Author's Accepted Manuscript

Development of a microchip europium nanoparticle immunoassay for sensitive point-ofcare HIV detection Jikun Liu, Bingchen Du, Panhe Zhang, Mohan Haleyurgirisetty, Jiangqin Zhao, Viswanath Ragupathy, Sherwin Lee, Don L. DeVoe, Indira K. Hewlett www.elsevier.com/locate/bios

PII: DOI: Reference:

S0956-5663(14)00324-8 http://dx.doi.org/10.1016/j.bios.2014.04.057 BIOS6761

To appear in:

Biosensors and Bioelectronics

Received date: 4 February 2014 Revised date: 23 April 2014 Accepted date: 29 April 2014 Cite this article as: Jikun Liu, Bingchen Du, Panhe Zhang, Mohan Haleyurgirisetty, Jiangqin Zhao, Viswanath Ragupathy, Sherwin Lee, Don L. DeVoe, Indira K. Hewlett, Development of a microchip europium nanoparticle immunoassay for sensitive point-of-care HIV detection, Biosensors and Bioelectronics, http://dx. doi.org/10.1016/j.bios.2014.04.057 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Development of a Microchip Europium Nanoparticle Immunoassay for Sensitive Point-of-Care HIV Detection Jikun Liu1*, Bingchen Du1, Panhe Zhang1, Mohan Haleyurgirisetty1, Jiangqin Zhao1, Viswanath Ragupathy1, Sherwin Lee1, Don L. DeVoe2, Indira K. Hewlett1* 1.

Laboratory of Molecular Virology, Center for Biologics Evaluation and Research, Food and Drug Administration, Bethesda, MD 20892.

2.

Department of Mechanical Engineering, University of Maryland, College Park, MD 20842

Abstract Rapid, sensitive and specific diagnostic assays play an indispensable role in determination of HIV infection stages and evaluation of efficacy of antiretroviral therapy. Recently, our laboratory developed a sensitive Europium nanoparticle-based microtiter-plate immunoassay capable of detecting target analytes at subpicogram per milliliter levels without the use of catalytic enzymes and signal amplification processes. Encouraged by its sensitivity and simplicity, we continued to miniaturize this assay to a microchip platform for the purpose of converting the benchtop assay technique to a point-ofcare test. It was found that detection capability of the microchip platform could be readily improved using Europium nanoparticle probes. We were able to routinely detect 5 pg/mL (4.6 attomoles) of HIV-1 p24 antigen at a signal-to-blank ratio of 1.5, a sensitivity level reasonably close to that of microtiterplate Europium nanoparticle assay. Meanwhile, use of the the microchip platform effectively reduced sample/reagent consumption 4.5 fold and shortened total assay time 2 fold in comparison with microtiter plate assays. Complex matrix substance in plasma negatively affected the microchip assays and the effects could be minimized by diluting the samples before loading. With further improvements in sensitivity, reproducibility, usability, assay process simplification, and incorporation of portable timeresolved fluorescence reader, Europium nanoparticle immunoassay technology could be adapted to meet 1

the challenges of point-of-care diagnosis of HIV or other health-threatening pathogens at bedside or in resource-limited settings. Keywords Europium nanoparticles; microchip; point-of-care diagnostics; HIV; resource-limited settings

*Correspondent authors: Dr. Jikun Liu and Dr. Indira Hewlett Laboratory of Molecular Virology CBER / FDA, Building 29B, Room 4NN16 8800 Rockville Pike Bethesda, MD 20892 Tel: +1-301-827-0795 Fax: +1-301-480-7928 E-mail: [email protected], [email protected]

2

1. Introduction As reported by UNAIDS, there were 34.2 million people living with HIV and 2.5 million incident cases globally, while 1.7 million people died from AIDS-related diseases. In HIV-infected population, 68% was in sub-Saharan Africa. 91% of Children under 15 years old carrying HIV virus were also living in this area. Despite the striking facts of the pandemic, however, it is very encouraging that with help of persistent global efforts against HIV/AIDS, the cost of antiretroviral therapy (ART) is continuously dropping and the essential treatment becomes more accessible to HIV-infected individuals residing in developing countries and resource-limited regions, resulting in remarkable declines in both incidence and death rates. (http://www.unaids.org/en/resources/presscentre/factsheets/) In typical ARTs, patients who are diagnosed HIV positive need to take antiretroviral drugs regularly to suppress the viral loads in their life spans. This life-long treatment scenario is similar to those for chronic diseases,(Colvin 2011; Wainberg and Jeang 2008) and its practice requires the combined use of HIV diagnostic techniques and sufficient supply of highly effective antiretroviral (ARV) drugs. A sensitive, specific and reliable HIV diagnosis technique plays an equally important role as ARV drugs in this strategy.(Wong and Hewlett 2010) Through continuous disease monitoring using HIV diagnosis technique, clinicians are able to determine the most appropriate timing of therapy to stop the progression of HIV disease in infected individuals and thus reduce their infectiousness to healthy populations. Enzyme-linked immunosorbent assay (ELISA) and nucleic acid test (NAT) are two major assay techniques widely employed for HIV/AIDS diagnosis in hospitals and laboratories.(Iweala 2004; Wong and Hewlett 2010) ELISA is commonly used for screening purposes. Its newest version, 4th generation HIV combo immunoassay, detects both HIV IgM/IgG antibodies and HIV capsid protein p24 antigen (detection limit: 4~10 pg/mL (Speers et al. 2005)) and thus enabling the detection of early HIV infection. NAT is by far the most sensitive and accurate HIV diagnosis platform (detection limit: 30~60

3

copies/mL (Busch et al. 2005)). It serves in early diagnosis of HIV infection in infants, monitoring infection progress, blood donor screening and confirmatory tests. A laboratory HIV diagnosis algorithm combining the combo immunoassay and NAT was recommended recently. (http://www.cdc.gov/hiv/testing/lab/guidelines/index.html) In the initial step, the combo immunoassay is used to screen for acute HIV-1 infection and for established HIV-1/2 infections. Positive specimens will receive further inspection using HIV antibody immunoassay to differentiate HIV-1 infection from HIV2 infection. Specimens positive on the combo immunoassay and negative on the antibody assay should be tested with NAT to confirm the infection. Unfortunately, application of both ELISA and NAT techniques requires abundant resources. The high cost, long assay time, complex instrumentation, requirement for well-maintained environment and highly trained professionals for testing limits access of these technologies to the less-developed world. As UNAIDS data indicate, most HIV-infected populations in need of ART are located in low- or middle- income countries and regions. Therefore, one must balance multiple factors in selecting a proper HIV diagnosis technique for use in these areas. Besides accuracy and reliability, a suitable HIV diagnosis technique for resource-limited settings (RLS) should also have attributes including low cost, short turn-around time, low reagent/energy consumption and simplicity. Portability and disposability are also desirable.(Allain and Lee 2005; Wong and Hewlett 2010) The single-use lateral flow assay (LFA)/immunochromatographic strip system is a widely accepted rapid point-of-care (POC) diagnostic platform in RLS.(Desai et al. 2011; Posthuma-Trumpie et al. 2009) The core of a LFA system for HIV detection is a filter paper or porous membrane strip carrying detection zone immobilized with anti-HIV antibodies. Presence of HIV antibodies in samples will initiate specific capture events in the zone and visualization of the capture is implemented by infusing gold nanoparticles (AuNPs) conjugated with anti-IgG/IgM antibodies through the detection 4

zone. No reader is necessary to detect the immuno-sandwiches formed since the color of AuNPs is visible to the naked eyes.(Wong and Hewlett 2010) Unfortunately, most LFA diagnosis platforms, including the recently developed rapid POC HIV antigen/antibody combo test, suffer from relatively low analytical sensitivity or high detection limit, an adversary factor that often compromises their performances in detection of early/acute HIV infections.(Desai et al. 2011; Gordon and Michel 2008) LFA platform alone is incapable of accurate quantitative analysis, large-scale multiplex pathogen detection and performing multi-step assays, thus further limiting their usage in diagnosis of HIV coinfections and identification of different HIV strains. Several solutions aiming at overcoming the problems haunting the current LFA systems have been proposed, which include using more sensitive bioconjugated probes, introducing flow control to core strip, integrating functional components such as detectors to LFA system, and developing a new generation of POC tools based on lab-on-achip/microchip (LoC) concept.(Chin et al. 2012; Corstjens et al. 2007; Gordon and Michel 2008) LoC is one subset of microelectromechanical system (MEMS) with emphasis on chemical/ biological processes.(Reyes et al. 2002) Shortly after its introduction in early 1990s the research on LoC rapidly attracted great attention and ignited widespread enthusiasm in broad areas such as chemistry, biology, medicine, material science, bioengineering and etc.(Manz et al. 1990; Reyes et al. 2002) The root of this excitement resides in the fascinating benefits LoC can offer. With downscaling of geometric dimension, diffusion distance shortens, surface-to-volume ratio increases and heat capacity reduces, leading to faster mass/heat transfer and thus higher reaction rate. Dimension shrinkage also brings extra advantages such as lower sample/reagent/energy consumption, lower waste generation and smaller instrument footprint. In addition, multiple functional modalities or same assay units can be miniaturized and packaged in one cartridge to perform complex assays or high-throughput parallel analysis, respectively. Recognizing the tremendous impacts and unique opportunities LoC will bring to global 5

health, investigators from both academia and industry have been working diligently to develop novel LoC diagnostic devices and transform proof-of-concept devices to practical POC tools.(Chin et al. 2012; Desai et al. 2011; Yager et al. 2006) To mention a few of such efforts, Sia’s group developed an immunoassay based microfluidic POC cartridge to simultaneously detect HIV and Syphilis and field testing in Rwanda has validated its effectiveness in RLS.(Chin et al. 2011) Kim et al introduced an immunofluorescence microtip chip system for rapid detection of Mycobacterium tuberculosis complex cells in sputum.(Kim et al. 2012) While Cheng et al employed a cell isolation microfluidic device to separate CD4+ T cells from 10 µL of blood and count the cell number through either microscopy and impedance sensing.(Cheng et al. 2009) Soh’s group reported a disposable microfluidic chip for sampleto-answer genetic analysis of H1N1 virus,(Ferguson et al. 2011) Lutz’s group demonstrated a fully automated microfluidic centrifugal disc system capable of detecting antibiotic resistance gene mecA of Staphylococcus aureus using isothermal recombinase polymerase amplification, (Lutz et al. 2010) and Lin’s group developed a highly sensitive microchip colorimetric assay for detecting a biomarker of Alzheimer’s disease using rolling circle amplification and G-quadruplex DNAzyme.(Lin et al. 2014) Some LoC devices are sufficiently mature and have been launched into market. Typical FDA approved examples include Abbott’s i-STAT® blood analyzer, Alere’s Pima® CD4 counter, Focus (Quest) Diagnostics’ 3MTM Integrated Cycler, and IQuum’s LiatTM rapid NAT analyzer. Most core assays for pathogen detection on LoC POC systems can be categorized into immunoassays and NATs, and the detection mechanisms are either label-free or requiring sensing probes. Label-free detection approaches are ideal for simple, rapid and multiplex pathogen diagnosis.(Emaminejad et al. 2012; Stern et al. 2010; Vasan et al. 2013; Yoon and Kim 2012) However, their accuracy, reliability, reproducibility and cost-effectiveness are yet to be examined in “real-world” applications. To date, detection based on visible, fluorescent, luminescent or electrochemically active

6

molecules sensing probes is still prevalent in most diagnostic assays. With rapid progress of nanotechnology, many nanomaterials possessing excellent physical and chemical properties have been discovered and introduced for high sensitivity detection, including quantum dots, gold/silver nanoparticle (NPs), carbon nanoparticles and carbon nanotubes. The Europium (III) chelate –dyed nanoparticles (Eu NPs) used in this work are synthesized composite nanospheres with an average diameter of 107 nanometers. Each Eu NP has a carboxylated polystyrene shell to encapsulate more than 104 high-quantum-yield Eu (III) ions chelated by β-diketones.(Harma et al. 2001) These lanthanide chelates have relatively high photostability and resistance to self-quenching effect, ( therefore, Eu NPs can emit stable and extremely high-intensity fluorescence under proper excitation.(Ukonaho et al. 2007) Additionally, the functionalized polymer shell provides great amount of chemical handles for bioconjugation and a protective barrier to prevent other strong chelating ligands in solution (e.g., EDTA) from destructing the highly fluorescent Eu (III) chelates. A large 300 nm Stokes shift exists in Eu NPs, allowing convenient separation of excitation from Eu NP emission wavelengths in measurements. When time-resolved fluorometry (TRF) is used in measurement, rapidly decayed background fluorescence from matrixes can be temporally decoupled from long-lifetime Eu NP fluorescence. All these features facilitate assays incorporating Eu NPs as probes to achieve high sensitivity. Harma et al reached a detection limit of 0.38 pg/mL (10 fM) in a prostate-specific antigen (PSA) model assay using streptavidin-conjugated Eu NPs (EuSA).(Harma et al. 2001) The same group repeatedly demonstrated great sensitivity gain via Eu NPs in a series of continuing studies.(Harma et al. 2008; Harma et al. 2010; Pihlasalo et al. 2011; Pihlasalo et al. 2012) A trend of exploiting highly sensitive Eu NPs on lateral flow immunoassay or LoC platforms also started to appear recently.(Cesaro-Tadic et al. 2004; Juntunen et al. 2012; Mantymaa et al. 2011; Xia et al. 2009)

7

Our group recently established ultrasensitive Eu-NP based immunoassays (ENIA) for detection of anthrax toxin and HIV-1 p24 antigen, with detection limits reaching low pg/mL or even sub-pg/mL levels.(Tang et al. 2009; Tang et al. 2010) We found in ENIA studies that due to its high sensitivity, lengthy multi-step signal amplification process such as those found in ultrasensitive bio-barcode nanoparticle assay (Tang et al. 2007; Thaxton et al. 2005) were unnecessary. Employment of EuSA in ENIA also obviated the use of labile catalytic enzymes crucial for sensitive detection in conventional ELISA. An added benefit is that the procedures of ENIA are similar to those of standard ELISA, allowing rapid adaption of this assay to regular laboratories and clinics. However, ENIA requires relatively large amounts of samples/reagents. Its relatively long assay time and usage of bulky auxiliary instruments also limit its direct adoption in RLS for pathogen detection. As the first step to transform ENIA to a practical POC technique, we miniaturized the assay on a prototype LoC/microchip platform and evaluated performances of the microchip ENIA (µENIA) in HIV assays. Several important factors that would affect µENIA are studied and the results are summarized in this work. Further optimization efforts necessary to develop µENIA to a mature assay platform for POC diagnosis of infectious diseases are also discussed in this work. 2. Experiments 2.1 Materials. Dow Corning Sylgard 184® polydimethylsiloxane (PDMS) kit was ordered from Ellsworth Adhesives (Germantown, WI, USA). Smooth-Cast 310® liquid polyurethane (PU) resin was purchased from Smooth-On (Easton, PA, USA). 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC), sulfo-N-hydroxysulfosuccinimide (sulfo-NHS), hydroxylamine, phosphate buffered saline solution (PBS), ELISA plate wash solution, Fluoro-Max® carboxylate-modified Europium nanoparticles (diameter: 107 nm), StartingBlock T20 (PBS) blocking buffer (PBST), bovine serum albumin (BSA), biotinylated rabbit anti-human IgG (H+L) secondary antibody (cat# 31786), 8

Neutravidin and NUNE MaxiSorp® 96-well ELISA plates were obtained from Thermo Fisher Scientific (Rockford, IL, USA). Pluronic® P123 (P123) and biotinylated rabbit anti-mouse IgG were purchased from Sigma-Aldrich (St. Louis, MO, USA). (Tridecafluoro-1,1,2,2-tetrahydrooctyl)-trichlorosilane was ordered from Gelest(Morrisville, PA, USA). Recombinant p24 protein standard and biotinylated HIV-1 p24 antigen detector antibody was purchased from Perkin-Elmer (Waltham, MA, USA). Anti-p24 monoclonal antibody (cat# C65690m) was acquired from Meridian Life Science (Memphis, TN, USA). Glass fiber pads and nitrocellulose membranes HF90 and HF180 were bought from Millenia Diagnostics (San Diego, CA, USA). Chromatography paper was purchased from Whatman (Sanford, ME, USA). 2.2 Preparation of bioconjuated Europium (III) nanoparticles. EDC/sulfo-NHS chemistry was used to covalently link primary amines of streptavidin to the carboxylic groups on the surface of Eu NPs. Briefly, Eu NPs were first added to 10 mM phosphate buffer (pH 7.0) with 10 mM EDC and incubated for 10 min, followed by activation with 10 mM sulfo-NHS in the same buffer for 20 min. After a buffer exchange with 10 mM carbonate buffer (pH 9.0), 2 mg/mL streptavidin in the carbonate buffer was added to the activated Eu NPs and the reaction was allowed to proceed for 2 h. The remaining active groups on EuNP surface were blocked with 10 mM hydroxylamine for 30 min. Bioconjugated Eu NPs were washed 5 times with 10 mM glycine buffer (pH 8.5) and stored at 4 oC. Buffer exchange and EuNP wash were implemented using NanoSep® centrifugal ultrafiltration devices with a molecular weight cutoff of 300kDa (Pall Life Sciences, Ann Arbor, MI, USA). All reactions were performed at room temperature. 2.3 µENIA of p24 antigen and HIV antibodies. Capture antibody or antigen was dissolved in a PBS (pH 7.2) solution and loaded to a freshly prepared microchip with a pipette gun. The device was incubated at 4 oC for at least 24 hours and the capture antibody or antigen coating solution was replaced

9

with PBST solution for blocking non-specific adsorption sites on the microreactor and microchannel walls. PBST solution was evacuated after incubation for 40 min at room temperature or 20 min at 37 oC. In step 1 of HIV-1 p24 antigen and HIV antibody assays, sample solutions were loaded to PBSTblocked microchips and the devices were kept at 37 oC for 15 min (antigen assay) or 30 min (antibody assay). In step 2, biotinylated secondary antibody solutions were filled in the devices following a wash step and the on-chip reactions were allowed to proceed at 37 oC for 30 min. In step 3, EuSA solution was loaded to the microchips after a second wash and the biotin-Neutravidin coupling reactions were incubated again at 37 oC for 15 min. The devices were subjected to a final wash and then placed on a modified rack for time-resolved fluorescence signal reading in a Perkin-Elmer 1420 Victor3V multilabel counter. Fluid pumping in the microchip assays was achieved manually using a rubber pipette filler (Thermo Fisher Scientific) connected to the devices through a section of PTFE tubing (1/32” ID × 1/16” OD, Cole-Plamer, Vernon Hills, IL, USA). 3. Results and Discussion 3.1 Fabrication of testing microchip. A large number of materials are available for microdevice fabrication. We conducted a series of screening experiments to identify the most appropriate substrate materials based on costs, fabrication easiness and optical properties. We finally chose PDMS as µENIA device substrate in our initial tests. (Microfabrication process can be found in Supplementary Material) Current PDMS testing microchip contains 12 independent microreactors and each one has a total volume of 22 µL. We initially employed a microreactor format having a 10-cm-long double spiral microchannel and a total volume of 0.9 µL in µENIA. Unfortunately, the microchips were unable to reliably detect sub-20 pg/mL level of target analytes in p24 antigen assay, possibly because the small microreactor did not have enough volume to contain a sufficient number of immunoreaction events 10

detectable by the reader in one sample loading. The sensitivity issue associated with this type of microchip may be solved by adopting a more sensitive yet costly reader and loading more samples via multiple manual injections or a precise pump to increase immunoreactivity in the devices. However, these approaches are apparently not favored in POC settings since they require complex instruments, intensive human intervention and long assay time. We finally modified the microreactor to a serpentine channel capable of accepting relatively large volume of sample in one loading. Our µENIA tests showed that this modification simplified sample/reagent loading and effectively enhanced assay sensitivity. 3.2 Preparation of microchip platform for ENIA. The first step to implement ENIA on LoC platform is to immobilize highly specific capture proteins on PDMS microchannel surface. Two major categories of approaches, chemical bonding and physical adsorption, are available for this purpose.(Kim and Herr 2013; Lin et al. 2014; Liu et al. 2011; Schramm et al. 1993; Zhou et al. 2010) The chemical approaches provide strong and durable biomolecule anchoring leading to improved assay performances, however, multiple steps are usually required in most methods and stringent quality controls are essential to achieve high immobilization density as well as uniform coverage. As a comparison, immobilization based on direct physical adsorption is a simpler and more reproducible method. Because of its convenience and reasonable assay results, we adopted the approach in this evaluation work to coat capture species to our disposable, single-use microdevices. PBS (pH 7.2) and sodium bicarbonate (pH 9.4) buffers are commonly used in antibody coating and we were able to coat antibodies to ELISA microtiter plates in both buffers. However, we found that no signals were detected in µENIA tests if sodium bicarbonate buffer was used as antibody coating buffer, whereas good results were repeatedly obtained when the coating buffer was replaced with PBS. We suspect that the basic condition of sodium bicarbonate buffer rendered the capture antibodies

11

negatively charged, which dramatically reduced their physical interaction with highly hydrophobic PDMS surface and thus total number of capture antibodies remaining on microreactor surface. Following studies on immobilization of capture species, we evaluated the blocking effect of 1% BSA, 1% BSA/0.1% P-123 mixture and PBST, a commercial ELISA plate blocker containing a small surfactant Tween 20 and a proprietary blocking protein. However, our assay results suggest that actually all the blockers can effectively depress non-specific adsorption of EuSA on microreactor surface in µENIA. However, we finally chose PBST simply because air bubbles could form and become trapped in microchips treated with BSA blockers more easily, which negatively affected reproducible sample/reagent loading and increased assay variation. 3.3 Optimization of µENIA performances. Because of its particularly important role in early detection of HIV infection prior to seroconversion and the commercial availability of pure protein standards, we adopted the HIV p24 antigen assay to study µENIA performances. Before assays, we optimized the concentration of EuSA by measuring signal-to-blank (S/B) ratio of 50 pg/mL of p24 in µENIA using serial dilutions of EuSA (Figure 1(A)). We found that signals from both µENIA reaction and blanks increased nonlinearly with EuSA concentration. The highest S/B ratio appeared at 1×109 particles/mL. However, when the EuSA concentration was higher than this value, the signal intensity of the blank increased more rapid than that of the µENIA signal, indicating that non-specific adsorption of the EuSA to PDMS surface increased. We used the optimal EuSA concentration in all on-chip p24 antigen assays and Figure 1(B) shows a typical result of µENIA of p24 antigen. 5 pg/mL (0.21 pM) of p24 can be reproducibly detected at an S/B ratio of 1.5 on the test microchips (limit of detection: 3.7 pg/mL or 0.15 pM), with a linear dynamic range extending from 0 to 500 pg/mL. However, the signal is indistinguishable from the blank signal when p24 concentration dropped below 1 pg/mL. The signal

12

variation in the same device was below 15% (relative standard deviation, RSD) and up to 25% (RSD) for different batches of devices. B

Fluorescence Intensity

A

R2 = 0.994 0

0

0.5

100

1

5

200

300

400

500

600

p24 Concentration (pg/mL)

Figure 1 (A) Optimization of EuSA concentration. (B) Typical µENIA of HIV p24 antigen; inset shows assay results in the range from 0 to 5 pg/mL (circled area in the Figure).

To investigate the impact of various factors and optimize the performances of µENIA, we carried out a series of experiments based on the microchip HIV p24 antigen assay. We first evaluated the impact of coating concentration of capture antibody on the final µENIA results. As indicated in Figure 2, only weak responses were detected from chips coated with low concentrations of the capture antibody (2 and 5 µg/mL). When the antibody concentration was increased to 10 µg/mL, strong signals were recorded in the µENIA. However, the calibration line deviated from linearity at 200 pg/mL point. Further increase of capture antibody concentration to 20 µg/mL extended the linear dynamic range to 500 pg/mL; while both µENIA signal intensity and analytical sensitivity (slope of calibration line) remained unchanged below 200 pg/mL as compared to those obtained from devices coated with 10 µg/mL of capture antibody. Since we were more interested in the detection of low pg/mL levels of p24 in this work and no significant µENIA improvements were obtained in this concentration range when higher coating

13

concentrations were used in device preparation, in most cases, we chose solutions with an antibody concentrate of 10 µg/mL to coat microchips for the p24 antigen assay.

Figure 2 Effect of antibody coating concentrations to µENIA of p24 antigen.

In regard to the optimization of reaction time, we varied the time of a particular step while other assay conditions remained the same in µENIA. The optimal time was then identified according to the highest analytical sensitivity of the assays (Figure 3). To our surprise, the figure of merit did not monotonically increase with reaction time in a 100-min long time frame as found in parallel ENIA tests on NUNE microtiter plates. Instead, it reached a constant number within a relatively short period of time. In step 1 of the on-chip p24 antigen assay, 15 min were enough to drive analytical sensitivity to its peak value. A similar trend was observed in the optimization of step 2 with an optimal time identified to be 30 min. In the final step, we found that a reaction time longer than 15 min did not offer any significant improvement in analytical sensitivity. The short length of time needed to reach the optimum reaction time may be induced by relatively small sample consumption of our microreactor and its relatively large surface area. Since the sample volume of a microreactor is only a quarter of that of a microtiter plate 14

well while the microreactor has relatively large surface area compared to the well, under the same conditions, a faster capture antibody-antigen reaction could be favored in the microreactor. Additionally, we observed an interesting phenomenon in both step one and two that reaction time longer than optimal did not help to increase analytical sensitivity, on the contrary, the value deteriorated slightly before reaching a plateau. We suspect that this phenomenon may arise over time from the detachment of surface-bound immunocomplexes and an increase in nonspecifically adsorbed EuSA to the exposed PDMS surface.

Figure 3 Reaction time optimization for each step of µENIA of p24 antigen. In step 1, both reaction times of step 2 and 3 were fixed at 15 min; in step 2, both reaction times of step 1 and 3 were both fixed at 15 min; in step 3, reaction times for step 1 and 2 were fixed at 15 min and 30 min respectively. Each datum is an average of 3 repeats.

Finally, we optimized microchip antibody assay conditions following the same procedures for of the microchip HIV-1 p24 antigen assay. We coated the devices with p24 antigen, used anti-p24 monoclonal antibody as a standard sample, and chose biotinylated rabbit anti-mouse IgG as the secondary/detection antibody. In a typical microchip antibody assay (Figure 4), 25.0 pg/mL (0.17 pM) of antibody can be detected at an S/B ratio of 1.9 and the limit of detection was estimated to be 10.6 15

pg/mL (0.07 pM). The detection sensitivity of on-chip antibody assay is actually equivalent to that of microchip p24 antigen assay in terms of molar concentration.

Fluorescene Intensity

R² = 0.994

0

20

40

60

80

100

120

Antibody Concentration (pg/mL)

Figure 4 µENIA of anti-p24 monoclonal antibody under optimal conditions. The microchip was coated with 5 µg/mL of p24 antigen in PBS. Reaction time was 30 min for step 1, 30 min for step 2 and 15 min for step 3. EuSA concentration was 2×108 particles/mL. Each datum is an average of 3 repeats.

3.4 µENIA in complex bio-matrix. To evaluate its potential in real-sample assays, we first challenged the microchip platform with p24-spiked plasma samples. Our initial results indicate that without any pretreatment, the complex bio-matrix species of plasma greatly distorted µENIA results. In these tests we observed that the background signal intensity surged at least 4-time higher than that of clean samples, and the responses from p24 serial dilutions in plasma significantly deviated from a line in the concentration range from 0-200 pg/mL. The results imply that strong non-specific adsorption of biomatrix species occurred on PDMS surfaces and their intense competition for EuSA with p24 immunocomplexes greatly interfered with µENIA. To alleviate impact of the plasma matrix effect, we diluted the spiked plasma in PBST before loading it to chip. The data shown in Figure 5(A) indicate when the dilution level reaches 10 the µENIA result of plasma is equivalent to that of typical clean 16

samples, while the performance of µENIA in the testing of a 5-time diluted plasma is 85% of that in a clean sample assay in terms of analytical sensitivity. Plasma matrix effects become significant at dilution factors lower than 5 and the performance drops considerably. Additionally, the lowest detectable p24 concentration at an S/B ratio of 1.5 increased from 5 to over 10 pg/mL when plasma samples were diluted less than 5 times. We therefore conclude that a dilution factor of 5 is a reasonable balance point between enhancing detection sensitivity and reducing matrix effects and may be adopted in future µENIA of p24 antigen in real clinical plasma samples using the current microchip platform. A

B

Fluorescence Intensity

Analytical Sensitivity (mL/pg)

35000

600 500 400 300 200 100 0

30000 25000 20000 15000 10000 5000 0

Clean Sample

10X

5X

3X

2X

1

2

3

4

5

6

7

8

9

N1

N2

Plasma Sample

Plasma Dilution Factor

Figure 5 (A) Effect of dilution to analytical sensitivity in µENIA of p24-spiked plasma samples. (B) µENIA of HIV antibodies in plasma samples. Plasma N1 and N2 are negative controls.

Following the on-chip p24 antigen assay, we evaluated the feasibility of applying µENIA to qualitative HIV antibody detection in plasma samples. Again, a specific dilution is necessary in analysis of these samples and we used a dilution factor of 100 in the assays to avoid matrix effects while reduceing sample consumption. Nine samples were randomly chosen from a pool of HIV positive plasma specimens (proved by ELISA assay and microtiter plate ENIA) and loaded to p24-coated microchip platform after being diluted in PBST doped with 0.55% TritonX-100. Plasmas of two healthy

17

persons were subjected to the same treatment as that for other HIV positive plasmas and injected into the microchip as negative controls. The µENIA result (Figure 5(B)) indicates all HIV patient plasmas show strong positive responses, with average signal-to-negative ratios (S/N) ranging from 5 (plasma 9) to a maximum of 61 (plasma 7). This qualitative result suggests that the microchip platform has a potential to be applied to detection of HIV antibodies from real patient samples. Furthermore, one may combine the HIV p24 antigen assay together with HIV antibody assay on current microchip platform for simultaneous detection of both HIV antibodies and antigen, an essential feature most of commercial 4th generation HIV combo assays have for early detection of HIV infection. 3.5 Future developments. Several improvements are required before validation with clinical samples and field tests. First, it would be beneficial to further improve the sensitivity of the PDMS microchip. Proper surface treatments as well as good blocking strategies could help to achieve this goal by increasing the density of immobilized bio-active capture species and depressing non-specific adsorption. Although many surface treatment and blocking methods are available for these purposes,(Zhou et al. 2010) there is still a compelling need to identify simple, robust, uniform and inexpensive ones with negligible effects on physical properties of substrates and microfabrication process. Second, to mimic POC assay practice in RLS settings, we only used a pipette gun and a rubber pipette filler in µENIA to manually manipulate samples and reagents in the devices. The operation unfortunately incurred much inconvenience and fatigue to operator, and caused assay variance as well. Integration of micropumps and microvalves to the core microchip would be of great help to improve usability of future µENIA platform. Integration of sample cleanup/processing modules adds additional functionalities to the microdevice and thus enhances its capability to handle real clinical samples in POC settings. µENIA shown in this work follows a time-consuming, standard multi-step sequential assay process. We are currently working on reduce the assay steps to one with the use of EuNP-labeled 18

detection antibodies. This assay process improvement would hopefully shorten the total assay time without sacrificing assay sensitivity. Relative large variation in µENIA assay results is another issue worthy of more efforts. We believe improvements in coating uniformity of capture species, consistency of sample/reagent loading and microchannel/microreactory geometry design would partially remedy the problem. As pointed out by Harma et al (Harma et al. 2001) and Ukonaho et al,(Ukonaho et al. 2007) a more fundamental source of Eu-NP assay variation may come from EuSA aggregates. These aggregates have broad size distribution and large steric hindrance that contribute to their relative low specific binding tendency and high non-specific binding possibility. Collecting non-aggregated EuSA using sizebased purification/separation methods for the assay could reduce µENIA assay variation. Using conditions that discourage nanoparticle aggregation in assays would also help for this purpose. Finally, the reader we are currently using is a benchtop TRF fluorometer for research purposes. An ideal TRF reader suitable for POC applications would have several features including high sensitivity, compact size, simplicity, low power consumption, and ability to upload diagnostic results through wireless networks to center laboratories, to list a few. Such a detector is crucial to implementation of µENIA at the bedside or in remote regions. 4. Conclusions We successfully implemented ENIA on a PDMS microchip platform in this work. Using bioconjugated Europium nanoparticle as probes, sensitivity of microchip immunoassay can be effectively enhanced without the employment of labile catalytic enzymes and help of complex signal amplification process. We reached a reasonably low detection level of 5 pg/mL in HIV-1 p24 antigen assay using µENIA. Meanwhile, we achieved a 4.5-fold reduction in sample/reagent consumption and a 2-fold reduction in assay time compared to conventional microtiter plate assay (Supplementary Material). We demonstrated both HIV antigen and antibody assays in plasma specimens using µENIA and found 19

that without other sample processing processes, dilution was necessary to overcome interference from the complex bio-matrix. Further efforts are needed to improve sensitivity, usability and assay reproducibility. Optimization of the assay to a rapid format and identification of a portable TRF reader suitable for the platform are essential to apply µENIA to POC HIV diagnosis in RLS settings. 5. Acknowledgements This work was supported financially by the National Heart, Lung and Blood Institute (NHLBI, Nano IAA-Y1-HB-8027-01) and Oak Ridge Institute for Science and Education (ORISE). We would like to thank Drs. Xue Wang and Krishnakumar Devadas for review of the manuscript. The findings and conclusions in this article have not been formally disseminated by the Food and Drug Administration and should not be construed to represent any Agency determination or policy. 6. References Allain, J.P., Lee, H., 2005. Expert Rev Mol Diagn 5(1), 31-41. Busch, M.P., Glynn, S.A., Wright, D.J., Hirschkorn, D., Laycock, M.E., McAuley, J., Tu, Y., Giachetti, C., Gallarda, J., Heitman, J., Kleinman, S.H., 2005. Transfusion 45(12), 1853-1863. Cesaro-Tadic, S., Dernick, G., Juncker, D., Buurman, G., Kropshofer, H., Michel, B., Fattinger, C., Delamarche, E., 2004. Lab Chip 4(6), 563-569. Cheng, X., Gupta, A., Chen, C., Tompkins, R.G., Rodriguez, W., Toner, M., 2009. Lab Chip 9(10), 13571364. Chin, C.D., Laksanasopin, T., Cheung, Y.K., Steinmiller, D., Linder, V., Parsa, H., Wang, J., Moore, H., Rouse, R., Umviligihozo, G., Karita, E., Mwambarangwe, L., Braunstein, S.L., van de Wijgert, J., Sahabo, R., Justman, J.E., El-Sadr, W., Sia, S.K., 2011. Nat Med 17(8), 1015-1019. Chin, C.D., Linder, V., Sia, S.K., 2012. Lab Chip 12(12), 2118-2134. Colvin, C.J., 2011. Global Health 7, 31. Corstjens, P.L., Chen, Z., Zuiderwijk, M., Bau, H.H., Abrams, W.R., Malamud, D., Sam Niedbala, R., Tanke, H.J., 2007. Ann N Y Acad Sci 1098, 437-445. Desai, D., Wu, G., Zaman, M.H., 2011. Lab Chip 11(2), 194-211. Emaminejad, S., Javanmard, M., Dutton, R.W., Davis, R.W., 2012. Lab Chip 12(21), 4499-4507. Ferguson, B.S., Buchsbaum, S.F., Wu, T.T., Hsieh, K., Xiao, Y., Sun, R., Soh, H.T., 2011. J Am Chem Soc 133(23), 9129-9135. Gordon, J., Michel, G., 2008. Clin Chem 54(7), 1250-1251. Harma, H., Dahne, L., Pihlasalo, S., Suojanen, J., Peltonen, J., Hanninen, P., 2008. Anal Chem 80(24), 9781-9786. Harma, H., Laakso, S., Pihlasalo, S., Hanninen, P., Faure, B., Rana, S., Bergstrom, L., 2010. Nanoscale 2(1), 69-71. Harma, H., Soukka, T., Lovgren, T., 2001. Clin Chem 47(3), 561-568. 20

http://www.cdc.gov/hiv/testing/lab/guidelines/index.html. http://www.unaids.org/en/resources/presscentre/factsheets/. Iweala, O.I., 2004. Contraception 70(2), 141-147. Juntunen, E., Myyrylainen, T., Salminen, T., Soukka, T., Pettersson, K., 2012. Anal Biochem 428(1), 31-38. Kim, D., Herr, A.E., 2013. Biomicrofluidics 7(4), 41501. Kim, J.H., Yeo, W.H., Shu, Z., Soelberg, S.D., Inoue, S., Kalyanasundaram, D., Ludwig, J., Furlong, C.E., Riley, J.J., Weigel, K.M., Cangelosi, G.A., Oh, K., Lee, K.H., Gao, D., Chung, J.H., 2012. Lab Chip 12(8), 1437-1440. Lin, X., Chen, Q., Liu, W., Li, H., Lin, J.M., 2014. Biosens Bioelectron 56, 71-76. Liu, W., Wei, H., Lin, Z., Mao, S., Lin, J.M., 2011. Biosens Bioelectron 28(1), 438-442. Lutz, S., Weber, P., Focke, M., Faltin, B., Hoffmann, J., Muller, C., Mark, D., Roth, G., Munday, P., Armes, N., Piepenburg, O., Zengerle, R., von Stetten, F., 2010. Lab Chip 10(7), 887-893. Mantymaa, A., Halme, J., Valimaa, L., Kallio, P., 2011. Biomicrofluidics 5(4), 46504-4650411. Manz, A., Graber, N., Widmer, H.M., 1990. Sensor Actuat. B-Chem. 1(1-6), 244-248. Pihlasalo, S., Kirjavainen, J., Hanninen, P., Harma, H., 2011. Anal Chem 83(4), 1163-1166. Pihlasalo, S., Kulmala, A., Rozwandowicz-Jansen, A., Hanninen, P., Harma, H., 2012. Anal Chem 84(3), 1386-1393. Posthuma-Trumpie, G.A., Korf, J., van Amerongen, A., 2009. Anal Bioanal Chem 393(2), 569-582. Reyes, D.R., Iossifidis, D., Auroux, P.A., Manz, A., 2002. Anal Chem 74(12), 2623-2636. Schramm, W., Paek, S.-H., Voss, G., 1993. ImmunoMethods 3(2), 93-103. Speers, D., Phillips, P., Dyer, J., 2005. J Clin Microbiol 43(10), 5397-5399. Stern, E., Vacic, A., Rajan, N.K., Criscione, J.M., Park, J., Ilic, B.R., Mooney, D.J., Reed, M.A., Fahmy, T.M., 2010. Nat Nanotechnol 5(2), 138-142. Tang, S., Moayeri, M., Chen, Z., Harma, H., Zhao, J., Hu, H., Purcell, R.H., Leppla, S.H., Hewlett, I.K., 2009. Clin Vaccine Immunol 16(3), 408-413. Tang, S., Zhao, J., Storhoff, J.J., Norris, P.J., Little, R.F., Yarchoan, R., Stramer, S.L., Patno, T., Domanus, M., Dhar, A., Mirkin, C.A., Hewlett, I.K., 2007. J Acquir Immune Defic Syndr 46(2), 231-237. Tang, S., Zhao, J., Wang, A., Viswanath, R., Harma, H., Little, R.F., Yarchoan, R., Stramer, S.L., Nyambi, P.N., Lee, S., Wood, O., Wong, E.Y., Wang, X., Hewlett, I.K., 2010. Clin Vaccine Immunol 17(8), 1244-1251. Thaxton, C.S., Hill, H.D., Georganopoulou, D.G., Stoeva, S.I., Mirkin, C.A., 2005. Anal Chem 77(24), 81748178. Ukonaho, T., Rantanen, T., Jamsen, L., Kuningas, K., Pakkila, H., Lovgren, T., Soukka, T., 2007. Anal Chim Acta 596(1), 106-115. Vasan, A.S., Mahadeo, D.M., Doraiswami, R., Huang, Y., Pecht, M., 2013. Front Biosci (Schol Ed) 5, 39-71. Wainberg, M.A., Jeang, K.T., 2008. BMC Med 6, 31. Wong, E.Y., Hewlett, I.K., 2010. HIV Therapy 4(4), 399-412. Xia, X., Xu, Y., Zhao, X., Li, Q., 2009. Clin Chem 55(1), 179-182. Yager, P., Edwards, T., Fu, E., Helton, K., Nelson, K., Tam, M.R., Weigl, B.H., 2006. Nature 442(7101), 412-418. Yoon, J.Y., Kim, B., 2012. Sensors (Basel) 12(8), 10713-10741. Zhou, J., Ellis, A.V., Voelcker, N.H., 2010. Electrophoresis 31(1), 2-16.

21

We developed a lab-on-a-chip platform for diagnosis of early HIV infection in multiple settings. Europium nanoparticles and time-resolved fluorescence detection were used in microchip assays. High sensitivity at pg/mL or attomole levels was achieved without the aid of amplification steps. Standards and patient plasmas were tested on the microchip platform.

22