Implementing Enzyme-Linked Immunosorbent Assays on a

Subscriber access provided by FORDHAM UNIVERSITY

Article

Implementing enzyme-linked immunosorbent assays on a microfluidic chip to quantify intracellular molecules in single cells Klaus Eyer, Simone Stratz, Phillip Kuhn, Simon K. Küster, and Petra S. Dittrich Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/ac303628j • Publication Date (Web): 06 Feb 2013 Downloaded from http://pubs.acs.org on February 22, 2013

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 free 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 accessible to all readers and 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.

Analytical Chemistry 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 18

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

Implementing enzyme-linked immunosorbent assays on a microfluidic chip to quantify intracellular molecules in single cells

K. Eyer, S. Stratz, P. Kuhn, S. K. Küster, P. S. Dittrich* ETH Zurich, Department of Chemistry and Applied Biosciences, CH-8093 Zurich (Switzerland)

*Corresponding author: Petra S. Dittrich ETH Zurich, Department of Chemistry and Applied Biosciences Wolfgang-Pauli-Str. 10 CH-8093 Zurich/Switzerland e-mail: [email protected] Phone: +41 44 633 68 93 FAX: +41 44 632 12 92

ACS Paragon Plus Environment

1


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

Page 2 of 18

ABSTRACT Cell-to-cell differences play a key role in the ability of cell populations to adapt and evolve, and are considered to impact the development of several diseases. Recent advances in microsystem technology provide promising solutions for single-cell studies. However, the quantitative chemical analysis of single-cell lysates remains difficult. Here, we combine a microfluidic device with the analytical strength of enzyme-linked immunosorbent assays (ELISA) for single-cell studies to reliably identify intracellular proteins, secondary messengers or metabolites. The microfluidic device allows parallel single-cell trapping and isolation in 625 picoliter microchambers, repeated treatment and washing steps, subsequent lysis and analysis by ELISA. Using a sandwich ELISA, we quantitatively determined the concentration of the enzyme GAPDH in single U937 cells and HEK 293 cells, and found amounts within a range of a few (1-4) attomol per cell. Furthermore, a competitive ELISA is performed to determine the concentration of the secondary messenger cyclic adenosine monophosphate (cAMP) in MLT cells in response to the hormone lutropin. We found the half maximal effective concentration (EC50) of lutropin to be an average of 2.51 ± 0.44 ng/ml. Surprisingly, there were large cell-to-cell variations for all supplied lutropin concentrations, ranging from 36 to 536 attomol cAMP for non-stimulated cells and from 80 to 1040 attomol cAMP for a concentration around the EC50 (3 ng/ml). Due to the high sensitivity and specificity of ELISA and the large number of antibodies available, we believe that our device provides a new, powerful means for single-cell proteomics and metabolomics.


2

Page 3 of 18

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


INTRODUCTION Organ function is maintained by the cooperation and interaction of cells from different tissues. Cells of the same tissue type rely on equal genetic repertoires and similar properties. Hence, it is reasonable to study large cell populations or tissue samples to obtain insights into fundamental cell processes including protein expression and metabolism of nutrients and drugs. However, many studies in the past have revealed individual cell differences within a large cell population1-5. Significant heterogeneities were found for gene expression6, for the concentrations of intracellular biomolecules such as proteins7-10, membrane receptors11-13, metabolites and cofactors14, 15 and for the kinetics of cellular processes16. These individualities are considered to be fundamentally important for cell adaptation and evolution17, but also play a key role in the emergence and treatment of diseases such as cancer18-21. Therefore, studies on the single-cell level are of high interest in biological and pharmacological research, particularly if these studies reveal the different cell responses after treatment with bioactive chemical substances. In recent years, many analytical platforms have been developed that facilitate the analysis of single living cells or the chemical composition of the cell content. While fluorescent activated cell sorting (FACS) is the gold standard for very high-throughput analysis of single living cells, the method is not straightforward for the quantification of intracellular or secreted compounds. The emergence of microfluidic platforms has promised novel analytical strategies for positioning, treatment and observation of single-cells. Several studies demonstrated the potential of microdevices for the analysis of DNA or RNA, e.g. in combination with polymerase chain reaction (PCR) to enhance the sensitivity of the measurements22-24. When targeting proteins and metabolites, the analysis gets far more difficult due to the lack of suitable amplification methods, the large number of different compounds present and their variations in chemical nature. Only a few microfluidic approaches have demonstrated the analysis of secreted proteins25-27 or of fluorescent proteins within the cell5. Most intracellular biomolecules are expected to be present in low copy numbers of around a few ten thousands28, hence the analytical method must have a high sensitivity in combination with a high specificity. Both requirements are nicely provided by immunoassays, particular in enzyme-linked immunosorbent assays (ELISAs), where the signal amplification is provided by the enzyme-catalyzed conversion of a substrate to a fluorescent product (or a chemiluminescent product). Immunoassays and ELISAs are powerful methods for identification and quantification of target molecules, but require several washing and incubation steps as well as surface immobilization. Combining ELISA and microfluidic devices has been recently presented for the analysis of secreted proteins from a cell culture (THP-1 cells)27 and single (immune) cells18. The latter device was employed for an immunoassay to identify intracellular proteins for the analysis of signaling pathways in tumor cells19. However, only relative quantification was possible and no enzymatic amplification was used to increase the signal for low abundance proteins. Previously, Quake and coworkers used flexible PDMS valves in combination with microfluidic channels29, 30. Based on these studies, our group recently presented a device for the determination of intracellular enzymes and cofactors on the single-cell level15. The detection of these molecules relied on specific fluorescent assays, which cannot be used for a wider range of biomolecules. Here, we introduce a more general method to quantify intracellular compounds by combining microfluidic cell trapping and isolation with the analytical power of ELISAs. Antibodies are immobilized on the surface


3


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

Page 4 of 18

of a microdevice, which captures the target molecules after cell lysis, and the addition of enzymelinked antibodies allows quantification of the bound target molecules by fluorescence microscopy. Since the cells are trapped in microsized hurdle structures in the center of a microchamber (Figure 1), they can be repeatedly treated and washed before lysis. After lysis and addition of the immunoassay compounds, the small volume of the microchambers prevents dilution of the target molecules and the fluorescent product and hence, guarantees a high sensitivity. In the following, we demonstrate the performance of the device by carrying out a sandwich ELISA, targeting the enzyme glycerylaldehyde 3-phosphate dehydrogenase (GAPDH), an essential enzyme in the glycolysis, which is expected to show little cell-to-cell variations. Furthermore, we used a competitive ELISA for the quantification of cAMP, which cannot be accessed in the standard sandwich formats. Here, intracellular cAMP and enzyme-linked cAMP compete for free binding sites at the immobilized anticAMP antibodies. MATERIALS AND METHODS A detailed description of the chip fabrication and assembling as well as fluid control is given in the supporting information. Materials SU-8 and developers for the resists were purchased from Microchem (Newton, MA, USA). 1H,1H,2H,2H-perfluorodecyl-dimethylchlorosilane was obtained from ABCR (Karlsruhe, Germany), and poly(dimethylsiloxane) (PDMS) (Sylgard 184) from Dow Corning (Midland, MI, USA). Amplex Red, streptavidin-Texas Red conjugate, blasticidin HCL, bovine serum albumin (BSA), Calcein-AM, cell dissociation buffer, Dulbecco's Modified Eagle Medium (DMEM), Glutamax 100x, Pen/Strep 100x, phosphate buffered saline (PBS), propidium iodide, RPMI 1640 and zeocin were all obtained from Invitrogen. Fetal bovine serum (dialyzed) was obtained from PAA laboratories (Pasching, Austria). GFP antibody, monoclonal anti-GAPDH antibody (mAB), polyclonal anti-cAMP antibody and polyclonal anti-GAPDH antibody horseradish peroxidase labeled (pAB-HRP) were obtained from GenScript (Piscataway, NJ, USA). Avidin was obtained from AppliChem (Gatersleben, Germany). Biotinylated BSA (bBSA), adenosine 3′,5′-cyclic monophosphate (cAMP), fluorescein, glyceraldehyde 3-phosphate dehydrogenase (GAPDH), magnesium chloride hexahydrate, potassium chloride, protein G-Biotin conjugate and Tetracycline HCl were all obtained from Sigma-Aldrich (Buchs, Switzerland). Dimethylsulfoxide (DMSO) was from Chemie Brunschwig AG (Basel, Switzerland). Green fluorescent protein (GFP) was obtained from MBL (Woburn, MA, USA). Hydrochloric acid and methanol were from VWR (Dietikon, Switzerland). Hydrogen peroxide 30% was obtained from MERCK Serono (Zug, Switzerland). MLT cells and luteinizing hormone (lutropin) were a generous gift from Merck Serono. Resorufin was obtained from ABCR (Karlsruhe, Germany). Tris(hydroxymethyl)aminomethane (TRIS) as well as Tween 20 were obtained from BIORAD (Richmond, CA, USA). Surface functionalization Binding sites were produced on the glass slide by micro-contact printing. Briefly, PDMS stamps containing posts of 2 µm height and 30 µm diameter were wetted with the bBSA/BSA solution. For the detection of GFP, 1% bBSA to BSA was used. For the competitive immunoassay, 0.057% bBSA to BSA was used. After half an hour of incubation, the stamps were thoroughly cleaned with ddH2O and dried with a nitrogen stream. After placing the stamps on a cleaned glass slide, they were exposed,


4

Page 5 of 18

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


together with the top part of the microfluidic chip, to oxygen plasma for 45 s at maximum power (18 W) in a plasma cleaner (Harrick Plasma Cleaner PDC-32G). After plasma treatment, the stamp was removed and the printed surface was aligned underneath the microchambers of the microdevice. The chip was then placed on a hot plate at 50°C for 30 minutes. To block the remaining surface, sterile-filtered BSA solution (4% (w/v)) was pipetted into the inlet reservoirs and filled into the chip by centrifugation (800 g, 5 min). The use of centrifugal force eases the complete filling of the microchannels. BSA was incubated for 1 hour and subsequently the chip was flushed with PBS. At this point, the chip was either used directly or could be stored for up to one week at 4°C without loss of activity. The reagents (avidin:streptavidin, protein G, antibody) were loaded into the reservoir and sucked into the chip using a syringe pump (neMESYS, Cetoni, Germany). The flow rate for the following steps was always 5 µl/min (see also Table 1). A 0.025 % (w/v) solution of avidin and streptavidin-Texas Red (10:1) was flushed through the chip for 30 minutes. The fluorescent tag was used to control the uniformity of the binding spots (Figure 1e, S2). Afterwards, the chip was washed with PBS for 10 minutes. Then a solution of biotinylated protein G (0.0025 % (w/v)) was flushed through for 30 minutes. After another washing step with PBS, the antibody was introduced (0.0001 % (w/v)). After 5 minutes, the flow was stopped for 15 minutes to allow the antibody to bind to the protein G and the device was again washed with PBS. Cell culture HEK 293 cells modified with the tetracycline repressor expression system were grown in DMEM supplemented with 10% FBS, 1% Glutamax, 1% Pen/Strep and the two selection antibiotics blasticidin HCl (10 µg/ml) and zeocin (100 µg/ml). U937 and MLT cells were grown in RPMI 1640 supplemented with 10% FBS (dialyzed), 1% Glutamax and 1% Penicillin/Streptomycin. On the day of the experiment, cells were harvested in enzyme free dissociation buffer (Invitrogen, with 25 µM Calcein AM) at a concentration of 0.3 million cells/ml and filtered through a Cell Tricks 20 µm filter (Partec GmbH, Görlitz, Germany). The cells were introduced into the chip at 20 µl/min for 5 minutes and isolated. For all cell lines, the lysis buffer consisting of 10 mM Tris HCl, 0.1 mM MgCl2, 1 mM KCl and 0.1% Tween 20 was supplied to the trapped cells. GFP expression was induced in HEK T-REx cells 12 hours before the experiment by addition of 5 µg/ml tetracycline to the medium. For the stimulation of MLT cells, the cells were cultivated in 6 well plates (300,000 cells/well). Media was changed to serum free RPMI supplemented with 500 µM IBMX and the indicated concentration of lutropin. All subsequently used solutions additionally contained 500 µM IBMX. The cells were incubated for 15 minutes at 37°C and afterwards washed with PBS and suspended in enzyme free dissociation buffer. Immunoassay for eGFP For the immunoassay, eGFP expressing HEK293 T-Rex cells were trapped, isolated and lysed on chip. Afterwards, fluorescence micrographs were taken of the closed microchambers on a TIRF microscope (Figure S1b). A 100x oil immersion objective (NA 1.47, HCX Plan Apo, Leica Switzerland) was used to create the evanescent field at the correct angle for total internal reflection (image area: 80 μm x 80 μm). The emitted fluorescence from the surface was collected by the same objective. For eGFP, the excitation wavelength was 488 nm, and fluorescence was passed through a 525 nm band pass filter.


5


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

Page 6 of 18

Fluorescence images were recorded using a front-illuminated EMCCD camera (iXon, Andor) with an exposure time of 200 ms. Images were recorded every 10 s for 20 min. Each image series was analyzed with ImageJ to determine the mean grey values of the printed spots and the interspot area over time. ELISAs (detection of GAPDH and cAMP) For the GAPDH-assay, the lysis buffer additionally contained 12.5 µg/ml HRP-labeled anti-GAPDH. For the cAMP-assay, 370 ng/ml cAMP-HRP was dissolved in the lysis buffer. After lysis, a solution of 20 mM HCl was flushed through the chip to inactivate non-specifically adsorbed HRP in the channels (outside the microchambers) for 15 minutes. The device was quickly washed with PBS. Afterwards, the freshly prepared substrate solution was introduced into the chip (0.1 mM hydrogen peroxide, 0.1 mM Amplex Red in PBS (stoechiometrical excess)). Chambers were opened quickly (500 ms opening time, 1 s delay time) and data was collected using a zoom microscope (Multizoom AZ100M, Nikon Corporation, long working distance objective 0.5x, optical zoom 2.7x). Fluorescent Micrographs were taken using a digital camera (Nikon digital SIGHT DS-Fi1, Nikon Corporation, Japan) and optical filters (for the detection of resorufin, an excitation at 555/40 nm and emission at 625/40 nm; for the detection of calcein an excitation at 470/20 nm and an emission at 550/50 nm) at exposure times of 30 seconds (resorufin) and 6 seconds (calcein). For the GAPDH ELISA, fluorescent micrographs were obtained every minute, whereas for the cAMP competitive ELISA micrographs were taken at 0, 1, 3, 5, 7, 9, 12, 15 and 18 minutes, respectively. To confirm the absence of any leakage from the chamber, a 50 µM fluorescein solution in PBS was introduced into the chip after the experiments. Fluorescent micrographs were obtained showing the absence of fluorescein fluorescence inside the chamber. To show that the valves were still functional, the chambers were quickly opened and micrographs taken afterwards showed the presence of fluorescein within the chambers. The chip including the microchambers was flushed with propidium iodine (10 µg/ml) in PBS to stain the DNA of the lysed cells. The chip was disconnected and transferred to an inverted microscope (IX70, Olympus, Switzerland), equipped with a mercury lamp and optical filters (excitation filter 525/30 nm and an emission filter at ˃ 600 nm), and fluorescence as well as brightfield images were recorded with a digital camera (EMCCD 887, Andor, Belfast UK) to ensure that single-cell lysates were analyzed (Figure S2). For the calibration, the lysis buffer with various GAPDH concentrations was introduced into the chip while the chambers were kept closed. The chambers were opened for 500 ms to introduce the antigen to the immobilized antibodies within the chamber. The assay was then performed as described above. To obtain the calibration curve for cAMP, the same procedure was performed with labeled (cAMP-HRP, 370 ng/ml) and unlabeled antigen (cAMP), dissolved in lysis buffer and mixed in different ratios off-chip. Data analysis Micrographs were analyzed using a custom MATLAB (MathWorks, Natick, MA, USA) script. Initially, the micrograph of the chip filled with fluorescein was loaded and used to determine the center of each microchamber within the image. These coordinates were recorded to automatically define the inner area of each microchamber from which the mean brightness was calculated and displayed as a function of time.


6

Page 7 of 18

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 fluorescence intensities were corrected in the following way: (i) There are slight deviations of the excitation light intensity within the microscope image. To account for them, a correction image of the microchip filled with fluorescein solution was taken after each ELISA. The measured micrograph is divided by this correction micrograph and normalized. (ii) Next, the linear part of the monitored curve was fitted to determine the slope. The fluorescence increased linearly in the time range of 5-10 minutes for the GAPDH assay and 7-18 minutes for the cAMP assay. (iii) To account for possible chipto-chip differences, data was further divided by the slope from chambers without cells and normalized. (iv) A data set was rejected if the R2 value was smaller than 0.99, i.e. only the linear curves are used. Only the microchambers that were enclosed by fully functioning and sealing valves, hosted only single-cells and where the cell was indeed completely lysed were analyzed. These microchambers were identified in a number of controls as summarized in Figure S2. RESULTS AND DISCUSSION Device design and operation The design of the device combines the ability to trap and isolate single cells in picoliter volumes with the possibility of rapidly exchanging the fluid surrounding the cell within a few hundred milliseconds whenever necessary (Figure 1). The device consists of two layers of poly(dimethylsiloxane) (PDMS) which are finally bonded on a microscope glass slide (Figure 1b). The core features are the chambers with the two central hurdles made of PDMS to physically trap the cells. In the current design, 60 of these chambers are ordered in alternate rows of 7 or 8 chambers on one microchip (Figure 1d, S4) and each row is connected via a microchannel. The height of the channels and chambers are 20 µm, which is sufficiently high to allow most mammalian cell types to pass and to prevent cellular stress due to shear forces. A ring-shaped valve (200 µm in diameter) surrounds the hurdles (Figure 1c). When actuated, it isolates the trapped cells from the surrounding solution in a volume of 625 picoliters. Every microchamber within a row can be opened and closed individually, which enables sequential supply of reagents to every single microchamber along the microchannel with virtually no crosscontamination. This is illustrated by trapping different food dyes in each column and releasing them sequentially (see Movie S1). To exchange the volume in the microchambers, the valve is opened for 500 ms, while the fluids are flushed through the channels at a total flow rate of 30 µl/min. This opening time is a compromise, being long enough to ensure complete fluid exchange, but short enough to prevent extensive flushing out of surface-bound assay compounds or analytes. Surface modification To enable ELISAs, the antibody against the target antigen was immobilized on the glass surface (Figure 1e). This was achieved by micro-contact printing of the biotinylated bovine serum albumin (bBSA) before bonding the glass to the fluid layer. After assembly, a filtered non-biotinylated BSA solution was centrifuged into the device and incubated for 2 hours to reduce the adsorption of other proteins to PDMS and glass. The protocol then included introducing avidin and biotinylated protein G (bPG) sequentially. This immobilization scheme via protein G allows great flexibility from this point onwards to finally immobilize any desired antibody on the surface (for protocol, see SI or Table 1). In principle, bBSA could be also introduced into the chip after assembling, which would result in a homogeneous immobilization on the glass surface and the PDMS channel walls. However, the


7


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

Page 8 of 18

patterning via micro-contact printing prior to chip assembly was preferred as it restricts the binding spots to the glass surface (especially convenient for TIRF microscopy, see SI). The quality of the bBSA immobilization pattern was conveniently tested by fluorescent microscopy for every microchip using fluorescently tagged streptavidin and allowed a good estimation and control over the number of binding spots for the antibody. Fluorescent micrographs confirmed the homogeneous fluorescence intensity within the spots and over all spots on the chip. Chamber to chamber variation was less than 2.5%, and chip-to-chip variation was negligible (less than 1%), demonstrating the reliability and reproducibility of the procedure (Figure S2 and S3a). If needed, the surface modification protocol could be interrupted at any time and the chip could be stored in the fridge for maximum one week without any loss of surface binding ability. Optimization of the general protocol using an immunoassay targeting eGFP Optimization of the protocol and first tests were performed with eGFP expressing HEK297 cells (see Figure S1). Here, anti-GFP antibody was immobilized on the glass surface. Next, the induced cells were flushed on the chip, trapped and isolated. Lysis buffer was flushed and the chambers were quickly opened to introduce the lysis buffer to the cells inside the chambers. For the assays, complete cell lysis is crucial, but lysis should not occur immediately. We therefore first optimized the lysis buffer. We found that cells are fully lysed by using a hypo-osmolar buffer containing 0.1% Tween 20 (see methods for composition) to assist the osmolar shock. The use of the relatively mild detergent, Tween 20, also helped to reduce background signals and was not expected to interfere with the assay protocols since concentrations up to 0.5% are regularly used in ELISA washing steps. When using the above-mentioned buffer, cell lysis occurs rather slowly (> 3 sec), so that the short opening of the chamber during introduction of the lysis buffer does also not result in a loss of analytes. The bright spots indicated that GFP was immobilized on the antibody-patterned spots, and hence confirmed that qualitative analysis of intracellular compounds by a single-cell immunoassay was possible (Figure S1b). Time-resolved measurements showed that the fluorescence signal was constant only after 10 minutes, i.e. the incubation time required after cell lysis. Furthermore, various control experiments were performed before and after each experiment to ensure data quality as described in the SI and Figures S2 and S3. Application to single-cell analysis - GAPDH Sandwich ELISA ELISAs are routinely used due to their high sensitivity given by the enzyme tag that serves as a signal amplifier (Figure 2a). In the following, we demonstrate the use of the microchamber array for singlecell ELISA with GAPDH as the target protein. GAPDH is an optimal target for evaluation of the method, since it is an essential enzyme in glycolysis and the number of copies per cell is expected to vary little within the cell population31. Furthermore, the oligomeric protein allows the binding of multiple antibody molecules, even if they recognize the same epitope. We determined GAPDH in U937 suspension cells as well as adherent HEK293 cells. For this, microchamber arrays with immobilized anti-GAPDH antibody (capture antibody) were prepared following the described protocol (see Materials and Methods, Table 1). Next, the cells were trapped and lysed. Additionally to the lysis buffer, the second enzyme-labeled antibody (detection antibody) was introduced. After closing the microchambers, the entire chip was flushed with hydrogen chloride


8

Page 9 of 18

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


(20 mM) to inactivate non-specifically bound enzyme-antibody conjugates on the channel walls32. After incubation, the chip was quickly washed with PBS and Amplex Red was introduced. The chambers were opened and due to the flow, unbound detection antibodies were washed away and Amplex Red was introduced to the chamber. The conversion of Amplex Red to fluorescent resorufin was monitored over time using a zoom microscope, which facilitates the recording of fluorescent images in 60 microchambers at the same time. From these measurements, the slope from the linear part of the kinetic reaction was determined. This slope is directly correlated to the concentration of the enzyme and hence, to the concentration of analyte (Figure 2c). First, we analyzed the U937 suspension cells (Figure 2d). Since they grow as single cells or small clusters, we could obtain a high percentage of single occupancy per chamber of 75%. Only single occupied microchambers were taken for data analysis. The results from 48 single U937 cells revealed a narrow distribution of GAPDH with an average value of 2.6 attomol, with minimal and maximal values of 2.0 and 3.2 attomol respectively. Next, we analyzed HEK T-REx cells according to the same protocol. We found that using this cell line, 26.7 % of the hurdles in the chambers were occupied by one cell. Data analysis of 46 cells revealed that HEK T-REx cells contained in average 2.5 attomol GAPDH, with maximal and minimal values of 0.9 and 4.1 attomol respectively. Interestingly, the value did not differ significantly when GFP production was induced in a subpopulation of cells by addition of tetracycline (2.7 +/- 0.4 attomol for stimulated cells, 2.5 +/- 0.9 attomol for non-stimulated). In general, data showed little variation in the amount of GAPDH in both U937 and HEK cells, indicating the function of GAPDH as a housekeeping protein. The current limit of detection (LOD) is as low as 970 zeptomol per microchamber for the ELISA described here. Compared to the standard ELISA protocols, the LOD is comparable in concentration and close to the theoretical limit of sensitivity given by the quality of the antibody (dependent on the binding constant to the target molecule)33. However, it should be emphasized that we have an excellent improvement concerning the absolute number of detected molecules by the factor of at least 1,000 due to the small volume of the microchambers. Further improvements of the LOD may be possible by using antibodies with a lower dissociation constant and by further optimization of the surface coatings to reduce the non-specific adsorption of the second antibody. cAMP competitive ELISA For the determination of smaller analytes with only one epitope, a sandwich format cannot be applied. Instead, competitive immunoassays are used to analyze smaller and more abundant molecules. We used this format for the determination of the second messenger cAMP in murine leydig tumor (MLT) cells (Figure 3a). The formation of cAMP is catalyzed by the enzyme adenylyl cyclase, which in turn is activated by neurotransmitters or hormones such as lutropin via the Gprotein coupled receptors (GPCR)34. In this competitive assay, anti-cAMP antibody was immobilized on the microchamber surface. Enzyme-labeled cAMP was added with the cell lysis buffer, which competes with the non-labeled cAMP (from the cytosol) for the binding sites. After washing and addition of the substrate Amplex Red, the fluorescence intensity was measured in each microchamber over time (movie S2). Again, the increase of fluorescence intensity taken from the initial linear phase depends on the concentration of HRP and therefore correlates to the intracellular compound (here the non-labeled ACS Paragon Plus Environment

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

Page 10 of 18

cAMP). After a short lag phase, the increase in fluorescence intensity was indeed linear and depended on the amount of enzyme present (Figure 3c). After approximately 45 minutes, the curves reached a plateau. This observation could not be explained by a complete conversion of the substrate, since the plateaus differed in fluorescence intensities. Instead, it is most likely caused by the inactivation of the enzyme, whose final amount per chamber is very low in this competitive assay (see additional control experiments in the SI and Figure S3b). The enzymatic function of HRP involves the formation of radicals and is therefore very prone to structural changes that could lead to enzymatic activity changes, i.e. destruction35. Based on these findings, we measured the fluorescence intensity at specific time points (all below 20 minutes) to obtain the increase of fluorescence intensity per minute. Furthermore, for a quantitative analysis, the number of binding sites as well as the concentration of enzyme-labeled cAMP had to be carefully optimized as these parameters determine the velocity of the fluorescence increase and hence, define the distinguishable concentration range. The influence of the bBSA density on the velocity of the enzymatic reaction is shown in Figure S3c. For the assay, different ratios of biotinylated to non-biotinylated BSA were printed. The surface was then modified using streptavidin, protein G and the cAMP antibody. Afterwards, only labeled antigen (cAMP-HRP, 370 ng/ml) was introduced into the chamber and incubated. After flushing the non-bound antigen away and introducing the reagents, the fluorescent reaction was monitored. As expected, the time needed for the full reaction correlates with the amount of binding sides (printed bBSA, Figure S3c). While higher ratios are not distinguishable due to the high velocity, lower percentages result in a slower reaction rate. Since image acquisition requires 30 seconds of exposure time, we chose a dilution of 0.057% bBSA in BSA. Using this density, the reaction needs around 30 minutes to complete, giving enough time for image acquisition as well as a short measurement time. For successful detection, we also had to optimize the amount of labeled antigen. From literature, we expected more than 100 attomol cAMP per cell36. For this amount, a concentration of 370 ng/ml cAMP-HRP (13.5 attomol per chamber) were ideal, with a dynamic range between 0-1500 attomol and a LOD of 75 attomol. Therefore, these parameters were set for the following measurements and used to obtain the calibration curve shown in Figure 3b. We performed cAMP measurements using MLT cells, which are adherent and hence, it was difficult to obtain high single occupancy of the cell hurdles. For a cell suspension of 0.3 million cells/ml and a cell loading flow rate of 30 µl/min, we found that after 5 minutes, a single-cell occupancy of 27 %. Out of 480 hurdles on 8 chips, 131 hurdles contained single cells. However, multiple cell occupancy often occurred (32 % of all hurdles) as well as a larger number of empty hurdles (41 %). Increasing the cell flushing time decreased the number of empty hurdles, but increased the multiple occupied chambers accordingly. Next, we prepared a doseresponse curve for cells stimulated with lutropin, which were additionally treated with 0.5 mM 3isobutyl-1-methylxanthin (IBMX) to inhibit the degradation of cAMP due to phosphodiesterases present in the cell. The addition of IBMX guaranteed constant cAMP levels over the measurement time and therefore simplified the protocol. For the stimulation curve, we analyzed the intracellular level of cAMP in single MLT cells for 8 different concentrations of lutropin. Our results show that MLT cells have a high basal level of cAMP (due to IBMX, average: 263 attomol), increasing to an average of around 900 attomol for lutropin concentrations above 20 ng/ml. These average cAMP levels inside the cell are surprisingly high, but in agreement with a similar work obtained from measurements of large cell populations36.


10

Page 11 of 18

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


Furthermore, the resulting dose-response curve of the average cell behavior exhibits a typical shape, with a half maximal effective concentration (EC50) of 2.51 ± 0.44 ng/ml, which is in agreement with literature. Hakola et al. found a dose response behavior of murine leydig tumor cells with the addition of lutropin with an EC50 between 1 and 3 ng/ml37. Others have reported an EC50 of 1.53 ± 0.18 ng/ml, but both values are within the range found in this study38. For the quantification of less abundant molecules, we tried to further lower the LOD of our competitive assay. This was achieved by using a high dilution of 0.0066% bBSA in BSA (less binding sites) and a low concentration of cAMP-HRP (5 ng/ml, i.e. 181 zeptomol per chamber). These settings were optimal for the detection of low cAMP concentrations (LOD as low as 500 zeptomol). However, discrimination of antigen concentrations higher than 5 attomol was not possible under these conditions. CONCLUSION We successfully combined immunological assays with a microfluidic platform for single-cell analysis. Intracellular compounds could be specifically, sensitively and quantitatively determined in cells before and after treatment with stimulating agents (amounts of attomoles to zeptomoles). The platform performs all of the required steps conveniently including long (hours) and short (seconds) incubation times and repeated washing steps. The subsequent analysis of the cell lysate is achieved without any further transport of the compounds and with only a little dilution. Cross-contamination is successfully avoided due to the arrangement of the chambers and their sealing properties. Furthermore, the device can be scaled up for further parallelization and could be adapted to commercial instruments such as a plate reader or a scanning microscope. Due to the flexibility of the immunological assay, the method can be applied to numerous proteins or metabolites of interest, provided that specific antibodies are available. We anticipate the future use of our method for (single-cell) proteomics and metabolomics as well as serving as a valuable tool for systems biology. ACKNOWLEDGMENTS The authors gratefully acknowledge Tom Robinson for careful reading of the manuscript, C. Bärtschi and H. Benz for the construction of the custom-build pressure control system. Furthermore, we would like to acknowledge the use of the clean room facility FIRST and of the Light Microscopy Center (LMC), both at ETH Zürich. The work was funded by Merck Serono and the European Research Council (ERC) under the 7th Framework Program (ERC Starting Grant, project no. 203428, nμLIPIDs). Supporting Information Available: This material is available free of charge via the Internet at http://pubs.acs.org.


11


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

Page 12 of 18

REFERENCES 1.

Walling, M.A.; Shepard, J.R.E. Chem. Soc. Rev. 2011, 40, 4049.

2.

Schmid, A.; Kortmann, H.; Dittrich, P.S.; Blank, L.M. Curr. Opin. Biotechnol. 2010, 21,12–20.

3.

Lecault, V.; White, A.K.; Singhal, A.; Hansen, C.L. Curr. Opin. Chem. Biol. 2012, 3-4, 381-90.

4.

Spiller, D.G.; Wood, C.D.; Rand, D.A.; White, M.R.H. Nature 2010, 465, 736–745.

5.

Cai, L.; Friedman, N.; Xie, X.S. Nature 2006, 440, 358–362.

6. Toriello, N.M.; Douglas, E.S.; Thaitrong, N.; Hsiao, S.C.; Francis, M.B.; Bertozzi, C.R.; Mathies, R.A. Proc Natl Acad Sci USA 2008, 105, 20173–20178. 7. Huang, B.; Wu, H.; Bhaya, D.; Grossman, A.; Granier, S., Kobilka, B.K.; Zare, R.N. Science 2007, 315, 81–84. 8.

Di Carlo, D.; Aghdam, N.; Lee, L.P. Anal Chem 2006, 78, 4925–4930.

9.

Lee, W.C.; Rigante, S.; Pisano, A.P.; Kuypers, F.A. Lab Chip 2010, 10, 2952–2958.

10. Salehi-Reyhani, A.; Kaplinsky, J.; Burgin, E.; Novakova, M.; deMello, A.J.; Templer, R.H.; Parker, P.; Neil, M.A.A.; Ces, O.; French, P.; Willison, K.R.; Klug, D. Lab Chip 2011, 11, 1256–1261. 11. Joensson, H.N.; Samuels, M.L.; Brouzes, E.R.; Medkova, M.; Uhlén, M.; Link, D.R.; AnderssonSvahn, H. Angew Chem Int Ed 2009, 48, 2518–2521. 12.

Najah, M.; Griffiths, A.D.; Ryckelynck, M. Langmuir 2012, 84, 1202–1209.

13. Yamamura, S.; Yatsushiro, S.; Yamaguchi, Y.; Abe, K.; Shinohara, Y.; Tamiya, E.; Baba, Y.; Kataoka, M. PLoS ONE 2012, 7, e32370. 14. Cecala, C.; Rubakhin, S.S.; Mitchell, J.W.; Gillette, M.U.; Sweedler, J.V. Analyst 2012, 137, 2965–2972. 15.

Eyer, K.; Kuhn, P.; Hanke, C.; Dittrich, P.S. Lab Chip 2012, 12, 765–772.

16.

Hanke, C.; Waide, S.: Kettler, R.; Dittrich, P.S. Anal Bioanal Chem 2012, 402, 2577–2585.

17. Agresti, J.J.; Antipov, E.; Abate, A.R.; Ahn, K.; Rowat, A.C.; Baret, J.C.; Marquez, M.; Klibanov, A.M.; Griffith, A.D. and Weitz, D.A. Proc Natl Acad Sci USA 2010, 107, 4004–4009. 18. Ma, C.; Fan, R.; Ahmad, H.; Shi, Q.; Comin-Anduix, B.; Chodon, T.; Koya, R.C.; Liu, C.C.; Kwong, G.A.; Radu, C.G.; Ribas, A.; Heath, J.R. Nat Med 2011, 17, 738–743. 19. Shi, Q.; Qin, L.; Wei, W.; Geng, F.; Fane, R.; Shin, Y.S.; Guo, D.; Hood, L.; Mischel, P.S.; Heath, J.R. Proc Natl Acad Sci USA 2012, 109, 419–424. 20.

Martin-Belmonte, F.; Perez-Moreno, M. Nature 2012, 12, 23–38.

21. Powell, A.A.; Talasaz, A.H.; Zhang, H.; Coram, M.A.; Reddy, A.; et al. PLoS ONE 2012, 7, e33788.


12

Page 13 of 18

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


22. White, A.K.; VanInsberghe, M.; Petriv, O.I.; Hamidi, M.; Sikorski, D.; Marra, M.A.; Piret, J.; Aparicio, S.; Hansen, C.L. Proc Natl Acad Sci USA 2011, 108, 13999–14004. 23. Leung, K.; Zahn, H.; Leaver, T.; Konwar, K.M.; Hanson, N.W.; Pagé, A.P.; Lo, C.C.; Chain, P.S.; Hallam, S.J.; Hansen, C.L. Proc Natl Acad Sci USA 2012, 109, 7665–7670. 24. Marcy, Y.; Ouverney, C.; Bik, E.M.; Lösekann, T.; Ivanova, N.; Martin, H.G.; Szeto, E.; Platt, D.; Hugenholtz, P.; Relman, D.A.; Quake S.R. Proc Natl Acad Sci USA 2007, 104, 11889–11894. 25.

Kortmann, H.; Kurth, F.; Blank, L.M.; Dittrich, P.S.; Schmid, A. Lab Chip 2009, 9, 3047–3049.

26.

Huang, Y.; Cai, D.; Chen, P. Anal Chem 2011, 83, 4393–4406.

27. Huang, N.T.; Chen, W.; Oh, B.R.; Cornell, T.T.; Shanley, T.P.; Fu, J.; Kurabayashi, K. Lab Chip 2012, 20, 4093-101. 28. Schwanhäusser, B.; Busse, D.; Li, N.; Dittmar, G.; Schuchhardt, J.; Wolf, J.; Chen, W.; Selbach, M. Nature 2011, 473, 337–342. 29.

Unger, M.A.; Chou, H.P.; Thorsen, T.; Scherer, A.; Quake S.R. Science 2000, 288, 113–116.

30.

Thorsen, T.; Maerkl, S.J.; Quake, S.R., Science 2002, 298, 580–584.

31.

Barber, R.D.; Harmer, D.W.; Coleman, R.A.; Clark, B.J. Physiol Genomics 2005, 21, 389–395.

32. 291.

Liu, G.; Amin, S.; Okuhama, N.N.; Liao, G.; Mingle, L.A. Histochem Cell Biol 2006, 126, 283–

33.

Zhang, H.; Wang. S.; Fang, G. J Immunol Methods 2011, 368, 1–23.

34.

Rich, T.C.; Karpen, J.W. Ann Biomed Eng 2002, 30, 1088-1099

35. Berglund, G.I.; Carlsson, G.H.; Smith, A.T.; Szöke, H., Henriksen, A.; Hajdu, J. Nature 2002, 417, 463–468. 36.

Rebois, R.V. J Cell Biol 1982, 94, 70–76.

37. Hakola, K.; Haavisto, A.M.; Pierroz, D.D.; Aebi, A.; Rannikko, A.; Kirjavainen, T.; Aubert, M.L.; Huhtaniemi, I. J Endocrinol 1998, 158, 441–448. 38. Jiang, M.; Lamminen, T.; Pakarinen, P.; Hellman, J.; Manna, P.; Herrera, R.J.; Huhtaniemi, I. MHR 2002, 8, 201–212.


13


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

Page 14 of 18

FIGURES

Figure 1: Design and operation of the microfluidic device for single-cell ELISA. a) Photograph of the device. The picture shows the fluid reservoir for the application of different reagents and fluid channels, both filled with black food dye, whereas pressure channels are filled with red food dye. b) Schematic of the different layers required for the fabrication of the microchip. Control and fluid layer are prepared from PDMS. The control layer is bonded to the fluid layer, which then is bonded onto a glass slide. c) Schematic of a microchamber. Fluid flow, (from left to right) is used to introduce and trap a cell inside the microhurdle feature and to deliver compounds to the cell. The trapped cell can be isolated from the environment by hydraulic actuation of a ring-shaped valve (grey color). Right: fluorescent micrograph. For demonstration, a trapped U937 cell was stained with calcein. The ringshaped valve was pressurized and the external solution was exchanged with 50 µM fluorescein. Scale bar: 150 µm. d) Brightfield micrograph of the chamber array. In total, the device consists of 60 hurdles and chambers. As shown by the different trapped food dyes, the chambers of one row can be actuated individually, therefore avoiding cross-contamination. Scale bar: 800 µm. e) TIRF microscopy image of the glass surface, patterned with fluorescently labeled avidin. The antibodies are immobilized on the bright areas as indicated in the drawings. Scale bar: 30 µm. The large white circle shows the inner chamber border.


14

Page 15 of 18

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 2: Sandwich ELISA for GAPDH quantification in single cells. a) Schematic of the GAPDH sandwich ELISA. The intracellular protein binds to the immobilized anti-GAPDH (capturing antibody) and to the secondary HRP-labeled antibody (detection antibody) that was added to the lysis buffer. After incubation, unbound detection antibody is washed away and bound detection antibody is detected via a fluorescence reaction. b) Measured curves for chambers containing one (black) or no HEK293 cell (red). After a lag period in the beginning, the fluorescence increase is linear between 5 and 10 minutes before the curves reach a plateau, indicating that the substrate is completely converted. c) Calibration curve showing the dependence of the slope of fluorescence increase with the GAPDH concentration in the chamber. Dashed line indicates the background signal at 7.01 ±

0.31. d) Histograms and box plots of absolute GAPDH levels for two cell lines (U937 suspension cells and adherent HEK cells).


15


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

Page 16 of 18

Figure 3: Second messenger quantification (cAMP) by a competitive ELISA. a) Schematic of the cAMP competitive ELISA. cAMP production in MLT cells is induced by the addition of lutropin (LH). Upon lysis, the secondary messenger binds to immobilized anti-cAMP capturing antibody on the surface. Labeled cAMP (HRP-conjugate) is added to the lysate and competes with cAMP for the binding spots on the surface. Unbound labeled cAMP is washed away and cAMP is detected via a fluorescence reaction. b) Correlation between cAMP content of the chamber and fitted linear slope of the kinetic curves. c) Fluorescence intensity in chambers without cells, non-stimulated cells and cells stimulated with lutropin (selected chambers). The linear region (grey) is used to determine the slope. d) Doseresponse curve for MLT cells at different LH concentrations (n=21 for 0.001 ng/ml, n=14 for 0.01 ng/ml, n=21 for 1 ng/ml, n=22 for 3 ng/ml, n=28 for 5 ng/ml, n=18 for 20 ng/ml, n=6 for 100ng/ml and n=6 for 500 ng/ml). Determined EC50: 2.51 ± 0.44 ng/ml. Red dots indicate mean values, and the red line represents the fitted dose response curve.


16

Page 17 of 18

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


TABLES Table 1: Surface modification protocol

Step

Time

Concentr ation

Flow rate

Flow-free incubation time

Other

[min]

(w/v) %

[µl/min]

[min]

BSA

60

4

-

PBS

10

Avidin/Streptavidin TR

30

PBS

10

Protein G, biotin

30

PBS

10

Antibody

20

PBS

10

5

Cells

~5

30

300,000 cells/ml, chamber closed

Lysis buffer

5

30

Chamber opened shortly

HCl

15

10

20 mM, chamber closed

PBS

2

30

Amplex Red solution

5

30

5 0.025

5

Ratio labeled : unlabeled 1:10

5 0.0025

5 5

0.0001

5


15

100 µM plus 100 µM hydrogen perxoide, chamber opened shortly

17


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

Page 18 of 18

For TOC only:


18

Implementing Enzyme-Linked Immunosorbent Assays on a

Recommend Documents