Tips: Acquisition & Sorting

Most of the instruments at our facility operate with Photomultiplier Tubes (PMTs). These amplify the light signal and convert it to an electronic signal. To increase the signal, the voltage per detector is increased. In a certain range, this is a linear process, but this range is different for each PMT. The 'best' voltage is the one where the best signal-to-background level is reached. This is (to be precise) determined by performing a so-called voltage walk and calculating the staining index for a fluorophore with different voltage settings. In reality, this step is often skipped, which is fine as long as the separation of the positive from the negative population is sufficient for the purpose. 
The PMT voltage does not only affect the specific signal. By increasing the voltage for a PMT, also the signals from dyes spilling over into this chanel, and thus the compensation value, are increased. This is (partly) why the samples and the compensation controls must be acquired with the exact same voltage settings. Below you can download a Technical Bulletin by BD, explaining some basics about compensation and setting up compensation controls in BD FACSDiva software. Come talk to the staff if you need further advice on how to compensate your panel.

The initial voltage settings for each PMT are detemined during the performance test (CS&T) of the instrument. Here, a standardized sample is used to find the detector voltage that leads to a certain separation of the populations in this sample (bead based). These settings of the latest CS&T are taken over when creating a new experiment. In general, it is not recommended to decrease these default values too much for the experiment, as this usually increases the measurement errors (coefficient of variation) - see also: Maecker et al. https://onlinelibrary.wiley.com/doi/full/10.1002/cyto.a.20333 

A threshold defines which properties an event must have in order to be seen as 'event'. Everything that is not fulfulling a certain threshold is therefore not visualized on the plots and not recorded. Setting meaningful thresholds is important:

Analysis of small particles will require decreasing the threshold to analyse events that are close to the 'noise'. If the threshold is set too low, the system is burdened with an excessive event rate from background, meaning small particles in the fluidics and electronic noise stemming from the detectors, increasing the electronic abort rate.

Increasing the threshold is particularly recommended for material with high background debris, such as primary samples (digested tissue, lysed blood). This does not only decrease the file size and makes post-acquisition analysis easer to handle for the software, but also frees up computing capacity of the instrument during the analysis/sort.
In FACS sorting, thresholding has another effect: every particle that is thresholded off, is not recognized as an event. Thus, this particle is also not seen as disturbing in the droplet. As an example: a little debris particle is sitting close to your target cell in the stream. If the debris particle is considered an 'event', this target cell will not be sorted depending on the Sorting Mode. If this debris is thresholded off and not 'seen' by the system, the target cell will be sorted. Thus, thresholding also impacts the effective purity and sorting yield. This is particularly important for downstream PCR, RNAseq, proteomics etc., as this debris sorted together with the target cell can include any cell fragments. 

By changing the flow rate, what is actually changed is the volumetric rate, so the volume measured in a given time. By increasing the flow rate, the sample core is increased, meaning that the sample stream is getting broader. This leads to a loss of sensitivity - a price one pays for faster acquisition. Increasing the flow rate with a given cell concentration also means increasing the event rate. This will eventually lead to increased electronic aborts or sort aborts. 

Increasing the cell concentration has the downside that more cells are aborted by the electronics, or have to be gated out as doublets ('physical doublets' and 'electronical doublets') and are lost for analysis/sorting. 

During sorting, the sort efficiency should be monitored (in the sort layout). While it is highly dependent on your cell type and sample material, the event rate (and also gating strategy) should be adjusted to keep the sort efficiency as high as possible (if you want to sort as many target-cells as possible...).

Unfortunately, there is no 'one size fits all' - come talk to the staff if you have questions. 

Cell aggregates will take longer to pass through the laser intercept than single cells. This affects the width and the area of the signal pulse, while the height stays roughly the same. Using a pulse geometry gate (such as FSC-A vs. FSC-H or FSC-W vs. FSC-H), doublets and larger cell aggregates can be eliminated.
Nuclear dyes (such as DAPI, Sytox dyes etc.) can be used for such a pulse geometry gate as well (for example as DAPI-H vs. DAPI-W), particularly convenient for populations with heterogenous size/complexity, as well as small particles (as SSC and FSC signals are less sensitive in that range). 

In sorting, it is particularly important that you include a doublet discrimination gate (two are preferable), as these mistakes cannot be indone and sorted doublets might decrease the purity of your sorted populations. 

If the area and height intensities of an event do not match (and an -A vs -H plot looks skewed), this means the Area Scaling Factor is not set correctly for the type of cell. This will have a negative impact on sorting quality. Please contact us if you need help setting this up. 

The crude recommendation is to use a nozzle that has a diameter at least 5-times the size of the target cell.
The nozzles are operated in different frequencies: while there are 87.000 drops per second formed with a 70μm nozzle, it is 30.000 drops per second with a 100μm nozzle in our instruments. This also affects the eventrate that can be used for sorting. In addition, it is important to remember that the different nozzle sizes are operated with different pressures - the smaller the nozzle, the higher the pressure that is applied. This can affect the post-sort cell viability or behavior of certain cell types. 

Choosing the correct sorting mode is crucial.

• 'Yield' mode will be rather liberal, the algorithm will sort your target event even though an unwanted event is in the same drop or close by. This means that almost all target cells are sorted, but that the purity will be decreased. 
• 'Purity' mode will abort drops that coincide with unwanted events, and also when an unwanted event only is close enough (meaning in the closer end of the adjoining drops). This will ensure the best purity of the sorted population, but will decrease overall yield. 
• '4-Way-Purity' is a sub-mode of 'purity'. This mode is necessary when sorting >2 populations at once, and ensures that residual charges from adjoining drops do not degrade the quality of the side streams. This mode results in decreased yield, but is necessary for three- or four-way sorting where precise deflection is required to not contaminate the inner tubes. To achieve best purity of your sorted samples, you should sort rare populations at the far-left and far-right position, and more abundant populations on the inner positions. 
• 'Single-Cell' mode is most conservative and will abort a drop even if it contains two target events. This is necessary for sorting of single clones, or single-cell-transcriptomics etc.  

The different algorithms are explained in more detail in the FACSAriaIII User Guide which can be downloaded from below. 

Cells will differ in their sensitivity and their ability to survive the sorting process (buffer, pressure, shearing force...).
Wise choice of Nozzle Size and Sample Buffer are crucial. In addition, the sorted cells must be sorted into a collection tube (preferably polypropylene) containing enough protein-containg buffer (preferably PBS/HBSS-based) for the cells to not hit the dry plastics.  

When sorting cells for subsequent culture we recommend that the media you place in the wells to capture the cells be buffered with a non-CO₂ based buffer, e.g. HEPES. If not the media will shift in pH, and this may be detrimental to your cells.

Of course, to get only viable sorted cells, it is necessary to include a dead-live discrimination dye to exclude non-viable input cells during the gating strategy. 

A best-practice cell sort includes a purity assessment of the sorted samples. 
For this purpose, sorted cells are re-introduced to the instrument and the percentage of target population is analysed. This allows for control of the quality of the sort (setup, gating strategy etc.) as well as for final assessment of the purity of the sample for downstream analysis. 
It is important to perform cleaning steps before re-introducing your sorted sample (sample line backflush and H₂0 run as a sample) in order to avoid carry-over.

If you are sorting your cells right into lysis buffer or other buffers that do not allow for re-analysis, the recommendation is to initially sort some cells into PBS/sorting buffer and use these for purity check.

Please note: Naturally, if a gate cuts through a homogenous population, the same population will distribute again when they are analysed the second time (probability distribution), and the population will spread on both sides of the gate border. This does not mean your sample is impure! Please feel free to contact me if you have any questions.  

When sorting >2 populations simultaneously, you achieve best purity of your sorted samples if sorting rare populations at the far-left and far-right position, and more abundant populations on the inner positions.

The Window Gate is the time interval during which a signal is greater than a pre-set threshold. To ensure that a voltage pulse is fully integrated for a given particle (and not just above the threshold), additional time is added to the initial window gate. This time is called the Window Extension.

Incorrectly set Window Extension affects the data and can lead to incorrect measurements. A Window Extension that is set too low does not fully integrate the signal of an event, and in addition increases the chance that the event is partly missed in the laser lines following the triggering laser (usually the 488nm laser) due to variance in the particle arrival time (laser delay). If the Window Extension is set too high, the windows assigned to the events will start overlapping, which will lead to an electronic abort of the events. This cannot be undone after acquisition. 

The aquisition software uses a default Window Extension. There are few applications where the Window extension can be or should be adjusted. We are happy to discuss this with you for your application.