Re: Fw: Going too fast?

David.C.McFarland@GSK.COM wrote (in response to Ray Hester):

>Commenting on your post reproduced at the bottom:
>
>If you are actually sorting this rare FITC+ population, which I will 
>assume is the goal since your setup is on a Vantage, then this is a 
>flawed approach.  Realize that a sorter needs to "see" the cells you 
>don't want as well as the cells you do, otherwise your purity will 
>be very poor.	An example is the investigator that used to come to 
>me with mammalian cultures with bacterial contamination.  He wanted 
>me to threshold out the bacteria, which greatly outnumbered the 
>cells of interest, and sort anyway.  This doesn't work.  Just 
>because you tell the instrument to ignore the bacteria, it doesn't 
>make them magically disappear and they will still be in great 
>numbers in the post-sort sample since they will be sorted along with 
>cells of interest.  In fact, I usually lower my threshold (based on 
>scatter) when sorting blood-derived cells to ensure that I see the 
>smaller particles (ie red cells/retics and platelets) that I don't 
>want contaminating my sample.	Of course, as you state, this can 
>really slow down the sort if your lower threshold value greatly 
>increases the events/sec.  One alternative might be pre-sort 
>enrichment before sorting.  We use magnetic separation a lot to 
>enrich rare cells before sorting.  In your case, pre-enrichment 
>using an anti-FITC magnetic bead may be a viable option, assuming it 
>doesn't quench.  (We actually use anti-PE mag sep a lot.)
>
>The incredibly high flow rate you mention will cause other issues as 
>well.	First, I can't believe you don't have clogging issues!	(I'm 
>assuming they are large cultured cells since you are using a large 
>nozzle tip and a smaller tip would allow you to sort 
>faster.)  Second, as good as the Diva electronics are, there has to 
>be a limit to how many events it can process per second.  (Sorry, I 
>don't know what it is off the top of my head).  If you exceed that 
>limit, cells will be truly ignored and you're basically wasting 
>sample and passing a lot of cells through that are not 
>characterized.  Then there is the issue with coincidence.  With such 
>a high concentration/flow rate and your relatively low sheath 
>pressure, I also can't imagine that you are analyzing single cell 
>events.  Lastly, when sorting in Diva you want to adjust your flow 
>rate such that only a few drops (3 to 5 works for me) out of every 
>ten are occupied and you only want to have one cell per 
>droplet.  You didn't provide your drop drive frequency, but I'm 
>certain that is much less than 2 million/sec, probably much less 
>than 100,000/sec.  This makes it obvious that virtually every drop 
>will be occupied, and multiply occupied at that.  This will greatly 
>disrupt droplet charging with the result being extensive fanning and 
>no real sort stream.
>
>Bottom line:  you're either going to have run slower or 
>pre-enrich.  I am not aware of a commercially available sorter that 
>can handle a couple of million events per second (yet).
>
>I think even if you could get it to sort in the manner you propose 
>that you would see very little enrichment of your target 
>population.	And honestly, I don't think this approach would 
>really even be suitable for simple flow analysis.
>
> >>> Ray Hester <rhester@jaguar1.usouthal.edu> 03/17/06 9:32 am >>> wrote:
>
>Hi,
>
>We have a Vantage with DiVa and wondered what the effect (if any) would
>
>be of	running FITC- stained cells at approx 2.5 x 10E6/sec but using a
>
>threshold set on FITC to effectively lower the particle rate to 50,000
>
>cells/sec., i.e., the cells below the threshold are mostly negative
>with
>regard to FITC. The sheath pressure is set at 30 psi and the sample
>differential at 0.8 using a 100 um nozzle.  We know the actual particle
>
>rate is approximately 2.5x10E6/sec based on cell counts, sample volume,
>
>and time required to deplete the sample.
>
>Is DiVa capable of 'ignoring' all of the events below the threshold so
>
>that the particle flow rate is truly 50k/sec as indicated in the
>software?

In my response to Ray I mentioned that thresholding on fluorescence 
could be used to increase the effective analysis rate. The situations 
in which we used it years ago were:
1) Differential white cell counting on whole blood; the analysis rate 
was 33,000 to 50,000 cells/second (yes, this could be done in the 
mid-1970s!) and scatter measurements were actually made at that rate 
for counting and sizing red cells and platelets. A nuclear 
fluorescence threshold was set to gate acquisition of three 
additional fluorescence parameters for white cells; since there is 
typically only one white cell per 1,000 red cells, we only measured a 
few dozen white cells/second.
2) Ploidy analysis of megakaryocytes in marrow, used to determine 
thrombopoietic activity in the isolation of a platelet-stimulating 
factor. An anti-platelet antibody and PI were used as stains; the 
threshold was set so that only events with DNA content 8C and above 
were collected. Megakaryocytes are typically less than 1% of 
nucleated cells in marrow, so this trick worked fine.

BD plays the same game in the FACSCount, which uses triggering on 
anti-CD3 fluorescence to count CD4+ or CD8+ T cells in unlysed whole blood.

As I suggested in my earlier posting, the analysis might be 
considerably more difficult at the extremely high cell concentrations 
required to do what Ray wanted to do.

I neglected to point out, as Dave and others have done, that it makes 
little sense to use this strategy if the objective is to sort cells. 
In analysis, cells below threshold are neglected; in sorting, cells 
below threshold that are in the same drop as the desired cell are 
sorted with it.

-Howard