used for monitoring the pH change of bacteria culture medium in a microfluidic chip.19,30 The obvious merit of using pH indicator is that the results can be read by naked eyes. However, most of pH indicators are dye molecules, which may exert some cytotoxicity and affect the growth of bacteria. In the present work, chitosan hydrogel cross-linked by GPTMS for pH sensing has been proved to be nontoxic and biocompatible.31,32 To further testify whether the hydrogel may affect the growth of bacteria, the bacterial sample was cultured for 24 h in a Petri dish coated with the chitosan hydrogel. A plate counting method was employed to evaluate the viability of bacteria. As shown in the Supporting Information (Figure S4), the chitosan hydrogel did not display any significant bactericidal effect. Another advantage of using hydrogel is that the real-time dynamic process of pH change can be monitored by FT-RIFS. In contrast, pH indicator can only tell the pH of culture medium at the initial and final stage. Accordingly, the chitosan hydrogel might be a good candidate as a pH sensing material for the measurements of biological samples. The integration of hydrogel sensing material with a microfluidic channel provides dual functions for pH monitoring and bacterial culture. It has been reported that the bacterial growth rate increases significantly in high surface-to-volume ratio microchannels,17 probably due to the high speed of oxygen supply in the microchannel. To further evaluate the scale effect on bacterial growth, E. coli culture with initial cell density of 1 × 107 CFU/mL was injected into the microfluidic pH sensor (volume ∼0.8 μL) and macrofluidic flow cell (volume ∼200 μL) placed with the same type of sensing chip, respectively. As shown in Figure 3, the response of EOT in the microfluidic channel is very quick even at the beginning of
Figure 3. Influence of chip volume on the optical response during bacteria incubation (initial cell density ∼107 CFU/mL). The increasing rate of EOT value in the microfluidic pH sensor (channel volume ∼0.8 μL) is significantly larger than that in a macrofluidic flow cell (channel volume ∼200 μL), in spite of the same initial cell density in both experiments.
bacteria incubation, indicating the fast pH change in the microenvironment. In contrast, the overall increase rate of EOT observed in the macrofluidic flow cell is much slower. There is no obvious optical response during the initial stage. A slight increase of EOT can only be found after incubation for at least 100 min. According to the discussion above, the pH change can reflect metabolic rate and consequently the growth rate of bacteria. Therefore, it can be inferred that the bacterial growth behavior in a macrofluidic cell almost coincides with that in conventional culture condition, in which the bacterial growth rate is very slow during its initial stage (usually called a lag phase). In contrast, fast bacterial growth in the initial stage may take place in the microfluidic chip. To further demonstrate the feasibility of using a microfluidic pH sensor to monitor the bacterial growth, E. coli suspensions with different initial cell densities (5 × 105, 2 × 106, 8 × 106 CFU/mL E. coli counts, respectively) were injected into microfluidic channels and cultured for a certain time duration, respectively. Nutrient medium without bacteria was also loaded into the microfluidic chip for the control experiment. As shown in Figure 4a, the EOT signal almost keeps constant in the control sample, whereas in the bacteria-containing samples, the increasing rate of EOT is approximately proportional to the initial cell density. The result is reasonable since the rates of glucose metabolism and the accumulation of organic acids are proportional to the initial bacterial cell density. Therefore, monitoring the increasing rate of EOT can not only characterize the growth rate of bacteria but also estimate the initial concentration of bacterial suspension. Figure 4b shows the original FT-RIFS data during the bacteria incubation. The peak marked with a straight dot line is corresponding to the optical thickness of the combined layer comprising porous silicon and chitosan hydrogel. We can find that only this peak displays a position shift during the bacterial incubation. A longer incubation time causes a larger red shift in the peak position. We think that measuring the peak shift has some obvious advantages over measuring the peak intensity especially in the application of biofluids detection, because the measurements of light intensity are more susceptible to be interfered by 2791
dx.doi.org/10.1021/ac303282j | Anal. Chem. 2013, 85, 2787−2794
Analytical Chemistry
Article
Figure 4. (a) Time traces of EOT changes during bacterial incubation in the microfluidic channel. The increasing rate of EOT value became larger with the increase of initial bacterial cell density. The black curve represents negative control (sample without bacteria). (b) The FT transform spectra of the microfluidic pH sensor at different time points during bacterial incubation with initial cell density of 8 × 106 CFU/mL.
Figure 5. (a) EOT change measured by the microfluidic pH sensor loaded with bacteria sample (initial cell density ∼1 × 106 CFU/mL) incubated in glucose medium spiked with different types of antibiotics (1.0 μg·mL−1). (b) The inhibitory fraction of the four types of antibiotics (1.0 μg·mLl−1) at different incubation times. The inhibitory fractions of azithromycin (AZM), amikacin (AMK), ciprofloxacin (CIP), and tetracycline (TC) are characterized by percentage of EOT change accounting for the total EOT change in the blank sample during the incubation.
of antibiotic can be estimated from the bacterial inhibition fraction (I) as expressed with the following equation:
the sample turbidity. For instance, the light scattering caused by the microorganism particles will remarkably interfere with the measurements of light intensity. In contrast, the variation in the sample turbidity will not affect the measurements of peak shift. Antibiotic Susceptibility Test. Antimicrobial susceptibility testing is required to determine the antibiotic susceptibility pattern of a pathogen. This provides clinically relevant information for prescribing more appropriate therapies. To demonstrate the applicability of microfluidic devices for AST, the microfluidic pH sensors were loaded with E. coli suspension cultured in glucose containing MHB spiked with four different types of antibiotics (ciprofloxacin, tetracycline, azithromycin, amikacin), respectively. The initial cell density in each sample is around 1 × 106 CFU/mL. Following the same methods described above, the increasing rate of EOT was monitored in real time. It has been well-known that antibiotics can inhibit the bacterial growth and decrease the rate of glucose metabolism. By measuring the rate of EOT increase, the bacterial inhibition effect of antibiotics can be determined. As shown in Figure 5a, compared with the blank sample (with bacteria but without antibiotic), the increasing rate of EOT value in all the antibiotic-spiked samples significantly decreases. A larger decreasing ratio in EOT value indicates a higher antimicrobial activity of the antibiotic. Accordingly, the antimicrobial activity
I=
ΔEOTblank − ΔEOTantibiotic ΔEOTblank
(1)
where the ΔEOTblank and ΔEOTantibiotic represent the EOT change in the blank sample and antibiotic-spiked sample at a specific time, respectively. As observed in Figure 5b, the overall order of the bacterial inhibition fraction is as follows: ciprofloxacin (CIP) > tetracycline (TC) > amikacin (AMK) > azithromycin (AZM). The antimicrobial activity of CIP is so strong that almost 100% inhibition fraction was achieved even at the beginning of culture stage. The results agree with the data determined by the conventional off-chip method based on the standard Clinical Laboratory and Standards Institute (CLSI) methodology (Supporting Information, Table S1). Furthermore, the real-time observation also provides us a set of dynamic data to evaluate the antimicrobial activity of each antibiotic in different stages. As shown in Figure 5a, within the initial stage (
