Rapid Bacteria Detection at Low Concentrations ... - ACS Publications

Jul 8, 2019 - Rapid Bacteria Detection at Low Concentrations Using Sequential Immunomagnetic Separation and Paper-Based Isotachophoresis ...
1 downloads 0 Views 1MB Size
Subscriber access provided by UNIV OF SOUTHERN INDIANA

Article

Rapid bacteria detection at low concentrations using sequential immunomagnetic separation and paper-based isotachophoresis Federico Schaumburg, Cody Carrell, and Charles S. Henry Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.9b01002 • Publication Date (Web): 08 Jul 2019 Downloaded from pubs.acs.org on July 17, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Rapid bacteria detection at low concentrations using sequential immunomagnetic separation and paper-based isotachophoresis Federico Schaumburg§, Cody S. Carrell†, Charles S. Henry†* § INTEC (Universidad Nacional del Litoral-CONICET), Predio CCT-CONICET, RN 168, 3000, Santa Fe, Argentina. † Department of Chemistry, Colorado State University, Fort Collins, Colorado 80523, United States. ABSTRACT: Detecting bacteria is important in the fields of human health, environmental monitoring and food safety. Foodborne pathogens alone are estimated to cause 420,000 deaths annually, with low-income regions affected most. Despite improvements in bacterial detection, fast, disposable, low-cost, sensitive, and user-friendly methods are still needed. Traditional methods for detecting bacteria rely primarily on cell culturing or PCR, which require highly trained personnel, a central laboratory and take several hours or even days to deliver results. Low cost methods like lateral flow immunoassays exist but frequently suffer from poor sensitivity and/or lack quantitative results. Here, a rapid method for detecting bacteria at very low concentrations is presented using two sequential preconcentration steps. In the first preconcentration step, sample is mixed with antibody-modified magnetic particles and free antibodies conjugated to β-galactosidase (β-gal). The target bacteria are isolated and concentrated using immunomagnetic separation. The isolated bacteria are then incubated with chlorophenol red-β-D-galactopyranoside (CPRG), which reacts with β-gal to produce chlorophenol red (CPR) in a bacteria concentration dependent manner. In the second step, CPR and CPRG are separated and focused using an isotachophoretic microfluidic paper-based analytical device, significantly improving the final detection limit relative to paper-based devices lacking the focusing mechanism. Moreover, CPR and CPRG form two visible color bands which act as test and control bands respectively, improving assay robustness. The method was tested with E. coli DH5-α, and successfully detected concentrations as low as 9.2 CFU/mL in laboratory samples, and 920 CFU/mL in apple juice samples, in approximately 90 minutes.

Detecting bacteria is critical in the medical, environmental and food safety fields. For example, bacterial meningitis, a disease which can become fatal 24 h after symptoms are noticed, causes over 200,000 deaths per year, with higher incidence in rural, low-resource areas.1 On the other hand, foodborne pathogens (most of them bacteria like Escherichia coli, Salmonella enterica or Listeria monocytogenes) are estimated to cause 600 million food-related illnesses and 420,000 deaths annually.2 Again, low-income regions are the most affected. Comprehensive solutions to these and similar problems require fast, disposable, low-cost, sensitive methods,3 capable of detecting in-field bacteria concentrations on the order of 1 CFU/mL or lower. For example, the World Health Organization recommends E. coli and thermotolerant coliform bacteria concentrations 𝛺𝐶𝑃𝑅𝐺 > 𝛺𝑃𝑇𝐸 (1) where the P superscript denotes the paper substrate. For that purpose, CPR and CPRG mobilities were measured in a 1 mm wide and l = 5 cm long laminated paper channel, in whose ends a Φ = 200 V potential difference was applied after fully wetting it with a 10 mM Tricine with 20 mM Bistris buffer solution. The experiments (repeated four times for each analyte) were filmed using a digital microscope and the velocity (v) of the CPR and CPRG peaks was determined. The electroosmotic flow (EOF),

E. coli growth and sample preparation E. coli DH5-α was used as the model bacteria in this work, and was purchased from Thermo Fisher Scientific and grown in LB Miller broth (Sigma-Aldrich, pH 7) overnight in a shaker at 37°C and 200 RPM. The bacteria concentration was quantified by serial dilution and plating on LB agar plates. The initial concentration of the stock E. coli solution used in this work was 9.2×108 CFU/mL. Serial dilutions of this stock solution were made using either LB broth or apple juice. E. coli detection using IMS combined with paper wells The enzymatic assay presented by Srisa-Art et al.22 for Salmonella typhimurium was modified for DH5-α detection. This method takes around 90 min and consists of several steps involving IMS, enzyme conjugation and colorimetric detection. First, 1 mL of the sample was incubated in a rotator (RotoFlex, Argos Technologies, Elgin, IL, USA) for 20 min with paramagnetic beads conjugated with the anti-E. coli antibody AB25823 (5 μL Magnetic Beads-Ab 20 mg/mL) in a microcentrifuge tube. Second, IMS was performed using a magnet (DynaMag-2 Magnet, Thermo Fisher Scientific Inc., Waltham, MA, USA) to isolate and concentrate the E. coli from the sample matrix by removing the supernatant and resuspending the content in 100 μL PBS. This step makes the assay highly specific because of the selectivity of the antibody. Third, the content of the tube was incubated for 20 min in a rotator with the biotinylated antiE.coli antibody AB20640 (100 μL Ab-Biotin 0.08 mg/mL). To remove unbound Ab-Biotin, the sample was washed twice using IMS and 1 mL PBS. Fourth, the content of the tube was incubated for 10 min in a rotator with streptavidin conjugated with β-galactosidase (Strep-β-Gal 4 μg/mL) and again, two washing steps were performed. The content was resuspended in 100 μL PBS. Finally, 10 μL of the content of the tube (the IMS processed sample) were incubated for 30 min in a 7 mm paper well with 10 μL of 1 mM CPRG, where the β-Gal catalyzed the cleavage of CPRG into CPR. CPR is dark red while CPRG is light yellow, and this difference can be detected by naked eye or image analysis. Figure 1a and 1b demonstrates these concepts. Additional details about this procedure can be found elsewhere.22 E. coli detection using IMS combined with ITP The first steps of the proposed method coincide with the first four from the previously described IMS-PAD assay, except that the IMS processed sample was resuspended in the TE solution (10 mM Tricine with 20 mM Bistris pH = 7.45) instead of PBS. These common steps are schematized in figure 1a. After that, in concordance with the IMS-Well method, 10 μL of the IMS processed sample were incubated in a microcentrifuge tube with 10 μL of 1 mM CPRG also dissolved in TE for three different incubation times. Finally, ITP was performed in the ITP μPAD, described in the following section, for 6 min to concentrate the produced CPR and lower the detection limit. Here, a 100 V electric potential was applied between the device electrodes, and 70 μL LE were pipetted in the LE reservoir (in contact with the anode). Once the full length of the channel was wet, 50 μL TE were pipetted in the TE reservoir (in contact with the cathode) and immediately after, the full content of the microcentrifuge tube (20 μL) were also pipetted in the TE reservoir. The colorimetric detection consisted of identifying the test and control lines. The ITP procedures were filmed using a digital microscope (Dino-Lite, Torrance, CA) for further image processing. The steps are schematized in figure 1c.

3

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

with a direction opposed to the electrophoretic migration direction of the anionic analytes, was also taken into consideration. The EOF was measured using the current monitoring method37 𝑃 and found to be 𝛺𝐸𝑂𝐹 = 3.33x10-9 m2/Vs. Mobilities in the paper 𝑃 𝑃 channel were calculated as 𝛺𝐶𝑃𝑅 = 𝛺𝐸𝑂𝐹 + 𝑣𝐶𝑃𝑅 𝑙/𝛷 and 𝑃 𝑃 𝛺𝐶𝑃𝑅𝐺 = 𝛺𝐸𝑂𝐹 + 𝑣𝐶𝑃𝑅𝐺 𝑙/𝛷, respectively. However, mobilities are typically tabulated for non-porous channels. Mobilities in porous media (𝛺𝑃 ) are lower than their non-porous counterparts (𝛺 𝑁𝑃 ) since the charged molecules move following intricated (instead of straight) trajectories. 𝛺𝑃 can be calculated starting from 𝛺𝑁𝑃 , using the macroscopic tortuosity (𝜏) of the media, through the relation 𝛺𝑃 = 𝛺𝑁𝑃 /𝜏 2 .38 This relation can be replaced in inequality 1 yielding, 𝑁𝑃 𝑁𝑃 𝑁𝑃 𝛺𝐿𝐸 > 𝛺𝐶𝑃𝑅 > 𝛺𝐶𝑃𝑅𝐺 > 𝛺 𝑁𝑃 (2) 𝑇𝐸 i.e the inequality also holds for non-porous channels. Thus, the electrolyte system was chosen taking into consideration nonporous channel mobilities. The CPRG mobility in an non-po𝑁𝑃 𝑃 rous channel was estimated as 𝛺𝐶𝑃𝑅𝐺 = 𝛺𝐶𝑃𝑅𝐺 𝜏2 = 𝑃 𝑁𝑃 𝑃 𝛺𝐶𝑃𝑅𝐺 𝛺𝐶𝑃𝑅 /𝛺𝐶𝑃𝑅 . Both calculated and estimated mobilities are summarized in Table 1. From the values of this table, the tortuosity of Whatman 1 can be calculated to be ~1.6. To satisfy inequality 2, the LE solution used was composed of 200 mM HCl and 400 mM Bistris (pH = 6.51), while a 10 mM Tricine with 20 mM Bistris (pH = 7.45) solution was used as the TE.27 In both cases Bistris acts as the counterion. When the electric field is applied, the TE undergoes an adaptation process in the channel, giving place to TE concentrations comparable to the LE concentration. However, in the reservoir, where the volume of liquid is around 80 times the volume in the channel, TE concentration remains fairly unchanged. Thus, the EOF is mainly controlled by the TE, since the LE has a high ionic strength reducing its contribution to the EOF. Another reason for choosing a highly concentrated LE is that the maximum analyte concentration achievable (i.e. the plateau mode concentration) is proportional to the LE anion concentration.39 A modified TE composition using MOPS is described below. Table 1: Non-porous channel and paper electrophoretic mobilities for the different species used. Species

pH

ΩNP [m2/Vs]

Tricine

7.45

-4.90x10-9

Bistris

7.45

2.10x10-9

6.51

8.02x10-9

CPRG

7.45

-16.21x10-9#

MOPS

7.45

-16.5x10-9

CPR

7.45

-19.53x10-9+

HCl

6.51

-65.86x10-9

CPRG and CPR peak separation and quantification The success of the two-band approach depends on the ability of resolving both the test and control bands. However, even though the mobilities of CPR and CPRG are significantly different, if no other anionic chemical species with a mobility between ΩPCPR and ΩPCPRG is present in the sample (the most common scenario), both bands will appear together and will be difficult to distinguish using the naked eye. To circumvent this problem, two different band separation techniques were tested. In the first approach a chemical spacer was employed. A color𝑃 𝑃 less chemical species S with a mobility 𝛺𝐶𝑃𝑅 > 𝛺𝑆𝑃 > 𝛺𝐶𝑃𝑅𝐺 can be included in the TE solution to act as a spacer, producing a separation of CPR and CPRG plugs. Also, if the spacer operates in plateau mode ITP, its concentration will be proportional to the distance between CPR and CPRG peaks, thus the band separation can be fixed to a desired value. A search of a spacer 𝑁𝑃 𝑁𝑃 such that 𝛺𝐶𝑃𝑅 > 𝛺𝑆𝑁𝑃 > 𝛺𝐶𝑃𝑅𝐺 was performed on the 40 PeakMaster 5.3 database and the buffer MOPS satisfies such a relation (see table 1). To find a proper spacer concentration, different modified TE solutions containing 10 μM CPR and CPRG, and MOPS buffer in concentration ranging from 500 μM to 4 mM, were tested. The second approach used for peak separation and to provide quantitative information was image processing. Images of the ITP μPAD obtained with a digital microscope were analyzed and it was found that the green channel intensities correspond practically to the CPR concentration only, thus filtering CPRG contributions. Hence, a calibration curve for CPR was obtained from the green channel. A more detailed analysis can be found in the supplementary information. E. coli detection in apple juice using IMS-ITP To test the proposed assay performance on real food samples, Kroger branded apple juice (Cincinnati, OH, USA) was purchased on 29/08/2018. A 110 μL aliquot of the E. coli culture Ci = 9.2x108 CFU/mL was diluted with 990 μL apple juice to obtain the first dilution, and further 10-fold serial dilutions were performed also using the apple juice until reaching 10 -8Ci. These set of samples were used with IMS-ITP protocol with no modifications.

ΩP [m2/Vs]

RESULTS AND DISCUSSION ITP μPAD characterization The μPAD-ITP system was studied by performing three sets of experiments (each one repeated four times) with probe samples composed of 10 μM CPR, 10 μM CPRG and 10 μM of each CPR and CPRG respectively, dissolved in TE (figure S1). It was found that optimum time for performing the ITP experiments is 6 min. After 6 min, the intensity of the CPRG and CPR peaks increased at a lower ratio and they are still far from the LE reservoir. Also, short durations are preferable because i) the overall time of the procedure decreases, ii) the peaks are narrower at shorter times because of mechanical dispersion 42 and iii) the wax defining the hydrophobic channel absorbs liquid and sample proportionally with time.43

-6.024x10-9* -7.26x10-9*

Mobilities and pH values obtained from Peakmaster 5.340 unless specified. *Measured values. 𝑁𝑃 𝑃 𝑁𝑃 𝑃 #Estimated as 𝛺𝐶𝑃𝑅𝐺 = 𝛺𝐶𝑃𝑅𝐺 𝛺𝐶𝑃𝑅 /𝛺𝐶𝑃𝑅

+Taken from Hopper et al.41 and corrected to the TE pH value with Peakmaster 5.3.

4

ACS Paragon Plus Environment

Page 4 of 9

Page 5 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry To obtain the LOD and linear range for the ITP μPAD, signal was measured for initial CPR concentrations (CPR0) ranging from 200 nM to 10 mM dissolved in TE. The CPR concentration profiles obtained after 6 min in the ITP channel are shown in figure S3a. It was found that the CPR LOD is 600 nM. Also, as CPR0 (x-axis in figure S3b) increases, the peak maximum increases until CPR0 ~100 μM where the system stops working in peak mode and starts working in plateau mode ITP. Hence, the maximum CPR peak concentration (y-axis in figure S3b) that can be obtained with this configuration is around 2.5 mM, as can be observed in figure S3. Once plateau mode is reached, the length of the plateau zone can be used as an indicator of CPR0. The ability of the system to focus the analyte into a plug can be measured with the preconcentration ratio (PR). PR when t = 6 min can be defined as CPR6/CPR0 and it is plotted vs. CPR0 in figure S2. There, it can be seen that PR takes values around 102 when the device works in peak mode ITP, and decreases with CPR0 when the device works in plateau mode ITP. It should be noted that when the ITP μPAD works in plateau mode, most of the theoretical models reported in the literature25,27,44 cannot be applied. From figure S2 it can be seen that, the ITP device works better when the sample is more diluted, i.e. the relative difference between the peak intensity and CPR0 is larger for lower concentrations.

Figure 2: a) CPRG and CPR peak separation obtained with MOPS 1 mM. b) Distance between CPR and CPRG peaks (black crosses), and half-width of CPR (gray squares) and CPRG (gray diamonds) versus MOPS concentration.

E. coli detection using IMS combined with paper wells The IMS-Well assay was tested on E.coli dilutions and a blank, giving the results summarized in figure 3d. In this figure CPR concentration was estimated as the inverted intensity of the green channel IG. The calibration curves for CPR cannot be used in this case, since they were obtained with laminated paper where the liquid is found within the paper fibers (the case of the ITP μPAD channel), while in the well assay the sample is suspended in the paper and above the paper in a droplet . In figure 3 the uncertainty is expressed as ±1 SD from the average value of the green channel of the image, when a selection covering the region of interest is considered. When the uncertainty on the measurement of the blank (gray horizontal bar in figure 3d) is compared with the rest of the measurements, it follows that the limit of detection of the assay is LODITP-Well=9.2x106 CFU/mL. It is important to note that the IMS-well assay was not optimized in this study, giving in a poor LOD. In previous IMS well assays, LODs of 102 CFU/mL were realized.22 The IMS-well assay was used as a baseline to measure improvement from ITP. E. coli detection using IMS combined with ITP 10 μL of the same IMS processed samples used for the IMSWell assay were incubated in a microcentrifuge tube with 1 mM CPRG for tinc = 10, 20 and 30 min, and used as an inputs for ITP. For each position in the longitudinal direction of the ITP channel, the IG average and SD were calculated transversally. The peak CPR concentration and its SD were plotted for each E. coli concentration. giving the results shown in figure 3a-c. As a general rule, longer tinc gave stronger CPR signals. The limit of detection, was found to be LODIMS-ITP = 9.2 CFU/mL, but tinc ≥ 20 min, gave peaks easily differentiated from the blank. The improvement introduced by the IMS-ITP assay when compared with the IMS-Well assay can be quantified by comparing the LOD of each method. The LODIMS-ITP was found to be 106 times better than LODIMS-Well. The total assay time for the IMSITP assay was 76, 86, and 96 min for tinc = 10, 20 and 30 min, respectively compared to 90 minutes for the IMS-Well assay.

Detection of the peaks For the chemical spacer approach, the colorless buffer MOPS, with a mobility between CPR and CPRG, was used to produce a separation between test and control bands to make it easier to visualize the two bands. Figure 2b shows the obtained CPRG and CPR peak separations obtained for the MOPS concentrations studied. As expected, because the spacer works in plateau mode, the distance between peaks increases with higher spacer concentrations. Unfortunately, the peaks lose definition as MOPS increases. The loss of definition is especially noticeable in the CPRG peak, which could be because its mobility is very similar to MOPS mobility and hence they mix at the plug interface. The loss may also be due to the modification produced by MOPS to the original TE composition, forming a mixed TE zone and reducing its focusing capabilities. The addition of 1 mM MOPS in the TE, was found to be the optimal concentration to maximize the peak separation (1 mm) and the resolution. For the algorithmic approach, digital images of a laminated Whatman 1 paper filled with a mixture of CPR and CPRG were used to determine the CPR concentration profile via image processing using the inverted green channel intensity IG. A model for IG as a function of CPR was found to be IG=2765.12xCPR0.4588 (see supporting information for details).

5

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 3. CPR peak intensity as a function of E.coli concentration obtained after performing ITP during 6 min, for three different incubation times: a) tinc = 30 min, b) tinc = 20 min, c) tinc = 10 min. d) CPR concentration estimation in the IMS-Well assay as the inverted color intensity of the green channel versus the concentration of E. coli.

The exceptional performance of the IMS-ITP assay, when compared to the IMS-Well test, is the result of amplifying the initial bacteria concentration. First, the preconcentration factor of the IMS procedure PRIMS is ~10 (initial sample 1 mL / final sample 100 μL). Second, the fact that more than one Ab-Biotin might bind to an E. coli means that more than one β-Gal might be bound to a single E.coli. Third, the fact that each β-Gal produces multiple CPR molecules. This is the reason why the longer t inc, the stronger the CPR signal. The number of CPR molecules produced depends linearly on tinc (10-30 min) and the enzyme units of β-Gal (~525 U/mg).45 Here, we suppose that CPRG is in excess. Fourth, ITP preconcentrates the CPR at a rate PRITP estimated in figure S2. Fifth, the rapid growth of IG with log(CPR) (figure S4a). According to figure S2, CPR0 can be concentrated up to ~200 times, which translates to a notable IG increase, that allows easy differentiation of the CPR plug from the blank. Also, CPRG plays two different roles in determining assay sensitivity and LOD. In the IMS-Well method CRPG creates a background noise that interferes with the CPR signal, while in the IMS-ITP assay CPRG is completely separated from CPR, meaning that no interference occurs, and forming a control band which provides an extra functionality to the test. Regarding bacteria quantification, a sigmoidal relationship between the sample bacterial load and the intensity of the CPR peak was expected, based on the sigmoidal nature of the original IMS assay,22 the kinetics of typical sandwich immunoassays,46 and the kinetics of the ITP μPAD (see figure S3b). However, even though the IMS-ITP method showed a general trend of larger CPR peaks for higher bacteria concentration, the correlations obtained were not clear enough to quantitatively estimate the bacterial load from the ITP peak intensity. A more intensive probabilistic study would be required to provide an absolute measurement of starting bacteria concentration.

Figure 4. ITP channels, with different E. coli concentrations and the blank when MOPS 1 mM is used as a chemical spacer between the CPR and CPRG peaks. The test and control zones are indicated with dotted rectangles. The experiments were repeated using a TE solution including MOPS 1 mM as a spacer. Because of the described loss in the peak intensities, the chosen incubation time was 30 min. Also, when MOPS is used as a spacer, the plugs move faster and the optimum time for measuring CPR and CPRG intensity was 4 min instead of 6 min. The larger velocity is probably due to a decreased EOF because of the increased ionic strength in the TE. Figure 4 shows the bands obtained in positive and negative (blank) samples. As predicted, the test and control bands can be appreciated using the naked eye in the two positive cases, while in the negative case only the control band is present.

E. coli detection in apple juice The IMS-Well and IMS-ITP assay was performed in apple juice contaminated with E. coli using tinc = 20 min. The results are summarized in figure 5. It can be seen that for the IMS-Well, 6

ACS Paragon Plus Environment

Page 6 of 9

Page 7 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry easy-to-build, disposable μPAD. The equipment required is also inexpensive and portable. It reduces to a magnet, a rotator and a 100 V DC-power source which can be replaced by a wall adapter. Although the improvement achieved by the ITP μPAD was demonstrated on a IMS assay, it could also be used to amplify the signal in any enzymatic colorimetric assay, like ELISA or lateral flow test. The IMS-ITP assay was compared with a recently reported method that combined IMS with a well-type μPAD. The LOD was improved by 6 orders of magnitude on laboratory samples and 5 orders of magnitude on apple juice samples. In order to maintain the same conditions from the IMS-Well assay for comparison, equal volume of IMS processed sample (10 μL) was used in the IMS-ITP assay. The sample volume could be increased to achieve further improvements given that the volume allowed by the ITP μPAD reservoirs is 70 μL. In spite of the remarkable capability shown by the IMS-ITP assay to detect contaminated samples, and even though the proposed method showed a general positive correlation between CPR peaks and bacterial load, no clear quantitative trend was found. The probabilistic study required for that purpose is out of the scope of this work. Regarding the duration of the IMSITP assay, it is approximately the same as the IMS-Well assay, but remarkably lower than the traditional methods. In addition to the enhanced bacteria detection, the IMS-ITP method incorporates a control mechanism that should improve robustness. For such purpose, a two-band approach was developed taking advantage of the ITP capability of forming focused plugs and the visible nature of both CPR and CPRG. Such an approach can be used to detect other microorganisms or substances, provided that β-Gal enzymes can be bound to the target. In this case, the same ITP system can be used (i.e. the same μPAD, LE, TE, protocol and image processing techniques) while some modifications will surely have to be made on the IMS procedure, like the antibodies and incubation times required. Moreover, the assay can be extended to other enzymesubstrate pairs and even other detection methods, like electrochemical or fluorescence. Finally, the IMS-ITP assay proved to be a promising candidate for fast, on-site bacteria detection in extremely low concentrations. Future research should focus on enabling multiplexed detection of bacteria, by attaching multiple antibodies, specific to several bacteria types, to the magnetic beads. This would mean a specificity loss for the method, but is required in applications like bacterial infection or food contamination detection. Also, future research should focus on further reduction of energy consumption, instrumentation and user intervention, by implementing the sample preparation (like pH correction) and IMS steps on the μPAD to achieve a self-contained assay. To serve this purpose, advantage can be taken from readily existing technology, such as mobile communication devices who offer complementary functions, like optical sensors, memory, computing power, energy sources and connectivity. All these, plus the attributes shown by the IMS-ITP assay are decisive for achieving the self-sufficient, hand-held devices required in sensitive areas like food safety, emergencies or low resource health settings in low income countries.

LODIMS-Well = 9.2x107 CFU/mL while for the IMS-ITP, LODIMS-ITP = 9.2x102 CFU/mL, meaning an improvement of 105. When compared with the laboratory samples in the previous section, the LODIMS-ITP was higher by two orders of magnitude. Because the pH of the apple juice is 3.4, while the pH of LB Miller broth is 7, it is likely that the pH of the apple juice interfered with the binding of the Ab to the E. coli in the first step of the IMS procedure, as discussed elsewhere.47-49 This could be circumvented by correcting the sample pH prior to the beginning of the procedure. We chose not to do this step here, but it constitutes a practical challenge to be addressed in the future. Again the total duration of the assays were similar (86 min for IMS-ITP versus 90 min for IMS-Well).

Figure 5. a) CPR concentration in the IMS-Well assay, estimated as the inverted intensity of the green channel, for apple juice samples with different E. coli concentrations and b) CPR peak concentration in mM in the IMS-ITP assay, for apple juice samples with different E. coli concentrations. In both a) and b), the light gray horizontal band indicates the blank uncertainty. c) Test and control bands obtained with the IMS-ITP assay, for an apple juice sample containing 9.2x107 CFU/mL of E. coli. The TE used contained MOPS 1mM.

CONCLUSIONS

ASSOCIATED CONTENT

Here, a fast, sensitive method for detection of bacterial contamination of food, environmental or biological samples at concentrations as low as 9.2 CFU/mL was developed and tested on a real food sample. The assay is based in the application of two sequential preconcentration steps (IMS and ITP) using a cheap,

Supporting Information Supporting figures for the kinetics of the ITP μPAD; the preconcentration ratio of the ITP μPAD; the limit of detection of 7

ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

the ITP μPAD for CPR; the algorithm for CPR and CPRG peak separation (PDF).

(20) Yamada, K., Shibata, H., Suzuki, K., & Citterio, D. (2017). Lab Chip, 17(7), 1206-1249. (21) Moghadam, B. Y., Connelly, K. T., & Posner, J. D. (2015). Anal Chem, 87(2), 1009-1017. (22) Srisa-Art, M., Boehle, K. E., Geiss, B. J., & Henry, C. S. (2017). Anal Chem, 90(1), 1035-1043. (23) Helmut Günzler, I. I., & Williams, A. (2001). Handbook of analytical techniques. Evolution, 1, 1-2. (24) Jung, B., Bharadwaj, R., & Santiago, J. G. (2006). Anal Chem, 78(7), 2319-2327. (25) Rosenfeld, T., & Bercovici, M. (2014). Lab Chip, 14(23), 44654474. (26) Li, X., Luo, L., & Crooks, R. M. (2015). Lab Chip, 15(20), 4090-4098. (27) Rosenfeld, T., & Bercovici, M. (2018). Lab Chip, 18(6), 861868. (28) Olsvik, O., Popovic, T., Skjerve, E., Cudjoe, K. S., Hornes, E., Ugelstad, J., & Uhlen, M. (1994). Clin Microbiol Rev, 7(1), 43-54. (29) Skjerve, E., Rørvik, L. M., & Olsvik, O. (1990). App Environ Microb, 56(11), 3478-3481. (30) Skjerve, E., & Olsvik, Ø. (1991). Int J Food Microbiol, 14(1), 11-17. (31) Chapman, P. A., Wright, D. J., & Siddons, C. A. (1994). J Med Microbiol, 40(6), 424-427. (32) Monceyron, C., & Grinde, B. (1994). J Virol Methods, 46(2), 157-166. (33) Kassimi, L. B., Gonzague, M., Boutrouille, A., & Cruciere, C. (2002). J Virol Methods, 101(1-2), 197-206. (34) Park, Y., Cho, Y. H., Jee, Y., & Ko, G. (2008). App Enviro Microb, 74(13), 4226-4230. (35) Schaumburg, F., Kler, P. A., & Berli, C. L. (2018). Sens Actuat B-Chemical, 259, 1099-1107. (36) https://inkscape.org/ (Accessed on 11-26-2018) (37) Wang, W., Zhou, F., Zhao, L., Zhang, J. R., & Zhu, J. J. (2007). J Chromatogr A, 1170(1-2), 1-8. (38) Kunkel, H. G., & Tiselius, A. (1951). J Gen Physiol, 35(1), 89118. (39) Chen, S., Graves, S. W., & Lee, M. L. (1999). J Microcolumn Sep, 11(5), 341-345. (40) https://echmet.natur.cuni.cz/software/peakmaster (Accessed on 11-26-2018) (41) Hopper, K. G., LeClair, H., & McCord, B. R. (2005). Talanta, 67(2), 304-312. (42) Urteaga, R., Elizalde, E., & Berli, C. L. (2018). Analyst, 143(10), 2259-2266. (43) Nguyen, M. P., Meredith, N. A., Kelly, S. P., & Henry, C. S. (2018). Anal Chim Acta, 1017, 20-25. (44) Khurana, T. K., & Santiago, J. G. (2008). Anal Chem, 80(16), 6300-6307. (45) https://www.sigmaaldrich.com/catalog/product/sigma/s3887 (Accessed 7/12/2018) (46) Christopoulos, T. K., & Diamndis, E. P. (1996). Theory of immunoassays. In Immunoassay (pp. 25-50). (47) Kamata, N., Enomoto, A., Ishida, S., Nakamura, K., Kurisaki, J. I., Kaminogawa, S. (1996). Biosci Biotech Bioch, 60 (1), 25-29. (48) Hughesjones, N. C., Gardner, B., Telford, R. (1964). Immunology, 7 (1), 72-&. (49) Borrok, M. J., Wu, Y. L., Beyaz, N., Yu, X. Q., Oganesyan, V., Dall'Acqua, W. F., Tsui, P. (2015). J. Biol. Chem, 290 (7), 4282-4290.

AUTHOR INFORMATION Corresponding Author *E-mail: [email protected]. Phone: +1-970-491-2852. ORCID Charles S. Henry: 0000-0002-8671-7728

ACKNOWLEDGMENT FS acknowledges CONICET and the postdoctoral fellowship granted by the Fulbright Program/Ministerio de Educación. CSH and CC acknowledge support from the National Wildlife Research Center of the USDA under grant ___.

REFERENCES (1) Dou, M., Dominguez, D. C., Li, X., Sanchez, J., & Scott, G. (2014). Anal Chem, 86(15), 7978-7986. (2) Havelaar, A. H., Kirk, M. D., Torgerson, P. R., Gibb, H. J., Hald, T., Lake, R. J., ... & Speybroeck, N. (2015). PLoS medicine, 12(12), e1001923. (3) Mabey, D., Peeling, R. W., Ustianowski, A., & Perkins, M. D. (2004). Nat Rev Microbiol, 2(3), 231. (4) World Health Organization. (2004). Guidelines for drinking-water quality (Vol. 1). World Health Organization. (5) US Food and Drug Administration. (2004). Guidance for industry: Juice HACCP hazards and controls guidance first edition. (6) Baskaran, S. A., Amalaradjou, M. A. R., Hoagland, T., & Venkitanarayanan, K. (2010). Int J Food Microbiol, 141(1-2), 126-129. (7) Kimball, D. A. (Ed.). (2012). Citrus processing: quality control and technology. Springer Science & Business Media (8) Costello, M., Rhee, M. S., & Kang, D. H. (2002). J Rapid Meth Aut Mic, 10(1), 19-25. (9) Madigan, M. T., Martinko, J. M., & Parker, J. (2010). Brock biology of microorganisms (Vol. 13). (10) Malorny, B., Paccassoni, E., Fach, P., Bunge, C., Martin, A., & Helmuth, R. (2004). App Environ Microb, 70(12), 7046-7052. (11) Singer, R. S., Cooke, C. L., Maddox, C. W., Isaacson, R. E., & Wallace, R. L. (2006). J Vet Diagn Invest, 18(4), 319-325. (12) Chen, J., Alcaine, S. D., Jiang, Z., Rotello, V. M., & Nugen, S. R. (2015). Anal Chem, 87(17), 8977-8984. (13) Hossain, S. Z., Ozimok, C., Sicard, C., Aguirre, S. D., Ali, M. M., Li, Y., & Brennan, J. D. (2012). Analytical and bioAnal Chem, 403(6), 1567-1576. (14) Yang, Y., Noviana, E., Nguyen, M. P., Geiss, B. J., Dandy, D. S., & Henry, C. S. (2016). Anal Chem, 89(1), 71-91. (15) Li, C. Z., Vandenberg, K., Prabhulkar, S., Zhu, X., Schneper, L., Methee, K., ... & Almeide, E. (2011). Biosens Bioelectron, 26(11), 4342-4348. (16) Wang, Y., Ping, J., Ye, Z., Wu, J., & Ying, Y. (2013). Biosens Bioelectron, 49, 492-498. (17) Jokerst, J. C., Adkins, J. A., Bisha, B., Mentele, M. M., Goodridge, L. D., & Henry, C. S. (2012). Anal Chem, 84(6), 29002907. (18) Adkins, J. A., Boehle, K., Friend, C., Chamberlain, B., Bisha, B., & Henry, C. S. (2017). Anal Chem, 89(6), 3613-3621. (19) San Park, T., Li, W., McCracken, K. E., & Yoon, J. Y. (2013). Lab Chip, 13(24), 4832-4840.

Table of Contents (TOC)

8

ACS Paragon Plus Environment

Page 8 of 9

Page 9 of 9 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

ACS Paragon Plus Environment

Page S-11