Sensitive Method for Determination of Protein and Cell Concentrations

Article pubs.acs.org/ac

Sensitive Method for Determination of Protein and Cell Concentrations Based on Competitive Adsorption to Nanoparticles and Time-Resolved Luminescence Resonance Energy Transfer between Labeled Proteins Sari Pihlasalo,* Pauli Puumala, Pekka Han̈ ninen, and Harri Har̈ ma ̈ Laboratory of Biophysics and Medicity Research Laboratory, University of Turku, Tykistökatu 6A, FI-20520 Turku, Finland ABSTRACT: A sensitive mix-and-measure method for the determination of protein and cell concentrations was developed. It is based on the competitive adsorption between the analyte and donor- and acceptor-labeled proteins to carboxylate-modified polystyrene nanoparticles. A high timeresolved luminescence resonance energy transfer (TR-LRET) signal is detected in the absence of the analyte due to the close proximity of the nanoparticle-adsorbed labeled proteins. The increased concentration of the analyte decreases the adsorption of the labeled proteins, leading to the loss of proximity and thus a decrease in the TR-LRET. The detection limit of the assay was 2.6 ng of proteins, which is higher than that of the most sensitive commercial methods. The method was also applied to cell counting, and 200 eukaryotic cells were measured in a microtiter well under the optimized conditions. The average coefficient of variation for both developed assays was approximately 10%, and the protein-to-protein variability for 11 different proteins was no more than 20%. The developed method requires no labeled particles, making the concept optimally applicable to varying targets as the material of the particle may be selected according to the application.

Q

require a heating step, and Lowry requires an incubation for 40 min and BCA an incubation even for 2 h. Fluorometric methods, such as OPA (o-phthaldialdehyde),10,11 CBQCA (3(4-carboxybenzoyl)quinoline-2-carboxaldehyde),12 and NanoOrange,13 are based on the detection of amino groups or the binding of a dye to detergent-coated proteins. The response of the OPA and CBQCA methods varies between different proteins due to the diverse amino group contents of proteins. All of these methods have relatively high sensitivity, 10−20 ng of protein, but also have experimental constraints as NanoOrange and CBQCA require a heating step, CBQCA requires a long incubation of 1.5 h, and OPA and CBQCA require the handling of toxic reagents. The electronic cell counting requires an investment in the instrument and a relatively large number of cells per assay. Microtiter plate readers are widely available, facilitating the adaptation of plate-based assays with the addition of reagents only. The assays detecting mitochondrial reductase (e.g., MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium)14 and MTS (3-(4,5-dimethylthiazol-2-yl)-5-(3-(carboxymethoxy)phenyl)-2-(4-sulfophenyl)-2H-tetrazolium)15), intracellular esterase,16 and acid phosphatase17,18 relate the activity of the enzymes to the presence of cells. The limits of detection for

uantification of proteins and counting of cells are routinely performed in laboratories. Traditionally, proteins are quantified with Kjeldahl’s method, and cells are manually counted with a hemocytometer using a light microscope. These methods are insensitive, time-consuming, and laborious. More advanced absorbance (UV280) measurement has been utilized in a cuvette or microtiter plate format. The counting with a hemocytometer has a subjective nature, which has been overcome with automated electronic counting chambers.1−4 Although the approaches are more user-friendly, the sensitivity or throughput is still limited. Several colorimetric and fluorometric microtiter plate assays have been developed to speed up these tasks and to improve the detection limits. However, they still have limitations in the experimental procedures, dynamic range, sensitivity, signal stability, and protein-to-protein or cell-to-cell variability. Generally, fluorometric methods have a higher sensitivity and dynamic range than colorimetric methods. Commercially available colorimetric protein quantification methods, such as Lowry,5 BCA (bicinchoninic acid),6 and Bradford,7 are based on the determination of peptide bonds or amino groups. The absorbance detection limits the sensitivity of these methods from submicrograms to several micrograms of protein in a sample, and the dynamic range is short.5−8 The response to different proteins varies significantly, especially for the Bradford and the Lowry methods.9 Furthermore, the experimental setups are relatively laborious and time-consuming, as Lowry requires two reagent additions, Lowry and BCA © 2012 American Chemical Society

Received: February 29, 2012 Accepted: April 26, 2012 Published: April 26, 2012 4950

dx.doi.org/10.1021/ac300597j | Anal. Chem. 2012, 84, 4950−4956

Analytical Chemistry

Article

these commercial methods are typically well above 100 cells in a sample. Because the cellular metabolic activity depends on the cell type and the medium,19 these assays are prone to cell-tocell variability and the incubation step is even 4 h long. Today, many cell counting methods are based on the detection of nucleic acid bound dyes, such as DAPI (4′,6-diamidino-2phenylindole) and SYBR Green. The dye fluorescence is enhanced at binding but varies between cells due to the differences in DNA content. The sensitivity is thus cell type dependent and ranges from a few tens to several thousands of cells.2,20,21 Errors in cell counting are measured, as DNA outside the cells is also detected. ATP (adenosine triphosphate)22,23 assay is sensitive and relatively fast without heating steps, and the dynamic range is long from single cells to 107 cells. However, the assay counts only metabolically active living cells, and the signal varies between cell types.23,24 The luminescence signal is not stable, which implies the need for timed measurements. As ATP is naturally ubiquitous, the performance of the assay and the reagents are also susceptible to contamination. Previously, we have reported competitive adsorption-based methods for the quantification of proteins and cell counting.25−28 The detection relied on the time-resolved luminescence resonance energy transfer (TR-LRET) between the acceptor-labeled protein and donor-labeled particles or the fluorescence quenching of the labeled protein by the quencher particles. These methods are entirely based on nonspecific interactions of the analyte with a solid surface, facilitating the measurement of both cells and proteins using the same assay principle. This provides highly cost-effective methods for daily laboratory work with a low protein-to-protein or cell-to-cell variability. In this paper, we report a novel assay with an alternative detection principle, where the TR-LRET between two labeled proteins adsorbed onto a nanoparticle is the basis for the signal outcome (Figure 1). A decrease in the TR-LRET signal is detected when analyte proteins or cells replace the labeled proteins on the nanoparticles. The main advantage of the method as compared to the previously developed assays is the high versatility and capability to adapt according to the application. No labeled particle is required, and thus, the material of the particle can be tuned and selected without major restrictions of the application. The applicability of the new method is shown for the quantification of 11 different proteins and eukaryotic Chinese hamster ovarian (CHO-K1) cells after optimizations.

Figure 1. Schematic illustration of the developed TR-LRET assay using donor- and acceptor-labeled proteins and unlabeled nanoparticles with a diameter of 61 nm. The labeled proteins adsorbed to the nanoparticles at close proximity to each other, and TR-LRET was detected in the absence of analytes (a). The luminescence signal decreased when the analyte protein (b) or cells (c) interacted with the nanoparticles and the adsorption of the labeled proteins was prevented. The protein solution or the cell suspension was excited at 340 nm, and the sensitized emission was monitored at 730 nm.

was synthesized and characterized according to the literature.29 Alexa Fluor 680 carboxylic acid succinimidyl ester was ordered from Molecular Probes (Eugene, OR), and PD-10 and NAP-5 gel filtration columns were obtained from GE Healthcare (Uppsala, Sweden). DELFIA enhancement solution was purchased from Wallac, Perkin-Elmer Life and Analytical Sciences (Turku, Finland). CHO-K1 cells were from the Laboratory of Biophysics. Dulbecco’s modified Eagle’s medium high glucose, L-glutamine, fetal bovine serum, and penicillin/ streptomycin solution were obtained from EuroClone (Bio West, rue de la Caille, Nuaille, France). Concentrated hydrochloric acid and sodium chloride were purchased from Sigma-Aldrich. Sodium hydroxide was from FF-Chemicals (YliIi, Finland), sodium hydrogen carbonate from Merck KGaA (Darmstadt, Germany), and phosphate-buffered saline (PBS) from Lonza (Basel, Switzerland). High-purity Milli-Q water was used to prepare all aqueous solutions. Methods. Labeling of γ-Globulin with Alexa Fluor 680 and Eu3+ Chelate. Two labeled γ-globulins were prepared. Alexa Fluor 680 carboxylic acid succinimidyl ester was conjugated to γ-globulin (γG−Alexa) as recommended by the manufacturer Molecular Probes. The degree of labeling (5.6) was determined from absorbance measurements. γG was also labeled with 9-dentate Eu3+ chelate {2,2′,2″,2‴-{[4′-(4‴isothiocyanatophenyl)-2,2′:6,6″-terpyridine-6,6″-diyl]bis(methylenenitrilo)}tetrakis(acetato)}europium (γG−Eu3+). The 9-dentate Eu3+ chelate (60 nmol) was added to 117 μL of 0.2 M carbonate buffer, pH 9.9, containing 1.3 nmol of γG. After overnight incubation, the free chelate was separated from the labeled γG by gel filtration with PD-10 columns in 100 μL fractions in PBS. The degree of labeling and protein concentration for each fraction were determined with UV280

■

MATERIALS AND METHODS Materials. Amino-modified polystyrene particles (240 nm) were ordered from Spherotech Inc. (Libertyville, IL) and silica particles (6840 nm) from Bangs Laboratories Inc. (Fishers, IN). Streptavidin (>97%) was purchased from BioSPA (Milan, Italy), and all other proteins (γ-globulins from bovine blood, 97%; thyroglobulin from bovine thyroid, 99%; histone from calf thymus, type II-A, high purity stated by the manufacturer; lysozyme from chicken egg white, 100%; catalase from bovine liver, 65%; albumin from bovine serum (BSA), >99%; albumin from chicken egg white, 97%; albumin from sheep serum, >98%; albumin from porcine serum, 99%; myoglobin from equine skeletal muscle, 95%) were purchased from SigmaAldrich Co. (St. Louis, MO). 9-Dentate Eu3+ chelate {2,2′,2″,2‴-{[4′-(4‴-isothiocyanatophenyl)-2,2′:6,6″-terpyridine6,6″-diyl]bis(methylenenitrilo)}tetrakis(acetato)}europium, was from the Laboratory of Biophysics (Turku, Finland) and 4951

dx.doi.org/10.1021/ac300597j | Anal. Chem. 2012, 84, 4950−4956

Analytical Chemistry

Article

proteins γG−Eu3+ and γG−Alexa) and large relative change of the TR-LRET signal in the presence of BSA (optimal adsorption for BSA). In the test, 35 μL of 0, 0.70, or 700 mg/L BSA in the universal buffer and 5.0 μL of nanoparticles (80 amol) in water were mixed, and 5.0 μL of the mixture of γG−Eu3+ (degree of labeling 4.0) and γG−Alexa (in total 0.1 pmol, γG−Eu3+:γG−Alexa ratio 1) was added. In addition, different buffers (5.0 mM glycine, pH 3.0; citrate, pH 3.0; PBS, pH 7.4; 5.0 mM glycine, pH 3.0, 0.10 M NaCl) were tested and compared to the universal buffer. Incubation Time. The kinetics of both assay steps was studied in two separate tests to define the optimal incubation times. In the first test, the time required for the adsorption of the protein was studied by adsorbing different concentrations of BSA to the nanoparticles (30 amol) in 40 μL of 5.0 mM glycine buffer, pH 3.0, with different incubation times before adding 5.0 μL of the mixture of γG−Eu3+ (degree of labeling 4.0) and γG− Alexa (in total 40 fmol, γG−Eu3+:γG−Alexa ratio 1.6). In the second test, the optimal incubation time for the adsorption of the labeled proteins was investigated. The luminescence signal was monitored as a function of time after the addition of the mixture of γG−Alexa and γG−Eu3+. Response for Different Proteins. The response of the developed method was tested for 11 proteins with different isoelectric points (4.5−11) and sizes (14−670 kDa). The protein calibration curves were run following an optimized, typical assay protocol in 5.0 mM glycine buffer, pH 3.0, with 30 amol of nanoparticles and 40 fmol of labeled proteins (degree of labeling 4.8 for γG−Eu3+, γG−Eu3+:γG−Alexa ratio 1.6). Total Cell Counting Assay. The cell concentration determinations were carried out similarly to the protein quantification. However, the optimized conditions for the assay were different. The optimal assay buffer was 50 mM glycine, pH 2.5. The cell samples in PBS were diluted 4-fold, and therefore, NaCl was present in the wells. The final concentrations of nanoparticles (30 amol) and labeled proteins (γG−Eu3+ with a degree of labeling of 4.8 and γG−Alexa, in total 50 fmol, γG−Eu3+:γG−Alexa ratio 1.6) were 0.6 pM and 1 nM, respectively. TR-LRET and the corresponding europium luminescence emission intensities were measured as described above. Total Cell Counting Assay Optimizations. Adsorption Buffer and pH. A universal buffer (see the components above) was used to study the effect of pH on the performance of the assay over a range of 1.5−9.0. The objectives were similar to those of the protein assay. In the test, 35 μL of 0, 500, or 10 000 CHO cells in the universal buffer and 5.0 μL of nanoparticles (30 amol) in water were mixed, and 5.0 μL of the mixture of γG−Eu3+ (degree of labeling 4.8) and γG−Alexa (in total 0.1 pmol, γG−Eu3+:γG−Alexa ratio 1.6) was added. The universal buffer was changed into a simpler buffer, glycine at pH 2.5. The concentrations of the glycine buffer (5.0−100 mM) and 50 mM glycine buffer with varying concentration of NaCl (0.009−2.5 M) were tested in two separate tests.

and luminescence measurements. For the luminescence measurements, the samples were diluted in DELFIA enhancement solution, as recommended by the manufacturer, except that the europium ions were first dissociated from the chelates in 0.1 M hydrochloric acid before mixing and enhancement with DELFIA. Fractions having the highest concentrations of the labeled protein were used in this study. Preparation of Carboxylate-Modified Polystyrene Nanoparticles. Carboxylate-modified polystyrene nanoparticles were prepared by emulsion copolymerization of styrene and acrylic acid as described previously by us.30 The mean diameter of the nanoparticles was 61 nm as determined with a Beckman Coulter N4 PLUS submicrometer size analyzer (Fullerton, CA). The concentration of the particles was calculated from the weight of the vacuum-dried dispersion. Preparation of Cell Suspension. CHO cells were cultured in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum, 2.0 mM L-glutamine, 1.0 × 105 U/L penicillin G, and 0.10 g/L streptomycin sulfate to passage 18-22. The cells were harvested by scraping in PBS. The harvested cells were washed with PBS by centrifugation (800 rpm, 5 min) with a Sorvall RT 6000D centrifuge (Thermo Fisher Scientific Inc., Waltham, MA) three times. The cell numbers were determined by counting the cells in a hemocytometer under the microscope. The cell suspension containing 1.0 × 107 cells/mL was used to prepare the dilutions for assays. Total Protein Quantification Assay. The protein concentration determinations using nanoparticle application were carried out in 384-well microtiter plates. In a typical assay, 35 μL of the sample in 5.0 mM glycine buffer, pH 3.0, and 5.0 μL of the carboxylate-modified polystyrene nanoparticles (30 amol, 61 nm in diameter) in water were mixed. After 20 min of incubation, a mixture of γG−Eu3+ (degree of labeling 4.0) and γG−Alexa (in total 40 fmol, γG−Eu3+:γG−Alexa ratio 1.6) was added in 5.0 μL of water. The final concentrations of nanoparticles and labeled proteins (γG−Eu3+ and γG−Alexa) were 0.6 pM and 0.8 nM, respectively. After 20 min of incubation, TR-LRET luminescence emission intensities were measured in a 50 μs window after a 75 μs delay time using a Victor2 multilabel counter (Wallac, Perkin-Elmer Life and Analytical Sciences) with 340 nm excitation and 730 nm emission wavelengths. Corresponding europium luminescence emission intensities were measured in a 400 μs window after a 400 μs delay time with 340 nm excitation and 615 nm emission wavelengths. Total Protein Quantification Assay Optimizations. Material and Size of Particles. The performance of the prepared carboxylate-modified polystyrene nanoparticles with a diameter of 61 nm was compared to that of two commercial particles: 240 nm amino-modified polystyrene particles and 6840 nm silica particles. In the test, 35 μL of 0 or 700 mg/L BSA in universal buffer (see components below), pH 3.0, and 5.0 μL of particles (30 amol of 61 nm, 2 amol of 240 nm, and 2 zmol of 6840 nm particles) in water were mixed, and 5.0 μL of the mixture of γG−Eu3+ (degree of labeling 4.8) and γG−Alexa (in total 40 fmol, γG−Eu3+:γG−Alexa ratio 1.6) was added. Adsorption Buffer and pH. A universal buffer containing 5.0 mM sodium tetraborate, citric acid, tris(hydroxymethyl)aminomethane, potassium dihydrogen phosphate, and potassium chloride was used to study the effect of pH on the performance of the assay over a range of 1.5−9.0. The objectives were to obtain a high TR-LRET signal in the absence of the sample protein BSA (optimal adsorption for the labeled

■

RESULTS AND DISCUSSION A new competitive nanoparticle-based method was developed for the quantification of proteins and counting of cells. The method utilizes the interaction of the sample protein or cells with nanoparticles and the TR-LRET between labeled proteins (Figure 1). In the absence of the analyte, the donor Eu3+ chelate labeled γ-globulin (γG−Eu3+) and the acceptor Alexa Fluor 680 labeled γ-globulin (γG−Alexa) adsorb to the 4952

dx.doi.org/10.1021/ac300597j | Anal. Chem. 2012, 84, 4950−4956

Analytical Chemistry

Article

nanoparticles and the TR-LRET is detected between the dyes in close proximity to each other. A decreasing signal was measured with increasing sample concentration, as the adsorption of the labeled proteins and TR−LRET were prevented. The assays were optimized for assay buffer and pH, incubation time, material and size of the particles, number of nanoparticles, and concentration of labeled proteins. The response of the assays was tested for eukaryotic CHO-K1 cells and 11 different proteins. Total Protein Quantification Assay. Carboxylate-modified polystyrene nanoparticles, 61 nm in diameter, were prepared for the development of the method. Their assay performance was compared to that of two commercial particles with different polymers, surface groups, and sizes, 240 nm amino-modified polystyrene nanoparticles and 6840 nm silica microparticles, by keeping the total surface area of the differently sized particles constant. The ratio between the LRET signals at BSA concentrations of 0 and 700 mg/L (B0/B) was 50 for the 61 nm nanoparticles and approximately 3 times lower for the 240 and 6840 nm particles. Therefore, the 61 nm nanoparticles were chosen for all further experiments due to the efficient adsorption of γG−Eu3+ and γG−Alexa and the highest B0/B. The results suggest that the polymer and size of the particle may be freely selected, making the method versatile and adaptable to different applications. The surface of the 61 nm nanoparticle contains carboxylate, sulfite, and sulfate groups (pKa(−COOH) = 4.7, pKa(−SO3H) = 1.9, and pKa(−SO4H) = −3), giving a negative surface charge to the nanoparticles. The adsorption efficiencies of γG−Eu3+ and γG−Alexa (Figure 2b) and sample protein BSA (Figure 2a) to the nanoparticles and the performance of the assay were studied as a function of pH in a universal buffer. The TR-LRET was measured in the absence and in the presence of different BSA concentrations, and B0/B was calculated (Figure 2). The effect of pH on the adsorption of γG−Eu3+ and γG−Alexa and the TR-LRET efficiency was investigated by measuring the TRLRET signal at zero concentration of BSA in varying pH values. The measured TR-LRET signal (normalization with the TRLRET signal in the presence of a high concentration of BSA) is presented as a function of pH in Figure 2b (TR-LRET/BSA, B0/B for 700 mg/L). The efficient TR-LRET was observed at low pH 2−3 and at pH 5−6, which is just below the pI 6.4−8.8 of γG.31 The observed effect of pH on the adsorption is mainly due to electrostatic interactions between the proteins and nanoparticles. At low pH 2−3, γG has a distinct positive total charge and covers well the nanoparticle possessing a negative surface chargethe attractive electrostatic interaction between opposite charges.32,33 The adsorption below the pI might be related to the maximal adsorbed amount obtained near the pI. The protein is uncharged at its pI, and thus, its solubility is decreased and electrostatic repulsion is reduced between proteins on the surface of the nanoparticles.34−36 The adsorption of sample protein was studied at a moderate concentration of sample BSA (Figure 2a). The most sensitive detection of BSA, the displacement of γG−Eu3+ and γG−Alexa, and the highest B0/B for the TR-LRET were obtained at low pH 2−4 (TR-LRET/BSA, B0/B for 700 μg/L in Figure 2a). Comprehensibly, this change in the TR-LRET signal was also detected as an increase in the luminescence of the donor Eu3+ chelate in the same pH range (Eu/BSA, B/B0 for 700 μg/L in Figure 2a). In this pH range, BSA has a positive net charge, as its pI is 4.90.37 According to these results, pH 3 was chosen for the protein quantification assay due to the optimal result for the

Figure 2. Effect of pH on the adsorption of BSA using the developed TR-LRET protein and cell assay: Normalized signal (B0/B for TRLRET and B/B0 for Eu signal) as a function of the adsorption pH. The test was carried out without analytes and with analytes at BSA concentrations of (a) 700 μg/L and (b) 700 mg/L or cell concentrations of (a) 500 and (b) 10 000 per well and nanoparticles with a diameter of 61 nm.

tests, and efficient adsorption is expected to different proteins due to their positive charge at low pH. A simple substitute for the universal buffer was investigated using 5.0 mM glycine, pH 3.0; 5.0 mM citrate, pH 3.0; PBS, pH 7.4; and 5.0 mM glycine, pH 3.0, containing 0.10 M NaCl. Similar or improved B0/B ratios were observed for buffers with pH 3.0 and clearly a lower ratio in PBS (data not shown). The highest B0/B (between 0 and 700 mg/L BSA) was measured for 5.0 mM glycine, pH 3.0, which was chosen for further experiments. The sensitivity and the signal-to-background ratio of the protein quantification assay was optimized for the number of nanoparticles, the concentration of labeled proteins γG−Eu3+ and γG−Alexa, and the ratio of γG−Eu3+ and γG−Alexa (data not shown). The sensitivity was improved, as the concentrations of nanoparticles and labeled proteins were reduced until the coefficient of variation for the signal increased excessively. The lowest limit of detection was reached at a concentration of 0.6 pM for nanoparticles and 0.8 nM for labeled proteins (γG−Eu3+ and γG−Alexa). The calculated total surface area of the nanoparticles and the maximal surface coverage of the labeled proteins on the nanoparticles are in the same range, which suggests that the nanoparticles are efficiently covered with the labeled proteins. The highest signal ratio 4953

dx.doi.org/10.1021/ac300597j | Anal. Chem. 2012, 84, 4950−4956

Analytical Chemistry

Article

measured between 0 and 700 mg/L was obtained with a γG− Eu3+:γG−Alexa ratio of 1.6. The kinetics of two assay steps was investigated to optimize the incubation times. Different concentrations of BSA were incubated with nanoparticles for varying times before the mixture of the labeled proteins was added (Figure 3a). The

Figure 4. Calibration curve of BSA measured with the developed nanoparticle-based (61 nm) assay. The data were fitted to a modified Hill function.

detection limit was 74 μg/L or 1.1 nM, corresponding to 2.6 ng in a microtiter well, and the average coefficient of variation was 11% for the signal. The sensitivity is clearly higher than for the most sensitive commercial methods, NanoOrange,13 CBQCA,12 and OPA,38,40 having a limit of detection of 10− 20 ng. Moreover, the sensitivity of the developed method was over 10000-fold higher than for the widely used Bradford7 assay reported in the literature. The saturation of the signal was observed at a concentration level of approximately 1 mg/L or 40 ng of BSA. The total binding area of the BSA molecules was calculated by assuming that the molecules are spherical with a binding area41 of 14 nm2 and organize into a hexagonal closepacked structure on the nanoparticles. This corresponded well to the available surface area of the nanoparticles in the wells, suggesting that the nanoparticles are efficiently covered with the sample protein molecules at approximately 1 mg/L BSA. The response of the assay was tested for 11 different proteins having different sizes, shapes, and pI values to evaluate the protein-to-protein variability (Figure 5). The molecular masses ranged from 14 to 670 kDa and pI values from 4.5 to 11. Different proteins were detected within 20% and individual proteins within 9% coefficient of variation in signal values. The coefficient of variation for different proteins is similar to or higher than those measured for the commercial and reported methods, 20% for the NanoOrange method,13 30% for the CBQCA12 and BCA methods,6 and even 40−50% for the Bradford method,9 as estimated from the literature data. These results suggested that the developed assay responds to different proteins in a nearly equal manner. Thus, the quantification of protein mixtures or proteins with unknown structures may be measured relatively accurately. The incompatibility of the assay for possible interfering agents was not tested. However, the efficacy of protein adsorption is not typically decreased in the presence of small molecules, as protein adsorption is considered an irreversible process and proteins as macromolecules have higher affinity for the surfaces than small molecules.42 Detergents have interfered in our earlier nanoparticle-based assays, as they increase the distance between the donor and acceptor.43,44 However, the high assay sensitivity allows the dilution of the sample to the assay buffer and simultaneous dilution of possible interfering agents. The sensitivity of the assay for detergents also states the

Figure 3. Kinetics of the protein assay for BSA. Nanoparticles with a diameter of 61 nm were mixed with varied concentrations of BSA in a 5.0 mM glycine buffer at pH 3.0. (a) After incubation for the indicated times, a mixture of γG−Eu3+ and γG−Alexa was added. The TR-LRET signal was measured 20 min after the last addition. (b) After incubation for 20 min and the addition of a mixture of γG−Eu3+ and γG−Alexa, the TR-LRET signal was monitored as a function of time.

normalized TR-LRET signal decreases during 20 min at a BSA concentration of 490 μg/L. Thus, the incubation time of 20 min was optimal for the high sensitivity and low variation of the assay. Thereafter, the TR-LRET signal was monitored after the addition of γG−Eu3+ and γG−Alexa (Figure 3b). The equilibrium was reached in approximately 20 min, which corresponds also to the time required for the adsorption of the sample BSA. The signal was stable for at least 30 min in all concentrations tested. The overall incubation time is adequately short compared to the incubation times reported for the most sensitive commercial methods: 30 min for NanoOrange13 and 90 min for CBQCA.12 Although the incubation time for the OPA38 assay is only 5 min, the reagent and the fluorescence signal stability limit the use of the assay, as the signal decreases after the maximum is reached.38,39 A calibration curve was run for BSA under the optimized assay condition and incubation time (Figure 4). The limit of detection was calculated from the standard deviation (2 SD) of the luminescence signal at zero BSA concentration. The 4954

dx.doi.org/10.1021/ac300597j | Anal. Chem. 2012, 84, 4950−4956

Analytical Chemistry

Article

Figure 6. CHO cell calibration curve measured with the developed nanoparticle-based (61 nm) assay. The samples in PBS were diluted 4fold to the assay buffer 50 mM glycine, pH 2.5. The data were fitted to a modified Hill function. Figure 5. Response curves of different proteins assayed with the developed nanoparticle-based (61 nm) method. The data were fitted to a modified Hill function. The average coefficient of variation (CV) for the signal is shown in the inset for assays of individual and all proteins at each measured concentration.

dyes.2,20,21 The number of cells required to adsorb all nanoparticles in an assay was calculated assuming that the nanoparticles are spherical and organize into a hexagonal closepacked structure on the surface of the CHO cells with a surface area45 of 1100 μm2. The calculated number of cells, 2400, is relatively well in accordance with the number of cells, 5000, at the saturation of the TR-LRET signal at high cell concentration.

versatility of the developed assay, as it can be applied for the quantification of detergents. Total Cell Counting Assay. The principle of the cell counting assay was identical to that of the protein quantification assay (Figure 1b). Due to the smaller size of the nanoparticles compared to the size of typical eukaryotic cells, the nanoparticles attached rather to the surface of the cells, which is quite opposite the protein assay, where sample proteins occupied the surface of the nanoparticles. The interaction of CHO cells with the nanoparticles and its effect on the sensitivity of the assay were studied as a function of pH in the universal buffer (TR-LRET/cells, B0/B for 500 cells in Figure 2a). The highest sensitivity was obtained at pH 2−3, and accordingly, glycine buffer, pH 2.5, was chosen for the assay. Our aim was to develop a cell counting method with minor sample pretreatments and dilutions enabling the measurement also at the low end of the concentration range. The cells are frequently harvested in PBS containing approximately 7 mM phosphate and 140 mM NaCl. The developed method was optimized for glycine buffer, and thus, we studied the effect of the concentration of glycine (5.0−100 mM) and NaCl (0.009− 2.5 M) on the sensitivity (data not shown) to dilute the sample minimally. A negative effect on the sensitivity was measured at glycine and NaCl concentrations above 50 and 40 mM, respectively. Therefore, the assay tolerated the 4-fold dilution of the cell samples from PBS into 50 mM glycine, pH 2.5, without a significant decrease in the limit of detection. The assay performance was estimated from the calibration curve of the CHO cells (Figure 6) measured in 50 mM glycine, pH 2.5. Samples mimicking PBS harvesting were prepared and diluted to 50 mM glycine, pH 2.5, and thus, NaCl salt was present. The limit of detection was approximately 200 cells as calculated from the standard deviation (2 SD) of the luminescence signal at a zero cell concentration, and the coefficient of variation was 11% for the signals. The sensitivity is similar to or higher than for the widely used commercial assays, such as acid phosphatase,17,18 MTT,14 and DNA

■

CONCLUSIONS We have demonstrated that proteins can be quantified and cells counted using a novel assay principle, where labeled proteins (donor and acceptor) and the analyte compete for the adsorption to nanoparticles. The methods have a simple mixand-measure protocol performed in a high-throughput microtiter assay format. The increase in the concentration of the analyte is detected as a decrease of the TR-LRET due to the prevented adsorption of the labeled proteins and the subsequent separation of the donor and the acceptor. Standard luminescence plate readers found in most laboratories are utilized, making the methods amenable for the end-users. As labeled particles are not required in the current method, the concept may be applied to match particles which may have more specificity to the adsorbing target analytes in contrast to the methods developed earlier by us.25−28,46 Moreover, the study on the effects of the particle surface properties, counting of particles, or estimation of the size of the particles can be potential applications for the method. The developed method has higher sensitivity than the existing commercial methods for the quantification of proteins and sensitivity equal to or higher than that of the widely used commercial assays for cell counting. The developed method requires no heating steps, which is in contrast to many commercial assays such as the CBQCA12 and NanoOrange13 methods. As the luminescence signal of the developed method is stable over time after the maximum signal is obtained, timed measurements are not demanded. This is in contrast to the OPA method suffering from instable reagents and fluorescence signal. The developed method with the new assay principle is potentially highly useful for the quantification of proteins, for the counting of cells, and for applications such as the study of the properties of the surfaces due to its high sensitivity, simplicity, and versatility. 4955

dx.doi.org/10.1021/ac300597j | Anal. Chem. 2012, 84, 4950−4956

Analytical Chemistry

■

Article

(29) Hakala, H.; Liitti, P.; Puukka, K; Peuralahti, J.; Loman, K.; Karvinen, J.; Ollikka, P.; Ylikoski, A.; Mukkala, V.-M.; Hovinen, J. Inorg. Chem. Commun. 2002, 5, 1059−1062. (30) Huhtinen, P.; Kivelä, M.; Kuronen, O.; Hagren, V.; Takalo, H.; Tenhu, H.; Lövgren, T.; Härmä, H. Anal. Chem. 2005, 77, 2643−2648. (31) Condie, R. M. Pure intravenous human and animal gamma globulins. U.S. Patent 4,296,027, Oct 20, 1981. (32) Hemmersam, A. G.; Rechendorff, K.; Besenbacher, F.; Kasemo, B.; Sutherland, D. S. J. Phys. Chem. C 2008, 112, 4180−4186. (33) Shen, D. Z.; Xue, Y. H.; Kang, Q.; Xie, Q. J.; Chen, L. X. Microchem. J. 1998, 60, 1−7. (34) Quiquampoix, H.; Ratcliffe, R. G. J. Colloid Interface Sci. 1992, 148, 343−352. (35) Duinhoven, S.; Poort, R.; van der Voet, G.; Agterof, W. G. M.; Norde, W.; Lyklema, J. J. Colloid Interface Sci. 1995, 170, 340−350. (36) Duinhoven, S.; Poort, R.; van der Voet, G.; Agterof, W. G. M.; Norde, W.; Lyklema, J. J. Colloid Interface Sci. 1995, 170, 351−357. (37) Conway-Jacobs, A.; Lewin, L. M. Anal. Biochem. 1971, 43, 394− 400. (38) www.piercenet.com/instructions/2160662.pdf, accessed May 5, 2012. (39) www.interchim.com/interchim/bio/produits_uptima/tech_ sheet/FT-UP02727(OPA).pdf, accessed May 5, 2012. (40) www.piercenet.com/instructions/2160663.pdf, accessed May 5, 2012. (41) Wright, A. K.; Thompson, M. R. Biophys. J. 1975, 15, 137−141. (42) Nakanishi, K.; Sakiyama, T.; Imamura, K. J. Biosci. Bioeng. 2001, 91, 233−244. (43) Härmä, H.; Laakso, S.; Pihlasalo, S.; Hänninen, P. Tenside, Surfactants, Deterg. 2010, 47, 40−42. (44) Härmä, H.; Laakso, S.; Pihlasalo, S.; Rana, S.; Bergström, L.; Hänninen, P. Nanoscale 2010, 2, 69−71. (45) Coorssen, J. R.; Schmitt, H.; Almers, W. EMBO J. 1996, 15, 3787−3791. (46) Pihlasalo, S.; Kulmala, A.; Rozwandowicz-Jansen, A.; Hänninen, P.; Härmä, H. Anal. Chem. 2012, 84, 1386−1393.

AUTHOR INFORMATION

Corresponding Author

*E-mail: sari.pihlasalo@utu.fi. Phone: +358 2 333 7069. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the European Commission (FP7 Collaborative Project NANOGNOSTICS-HEALTH-F5-2009242264) and the Graduate School of Chemical Sensors and Microanalytical Systems.

■

REFERENCES

(1) Coulter, W. H. Means for counting particles suspended in a fluid. U.S. Patent 2,656,508, Oct 20, 1953. (2) Stewart, N. T.; Byrne, K. M.; Hosick, H. L.; Vierck, J. L.; Dodson, M. V. Methods Cell Sci. 2000, 22, 67−78. (3) Givan, A. L. Flow Cytometry: First Principles, 2nd ed.; John Wiley and Sons, Inc.: New York, 2001. (4) Schärfe, J. BIOspektrum 2004, 10, 334−336. (5) Lowry, O. H.; Rosebrough, N. J.; Farr, A. L.; Randall, R. J. J. Biol. Chem. 1951, 193, 265−275. (6) Smith, P. K.; Krohn, R. I.; Hermanson, G. T.; Mallia, A. K.; Gartner, F. H.; Provenzano, M. D.; Fujimoto, E. K.; Goeke, N. M.; Olson, B. J.; Klenk, D. C. Anal. Biochem. 1985, 150, 76−85. (7) Bradford, M. M. Anal. Biochem. 1976, 72, 248−254. (8) www.piercenet.com/instructions/2160129.pdf, accessed May 5, 2012. (9) www.piercenet.com/files/1602063_PAssayHB_122910.pdf, accessed May 5, 2012. (10) Roth, M. Anal. Chem. 1971, 43, 880−882. (11) Lorenzen, A.; Kennedy, S. W. Anal. Biochem. 1993, 214, 346− 348. (12) You, W. W.; Haugland, R. P.; Ryan, D. K.; Haugland, R. P. Anal. Biochem. 1997, 244, 277−282. (13) Jones, L. J.; Haugland, R. P.; Singer, V. L. BioTechniques 2003, 34, 850−861. (14) Mosmann, T. J. Immunol. Methods 1983, 65, 55−63. (15) Cory, A. H.; Owen, T. C.; Barltrop, J. A.; Cory, J. G. Cancer Commun. 1991, 3, 207−212. (16) Karmazsin, L.; Balla, G.; Szollosi, J. Acta Paediatr. Hung. 1979, 20, 249−253. (17) Yang, T.-T.; Sinai, P.; Kain, S. R. Anal. Biochem. 1996, 241, 103−108. (18) Connolly, D. T.; Knight, M. B.; Harakas, N. K.; Wittwer, A. J.; Feder, J. Anal. Biochem. 1986, 152, 136−140. (19) Zhang, Z. D.; Cox, G. Biochem. Mol. Biol. Int. 1996, 38, 431− 436. (20) Gillies, R. J.; Didier, N.; Denton, M. Anal. Biochem. 1986, 159, 109−113. (21) Jones, L. J.; Gray, M.; Yue, S. T.; Haugland, R. P.; Singer, V. L. J. Immunol. Methods 2001, 254, 85−98. (22) Campbell, A. K. ChemiluminescencePrinciples and Applications in Biology and Medicine, Ellis Horwood Ltd.: Chichester, U.K., 1988; Chapter 5, pp 267−314. (23) Crouch, S. P. M.; Kozlowski, R.; Slater, K. J.; Fletcher, J. J. Immunol. Methods 1993, 160, 81−88. (24) Lundin, A.; Hasenson, M.; Persson, J.; Pousette, Å. Methods Enzymol. 1986, 133, 27−42. (25) Härmä, H.; Dähne, L.; Pihlasalo, S.; Suojanen, J.; Peltonen, J.; Hänninen, P. Anal. Chem. 2008, 80, 9781−9786. (26) Pihlasalo, S.; Kirjavainen, J.; Hänninen, P.; Härmä, H. Anal. Chem. 2009, 81, 4995−5000. (27) Valanne, A.; Suojanen, J.; Peltonen, J.; Soukka, T.; Hänninen, P.; Härmä, H. Analyst 2009, 134, 980−986. (28) Pihlasalo, S.; Pellonperä, L.; Martikkala, E.; Hänninen, P.; Härmä, H. Anal. Chem. 2010, 82, 9282−9288. 4956

dx.doi.org/10.1021/ac300597j | Anal. Chem. 2012, 84, 4950−4956