Optimizing optical flow cytometry for cell volume-based sorting and analysis

Abstract

Cell size is a defining characteristic central to cell function and ultimately to tissue architecture. The ability to sort cell subpopulations of different sizes would facilitate investigation at genomic and proteomic levels of mechanisms by which cells attain and maintain their size. Currently available cell sorters, however, cannot directly measure cell volume electronically, and it would therefore be desirable to know which of the optical measurements that can be made in such instruments provide the best estimate of volume. We investigated several different light scattering and fluorescence measurements in several different cell lines, sorting cell fractions from the high and low end of distributions, and measuring volume electronically to determine which sorting strategy yielded the best separated volume distributions. Since we found that different optical measurements were optimal for different cell lines, we suggest that following this procedure will enable other investigators to optimize their own cell sorters for volume-based separation of the cell types with which they work.

Figure 1. Cell size is better approximated by SSC-A rather than FSC-A.

(A) Sort gates depicting the generalized sort scheme whereby the upper and lower 10% of the intensity distribution for the parameter of interest (FSC-A is shown in the figure) were sorted on a FACSAria. (B) L1210 (upper panels), and FL5.12 (middle panels) mouse cells, and HL60 human cells (lower panels) were sorted by FSC-A (left panels) and SSC-A (right panels) using the gating scheme shown in panel A. Volume distributions measured on a Coulter Counter are shown for cells isolated from the lower (red) and upper (black) sort gates. The overall quality of the size separation was estimated by two measures, the percent overlap and the difference in femtoliters (fL) between medians (Δ median) of the size distributions of the two sorted populations.

Figure 2. The signal width improves size separation by FSC.

L1210 (upper), FL5.12 (middle), and HL60 (lower) cells were sorted by FSC-W. Size distributions, percent overlap and Δ median, were measured and calculated as in Figure 1.

Figure 3. The cell's autofluorescence can indicate its size.

(A) Bivariate plots of L1210 autofluorescence signal area versus SSC-A (i-iv) or FSC-W generated on FACSAria (v-viii). Autofluorescence was elicited utilizing the 3 indicated excitation wavelengths, and subsequently measured at the depicted bandwidths. Linear correlations between cellular autofluorescence and light scatter are summarized in Table 2. (B) L1210 (upper), FL5.12 (middle) and HL60 (lower) cells were sorted by autofluorescence intensity, elicited by 405 nm wavelength excitation and measured at 425-475 nm (450/50-A). Size distributions, percent overlap, and Δ median were measured and calculated as in Figure 1B.

Figure 4. Cell size is better approximated by a combination of optical parameters.

(A) L1210 cells were sorted utilizing a sequential boolean gating strategy. Gates P1 and P2 (see upper left panel) included the upper and lower 20% of the FSC-W distribution. The subsequent sort gates P3, P4 and P5, P6 (see lower left panels) were based on the upper and lower 20% of the SSC-A distributions of the “low” (P1 gate) and the “high” (P2 gate) FSC-W populations. The size distributions of the four sorted populations were analyzed on the Coulter Counter. Depicted are the median size values of each population. (B) Sort gate operations performed in (A) were repeated utilizing a bivariate gating strategy resulting in 4 gates with a final 4% of the total population in each (left panel). The size distribution of the four sorted populations (right panel) was measured as in (A). (C) A sequential boolean gating strategy, similar to that used to set sort gates in (A), was utilized for FSC-W and 450/50-A autofluorescence parameters (left panel) and for SSC-A and 450/50-A autofluorescence parameters (right panel). The size distribution and the median size values are depicted. (D) L1210 cells were sorted by 594 nm-excited 660/20-A autofluorescence (660/20-A; APC-A). The upper and lower 10% of the distribution were sorted for size measurement (see Figure 1 for details). (E) Experiment described in 4C repeated for FSC-W and 660/20-A autofluorescence parameters (left panel) and for SSC-A and 660/20-A autofluorescence parameters (right panel).