Why does cell sorting impair cell viability? We have run some experiments to compare our VACS technology with the conventional technology, which we call CICS (pronounced “kicks”: read more to find out why…). Now we need your feedback: Are we thinking along the right lines? What other experiments should we do?

Verity Jackson, Samson Rogers

How to sort sensitive cells is an important experimental issue for many cell biologists, but the mechanisms causing cell death, damage or loss of function during cell sorting have so far not been systematically investigated. Cytometrists disagree when they are asked to point to the most likely culprit. Till now, one type of technology has dominated the cell sorting field, meaning that biologists haven’t had much choice over which cell sorting technology to use. But with the rise of microfluidic cell sorters, the availability of new methods to study cell stress responses, such as single cell transcriptomics, and new clinical applications of cell sorting on the horizon, we think now is the time to study the issue more deeply.

With viability in mind, we’ve conducted a few experiments at Cellular Highways to compare our new technology, vortex-actuated cell sorting (VACS) with the conventional technology. We present our results not as a definitive study but as part of an ongoing project. We are now asking for feedback from cell biologists and cytometrists: Do our results and conclusions fit with your experience? Are there other experiments that we should do? Do you have any ideas or tips for us?

Nomenclature for the conventional technology “CICS”

Firstly, some nomenclature: What should one call the conventional technology? It is commonly known as FACS, but this acronym is a trademark of Beckton, Dickinson and Company (BD). Since the acronym stands for Fluorescence-Activated Cell Sorting, it refers to the cytometry bit (fluorescence activation) that also describes any other type of cell sorter that uses fluorescence. Yet the acronym doesn’t describe the technology for deflecting cells at all, i.e. the electrostatic deflection of droplets created by stimulation of the Plateau-Rayleigh instability of a pressurised stream flying out of a nozzle and into the air. However, the latter is quite a mouthful, and cytometrists already use the expression “stream-in-air” to refer to systems where the laser impinges the stream while it travels through the air, as opposed to “stream-in-glass” systems where the laser impinges the stream in a cuvette, despite the fact that the stream later travels through air to be sorted. Actually, the historical origin is that Mack Fulwyler adapted this technology from Dick Sweet’s continuous inkjet, which of course sorted ink droplets rather than droplets containing cells. For these reasons, we refer to this generic conventional technology as “continuous inkjet cell sorting” (CICS – pronounced “kicks”), since this reflects the technology’s origin. We will only use the acronym FACS to refer to results specifically obtained on commercial FACS instruments.

Two cell lines that are known to be sensitive to conventional cell sorting are fine in VACS

No cell sorter wins prizes for being able to sort lymphocytes with high viability. Lymphocytes are among the hardiest of cells: as cytometrists sometimes joke, lymphocytes are almost as hardy as beads.

Many types of cells are known to be sensitive to CICS, however. Even if we ignore cells that are large or have obviously delicate structures on their surfaces, many common mammalian cell lines of less than 20 microns diameter show damage or stress after sorting. Often, the damage is subtle and not obvious when using a viability stain immediately post-sort. However, other effects become apparent later, for example a lack of growth in subculture or phenotypic changes.

We focused on two common cell lines, CHO-K1 and THP1, which are known to be sensitive to CICS, according to a previous paper on this subject [1].

Cells were sorted on the same day on our VACS experimental rig and on two CICS instruments, both BD FACSArias: a BD FACSAria III with a 70 µm nozzle, sample rate 1.0, sheath pressure 70 psi, and a BD FACSAria Fusion with a 100 µm nozzle, sample rate 2.0, sheath pressure 20 psi. In the VACS process, cells were suspended in PBS (phosphate-buffered saline) throughout, while in the CICS process, the cells were input in PBS, diluted by BD FACSFlow media in the sheath, and landed in collection tubes partially filled with PBS. In both processes, we deflected cells at random into the sort collection output. For each cell type, control cell batches were kept in PBS.

When put back into culture, CICS-sorted cells lagged for a couple of days before resuming normal growth. This could be because the cells were stressed or shocked in some way, delaying their replication for a day or two, or because some proportion of the cells were no longer replication-competent after the CICS sort. However, VACS-sorted cells resumed growth without delay, replicating at the same rate as control cells (results in Figure 1).

For the THP1 experiments, we also used a viability stain (Propidium Iodide) immediately post-sort (Figure 2). At first glance, it is difficult to see an obvious difference in viability between the sorting methods, since the viable fractions are all close to 100%. But on quantitative examination, there is a difference: approximately 1% of the cells had their membranes damaged by CICS, but less than about 0.1% by VACS.

Another effect that we observed was that light scattering measurements of the cells are affected by sorting (Figure 3). The CICS-sorted cells show a decrease in light scatter, which is not shown by the VACS-sorted cells. We don’t yet know what has caused this.

So, we have observed three effects of sorting on two common cell lines, of which the lag in growth was the most profound. In each case, cells were impacted by CICS, but VACS-sorted cells behaved similar to control cells.

Interestingly, the lag in growth was quite similar for both 70 µm and 100 µm nozzle runs on CICS, even though the pressure was far higher for the 70 µm (about 5 bar) than the 100 µm (about 1 bar). For comparison, the experimental VACS chips are fed with a pressure of about 3 bar. So, we observed no correlation between pressure and growth lag for either cell type, neither between CICS and VACS nor between different nozzle sizes on CICS. Moreover, on CICS, the 100 µm (with a higher sample flow rate) was not gentler than the 70 µm nozzle (with a lower sample flow rate), which runs contrary to the assumptions of many cytometrists.

How do we make sense of these observations?

Is peak mechanical stress the most important physical determinant of cell damage?

Living cells are diverse: lots of physical phenomena could potentially affect them, including pressure, shear or extensional flow. So, let’s narrow the scope of the discussion to common cell types that are used in vitro, such as suspension cell lines, rounded trypsinized cell lines or primary leukocytes, and to effects that can be measured within a couple of days, i.e. viability, growth, phenotype.

Speaking to cytometrists, there is no consensus about what causes damage to these cell types in a flow sorter. Some point to peak pressure or rate of decompression, others to shear rate or chronic exposure to flow or pressure.

In the results above, we have seen no obvious correlation between cell damage and pressure. Instead we look at a study by Mollet et al. [1], who investigated a measure that is more commonly correlated with cell damage in the bioreactor literature – the energy dissipation rate (EDR). EDR can be interpreted as a measure of peak mechanical stress that a cell experiences, combining both extensional and shear flows. Importantly, EDR has been used successfully to predict cell damage in fluid flow in different types of instruments [2]. EDR is also a convenient physical quantity, since it is a single scalar parameter that can be obtained from the fundamentals of fluid dynamics, for both laminar and turbulent flows, and accounts for both shear and extensional components [2].

Mollet at al. studied cell damage in CICS in three ways: by experiment, by simulation of EDR, and by experimental use of a flow through a model nozzle with varying the flow rate [1]. They found that cell damage increases rapidly above some threshold EDR, depending on cell type, producing something like a hockey-stick graph (Figure 4).

For CICS nozzles, Mollet et al. [1] simulated the EDR that cells experience as they flow through a nozzle. They found that the cells experience a brief moment of intense shear and extensional flow at the point they enter the nozzle (Figure 5). Interestingly, the peak EDR does not only depend on nozzle size and sheath pressure, but also the differential pressure (i.e. sample flow rate). This determines the core diameter in the sheath flow: a higher differential pressure results in higher shear as cells pass through the nozzle further from the axis.

We estimated the EDR that cells experience in VACS as follows. For laminar Poiseuille flow, EDR = mg², where m is the viscosity (1 mPas for water or PBS), and g is the shear rate. Considering the flow velocity, the flow profile and the position of the cells within the flow, we estimate that the cells see a peak EDR = 1.3×10⁵ W/m³ in our experimental VACS chips. This is not likely to make VACS the gentlest of microfluidic cell sorters, but only confirms that it seems to be much gentler than CICS.

We then compared the EDR values for the cell sorting techniques (CICS and VACS) with the threshold levels for cell damage (CHO-K1 and THP1), according to the data of Mollet et al., as shown in Figure 6. The EDR for CICS in the typical experimental set-up we have used is similar to or higher than the damage threshold for these cell types. The EDR for VACS is significantly lower.

Although not conclusive, our results are consistent with the idea that the main physical phenomenon that damages cells during sorting is the peak mechanical stress, as quantified by EDR [1, 2]. Our results certainly would not support any simple relationship between pressure and cell damage.

We want your feedback: Did we miss any definitive studies in the literature? Did we do the right experiment, and did we interpret it in the right way? Please get in touch with us: hello@cellularhighways.com

[1] Mollet et al. 2008, Biotechnology and Bioengineering, Vol. 100, No. 2

[2] Chalmers 2015, Current Opinion in Chemical Engineering, 10:94–102