[Cytometry] (Summary + Thank you) Questions on how to improve big cell sorting--Aria

Mok Michelle michelle_mok at bsf.a-star.edu.sg
Tue May 26 00:15:01 EDT 2009

I would like to express my gratitude to those who replied to my thread. 
Thank you all for the corrections, suggestions, and sharing your experiences with me.
It has been very useful.

Thank you!


Original Question:

Dear Flow-ers,

I'm a technical operator of 2 FACSAria in a busy service lab. We tend to receive a lot of cell line samples for sorting. While our practice so far has been to perform sorting conservatively to ensure cell survival, we are now quizzed on how to improve productivity because everyone (the clients) just takes a really long time to get enough cells out of our sorter.

Our cell lines are mostly large cells CHO, HEK, HEPG2, HUH7, Adipocyte progenitors, Dendritic cells, HeLa cells, Differientiated and non-differentitated ES Cells, EBs etc (I have no cell culture experience so I'm not sure of the exact dimensions of these types of cells). We ask our clients to give us a prep of 4 million cells/mL. We have so far always sorted on a Low (20 PSI) with a 100 nozzle. We start at a Flow Rate of 3.0, then we tweak the flow rate depending on how the cells survive at the threshold obtained at Flow Rate 3. We also try to achieve an abort rate of <20% on average.

So far, immortalize cancer cell lines are able to survive up to Flow Rate of 6. While Dendritic cells, HeLa cells and lots of cells from fresh embryos tend to be more fastidious and will die at a Flow Rate any higher than 2. Although while using these guidelines, we are able to ensure survivability of most cell lines we receive, we have a very low sorting output. 1mL (4million cells) at Flow Rate 3.0 takes 45mins to sort, while at Flow Rate 2 that same 1mL sample can easily take 1.5hrs to sort.

We would like to be able to sort faster or somehow churn out more cells/sorting hour. Given the nature of HeLa  and embryo cells which seem to die very easily, perhaps I won't be able improve sorting those cells. But I'm hoping that maybe the smaller cells like ES cells, Monocytes can be sorted faster? Does anyone have any tips to share? I have heard from my users that there are FACS machines that can do 1mL sorts of cell line samples in 10 - 20mins, so I'm wondering if I'm actually being overtly cautious with the guidelines we used.

Thank you and best regards,
Michelle Mok

Replies (All identities have been kept confidential):

Just one suggestion - if you ask your clients to provide their cells at concentrations of 10-15 million per ml, then you can keep the flow rate very low (which seems to be a survival factor for some cells), but process more cells in the same amount of time as your previous settings.  
\When cells are provided at these high concentrations, I often ask the client to provide a more dilute sample of the same cells for setting the instrument and the gates.  My main concern of having cells at the high concentration would be of them settling out, but the Aria has a mixer, and you can always unload the sample tube periodically to give it a gentle 'finger vortex'.

I've sorted Dendritic cells in the past on my MoFlo,(prior to the availability of the direct markers)

I would make the concentration 8-10 M /ml. 

Also I sort everything on a 70um tip. Haven't seen adverse effects from that.

A major point, I've put up to 50% FCS in the catch tube to enhance survival.

The best wayt to speed up a sorter is to increase the concentration. you might increase the abort rate, but if you can get enough cells for the customer, what do they care? They might think, what the heck I'm saving money the sort took less time.

Cell counts are important to be accurate otherwise you never know what to expect, always make sure hemocytometer users are using the ACTUAL HEAVY cover glass, and NOT a cover slip. The cover slip will distort downward and make the cell counts completely wrong. I've seen that many times over the years.

We, too, are a busy core flow lab that does sorting.  Yes, sorting low frequency, large cells does take a long time.  
We require that investigators schedule sorts by blocks of time (1 hour, 2 hours, etc.).  They are required to make the calculations prior to submitting samples so they know how long it will take to reach their goal of collected cells.  By requiring them to do the calculations, they know how many cells they need to submit to reach that goal, and sometimes they realize their goal is not obtainable.  This has helped by making the investigators aware of all the variables that affect sorting as well as eliminate open-ended sorts ("collect as many as possible", or "just sort the entire sample").  We do not accept sorts without an end-point specified.

Therefore, sorting is performed until the desired number of cells is collected or until the scheduled sort time elapses. In this way, we can schedule more than 1 sort in a day and (with no unforeseen problems) get home at a decent hour (or at least know ahead of time how much O.T. will be required).

The following is what we require our investigators to do when they schedule sorts:

Calculate the time needed for your sort, and the number of cells you need to submit:
A) Collection rate in cells/hr = (%viability) x (%expression) x (25.0 x 10^6 events/hr*)
*25.0 x 10^6 events/hr = our average sort rate using the 100 um nozzle.  This will vary with flow rate.

B) #of cells to submit = (#cells desired) ÷ (% expression)

C) Time your sort will take = (#cells desired) ÷ (collection rate determined in A)

For example:
An investigator wants to sort 1.0 x 10^6 GFP+CD34+ cells that are 2.0% positive of total cells in the culture, and the pre-sort cell viability is 90% :

The investigator would have to submit 50 x 10^6 cells, and the sort would take 2.2 hours:

A = (0.9) x (0.02) x (25.0 x 10^6 events/hr) = 450,000 cells per hour can be sorted

B = (1.0 x 10^6 cells) ÷ (0.02) = 50 x 10^6 cells need to be submitted

C = (1.0 x 10^6 cells) ÷ (450,000 cells/hr) = 2.2 hours to complete the sort

Most of those cell lines will pass easily through the 70um nozzle so then you can use the high speed option. However, it is really important that you keep the Agitation option on during the process and that you PRE-FILTER all samples through 40uM mesh. I wouldn't put dendritic cells through the 70uM nozzle though.

All of what Tim says is great.

1> It is the system pressure and nozzle size that does the harm (shear factors) and not the sample pressure.  Tim is correct and stay away from the high flow rate setting.  If scatter goes to crap then you are running at too high a sample pressure.  BD has an 85u and a 130u nozzle.  The 85 will help the throughput.   I run it at 25 PSI with a DDF of ~45k.  Not much change in PSI so some cells will like it ok.  You can also try the 100u nozzle at 25 PSI and a DDF of ~36K.  All nozzles are diff.  Just play with them a bit. We sort most cell lines using the 100u and can sort through about 20 million per hour.  I like the slow and good method.  If cells are to be used for RNA then I will use the 70 at 50 PSI. and DDF of 78k.

2> Temp is important.  Some are not tolerant to 4 degrees.  Some like to be kept cold during and after sorting.  So keep the sort collection holders cold.

3> We use Sterile Dulbecco's PBS, a good stable buffer.  No problems seen.

4> I ask the user to provide media that the cells are grown in for this purpose.  I usually put about 2mls in a 15 ml tube and .5 in a 5ml tube for collection.

PS: Flow rates are different on each Aria.  A 2 on one may be a 3 on another. To get a handle on this, put 1 ml of water in a 5ml tube.  Weigh it and record that number.  Put the tube on the instrument and run for 15 min at a flow rate you like.  Remove the tube and weigh it again.  1 ml water = 1 gram.  At least in theory.  Multiply the diff in weight by 4 and you will get the volume/hr.  I think I got the math right here.

Sample flow rate has very little effect on cell viability...you have
other potential issues probably related to sample prep and buffer

Increasing the flow rate to 6 is not the way to increase throughput.  All
you are doing is increasing the core-stream, and thus increasing the
potential for coincident events, as well as potentially affecting the
fluorescent signal of cells that are out of the optimal focus of the lasers.
The pressure differential should not be killing the cells.

Your pressure is dictated by the nozzle size.  So for lymphocytes, you can
run with the 70 micron nozzle at rates of over 20,000 cells/second (72
million events an hour) with good viability.  You can empirically determine
if the cells will go through a 70 micron nozzle by sorting some unlabeled
cells and watching the side-streams.  If the cells are too big, then the
side streams will be unstable and you'll see a lot of fluttering in the
camera.  I prefer to do this, rather than rely on a microscopic measurement
- as we're talking about cells in solution vs. on a slide.  Besides, I think
it's quicker and easier.

What is your frequency?  If, for example your frequency is 20 -- that means
you are generating 20,000 drops a second.  If you try for a conservative 1
cell every 4 drops, your max rate should be about 5,000 events/second (or 18
million cells/hour).  So have the investigators concentrate their samples to
20 million cells/ml, and you should increase throughput.  Based on your
numbers, you can sort a ml of sample about 45 minutes at a differential of

Secondly, determine if the cells are sensitive to temperature.  Some cells
will die if kept at 4C, others are fine at that temp.

Third, check your buffers -- what are you using for sheath?  Make sure that
it's not killing the cells.  We've had that problem before, and some cells
are more sensitive than others.

Fourth, check what the cells are being caught/held in.  That buffer could be
causing issues -- and are the investigators coating their tubes to minimize
electrostatic attraction of the cells to the plastic?

    We often sort large (>30um) cells for culture and have found the same problem. You appear to be doing everything right (low pressure, low pressure differential, large tip) but the rule of thumb here is to ensure that the diameter of the cell is less than 1/4 the diameter of the nozzle tip aperture- this magic number appears to minimize the shear stress and other physical pounding the cell might take through hydrodynamic focusing, piezoelectric oscillation, 'scraping' on the walls of the sample injection lines, etc. The further you can get the cell away from the interface between the jet and the inner core the better. I think the FACSAria has a 150um nozzle tip available? If you are sorting into tubes (not wells) another tip would be to test the sort stream on the side of the tube before actually running the sample to get the incident angle minimized. This way the cell 'rolls' down the side of the tube instead of splashing to a halt. Then again I sort phytoplankton cells so I laugh whenever someone mentions Chinese Hamster Ovaries ;) Best of luck!

More information about the Cytometry mailing list