The development of cell lines that express a specified protein of interest is critical to the generation of biopharmaceutical molecules such as monoclonal antibodies. To establish a cell line, single cells are seeded in wells and allowed to proliferate to form a colony, which can be further expanded after assessing productivity. Instrument technologies exist to seed single cells in a microplate format, including flow cytometry and singlecell printing devices. A more traditional approach to obtain single-cell wells is known as limiting dilution. In this technique, a series of serial dilutions arrives at singlecell wells along with a distribution of void wells and multiple-cell wells at a predicted statistical probability.
Assessing clonality is key to the establishment of a cell line and evidence of monoclonality is required by regulatory agencies to get a biopharmaceutical drug to the marketplace. The traditional and most accepted approach involves the visualization of wells using transmitted white light (brightfield) on day 0 to confirm the presence of a single cell. However, the definitive identification of a single on day 0 is not without challenge, as cellular debris and well artifacts can be easily mistaken for cells. Consequently, cell line developers typically evaluate cells at the colony stage and trace back the origins of the colony to confirm monoclonality.
An alternative method to assessing clonality involves fluorescently labeling the cell population prior to seeding single cells. Here, we outline the differences between a transmitted white light (WL) workflow and one involving both WL and fluorescence (FL). We primarily focus on the benefit of using fluorescence dyes to more easily automate single cell detection while simultaneously establishing clonality more conclusively. There are additional fluorescence techniques (GFP expression systems, for example), which have further benefits, but are not discussed in detail here because their workflows are very similar. However, when considering a fluorescence expression workflow, it is important to evaluate the sensitivity of the imaging instrument as dyes tend to be much brighter than expression-based approaches. In this application note, we use the viability dye calcein AM (CAM), a molecule that fluoresces green only after translocating the membrane of living cells. We lay out a basic approach to label cells for reliable detection of single cell wells while maintaining high rates of viability.
Materials and methods
Using calcein AM dye to label cells before limiting dilution
In order to develop a workable cell staining protocol, we used the EarlyTox™ Cell Viability Assay Kit (Molecular Devices, cat. # R8343). In this kit, calcein AM (CAM), a small cell-permeable molecule, can be employed as marker for cell viability. Once CAM translocates the plasma membrane, the AM moiety is cleaved by esterases in the cytosol, resulting in a green fluorescent calcein molecule. Calcein presence in the cytosol is transient and viable cells typically eject this molecule gradually over a span of hours.
Labeling cells and preparation for limiting dilution
For limiting dilution, CHO-K1 cells (ATCC) were used. CHO-K1 cells were grown in adherent conditions in 35-mm dishes to 50–75% confluence. To prepare for labeling, cells were washed with PBS then incubated with serum-free media containing the specified calcein AM fluorescent dye concentration for 40 minutes at 37°C. Cells were then washed 3 times with PBS and lifted using trypsin and prepared for counting. To eliminate clumping, cells were filtered through a 40-μm strainer (Corning, cat. # 431750) prior to counting (Countess, ThermoFisher Scientific). Cells were resuspended to 1 x 106 cells/mL in serum medium (10% FBS) and diluted sequentially to achieve a 0.5 cells/well concentration and plated in 384-well plates (Greiner, cat. # 781182).
Scanning plates in WL and FL using the CloneSelect Imager (CSI)
Plates were imaged on a CSI modified to include fluorescence capability. Imaging occurred within 30 minutes of seeding to allow cells to settle and adhere to the well bottom. Microplates were imaged using the combined fluorescence and brightfield mode. Fluorescence images were captured with a 300 ms exposure. A positive control well containing 1000 cells was used to optimize focal height prior to plate scan, and the cell count feature was used automate to cell counts. Plates were subsequently scanned using only the brightfield standard resolution scan mode on days 0, 1, 2, 3, 6 and 10.
To quantify raw outgrowth, single-colony wells were counted on day 10 and displayed as a percentage of the total wells. Single colony-containing wells were then evaluated on a well-by-well basis for the presence of single cells on day 0 to determine clonal outgrowth, displayed as the percentage of single cells that subsequently formed colonies.
Instrument overview and workflow
Confirming that a cell type expressing a biologic originated from a single cell progenitor is a critical requirement for cell line development and regulatory approval. Two general imaging approaches can be used to determine whether a cellular population is monoclonal in origin. The first approach assesses cells in a label-free fashion, whereas a new emergent method uses fluorescence to detect the presence of cells. Both approaches involve a sequence of acquisition analysis and reporting steps.
The label-free approach to assessing monoclonality
A standard limiting dilution of 0.5 cells per well is expected to yield 30% single cell-containing wells, however identifying those single-cell wells at day 0 is difficult. Debris and well artifacts can easily be mistaken for cells, making the process subjective and error-prone. As a result, plates are typically scanned in white light repeatedly over the course of several days, beginning with day 0, to monitor the growth of cells into colonies (Figure 1, left column). Once prominent colonies have formed (usually within 6–10 days), they can be tracked back to day 0 to determine the presence of a single cell. This process can be time-consuming, but the Loci Count feature on the CloneSelect Imager helps improve this process by automating the detection of single colonies (yellow box) and linking various time points of a plate scan. Meanwhile, the Report Generator feature allows the identification of cell regions (green box in Figure 1) and debris regions (not shown), and the exporting of a monoclonality report for submission to regulatory agencies. Wells where single colonies form and can be shown to originate from a single cell are designated as monoclonal and carried forward to assess productivity (titer) and scale-up.
Figure 1. Comparing the transmitted white light (WL) and fluorescence (FL) workflows in assessing monoclonality.
The fluorescence approach to assessing monoclonality
An alternative approach to the assessment of monoclonality involves using fluorescence, a powerful biological tool that generates high signal-to-background (Figure 1, right column). One method involves using a fluorescent probe to stain cells expressing the biopharmaceutical target protein of interest. The stained cells are then seeded into microplates using a limiting dilution approach. Scanning a plate on the fluorescence-capable CSI allows acquisition of high-resolution images in fluorescence and white light. One advantage of using fluorescence is that plates can be analyzed for clonality immediately after seeding (on day 0). In comparison, the white light workflow requires waiting until prominent colonies have formed about 6 to 10 days post-seeding. The use of fluorescence also enables accurate determination of cell counts based on thresholding, automating an otherwise laborious process (Figure 2). Finally, one can use fluorescence probes that only stain viable cells, eliminating the risk of dead cells, debris, or well artifacts from erroneously contributing to cell count. A white light overlay provides added assurance that what is detected in fluorescence in fact appears as a cell in brightfield.
Figure 2. Fluorescence workflow enables automated cell per well counting and detection of the cell regionFluorescence scan using a fluorescence-capable CSI shows a color-coded pattern for cells per well. Void wells are white, single-cell wells are green, asymmetrical or dividing-cell wells are yellow, and multiple cell wells are red. Whole well and detailed view of (B) single cell region (C) multi-cell region, (D) asymmetrical, diving cell region (E) void regions containing a cell artifact. Note that artifact regions, which contain structures that appear indistinguishable from cells in white light, will not fluoresce and as such will automatically be eliminated from the fluorescence analysis.
Automated cell counting per well using fluorescence for monoclonality identification
Detection of fluorescence signal from stained cells enables the automated detection and count of cells in an objective and reproducible manner. Single-cell wells are marked green while multiple cell-containing wells are marked red. Dividing cells are marked yellow based on a pre-specified x and y-axis symmetry ratio while void wells are labeled white. Example images of each classification type are shown in Figures 2 B–E.
Using the fluorescence workflow to determine the optimal Calcein-AM concentration
A critical parameter in developing a reliable labeling protocol prior to limiting dilution is ensuring that the cells are adequately labeled for detection using the fluorescence channel on the CSI. An ideal staining condition would have reproducible low concentration sufficient to label all or most cells without the cytotoxic effects as single cells proliferate to form colonies.
Three concentrations of CAM viability dye were compared under identical staining conditions. The concentrations tested were 5 μM as a high staining range, 1 μM as a medium level stain and 0.5 μM as a low-level stain. To confirm that the staining method works, we set up an initial staining ~1000 cells per well using the three concentrations of dye. We scanned those wells using fluorescence and brightfield on the CSI under identical exposure conditions. A representative subset of a well is shown in Figures 3A, B, and C. One can observe that the majority of cells are detectible at the 5 μM (Figure 3A) and 1 μM CAM conditions (Figure 3B), but not at the 0.5 μM CAM condition (Figure 3C).
We then performed a limiting dilution at 0.5 cell per well using the stained cells then scanned the plates in fluorescence and brightfield using the CSI under identical exposure settings. The fluorescence-based cell count automatically calculates the number of cells per well, as shown in Figures 3D–F. From this, one can observe that the 5 μM and 1 μM CAM conditions appear to show a similar distribution of cells per well from limiting dilution (Figure 3D–E). Meanwhile, the 0.5 μM CAM condition shows only a small number of wells containing single and multiple cells (Figure 3F). This low number of cells detected is consistent with the notion that using 0.5 μM CAM is not sufficient to stain the majority of the cells of interest to a detectible level on the CSI. It is important to note that colonies still formed in this plate (data shown in Figure 4), indicating that cells were present, but not detectable with fluorescence.
To establish that the level of staining is sufficient to detect the majority of cells, we counted the total number of void wells, single-cell wells and multiple-cell wells for all the 3 concentrations of CAM. Under 5 μM CAM, 58.2% of the wells were devoid of cells, whereas 31.3% contained a single cell, and 10.5% (n=380 wells) contained multiple cells. Meanwhile under 1 μM CAM staining levels, 58.7% of wells contained no cells, 30.3% contained a single cell, and 11.1% (n=760 wells) contained multiple cells. Finally, under 0.5 μM CAM staining levels, 96.6% of the wells contained no cells, 2.1% of the wells contained a single cell and 1.3% (n=760 wells) of the wells contained multiple cells. Comparing the cell per well valence distribution with theoretical values for a 0.5 cell per well limiting dilution shows that the 5 μM and 1 μM staining conditions are very similar to the predicted values, whereas those of 0.5 μM differ significantly (Figure 3G). A quantification of the difference shows that the 5 μM and 1 μM conditions diverge by <5% whereas that of 0.5 μM diverges by >90% from the predicted values (Figure 3H).
Figure 3. Finding a suitable concentration of calcein AM to optimize detection at the single cell level. (A) High-density wells containing cells stained with 5 μM CAM shows consistent detection of cell signal. (B) High-density wells containing cells stained with 1 μM CAM shows reliable detection of cell signal. (C) High-density wells containing cells stained with 0.5 uM CAM shows that only a minority of cells can be detected based on the low fluorescence intensity signal. (D–F) Fluorescence scan of plate using CSI from limiting dilution using 5 μM, 1 μM, and 0.5 μM CAM cell staining. (G) Cell per well count tally for the 5, 1, 0.5 μM CAM cell staining and comparison with theoretical values from limiting dilution at 0.5 cells per well. Comparison of single cell counts (dashed line). (H) Percent divergence of observed and predicted single cell data. >90% for 0.5 μM CAM compared with <5% for 1 and 5 μM CAM.
Using the White-Light workflow to assess the effect of CAM on viability
Fluorescent probes that stain living cells are notorious for having deleterious effects on cell viability. As such, we sought to test the effect that CAM has on outgrowth of clones after limiting dilution. We evaluated colony outgrowth from the cells that were stained with 0.5 μM, 1 μM, and 5 μM CAM prior to limiting dilution. To control for this process, a similar batch of cells was mock-stained using vehicle prior to setting up a limiting dilution in identical fashion. The plates were scanned using white light on the CSI on day 0 and a number of subsequent days (days 1, 2, 3, 6) until they formed prominent colonies (day 10). Representative plates of clonal outgrowths for each condition at day 10 are shown in Figure 4A. A quick comparison of day 10 colonies shows that the 5 μM condition (Figure 4A, lower right panel) has far fewer colony outgrowth compared to other staining conditions and label-free control. We tallied the number of wells that formed single-colonies for each condition. The label-free control derived colonies showed 26.3% ±1.9 (n = 1196 wells) single-colony outgrowth. Meanwhile 29.6% ± 3.2 (n=760 wells) and 30.1% ± 3.5 (n=760 wells) showed single-colony outgrowth for the 0.5 μM and 1 μM staining condition. In contrast, only 10.3% (n=380 wells) of wells grew single-colonies for the 5 μM staining condition, a 2.5-fold decrease (Figure 4B). This decrease in the percentage of outgrowth is consistent with cytotoxicity of CAM at higher concentrations, which result in a dramatic effect on proliferation from single cells. In comparison, the 0.5 μM, 1 μM, and label-free control show no discernible differences in the number of raw single-colony outgrowth.
Another observation is that the colonies derived from 5 μM staining appear to be smaller compared with colonies derived from label free or lower CAM concentrations. Representative day-6 colonies from each condition are shown (Figure 4C). A quantification of surface area of day 6 colonies shows that 5 μM CAM-stained cells are ~40% smaller compared with their label-free counterparts. Meanwhile, there was no measurable difference in colony size between 0.5 μM and 1 μM CAM-stained colonies and the label-free control. Therefore, the use of CAM dye at a high concentration results in a dramatic decrease in clonal outgrowth and smaller colonies compared with label-free control.
Figure 4. Measuring the effect of different calcein AM concentrations on viability. (A) Day 10 clonal outgrowth of representative plates from label-free, 0.5 μM, 1 μM, and 5 μM calcein AM stained cells before limiting dilution. (B) Quantification of raw outgrowth of single colony-containing wells. (C) Representative single colony-containing wells at day 6. (D) Quantification of colony size (surface area) shows that colonies derived from 5 μM CAM cell staining are ~40% smaller in size compared with label-free and lower CAM cell staining conditions. (E) Quantification of single colony wells derived from single cells shows consistent data to raw outgrowth percentages outlined in (B).
Using the combined WL and FL workflows to assess the effect of CAM on viability of colonies derived from single cell wells
So far we’ve discussed the viability of clones at the population level. We used this simple approach in order to optimize the concentrations of CAM to test. However, this data does not directly measure the viability of single cells. Hence, we wanted to confirm that the observed single-colony outgrowth percentage is truly accurate by looking at colony formation derived from single cells on a well-by-well basis. This was accomplished by calculating the percentage of single-cell wells that were able to form a colony. From this more direct comparison, 75% ±8 (n=1520 wells) of single-cell wells form colonies in the 1 μM CAM staining condition. Similarly, 73.6% ±9.2 (n=1196 wells) of single-cell wells form colonies for the label-free control. Meanwhile, a mere 21.4% of single-cell wells form colonies for the 5 μM CAM staining condition (Figure 4E). The consistency of raw outgrowth of single colonies and colonies from single-cell wells shows that there is no effect on cell viability using the CAM viability dye at the 0.5 or 1 μM staining concentrations
Guidelines on optimizing calcein AM dye concentrations
An ideal concentration of dye balances the ability to detect a majority of labeled-cells with minimal cytotoxic effects on clonal outgrowth. We tested a few concentrations, namely 0.5, 1, and 5 μM, for the viability dye calcein AM. We found 1 μM to be the ideal concentration based on the detection and viability readouts. A similar assay optimization step would need to be performed based on the dye, cell type, imaging system, and other experimental conditions.
Traditionally, the process of establishing a clonal cell line for use in the manufacturing of monoclonal antibodies has avoided the use of fluorescence reagents because of concerns over cytotoxicity and genetic alterations that may occur. The development of inexpensive and accessible next-generation sequencing technologies though has alleviated many of the concerns surrounding genetic alterations, providing an opportunity to use fluorescence dyes if the issues surrounding cytotoxicity can be overcome. Here, we demonstrate an optimized workflow using the fluorescence reagent, calcein-AM, in conjunction with a fluorescence-capable CSI that shows similar viability to label-free conditions while simultaneously providing high assurance of clonality. In this workflow, we found that a 1 μM concentration of calcein-AM is ideal for detection of single cell on the CSI while not inducing any cytotoxicity.
The benefit to using a fluorescence approach for confirming monoclonality is that it allows for the automated detection of single cells, significantly reducing the number of hours required to manually identify single cells. The analysis can be immediately performed on day 0 as well, providing information on clonality much earlier in the workflow. As with any approach, there are trade-offs to using fluorescence. Most obvious, it requires additional sample prep and finding the appropriate concentration of dye requires optimization across a number of parameters (cell type, imaging system, media, etc.). The imaging of microplates in fluorescence is inherently slower than in brightfield and must be considered when high throughput workflows are desired. Determining whether brightfield or fluorescence imaging is most appropriate will then depend upon the individual workflow. Table 1 summarizes some of the key differences between both workflows to help guide the choice between the two workflows. The CloneSelect Imager with fluorescence is able to meet the needs for either workflow though as shown in this application note, alleviating the need to make a choice between instruments when choosing a particular workflow.
|When is clonality established?||Later in the workflow (typically day 6–10)||Immediately in the workflow (day 0)|
|How is clonality established?||Software automatically detects colonies on day 10. They are manually tracked back to day 0 to visually confirm a single cell.||Software automatically detects single cells on day 0 and are then manually reviewed to confirm|
|Accuracy of automated single cell detection||Low – medium accuracy depending upon plate type, cell type, etc.||High accuracy|
|Sample prep||Minimal||Requires addition of dye and incubation time|
|Assay optimization||Concentration of dye must be optimized based on cell type, incubation time, and sensitivity of imaging system||None required|
|Time needed to acquire images of entire 96 or 384-well plate||<90 sec on CloneSelect Imager||<12 minutes on CloneSelect Imager in WL+FL mode|
|Time needed to review images for presence of single cell||Hours to days||Minimal – software performs automatically|
Table 1. Comparing the WL and FL workflows to assessing monoclonality.