An optical biosensor for detection of pathogen biomarkers from Shiga toxin-producing Escherichia coli in ground beef samples

An optical biosensor for detection of pathogen biomarkers from Shiga toxin-producing Escherichia coli in ground beef samples

Loreen Lamoureuxa,b,d, Peter Adamsc, Afsheen Banisadrb, Zachary Stromberge, Steven Gravesa,

Gabriel Montanoc, Rodney Moxleye, Harshini Mukundanb,d aUniv. of New Mexico, Ctr. for Biomedical Engineering (USA); Los Alamos National Lab:

bPhysical Chemistry and Applied Spectroscopy, cCtr. for Integrated Nanotechnology (USA); dThe New Mexico Consortium (USA); eUniv. of Nebraska-Lincoln, School of Veterinary Medicine and

Biomedical Sciences (USA)

ABSTRACT

Shiga toxin-producing Escherichia coli (STEC) poses a serious threat to human health through the consumption of contaminated food products, particularly beef and produce. Early detection in the food chain, and discrimination from other non-pathogenic Escherichia coli (E. coli), is critical to preventing human outbreaks, and meeting current agricultural screening standards. These pathogens often present in low concentrations in contaminated samples, making discriminatory detection difficult without the use of costly, time-consuming methods (e.g. culture). Using multiple signal transduction schemes (including novel optical methods designed for amphiphiles), specific recognition antibodies, and a waveguide-based optical biosensor developed at Los Alamos National Laboratory, we have developed ultrasensitive detection methods for lipopolysaccharides (LPS), and protein biomarkers (Shiga toxin) of STEC in complex samples (e.g. beef lysates). Waveguides functionalized with phospholipid bilayers were used to pull down amphiphilic LPS, using methods (membrane insertion) developed by our team. The assay format exploits the amphiphilic biochemistry of lipoglycans, and allows for rapid, sensitive detection with a single fluorescent reporter. We have used a combination of biophysical methods (atomic force and fluorescence microscopy) to characterize the interaction of amphiphiles with lipid bilayers, to efficiently design these assays. Sandwich immunoassays were used for detection of protein toxins. Biomarkers were spiked into homogenated ground beef samples to determine performance and limit of detection. Future work will focus on the development of discriminatory antibodies for STEC serotypes, and using quantum dots as the fluorescence reporter to enable multiplex screening of biomarkers. Keywords: Shiga toxin-producing Escherichia coli (STEC), lipopolysaccharides (LPS), planar optical waveguide biosensor, membrane insertion assays, amphiphilic pathogen biomarkers

1. INTRODUCTION Every year, in the United States, approximately 265,000 people1 develop illnesses associated with STEC infection. A large number of these cases originate from consumption of contaminated food products, with cattle being a predominant carrier. Although STEC infections typically culminate in recovery, it can result in the development of severe and life threatening complications, such as hemolytic-uremic syndrome2,3. Currently, the United States Department of Agriculture, Food Safety and Inspection Service requires screening of raw, non-intact beef for multiple serotypes of STEC4. However, detecting these pathogens and discriminating different serotypes from each other, let alone from non-pathogens, is problematic for several reasons. Current methods such as latex agglutination, immunomagnetic separation, and culture on differential media, all suffer from higher degrees of non-specificity compared to methods such as polymerase chain reaction (PCR) and biomarker screening. Although sensitive, PCR-based methods can detect residual nucleic acid contaminants, resulting in false positives, and is time consuming, labor intensive and laboratory-dependent. Using biomarkers for the direct detection of pathogens is a common method used in many diagnostic applications. For STEC, primary biomarkers include Shiga toxin (Stx) and lipopolysacharide (LPS), among others. Stx is the primary virulence factor expressed by STEC strains. The ability to differentiate between Stx1 and Stx2 is also critical as STEC can express one or both toxins5. Serotyping of STEC is based on identification of the O-, H- and K-antigens6, or alternatively, just the O- and H-antigens. In the case of the O-antigen (O-Ag), identification is based on differences in the chemical signature of LPS, a lipoglycan associated with the bacterial outer membrane. LPS is amphipathic is nature,

Frontiers in Biological Detection: From Nanosensors to Systems VII, edited by Benjamin L. Miller, Philippe M. Fauchet, Brian T. Cunningham, Proc. of SPIE Vol. 9310, 931004 · © 2015 SPIE

CCC code: 1605-7422/15/$18 · doi: 10.1117/12.2079658

Proc. of SPIE Vol. 9310 931004-1

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 04/13/2015 Terms of Use: http://spiedl.org/terms

with a highly conserved hydrophobic tail and a highly variable hydrophilic polysaccharide O-Ag group, which extends outside of the bacterial cell wall. The O-Ag of LPS is significantly different between serogroups of STEC, while the hydrophobic portions remains conserved. Most of the current methods for biomarker detection are based on enzyme-linked immunosorbent assays or lateral flow assays, neither of which are optimized for detection of lipoglycans. These methods also suffer from non-specificity, primarily because of cross-reactivity associated with the reporter antibodies used for detection. Furthermore, biomarkers of STEC serogroups are structurally very similar, increasing the likelihood of cross-reactivity and false positives associated with such assays. However, development of specific targeted antibodies, and use of a suite of biomarkers that together offer greater resolution for diagnosis may help overcome this problem. For instance, the simultaneous detection of Stx and LPS will offer more accurate diagnosis of STEC than either one alone. Use of antibodies targeting the O-Ag of LPS can allow for reliable identification of STEC subtypes in a rapid format. Rational use of pathogen biomarkers using targeted antibodies to achieve rapid discriminatory detection of STEC serotypes in contaminated beef is the ultimate goal of our project. Detection of pathogen biomarkers that are typically present in very low concentrations, in a complex background, requires a sensitive platform. The waveguide biosensor developed at Los Alamos National Laboratory, utilizes the principle of evanescent sensing to create an intense optical field for the ultra-sensitive detection of fluorescent antibodies7,8. By performing biological detection within the sensing field, the platform maximizes the sensitivity of fluorescent reporters, while minimizing the excitation of background signal associated with the sample matrix8-10. The detection surface within the field is functionalized with a biotinylated lipid bilayer to facilitate biodetection. Several assay formats have been evaluated on the waveguide platform9,10-12, but those of relevance to this work are membrane insertion of amphiphiles and sandwich immunoassays10-12. Membrane insertion is a novel assay format developed by our team wherein the waveguide surface is functionalized with a supported lipid bilayer, allowing for the selective capture of amphiphiles, which are detected using a single fluorescently labeled reporter antibody. This allows for the presentation of the amphiphilic biomarker in a near in vivo conformation, facilitating exquisite sensitivity of detection. This presentation also allows for targeting of the discriminative O-Ag region of the biomarker, and therefore a unique and sensitive way to serotype STEC, in addition to confirmed identification of the pathogen..

2. METHODS

2.1 Waveguide preparation Single mode planar optical waveguides with a thin film (10 nm) coating of SiO2 (nGimAT), and a 10 nm SiO2 film was deposited (Spectrum Thin Films) to facilitate functionalization for biological assays8-11. Waveguides and glass coverslides were cleaned as previously described9,11 by bath sonication for 5 min, first in chloroform, followed by ethanol, then water. Waveguides and coverslides were dried under argon and exposed to UV-light and ozone for 40 min. Flow cells were constructed from cleaned waveguides and coverslips using a silicone gasket (Grace Bio-Labs) to bond the two pieces together and establish a sealed chamber9,11. Flow cells were immediately assembled after cleaning followed by injection with lipid micelles (Section 2.2) and incubation overnight, to facilitate stable bilayer formation. 2.2 Micelle preparation For membrane insertion experiments, 1% biotinylated 1,2-Dioleoyl-sn-glycero-3-phosphocholine (DOPC, Avanti® Polar Lipids) micelles were prepared by probe sonication as previously described9,11. In brief, DOPC, in chloroform and 1% cap-biotinyl sodium salt (Avanti®) were deposited into glass tubes, and chloroform was evaporated under an argon stream. Lipids were rehydrated in phosphate buffered saline (PBS) and stirred for 2 hr at room temperature (RT), 120 rpm on an orbital shaker, then 10 freeze-thaw cycles using liquid nitrogen and a warm water bath. Finally lipids were probe sonicated for 6 min (1.0 s pulse on, 1.0 s pulse off) using a Branson ultrasonic generator. 2.3 Lipopolysaccharides, beef samples, and antibodies Lipopolysaccharides from six strains of non-O157 STEC (DEC10B [O26:H11], B8227-C8 [O45:H2], MT#80 [O103:H2], 0201 9611 [O111:H11], GS G5578620 [O145:NM], and TY-2482 [O104:H4]) were selected and prepared



by hot phenol extraction and purified, then tested for activity (unpublished data). LPS for control groups O111:B4 were purchased from Sigma Aldrich and O157:H7 from List Labs. For membrane insertion assays, LPS crude extracts were prepared in 5 mg/mL stocks, and bath sonicated for 15 min, then diluted to the working concentration in PBS and re-sonicated for 15 min prior to experiment. Simple Truth® organic ground beef (Kroger Stores), was flash frozen in liquid nitrogen, and lyophilized under high vacuum on a Schlenk line for 48 hr. Dried material was crushed to a fine powder using mortar and pestle and suspended in lysis buffer (2mM sucrose, 10mM HEPES, 25mM KCl, 1mM EDTA, 10% v/v glycerol, 5 mg/mL concentration). The suspension was alternately vortexed (1 min) and bath sonicated (30 s) to eliminate aggregates. Samples were diluted to 1 mg/mL with PBS before use. Beef homogenate was used as a negative control, to evaluate background fluorescence and cross-reactivity in a crude matrix, and also as a suspension buffer for LPS measurements. Reporter antibodies for LPS were polyclonal antibody (pAb) anti-E. coli LPS (pAb E. coli, Bioss Antibodies), pAb anti-LPS O157 (pAb O157, LifeSpan Biosciences), and pAb anti-E. coli LPS O104 (pAb O104, Abraxis). All reporter antibodies were fluorescently labeled with Alexa Fluor® 647 (af647, LifeTechnologies). Monoclonal antibody (mAb) anti-Stx2 subunit A (mAb Stx2a, Toxin Technology) was biotinylated using an EZ-Link™ Sulfo-NHS-Biotinylation kit (Pierce) and used as a capture antibody in sandwich immunoassays, and pAb Stx2 subunit B (mAb-Stx2B, BEI Resources), was labeled with af647 and used as reporter. 2.4 LPS membrane insertion assays Concentration dependent LPS insertion assays were established in triplicate using LPS O157. Flow cells were prepared as described above, inserted into the platform, and blocked for 1 hr with 2% bovine serum albumin (BSA), then rinsed with wash buffer (0.5% BSA/PBS). Light (635 nm, incident power of 440-443 µW) was coupled into the diffraction grating and response signal adjusted for maximum peak intensity. Background signal was recorded, and the flow cell incubated (90 min) with the reporter antibody (25nM pAb O157-af647) to determine any non-specific binding (NSB). Flow cell was rinsed with wash buffer between measurements. LPS (prepared as in section 2.3) was incubated for 2 hr allowing its association with the supported lipid bilayer. Unbound LPS was removed by washing. Subsequently, the reporter antibody was incubated for 90 min facilitating binding to the LPS associated with the surface. After rinsing, the fluorescence signal associated with antibody bound to LPS, captured on the bilayer, was measured using a spectrometer (USB2000, Ocean Optics) interface associated with the instrument. Membrane insertion assays were performed with 200 μg/mL LPS from serogroups O26, O45, O103, O104, O111, and O145, using pAb O157-af647 as the reporter antibody. This approach simply exploits the enhanced cross-reactivity of a polyclonal antibody to the conserved regions of different serogroups of LPS. However, we raised the hypothesis that by use of antibodies specific for a particular LPS serogroup, we could potentially enhance the sensitivity and selectivity of detection by targeting the O-Ag region. To evaluate this, LPS O104 was tested using 25 nM pAb O104-af647 as the reporter, and compared to the signal using the non-specific pAb O157-af647. To determine detection of LPS O111:B4 in beef homogenates, NSB was measured using the reporter antibody, 80nM pAb E. coli – af647, 90 min. LPS (200 μg/mL) was mixed with 1 mg/mL beef homogenate sample and incubated in the flow cell for 2 hr, then rinsed. Specific signal was recorded after 90 min incubation with the reporter antibody. 2.5 Shiga toxin immunoassays Stx2 was detected by sandwich immunoassay using 1% biotinylated DOPC lipid bilayers. Waveguides were prepared and the background recorded. 2 μM streptavidin (Pierce) was incubated for 10 min and rinsed, followed by 25 nM pAb Stx2B-af647 (90 min) to measure NSB. Capture antibody (25 nM biotinylated mAb Stx2a) was incubated for 20 min, followed by 100 μg/mL Stx2 (60 min). The specific signal was recorded after a 90 min repeat incubation with the reporter antibody. 2.6 Data processing Limit of detection (LoD) for waveguide data was processed in MS Excel, by taking the maximum average NSB for each assay and determining the standard deviation (σ) of the replicates and adding 3σ, then multiplying by the sample concentration, and dividing by the maximum specific signal for that concentration (see equation). Data was graphed between wavelengths 625-825 nm using MS Excel.



UAA03

.000~M»MN643, ltlNl8ftfttt tt075 725 773

swim& Iim)

LoD = NSB + 3σ [Sample]Speci ic

3. RESULTS

3.1 LPS Insertion Assays LPS O157 (6.25 μg/mL) was detected by membrane insertion (figure 1) with a LoD = 6.23 µg/mL. The association of LPS with the lipid bilayer is not linear, as is expected, because high concentrations of LPS can change the dynamics of the membrane and saturation owing to their self-assembly and association with the membrane14,15. For these reasons, while higher concentrations of LPS (200 µg/mL and 100 µg/mL) distinctly demonstrate a higher signal:NSB ratio, the relationship between concentration and signal is more linear at lower concentrations. However, in the concentration range physiologically relevant for detection in contaminated food16,17, our assay is linear. Detection of LPS O111:B4 in ground beef homogenate (figure 2) showed an increased NSB of the antibody as compared to signal intensities recorded in buffer (data not shown). Still, the results are clearly positive, and our enhanced sensitivity allows us to detect LPS in a complex matrix with minimal sample processing. We are currently working on optimizing our insertion assay for all LPS serogroups. Then, it will be possible to proceed with detection in blind samples, and eventual incorporation into agricultural testing methods.

Figure 1. Detection of LPS O157 by membrane insertion. Each line is the specific signal associated with the respective concentration of LPS. NSB is the non-specific binding of the reporter pAb O157-af647.



LINO

soon

a _I

OM

aas

as.... ,,,

e

Figure 2. Dcase is pAb

Assays for thcross-reactivO104 was obfurther optimmembrane inmore specificthe developminterest are se

Figure 3. SO157-af64

Detection of 200b E. coli-af647.

he different serity with the co

bserved despitemization, but alsnsertion assay pc antibody dou

ment of specificerogroups O26

Serogroup speci

47, that is not spe

µg/mL LPS O1

ogroups of LPnserved region

e this. Howeverso more specifperformed bothubles the signalc, targeted antib6, O111, and O

ific detection ofecific to the LPS

11:B4 spiked in

S were developns of the LPS ar, discriminativfic antibodies. Fh with pAb O1l, enhancing sebodies for STE

O145.

f LPS. DetectionS O-Ag.

n 1mg/mL beef h

ped using the pantigen. Discrimve assays for seFor instance, fi57 and pAb O

ensitivity and aEC serogroups,

n of six subgrou

homogenate. Rep

pAb O157-af64minative detecterogroups O26

figure 4 depicts104. Results deccuracy of dete, for discrimina

ups of LPS usin

porter antibody i

47 (figure 3), ation of serogro

6, O111, and Os the comparisoemonstrate thaection. We areative detection

ng the indiscrim

in this

and exploiting oups O45, O10

O145 not only non of an LPS Oat in this case u currently work

n of LPS. Of sp

minate pAb

its 3, and

need O104 using a king on

pecific



Figure 4. CO157-af647non-specific

3.2 Shiga toxThe waveguineeded to detagainst the suamphiphilic m

Figure 5. DStx2a, and

Virulence biodirect detectimethods eith

Comparison of L7. Using a specic pAb O157-af64

xin immunoaside sandwich atermine the Loubunits of bothmembrane rece

Detection of Shi

d a pAb against S

omarkers of STion of pathogenher suffer from

LPS detection wific LPS O104 a47 reporter. Dep

ssays assay for Stx2 (oD for this assah Stx1 and Stx2eptor Gb3 to m

iga toxin 2 by saStx2b as a report

TEC, such as Lns in contamin

m non-specificit

while using an Oantibody doublepicted NSB for th

(100 μg/mL) reay. We are curr2, as well as ex

maximize sensit

andwich immunter.

4. CO

LPS and Stx arnated samples. ty in detection

O-Ag specific ps the detection his graph is from

esulted in a sigrently working xploring an altetivity

noassay. Capture

ONCLUSION

re released intoWhile many a

n or the inabilit

pAb O104-af647signal as compa

m pAb O104-af6

nal:NSB of 5 (to optimize th

ernative memb

e of Stx2 uses a

NS

o the infected happroaches for ty to accuratel

7 and a non-speared to detection

647

(figure 5). Furthe assay using mrane insertion

biotinylated mA

host, and therebdetection and

ly diagnose act

ecific pAb n with the

ther replicates amultiple antibostrategy, using

Ab against

by ideal candidserotyping exitive infection.

are odies g the

dates for ist, these Further,



methods developed for protein biomarker detection are traditionally applied to lipoglycans like LPS, resulting in poor sensitivity of detection. Herein, we present a new approach for the detection of amphiphilic lipoglycans like LPS, by membrane insertion. By performing membrane insertion assays for LPS on the waveguide biosensor platform, and coupling that to detection of protein virulence factors of STEC, we can accurately detect STEC with the requisite sensitivity, and discriminate between serotypes of STEC in beef lysates. Additionally, we are investigating the optimization of selective functional surfaces and highly specific antibodies against LPS O-Ags to increase the sensitivity. This will minimize cross-reactivity and enhance the specificity of the assay. Our work represents the first steps towards direct biomarker detection of STEC in beef lysates to mitigate the outbreaks of this foodborne pathogen.

5. ACKNOWLEDGEMENTS

The authors would like to thank several people who have dedicated time towards helping with the ideas and development of this paper, as well as assisting with data analysis and training. Thanks to Dr. Basil I Swanson, Karen Grace and W. Kevin Grace, of LANL, who developed the platform. Thanks to Dr. Swanson and Dr. Chaudhary, who together with Dr. Mukundan, invented the membrane insertion assay. Many thanks to Drs. Andrew Shreve, DJ Perkins, Aaron Collins, and Pearlson Prashanth. Additionally we would like to thank Mr. Aaron Anderson, Mr. Matthew Rush, Mrs. Gentry Lewis, and Ms. Kirstie Swingle. Thanks to the STEC Center at Michigan State University, as well Dr. John Luchansky for provision of STEC strains. The following reagent was obtained through BEI Resources, NIAID, NIH Polyclonal Anti-Shiga Toxin 2 Subunit B (IgG, Rabbit), NR-9352. This project was supported, in part, by Agriculture and Food Research Initiative Competitive Grant no. 2012-68003-30155 from the USDA National Institute of Food and Agriculture.

6. REFERENCES

[1] Scallan, E., Hoekstra, R.M., Angulo, F.J., Tauxe, R.V., Widdowson, M.-A., Roy, S.L., Jones, J.L., and Griffin, P.M., “Foodborne illness acquired in the United States—major pathogens,” Emerging Infectious Diseases 17(1), 7–15 (2011).

[2] Scallan, E., Jones, T.F., Cronquist, A., Thomas, S., Frenzen, P., Hoefer, D., Medus, C., Angulo, F.J., and Angulo, F.J., “Factors associated with seeking medical care and submitting a stool sample in estimating the burden of foodborne illness.,” Foodborne pathogens and disease 3(4), 432–438 (2006).

[3] Ethelberg, S., Olsen, K.E., Scheutz, F., and Jensen, C., “Virulence factors for hemolytic uremic syndrome, Denmark,” Emerg Infect … (2004).

[4] [Shiga Toxin-Producing Escherichia coli in Certain Raw Beef Products], 9 part 416, 417, 430, 31975–31981 (2012).

[5] Paton, J.C., and Paton, A.W., “Pathogenesis and diagnosis of Shiga toxin-producing Escherichia coli infections,” Clinical Microbiology Reviews (1998).

[6] Wang, L., Wang, Q., Reeves, P.R., 2010. The Variation of O Antigens in Gram-Negative Bacteria, in: Subcellular Biochemistry, Subcellular Biochemistry. Springer Netherlands, Dordrecht, pp. 123–152. doi:10.1007/978-90-481-9078-2_6

[7] Duveneck, G.L., Abel, A.P., Bopp, M.A., and Kresbach, G.M., “Planar waveguides for ultra-high sensitivity of the analysis of nucleic acids,” Analytica Chimica … (2002).

[8] Martinez, J.S., Grace, W.K., Grace, K.M., Hartman, N., and Swanson, B.I., “Pathogen detection using single mode planar optical waveguides,” Journal of Materials Chemistry 15(43), 4639 (2005).

[9] Mukundan, H., Anderson, A.S., Grace, W.K., Grace, K.M., Hartman, N., Martinez, J.S., and Swanson, B.I., “Waveguide-Based Biosensors for Pathogen Detection,” Sensors 9(7), 5783–5809 (2009).

[10] Mukundan, H., Xie, H., Anderson, A.S., Grace, W.K., Shively, J.E., and Swanson, B.I., “Optimizing a Waveguide-Based Sandwich Immunoassay for Tumor Biomarkers: Evaluating Fluorescent Labels and Functional Surfaces,” Bioconjugate Chemistry 20(2), 222–230 (2009).

[11] Mukundan, H., Kumar, S., Price, D.N., Ray, S.M., Lee, Y.-J., Min, S., Eum, S., Kubicek-Sutherland,



J., Resnick, J.M., et al., “Rapid detection of Mycobacterium tuberculosis biomarkers in a sandwich immunoassay format using a waveguide-based optical biosensor.,” Tuberculosis (Edinburgh, Scotland) 92(5), 407–416 (2012).

[12] Mukundan, H., Xie, H., price, D., Kubicek-Sutherland, J.Z., Grace, W.K., Anderson, A.S., Martinez, J.S., Hartman, N., and Swanson, B.I., “Quantitative Multiplex Detection of Pathogen Biomarkers on Multichannel Waveguides.,” Analytical Chemistry 82(1), 136–144 (2009).

[13] Sakamuri, R.M., Capek, P., Dickerson, T.J., Barry, C.E., Mukundan, H., and Swanson, B.I., “Detection of stealthy small amphiphilic biomarkers.,” Journal of Microbiological Methods (2014).

[14] Adams, P.G., Lamoureux, L., Swingle, K.L., Mukundan, H., and Montaño, G.A., “Lipopolysaccharide-Induced Dynamic Lipid Membrane Reorganization: Tubules, Perforations, and Stacks,” Biophysical Journal 106(11), 2395–2407 (2014).

[15] Hamley, I.W., “Self-assembly of amphiphilic peptides,” Soft Matter (2011). [16] Berczi, I., Bertok, L., and Bereznai, T., “Comparative studies on the toxicity of Escherichia coli

lipopolysaccharide endotoxin in various animal species,” … journal of microbiology (1966). [17] Cohen, J., “The immunopathogenesis of sepsis,” Nature (2002).



An optical biosensor for detection of pathogen biomarkers from Shiga toxin-producing Escherichia coli in ground beef samples

Documents