Sorting of cells of the same size, shape, and cell cycle stage for a single cell level assay without staining
© Ohnuma et al; licensee BioMed Central Ltd. 2006
Received: 21 April 2006
Accepted: 22 June 2006
Published: 22 June 2006
Single-cell level studies are being used increasingly to measure cell properties not directly observable in a cell population. High-performance data acquisition systems for such studies have, by necessity, developed in synchrony. However, improvements in sample purification techniques are also required to reveal new phenomena. Here we assessed a cell sorter as a sample-pretreatment tool for a single-cell level assay. A cell sorter is routinely used for selecting one type of cells from a heterogeneous mixture of cells using specific fluorescence labels. In this case, we wanted to select cells of exactly the same size, shape, and cell-cycle stage from a population, without using a specific fluorescence label.
We used four light scatter parameters: the peak height and area of the forward scatter (FSheight and FSarea) and side scatter (SSheight and SSarea). The rat pheochromocytoma PC12 cell line, a neuronal cell line, was used for all experiments. The living cells concentrated in the high FSarea and middle SSheight/SSarea fractions. Single cells without cell clumps were concentrated in the low SS and middle FS fractions, and in the higher FSheight/FSarea and SSheight/SSarea fractions. The cell populations from these viable, single-cell-rich fractions were divided into twelve subfractions based on their FSarea-SSarea profiles, for more detailed analysis. We found that SSarea was proportional to the cell volume and the FSarea correlated with cell roundness and elongation, as well as with the level of DNA in the cell. To test the method and to characterize the basic properties of the isolated single cells, sorted cells were cultured in separate wells. The cells in all subfractions survived, proliferated and differentiated normally, suggesting that there was no serious damage. The smallest, roundest, and smoothest cells had the highest viability. There was no correlation between proliferation and differentiation. NGF increases cell viability but decreases the proliferative ability of the PC12 cells.
We demonstrated a pretreatment method to collect well-characterized, viable, single cells without using fluorescent labels and without significant damage to the cells. This method is quantitative, rapid, single-step, and yields cells of high purity, making it applicable for a variety of single-cell level analyses.
Recent technical developments have enabled many phenomena to be detected at the single-cell level. For example, improved amplification of RT-PCR now enables single-cell analysis, while the micro-fabrication technique allows better control of cell positioning, morphological analysis, and examination of intercellular contacts between cells [1–3]. Conventional single-cell level measurement techniques such as patch clamping and time-lapse microscopy have also been improved by increasing the number of simultaneous measuring points, thus enabling high-throughput measurement without specialist technical knowledge .
Despite these advances, the simple application of these techniques to a conventional sample is not sufficient. Although heterogeneity within a cell population may have a negligibly small effect on the analysis in population measurements because of population averaging, it becomes prominent in single-cell level analysis. Thus, improvements in sample purification as well as measurement techniques are required for revealing new phenomena. Because a clonal cell line consists of genetically identical cells, it is suitable for single-cell assays. However, properties within a clonal line are not truly homogeneous because of differences in cell cycle, phenotypic variance, and cell damage caused by manipulations. Selection of cells based on empirical knowledge is necessary for meaningful data acquisition at the single-cell level, despite the bias that might be introduced by such a selection process. Thus, a quantitative and rapid sample-pretreatment method of selecting well-characterized and homogeneous cells from a cell line is important for high-quality data acquisition in single-cell assays.
A cell sorter is suitable for the quantitative and high-speed manipulation of individual cells and can be used to separate a specific type of cell based on a composite of flow cytometry (FCM) parameters. A cell sorter is routinely used to measure and sort particular type of cells from dispersed tissue or other mixtures of cell types. Cells can be measured and sorted using both fluorescent labels and light scattering. Measuring the fluorescence of a specific label is useful because labels can be used to amplify cell features. However, the labelling will be different for each application and for each sample. Moreover, labeling is not uniform and may even damage the cells. On the other hand, light scattering reflects general cell properties such as size and shape [5, 6], but is free from the above disadvantages.
Living cell fractions
To determine the fractions containing living cells by light scattering, PC12 cells were stained with the fluorescent markers calcein AM and propidium iodide (PI). Calcein AM is a fluorogenic esterase substrate that is hydrolyzed to a green fluorescent product in living cells; thus, green fluorescence was an indicator of cells with esterase activity as well as an intact membrane to retain the esterase products. PI is a red fluorescent nucleic acid stain that is only able to pass through the compromised membranes, thus it marks dead cells. In Fig. 1, each blue dot identifies the calcein-positive and PI-negative cells as living cells, whereas each red dot represents calcein-negative or PI-positive cells, which were assumed to be dead cells.
The fact that fractions containing red and blue dots were clearly separated in the scatter profiles in Fig. 1 suggests that using four light scatter parameter enables better gating out of the dead cell. As expected, the red dots were concentrated in the low FSarea fraction and the low to medium SSarea fractions, which may contain cell fragments. We found that red dots were also concentrated in the upper fraction of FSarea-FSheight profile and especially in the upper fraction of the SSarea-SSheight profile. Because dead cells may contain structures with increased light scattering properties, such as vacuoles or aggregated proteins, the height-to-area signal ratio in this fraction was high (see Fig. 2).
It is difficult to completely disperse cells mechanically to obtain truly single cells; usually the dispersed suspension contains many cell aggregates. Extensive enzyme digestion and trituration may aid in breaking up cell clumps, but will tend to cause cell damage. In addition, small parts of the cells may re-aggregate. When the PC12 cells were dispersed using trituration by pipette in calcium- and magnesium-free phosphate-buffered saline (PBS-), more than half of the cells remained as aggregates (see also ). Filtration and density-gradient centrifugation are easy way to remove big aggregates, and cell aggregates can be discriminated using a cell sorter to gate out particles that have a large peak and/or prolonged (broad) signal. However, these methods cannot be used to distinguish between cell doublets and large single cells. Here, we hypothesized that a cell doublet may exhibit a lower height-to-area signal ratio than a single cell of the same size as shown in Fig. 2. Although the longer axis of the doublet will not always be positioned parallel to the flow, the accuracy of doublet detection can be increased by using both FS and SS, which are at right angles to each other.
We tested this working hypothesis by sorting cells based on the FSarea-FSheight profile or the SSarea-SSheight profile. The nuclei of the sorted cells were then stained with SYTO24, a cell-permeable green fluorescent nucleic acid stain, and the nuclei in each particle were counted. Fig. 3A shows the FSarea-FSheight results. The single cell ratio decreased as FSarea increased. However, even among low FSarea fractions, the lower FSheight-to-FSarea ratio fractions contained larger numbers of doublets. The same findings were also seen in the SSarea-SSheight profiles (Fig. 3B). Although late M phase cells were counted as doublets using this method, this limitation is unlikely to have a large impact on the results. We concluded that our working hypothesis was correct.
Twelve single live-cell subfractions in the FSarea-SSarea profile
Size and shape of cells in the twelve subfractions
As shown in Fig. 5B, the cell diameter increased with the increasing SSn and FSn values. We fit the d a as a function of SSn and FSn. The fitting function is as follows,
where a and b were the fitting parameters. The a value was 3.2, suggesting that SSn has a linear relationship with cell volume. Actually, the cell volume at (SSn, FSn) = (2.00, 1.24) was about twice as large as that at (SSn, FSn) = (1.00, 1.24) as shown in Fig. 5B. The b value was 6.8, suggesting that FSn has a weaker positive correlation with cell diameter. Elongation (e) also increased with the increasing SSn and FSn (Fig. 5C), but depended more on FSn than on SSn. The e at the largest SSn is larger than at other values. At middle and low FSn, although the d a were quite different, all the e were the same (P > 0.1) suggest that the larger cells are not necessarily the elongated cells. There may be two different mechanisms between volume and elongation.
The roughness of the cell surface, P/A, exhibited different dependencies from the cell diameter and elongation, increasing with SSn but decreasing with FSn. The cell, which had a high P/A value, had bubbles around it (Fig. 5Ad) and appeared non-viable. The Pearson's correlation coefficient for the P/A value compared to the ratio of calcein-positive and PI-negative cells (living cells) in the twelve subfractions was significantly smaller than zero (p < 0.05) also supports the view.
Cell cycle in the twelve subfractions
Single cell assay in the twelve subfractions
We performed a fundamental single cell assay to verify our sorting method developed for doing single-cell assays, and to determine the fundamental properties of single cells. The single sorted cells of all twelve subfractions were cultured in separate wells of a 96-well plate and tested for viability, proliferation, and differentiation potential.
One might expect a negative correlation between proliferation and differentiation potential in these cells; however this was not the case. The differentiation ratio and proliferation rate were similar in all subfractions. It is notable that viability at subfraction (1.00, 1.24) was significantly higher than for the other subfractions (Fig. 7C). Because this subfraction contained the smallest, roundest, and smoothest cells (Fig. 5), this was an intuitive result. It is also notable that the cells with NGF showed high viability but a low proliferation rate. The VbNGF values were significantly larger than the VbSER values (P < 0.05), and the average VbNGF to VbSER ratio was 1.3 (Fig. 7C). Although the addition of NGF to serum-free medium, in which the cells can not survive, dramatically improves cell viability , the effect of adding NGF to serum-supplemented medium, in which the cells can survive and proliferate, has not been measured. Generally, cell viability assays are performed using mass cultures, and viability factors cannot be distinguished from proliferation factors. On the contrary, our single cell assays can evaluate viability and proliferation independently. We demonstrated directly that NGF increases cell viability but decreases the proliferative ability of the PC12 cells when cultured in the presence of serum.
This study demonstrated a non-staining sample-pretreatment method to select well-characterized homogeneous cells from a cell line for single-cell level analysis. Although a cell line is genetically identical, it is morphologically heterogeneous due to aggregation, cell cycle, and phenotype variance. We used both peak height and area of light scattering parameters to sort PC12 cells into live single-cell fractions, without the use of a specific label. Although it is usually difficult to discriminate between single large cells and small cell aggregates, we successfully separated these classes using the height-to-area signal ratio of both FS and SS. The distinction methods were versatile enough to apply to both living and ethanol-fixed cells. Moreover, this discrimination method is applicable not only to pretreatment steps but also in the data acquisition stage.
We showed that the FSn is closely related to both the degree of cell elongation and the amount of DNA per cell, and that the SSn was proportional to the cell volume. This is in contrast to the commonly accepted association between light scattering and cell morphology that FS represents the size and SS represents internal structure (cf. owner manual of ALTRA). Light scattering profiles are closely related to morphological features including size, shape, surface detail, and intracellular structure of the particle [6, 9, 10, 12]. The dependency, however, varies according to the structure of the particles and the acquisition method applied [9, 13]. The commonly accepted association was based mainly on the analysis of different-size beads or different kinds of hemocyte using narrow angle analysis [9, 10]. However, current machines use wide angles; for example, the FS and SS of ALTRA are the summation of light scattering data from -23° to 23° and from 42° to 138° from the laser axis, respectively. Moreover, we tested an adherent cell line, in which the component cells are almost of the same structure and size. Thus, it is not surprising that our results differed from the accepted association. One possible explanation of our results is that the refractive index of the cell nuclei differs slightly from that of the cytosol and that the cell contains many small particles that reflect light and are distributed homogeneously (Fig. 2D). Diffraction and refraction are the main components of forward scatter and thus are affected by the size of the nucleus, which is roughly in proportion to the DNA content of the cell. On the other hand, reflection from small particles is the main component of side scatter. Because SSn or SSarea is the time integration of side scatter, it is proportional to the number of particles in the cell and as a consequence, is proportional to cell volume.
Light scattering can be used to gain quantitative criteria of cell appearance, which is a good indicator of cell properties. In experiments where data are acquired using a microscope, such as patch clamping, cells must be selected based on appearance to acquire meaningful data. Based on microscopic observation, cell aggregates, dead cells, and less vital cells can be excluded and cells from which data can be acquired, are selected. This empirical selection, however, may introduce an unknown bias to the experiments. Our single-cell assay showed that the roundest and smoothest cells have high viability. This intuitive result suggests that cell selection using light scattering data is consistent with empirical selection based on cell appearance, but without introducing unknown bias.
In the single-cell assay we used the auto-clone option. Because the conventional limiting dilution method is stochastic, there is no guarantee that individual cells are in separate wells. However, auto-clone enabled us to plate a precise number of cells directly into a separate well. Thus, this option would also be useful for single-cell assays of cell extracts, such as single-cell RT-PCR. This option also enabled us to assess cell viability (survival ratio) and cell proliferation independently, which is difficult to achieve in conventional mass culture. Finally, the single-living-cell assay results confirmed that the cell sorting method used in this study caused no significant damage to the cells, as the sorted single cells in the individual wells survived, proliferated and differentiated normally.
We first demonstrated here that the live single-cell can be distinguished from cell doublets or aggregates from the height-to-area signal ratio in the light scatter parameter. We also demonstrated that the cell volume is estimated by SSarea, while the elongation and the level of DNA in the cells is estimated by using FSarea, which suggests that elongation of cell occurs at the late stage of cell cycle. Then we performed a single cell assay using auto-clone option. The single isolated cells survived, proliferated and differentiated normally, suggesting that no serious damage was caused by the method. We also show the relationship among basic properties of cells such as morphology, cell cycle, viability, proliferation and differentiation. Viability of a cell is estimated from the scattering data separately from the proliferation.
These results suggest that using a cell sorter as a pretreatment method for single-cell assay has several significant advantages. First, the light scatter profiles provide quantitative indicators of microscopic cell appearance, thereby reflecting empirical selection without introducing unknown bias. Moreover, fluorescent dyes such as GFP can be included into the technique if required. Second, because the cells can be sorted based on DNA content and fine differences in cell size, cell volume dependent characters, such as whole-cell expression levels and currents, can be measured with low variance. Third, the auto-clone option enables an exact number of selected cells to be plated into separate wells. Finally, all of these treatments can be done at once and is rapid and less invasive to the cells. Thus, cell sorting as used in this study provides a valuable pre-treatment step for single-cell level assays. This method can be applied not only to various types of single-cell assay but also to single-cell level analysis to investigate complex systems, such as the dynamics of multicellular interactions in which a small difference in the initial state can cause a large difference in the final state.
PC12 cells were obtained from the Riken Cell Bank (Cat #: RCB0009; Ibaragi, Japan). PC12 cells were cultured according to the method described by Green et al. . Briefly, PC12 cells were grown on collagen I-coated plastic dishes (Biocoat; Becton Dickinson, NJ, USA) at 37°C in a humidified incubator in an atmosphere of 5% CO2. The normal culture medium consisted of 85% Dulbecco's modified Eagles' medium (DMEM; Invitrogen Corporation, Carlsbad, CA, USA), 10% heat-inactivated horse serum (Invitrogen), 5% heat-inactivated newborn calf serum (Invitrogen), 50 units/ml penicillin and 50 μg/ml streptomycin (Invitrogen), and 4 mM L-glutamine (Invitrogen).
Flow cytometry and cell sorting
The culture medium was first replaced with ice-cold calcium- and magnesium-free phosphate-buffered saline (PBS-) containing 1 mg/ml bovine serum albumin (BSA), then the cells were detached from the dish by forceful aspiration of the solution through a 1-ml plastic pipette. The cell suspension was then filtered through a 40-μm nylon mesh (Becton Dickinson) to remove large cell aggregates. Preliminary experiments showed that the FCM parameters depended on temperature, therefore all solutions and samples were kept on ice during subsequent procedures. In experiments that included cell staining, fluorescent dyes were added to the suspension at this point. Preliminary experiments showed that the FCM profile drifted about 20 minutes after harvesting and became relatively stable thereafter. Thus, the suspension was kept ice cold for 30 minutes after harvesting, and then FCM acquisition and cell sorting were performed within 40 minutes using an EPICS ALTRA (Beckman Coulter, Inc., Miami, FL, USA). All measurements were made using an argon laser at 488 nm. The beam size at the focal plane was 6 μm in height and 100 μm in width. The cells were sorted according to four parameters, which comprised the peak height and area of forward scatter (FSheight and FSarea) and side scatter (SSheight and SSarea). The peak shape (peak width or time of flight), which is the transit time of a particle in the laser beam plane, is also a basic parameter. However we did not use the peak shapes because it was sensitive to the operation and was relatively unstable. PMT 2 was used to measure the SS because it had both area- and peak-value output.
The cell sorting settings were as follows: flow chip diameter, 100 μm; sheath solution, IsoFlow (Beckman Coulter); sheath pressure, 8.8 – 9.2 PSI; crystal frequency, 19.0 – 21.5 kHz; crystal drive, 15–30%; delay, 25 – 32 drops; sort mode, 3; coincident abort, on; and pulse pile up (PPU) sensitivity, 4–10 μs.
Cell size and shape
To evaluate the size and shape of the cells, phase-contrast images were captured using a digital camera (Coolpix 995; Nikon, Tokyo, Japan) placed on an inverted microscope (TS100; Nikon). After importing the images to a computer, an outline of each cell image was traced by hand. The perimeter (P), area (A), and the major and minor axes of the best-fitted ellipse (d maj and d min ) for the outlined cells were then measured using image analysis software (Scion Image; Scion Corporation, Frederick, ML, USA).
To estimate the cell diameter, d maj , d min and the equivalent circular diameter () and equivalent area diameter () were used. However, because these four values had a strong correlation (Pearson's correlation coefficients were greater than 0.96), only d a is shown as data. To quantify the cell shape, ellipticity () and the perimeter to surface area ratio () of the cells were calculated. In the case of a circle, e = P/A = 1, whereas e and P/A inequality give e ≥ 1 and P/A ≥ 1. In general, e is an indicator of the degree of elongation and P/A is an indicator of cell surface roughness.
To determine the fractions containing living cells, 100 nM of Calcein AM (c3100; Molecular Probes, Inc., Eugene, OR, USA) and 100 ng/ml of propidium iodide (PI, A4378; Sigma-Aldrich, St. Louis, MO, USA) were used. Calcein and PI signals were measured using a 525-nm bandpass filter and a 675-nm bandpass filter. Because the Calcein and PI fluorescence overlap was negligible, the fluorescence signals were not compensated.
To counting nuclei, the live cells were stained by 5 μM of Syto 24 (S7559: Molecular Probes, Inc).
To assess the cell cycle stage, total DNA in each lot of PC12 cells was measured using FCM. The cells were suspended in 300 μl of ice-cold PBS- and 700 μl of -20°C ethanol was added to the suspension while vortexing gently. The fixed cells were stored at -80°C. The cells were washed in PBS- and re-suspended in PI staining solution containing 20 μg/ml of PI, 0.1% (v/v) Triton X-100, and 200 μg/ml of DNase-free RNase in PBS-. The cells were incubated for 30 minutes at room temperature and were kept ice cold before FCM acquisition. PI signals were measured using a 675-nm bandpass filter.
Viability, differentiation and proliferation potential of single cells
PC12 cells were divided into fractions according to the twelve-subfraction separation (Fig. 4C). Single cells were plated into the individual wells of a 96-well plate using the auto-clone option of the ALTRA cell sorter. The success rate for plating one cell per well was more than 90%. Each well was precoated with collagen I (Iwaki, Tokyo, Japan) and contained 100 μl of normal culture medium.
Two day after sorting, half of the wells were given an additional 100 μl of culture medium containing 100 ng/ml of NGF without serum, and the other half were given culture medium with serum. Thus, half of the wells contained 50 ng/ml NGF, 5% HS and 2.5% NCS ("NGF" condition), and the other half contained normal culture medium ("SER" condition). Half of the culture medium was changed every 2 to 4 days. The cells were cultured from eleven to thirty days after sorting. Cell viability (Vb) or survival ratio was the ratio of the number of wells containing any cells to the number of single-cell plated wells. The differentiation potential for the surviving cells was the average ratio of neurite-bearing cells per total cells in a well. The proliferation potential for the surviving cells was the logarithm of the average cells in a well to the base 2.
This work was supported by a research fellowship (13–83203) from the Japan Society for the Promotion of Science for Young Scientists, a 21st century COE grant from the Japan Society for the Promotion of Science. The authors would like to thank Mr. Yasuhiko Nagasaka from Beckman Coulter Inc. for his technical assistance and helpful advice regarding flow cytometry and cell sorting.
- Sucher NJ, Deitcher DL, Baro DJ, Warrick RM, Guenther E: Genes and channels: patch/voltage-clamp analysis and single-cell RT-PCR. Cell Tissue Res. 2000, 302: 295-307. 10.1007/s004410000289.View ArticlePubMedGoogle Scholar
- Tooker A, Meng E, Erickson J, Tai YC, Pine J: Biocompatible parylene neurocages. Developing a robust method for live neural network studies. IEEE Eng Med Biol Mag. 2005, 24: 30-33. 10.1109/MEMB.2005.1549727.View ArticlePubMedGoogle Scholar
- Cai L, Friedman N, Xie XS: Stochastic protein expression in individual cells at the single molecule level. Nature. 2006, 440: 358-362. 10.1038/nature04599.View ArticlePubMedGoogle Scholar
- Guo L, Guthrie H: Automated electrophysiology in the preclinical evaluation of drugs for potential QT prolongation. J Pharmacol Toxicol Methods. 2005, 52: 123-135. 10.1016/j.vascn.2005.04.002.View ArticlePubMedGoogle Scholar
- Salzman GC, Crowell JM, Martin JC, Trujillo TT, Romero A, Mullaney PF, LaBauve PM: Cell classification by laser light scattering: identification and separation of unstained leukocytes. Acta Cytol. 1975, 19: 374-377.PubMedGoogle Scholar
- van den Engh GJ, Visser JW, Trask B: Identification of CFU-s by scatter measurements on a light activated cell sorter. Experimental hematology today. Edited by: J. BS and P. LG. 1979, New York, Springer Verlag, 19-26.Google Scholar
- Greene LA, Tischler AS: Establishment of a noradrenergic clonal line of rat adrenal pheochromocytoma cells which respond to nerve growth factor. Proc Natl Acad Sci U S A. 1976, 73: 2424-2428. 10.1073/pnas.73.7.2424.PubMed CentralView ArticlePubMedGoogle Scholar
- Greene LA, Farinelli SE, Cunningham ME, Park DS: Culture and experimental use of the PC12 rat pheochromocytoma cell line. In Culturing Nerve Cells 2 nd edition. Edited by: Banker G and Goslin K. London, England, The MIT Press; 1998:161-188.Google Scholar
- Brunsting A, Mullaney PF: Differential light scattering from spherical mammalian cells. Biophys J. 1974, 14: 439-453.PubMed CentralView ArticlePubMedGoogle Scholar
- Benson MC, McDougal DC, Coffey DS: The application of perpendicular and forward light scatter to assess nuclear and cellular morphology. Cytometry. 1984, 5: 515-522. 10.1002/cyto.990050513.View ArticlePubMedGoogle Scholar
- Greene LA: Nerve growth factor prevents the death and stimulates the neuronal differentiation of clonal PC12 pheochromocytoma cells in serum-free medium. J Cell Biol. 1978, 78: 747-755. 10.1083/jcb.78.3.747.PubMed CentralView ArticlePubMedGoogle Scholar
- Mullaney PF, Dean PN: The small angle light scattering of biological cells. Theoretical considerations. Biophys J. 1970, 10: 764-772.PubMed CentralView ArticlePubMedGoogle Scholar
- Meyer RA, Brunsting A: Light scattering from nucleated biological cells. Biophys J. 1975, 15: 191-203.PubMed CentralView ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.