Materials and methods
Cell culture
Human embryonic kidney cells (HEK 293 cell line, ATCC CLR-1573, Rockville, MD) were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Gibco, MD, USA) supplemented with 10% fetal bovine serum (FBS; Vitrocell, São Paulo, Brazil). Cells were cultured at 37 °C under a humidified atmosphere containing 95% air/5% CO2.
Cells staining with CFSE
Cells were labeled using the CellTraceTM CFSE Cell Proliferation Kit (Thermo Fisher Scientific, Waltham, MA, USA; C34554), according to the manufacturer’s instructions, with a few modifications. Cells (1 × 106) were washed with PBSA (Phosphate- buffered saline without calcium and magnesium), ressuspended in CFSE solution in PBSA (5 µM CFSE—1 mL final volume) and incubated for 20 min at 37 °C on a side-to-side shaker. A volume (9 mL) of 10% FBS DMEM was then added and the cells were incubated for 5 min at 37 °C in a side-to-side shaker in order to allow free CFSE to bind to serum proteins and improve free CFSE elimination. Labeled cells were centrifuged, ressuspended in 10% FBS DMEM and seeded for the growth curve experiment.
Growth curves
CFSE stained cells (5 × 104) were seeded onto 35 mm wells in 10% FBS DMEM, with culture medium change every other day. Triplicate wells were harvested by trypsinization at the indicated time points and the cells were fixed in 1 mL final volume of 3.7% formaldehyde. Growth curves were generated using four different approaches, namely: manual cell counting using the Neubauer Chamber, automatic cell counting using the Coulter Counter Analyzer (Beckman Coulter), automatic cell counting using the Accuri C6 Cytometer (BD Biosciences) and analysis of CFSE signal decay, also using the Accuri C6 Cytometer (BD Biosciences).
Automatic cell counting using the Accuri C6 Cytometer (BD Biosciences)
An end point acquisition stop was set at 100 µL for absolute cell counting. Cells were gated apart from debris in an SSC-A × FCS-A plot and the number of events were multiplied by 10 to yield the total number of cells per milliliter.
CFSE signal measurement using the Accuri C6 Cytometer (BD Biosciences)
Cells were gated apart from debris in a SSC-A × FCS-A plot. Single cells were then gated apart from the doublets and clumps in a FCS-H × FSC-A plot. An end point acquisition stop was set at 4000 events inside the single cells gate, from which the CFSE median fluorescence intensity (MFI) was determined. The MFI values were plotted as a function of time to analyze the kinetics of CFSE decay. Next, the inverse of MFI (values were raised to the power of − 1; MFI−1) was plotted as a function of time in order to change the plot from descendent exponential into an ascendant exponential.
Doubling time calculation
The cell specific growth rate (µ) was determined from the slope of the natural logarithm of cell count or MFI−1 as a function of time and doubling time (DT) using the formula [DT = ln 2/µ] [12].
Statistical analysis
Statistical analysis of the coefficient of variation of three replicates of the cell growth curves, determined by the different methods, was carried out by paired Wilcoxon tests. Analysis of CFSE MFI intensity and number of cells from time points 144 h to 168 h, in the presence or absence of 20 µg/mL Mitomycin C, was carried out using the t test.
Results
HEK 293 cells were stained with CFSE, seeded onto several wells and cultured for 7 days. Cells were harvested at different time points and the CFSE MFI was determined using the Accuri C6 Cytometer (BD Biosciences; Fig. 1b). The CFSE MFI from each time point was plotted as a function of time rendering an exponential descendent curve (Fig. 1c). Then, the inverse of CFSE MFI measurements were plotted as a function of time, transforming the descendent exponential plot into an ascendant plot and fitting very closely the conventional cell growth curve (Fig. 1d).
Furthermore, the absolute number of cells in the same samples were determined using different types of equipment and plotted as counting-based cell growth curves. The counting-based curves were generated by using three different counting devices, namely: the Neubauer chamber, the Coulter Counter Analyzer Cell Counter (Beckman Coulter) and the Accuri C6 Cell Counter (BD Biosciences). The fluorescence-based curve was plotted together with the counting-based ones, confirming that the curves were similar (Fig. 2).
We quantitatively analyzed the results obtained using the different methods by comparing the doubling time values obtained from each of the curves. The fluorescence-based method delivered the doubling time 18 h 56 min, slightly lower than those calculated from the counting-based method (Neubauer chamber—20 h 41 min; Cell Counter—20 h 05 min and Accuri C6—20 h 16 min; Additional file 1: Table S1).
The accuracy of cell growth curves generated by counting-based method depends on the precision in the determination of the total number of cells. On the other hand, this is not a requirement for fluorescence-based method, since the MFI of a small sample of single cells reveals the MFI of the whole population (Fig. 3a). The presence of cellular debris and cell clumps had no interference on MFI determination, as well as the number of cells left behind during cell harvesting. Nonetheless, these factors are important interferences on counting-based methods (Fig. 3b).
Furthermore, the values variation among technical replicates were lower in fluorescence-based method when compared to the counting one. This is quantitatively demonstrated by comparing the coefficient of variation of the triplicate measurements for each time point of the curve (Paired Wilcoxon test; Fig. 3c). This comparison also indicates the higher accuracy of fluorescence-based method.
Finally, we investigated the two last time points (144 and 168 h) to address the reliability on matching CFSE signal decay and cell proliferation. At these late time points, the CFSE has been maintained in conditions prone to spontaneous CFSE degradation for a long time (cytoplasmatic metabolism, 37 °C temperature). CFSE degradation regardless to cell division would abrogate the synchrony between fluorescence decay and cell proliferation. To address this question, the MFI of these last two time points were compared using Mitomycin C to halt cell proliferation. The number of cells did not increase upon Mitomycin C treatment and no CFSE signal decay was observed, confirming that CFSE decay matches to cell proliferation even after 7 days under cell culture conditions (Additional file 1: Figure S1). Since CFSE signal decays over time, this method may have limitations for longer-term studies, and this validation is requested for these cases.
Discussion
The growth curve plot is the most common analysis to characterize in vitro cultured cells under different conditions, such as genomic manipulations [10, 14], presence of biomaterial [15] and treatment with chemical compounds [16]. Here we compared cell growth curves plotted by counting- and fluorescence-based methods.
Counting-based methods constitutes a laborious task, in addition to be influenced by cell loss during cells harvesting, which directly affects the total number of counted cells. Moreover, cellular debris and cell clumps increase the underestimation of total number of cells in cell-counting based growth curves. On the other hand, fluorescence-based method constitutes the usage of fluorescence cell tracers to stain cells (Fig. 1) and tracking proliferating cells by analysis of fluorescence decay [13]. As a result, the fluorescence of a cell population decreases as a function of time as cells proliferate, allowing this methodology to be employed to assess cell proliferation and to determine cell lines doubling time [11, 14]. This method is based on assessing the fluorescence signal of a sample of single cells from a cell population (Fig. 3a), therefore it is not influenced by cell loss during cell harvesting, cellular debris or cell clumps.
These interferences in counting-based method lead to an underestimation of the number of cells. It biases the cell growth curve towards a lower inclination (Fig. 2), which favors a longer doubling time (Additional file 1: Table S1). Three main interferences are detailed below (Fig. 3b).
First, the counting-based method is influenced by cell loss during cell harvesting from the culture plate. Cell growth curves usually start with a few cells in the first time points and a much higher number of cells at the last time points, reflecting a comprehensive representation of cell proliferation dynamics. The coefficient of variation among the three replicates in the first time points, in the case of the counting based methods, is much greater than that of fluorescence-based method, because any cell loss during harvesting has a great influence, considering the low number of cells present at these first time points (Fig. 3c). Losing cells during harvesting has no interference in the fluorescence method, considering that a small sample of harvested cells is sufficient to address the CFSE MFI of single cells at any time point. Second, cells which are disrupted into cellular debris are not taken into consideration in the counting-based methods, since they are out of the range of detection, underestimating the total number of cells. Third, whether or not the doublets and cell clumps are considered, the cell counting-based methods are biased towards underestimation of the total number of cells, especially in the last time points, when the cellular density is higher.
Furthermore, the lower variation among measurements of the same time points in fluorescence-based methods is another advantage in comparison to counting-based methods (Fig. 3c). The development of accurate techniques to increase the precision of cell growth curves is important to provide reliability in determination of Lag, Log, and Stationary phase transitions. Moreover, it is also crucial for characterization of cell lines used for production of recombinant proteins in the Biotechnology Industry. Moreover, cell tracers with different spectra may be used to stain different cell lines, which allows the analysis of several cell lines simultaneously [11]. Therefore, another advantage of fluorescence-based method is the possibility of analyzing co-cultured cells, which is not possible by counting-based methods.
We conclude that the generation of growth curves of adherent cells by fluorescence-based methods have three main advantages in comparison to counting-based methods. First, fluorescence-based method does not overestimate doubling time because it is not influenced by cell loss during harvesting, debris and clumps. Second, the variation among different measurements of the same time points is lower, which increases the reliability on Lag, Log and Stationary phase transition. Third, this method allows analysis of co-cultured cells.
Limitations
-
The fluorescence methods are more expensive because it requires both fluorescence dye and the cytometer equipment.
-
The fluorescence-based method requests knowledge and training on cytometry to be carried out.