Krista McCutcheon , David Fei
Department of Analytical Sciences, Genentech Inc., South San Francisco, CA 94080-4990
Viability assays are crucial tools in the development of many drugs, particularly of cytotoxic agents. To best characterize potential compounds, viability assays must ideally be able to reproducibly and accurately report small percentage changes in viability, provide statistically significant quantification of biological activity, and account for changes in cell number during the assay. An analysis of cell number interference on MTT, alamarBlue™, and Guava® ViaCount® viability assays was performed in the development of a quantitative CD40 ligation assay. MTT and alamarBlue assays both showed significantly large statistical variations in viability measurements due to changes in cell numbers. Guava® ViaCount®’s ease of use and high reproducibility provide significant advantages over current viability or cytotoxicity assays where changing cell numbers can be problematic.
The investigation of cytotoxic therapeutic agents to treat malignant cells or other disorders of cell proliferation is a very active field of drug development. Viability assays that have the statistical power to quantitate biological activity, differentiate the potency of compounds and indicate stability are the most valuable. For this purpose a sensitive, reproducible, accurate, and dose-dependent cell-based assay is required.
A number of different methods are currently used to measure cell viability by indirect means of proliferation and metabolism (e.g., tetrazolium dyes such as MTT1, 3[H]-thymidine, or BrdU uptake2, and alamarBlue3). Both MTT and alamarBlue can be carried out in a 96-well, high-throughput format and have been used in quantitative cytotoxicity assays.1,3 A critical assumption used in indirect assays of viability, is that cell number will decrease in a reproducible manner, inversely and linearly proportional to the dose of the drug. This assumption would fail in cases where cell number is highly variable over the time course of the assay. For instance: if a small number of responsive cells are masked by a majority of proliferating cells; if the drug affects cellular aggregation or adhesion (properties which also are known to indirectly affect cell growth); if the drug has a nonlinear effects on cell number (e.g., by causing apoptosis or changing the cell cycle);or if the target cell line itself has irregular growth properties. Changing target cell lines or obtaining separate cell number data (e.g., using protein assays, coulter counters, TruCount™ beads or a haemocytometer), for each well in an assay, would often not be feasible. Guava® ViaCount® offers a unique alternative to measuring viability, enumerating both viable and non-viable cell populations, at the same time, in one sample. Dual labeling with two dyes of different fluorescent wavelengths and membrane permeability, are used to discriminate the viable and non-viable cell populations. One dye, PM2 (permeable to all cells) and the other, PM1 (permeable only to cells with a compromised cell membrane) enable viability data to be expressed as a percentage of total cell number in each sample well of the assay. Using Guava ViaCount, the direct measurement of non-viable cells, and the normalization of each data point to total cell number, corrects for inter-well variability introduced by cell number, improving intra- and inter-assay precision.
We report the impact of cell number interference on the MTT, alamarBlue, and Guava® ViaCount® assays, measured during the development of a quantitative bioassay for a humanized monoclonal antibody directed against human CD40. On malignant B cells, ligation of CD40 can result in direct inhibition of cell growth with or without apoptosis. The outcome and extent of CD40 activation on cells is complex, and may be regulated by cell phenotype, the strength of the CD40 signal, and the presence of co-factors.4-6 Both the pleiotropic anti-CD40 drug activity and the irregular growth properties of the malignant B-cell line used in the assays contributed to cell number interference in the MTT and alamarBlue assays. We demonstrate how the ability of the Guava® ViaCount® assay to normalize viability to cell number offered a major improvement in the monoclonal antiCD40 bioassay variability. The relative ease of the assay and our statistical results suggest that a viability assay using the Guava PCA and Guava® ViaCount® reagent may provide a significant advantage over other bioassays in current use, particularly in cases where intra- and interassay cell number is a problem.
Humanized anti-CD40 monoclonal antibody (IgG1) is a Genentech reagent (South San Francisco, CA, USA). Human IgG1κ was obtained from Sigma (St. Louis, MO, USA). Anti-human Fcγ, used for cross-linking the humanized antibodies to the culture plates, was obtained from Jackson ImmunoResearch (Westgrove, PA, USA). The goat anti-human IgG-FITC conjugate, used for FACS analysis, was purchased from BD PharMingen (San Diego, CA, USA). Guava ViaCount reagent was purchased from Guava Technologies (Hayward, CA, USA), MTT reagent was purchased from ATCC (Rockville, MD, USA), and alamarBlue was purchased from Trek Diagnostic Systems, (Westlake, OH, USA).
The Raji human cell line was obtained from ATCC (Rockville, MD, USA), and was grown in culture media conditions recommended by ATCC with 100 µg/mL penicillin/streptomycin. The cells were analyzed for homogeneity and uniformity of CD40 expression by flow cytometry. One million cells in growth media were centrifuged at 1,200 x g and re-suspended in 0.25 mL block buffer (1% BSA, PBS) for 1 hour on ice. Block buffer was removed and 0.25 mL of 10 µg/mL anti-CD40 monoclonal antibody or isotype-matched human IgG1κ in block buffer was added for 1 hour at ambient temperature. The cells were washed 3 times in PBS, followed by adding a 1:10 dilution of goat anti-human IgG-FITC in block buffer for 30 minutes at ambient temperature. The stained cells were washed three times in PBS, resuspended in 0.4 mL of PBS, and analyzed by flow cytometry on a BD FACSCaliber (Becton Dickinson, San Jose,CA).
Cell morphology was examined under a Nikon Diaphot 200, inverted light microscope, using a Phase I, DL 40x magnification lens.
A dilution series of anti-CD40 monoclonal antibody was cross-linked to high protein binding plates through the constant region of the heavy chain. Unbound anti-CD40 monoclonal antibody was removed by washing, and 50,000 Raji cells were added to each well. After 72 hours, the cells in each well were mixed, and aliquots were removed for staining with the MTT, alamarBlue, or Guava ViaCount reagents.
High protein binding, 96-well, Costar plates were coated with 0.1 mL, 30 µg/mL, goat anti-human Fcγ in 50 mM sodium bicarbonate, pH 9.6, overnight in a humidified 5% CO2, 37 °C incubator. Unbound goat antihuman Fcγ was removed with two 0.2 mL washes using sterile PBS. Coated plates were blocked with 0.3 mL, 0.5% BSA, PBS for 1 hour at ambient temperature. Blocking solution was removed and the plate washed once with 0.3 mL PBS. A dilution series of anti-CD40 monoclonal antibody (4 to 0.002 µg/mL) was prepared in PBS, and 0.1 mL applied on the plate with gentle agitation for 4 hours at ambient temperature. Unbound antibody was removed, and the plate was washed twice with 0.2 mL PBS. Raji cells were seeded into growth medium (RPMI 1640, 10% FBS, 2 mM L-glutamine, 1.5 g/L sodium bicarbonate, 4.5 g/L glucose, 10 mM HEPES, 1 mM sodium pyruvate, and 100 µg/mL penicillin/streptomycin) at less than 1 x 106 cells/mL, and grown in a humidified 5% CO2, 37 °C incubator for 24 hours before the assays were performed. Raji cells grew in reversible cell aggregates, and to maintain responsiveness in the assay, the cells were routinely split before the density exceeded 1 x 106 cells/mL. On the assay day, a suspension of 0.25 x 106 cells/mL of Raji cells was prepared in assay media (growth medium, containing 2% FBS), and 0.2 mL was added to each well to give 0.5 x 105 cells/ well. It was necessary to assay the cells in medium with reduced serum (1-5%). Cells assayed in 10% serum did not respond well in the assay. The cells are incubated with the Fc-cross-linked anti-CD40 monoclonal antibody for 72 hours in a humidified 5% CO2, 37 °C incubator. Raji cells typically grow in suspension. However, to make sure we assayed a population of cells representative of each well, we pipetted up and down in the wells 10 times before transferring cells for viability assays.
The alamarBlue assays were performed by adding 50 µL of alamarBlue to 50-100µL cells for 2–16 hours in a humidified 5% CO2, 37 °C incubator. Fluorescence was measured using an excitation wavelength of 530 nm and an emission wavelength of 590 nm. alamarBlue acts as an oxidation-reduction indicator that fluoresces in response to the chemical reduction of growth medium resulting from cell growth. Signal is quantified at 590 nm with a fluorometer.
MTT assays were performed as described in the product manual using 50 µL of cells. After 6 hours in a humidified 5% CO2, 37 °C incubator, detergent was added at RT and the signal measured from 6–24 hours. The yellow tetrazolium salt (MTT) is reduced in metabolically active cells to form insoluble purple formazan crystals, which are solubilized by the detergent. The color is then quantified by spectrophotometric means (570 nm).
Guava® ViaCount® assays were performed by transferring 50 µL of cells in media into microtitre tubes containing 150 µL of Guava® ViaCount® reagent. After incubation for 5 minutes at room temperature in the dark, viability and cell count measurements were taken using the Guava® Personal Cell Analysis System (PCA™). Viability data was expressed as a percentage of total cell number.
Percent viability (Guava® ViaCount®) or optical density (alamarBlue or MTT) was plotted on the y-axis against a logarithmic scale of anti-CD40 monoclonal antibody concentration (µg/mL) on the x-axis. Kaleidograph software was used to fit a 4-parameter logistic curve, according to the equation:
y = [(m1-m4)/ (1+(m0/m3)m2)] + m4
where m1= the estimated response at zero dose (minimum); m2= curvature parameter; m3= the response at 50% maximum response (IC50); and m4= the estimated response at infinite dose (maximum).
Anti-CD40 monoclonal antibody was observed to have multiple effects on the Raji cell line. In addition to decreasing cell viability, dose-dependent effects on cell number, homotypic aggregation, and cell spreading were also observed (Figure 1). Each effect had its own dose-dependence and time course of activity in response to the monoclonal anti-CD40. The contribution of cell number to assay variability was found to be significant, and normalization of viability data to cell number was essential to developing a reproducible assay. The cell number data (Figure 1A) fit poorly to a 4-parameter curve (correlation term, R = 0.88), and the coefficient of variation (CV) of individual points across the anti-CD40 antibody concentration range, was unacceptably high (8 - 30%).
Figure 1.Pleiotropic effects of the humanized anti-CD40 monoclonal antibody on Raji cells. Raji cells were incubated with humanized anti- CD40 monoclonal antibody for 72hrs. Samples were either assayed by Guava® ViaCount® A), or examined under an inverted microscope at 40x magnification (B, 1 μg/mL anti-CD40 antibody is shown). The effect of the drug on cell number, cell aggregation, and cell spreading are illustrated.
A comparison of typical results from alamarBlue, MTT and Guava® ViaCount® assays is shown in Figure 2. Across the dose-response curves, the intra-assay variability was 3–57% in the alamarBlue assay, 8–30% in the MTT assay, and 1–5% in the Guava® ViaCount® assay. Furthermore, alamarBlue and MTT assays on some days were unable to detect any change in viability. Guava® ViaCount® showed an inter-assay variability of 4–12% over 10 assays.
Figure 2.Dose-dependent inhibition of Raji cell viability by humanized anti-CD40 monoclonal antibody. Raji cells were incubated with humanized anti-CD40 monoclonal antibody for 72 hours. alamarBlue (A), MTT (B), or Guava® ViaCount® (C), from duplicate samples within one plate are shown, demonstrating variability within an assay.
A viability assay for the humanized anti-CD40 monoclonal antibody was qualified using Guava® ViaCount®. An inhibition curve with sensitivity over 1.5 log units was developed on Raji cells from 0.01 to 2 µg/mL antiCD40 monoclonal antibody. The mean 50% inhibitory concentration from ten independent assays, with cells from passages 10 to 22, was 0.21 ± 0.03 µg/mL (with a coefficient of variance of 12.3%). Precision was evaluated by creating three controls at the high (2 µg/mL), mid (0.3 µg/mL), and low (0.1 µg/mL) dose points of an inhibition curve. From 12 replicates in a plate, the percent viability of low, mid, and high controls was 57.8 ± 2.26% (CV= 3.9%), 45.2 ± 4.53 % (CV= 10.4%) and 32.8 ± 1.97 (CV = 6.0%), respectively. An intrasample CV of 2.9% was obtained in the Guava® ViaCount® assay when the mid control sample was measured repeatedly six times. The stability of the Guava® ViaCount® signal was tested using a row of 12 mid control samples measured after 10 minutes and one hour. The coefficient of variation between the 10 minute and 1 hour samples was less than 4.5% and the measurements showed no time-dependent drift. Using the entire 96-well plate, the intra-plate variation, determined using an average IC50 dose, was less than 5%. Accuracy was tested by measuring the recovery of a sample prepared at 75% of the average IC50 dose. The percent viability of the recovery sample was compared to 75% of the IC50 value obtained from the doseresponse curve in the same assay. The percent viability of the recovery sample was 103% of the expected value.
The Raji cell line used in these assays appeared homogeneous by FACS analysis: side scatter and forward scatter parameters showed a morphologically similar population; 100% of the live cells were consistently CD40+; and the level of CD40 expression was uniform (not shown). Despite this apparent cell uniformity, antiCD40 monoclonal antibody was observed to not only decrease cell viability but also to have dose-dependent and irregular effects on cell number and morphology. Changes in input cell number and assay media, sub-cloning the cell line, and synchronizing the cell cycle, were all unable to improve the assay variability. As a control, purified human IgG1κ was tested in the dose range of the assay and showed no effect on Raji cells. The observed morphological changes in Raji cells in response to anti-CD40 antibody in our assays have been reported in related studies.7-13
Cross-linking of the anti-CD40 monoclonal antibody via Fc has been found to optimize the inhibitory signal, CD40 at the cell surface.14,15 Examination of the wells under the microscope showed no evidence that the dose-dependent decrease in cell numbers or viability measured in these assays was caused by a proportional increase in cells being irreversibly captured by the anti-CD40 monoclonal antibody coat. The mechanism underlying the reduced cell numbers in this assay was examined but our data was inconclusive. AntiCD40 antibody has been observed in other studies to inhibit cell growth, with or without apoptosis.4-6
Flow cytometry analysis to assess viability and determine chemotherapeutic IC50 values on human leukaemic cell lines has previously been found to be comparable to the MTT assay.16 The ability of the Guava® PCA to use smaller sample volumes, fewer events, and 10-fold lower cell numbers than traditional flow cytometers is an advantage when cell culture volume and time is valuable. Furthermore, we found normalizing viability to cell number increased the confidence of IC50 determinations. This normalizing was facilitated by the Guava® ViaCount® software, which reports absolute cell counts and concentration for all samples. For this purpose, the Guava® ViaCount® method is much simpler than other methods, such as MTT or alamarBlue, that require a second assay to normalize data to cell number (e.g., coulter counters, TruCount beads, or a hemocytometer).
When used on the Guava PCA-96 system, the Guava® ViaCount® assay permits high throughput data acquisition directly from 96-well plates, creating a simple way to screen compounds or antibodies for cytotoxicity. Our data support that the Guava ViaCount assay is able to generate a sensitive, reproducible, and accurate quantitation of the anti-CD40 monoclonal antibody’s biological activity. This assay will be a valuable tool in the development of assays measuring the drug’s potency and stability. Other drug-cell line combinations may benefit from this type of analysis, particularly when there is a need to isolate the activity of a drug on cell viability from its non-linear effects on cell number.
We thank Lori O’Connel and Leonard Presta for engineering the humanized anti-CD40 antibody used in this study. We also thank Eleanor Canova-Davis and Maureen Beresini for helpful comments and advice.