Flow Cytometry QUERY!

Howard Shapiro hms at shapirolab.com
Tue Jan 30 23:28:03 EST 2007

Deepika Gaddam wrote-
> I am a graduate student in the department of Biology at UNC- 
> Charlotte, NC. We have a BD FACSCalibur System in our department  
> which I want to use for a specific purpose for my project. I have  
> also contacted BD Biosciences service support and also the  
> scientific advisor of the company, they gave an encouraging reply  
> but again that could not solve the issue I am facing. I wish to  
> explain very briefly the problem I am facing in achieving a very  
> crucial step in my research. I will really appreciate if I get any  
> encouraging suggestions.
> Research Details
> I have performed homologous lambda recombineering to introduce a  
> gfp gene variant into E.coli strain. I am not using any antibiotics  
> as selection markers to select for the recombinant, and I am only  
> screening the recombinant by a change in the phenotype (green  
> fluorescence emitted by the GFP). This change in the phenotype will  
> be observed using a Blue LED attached to an interference filter,  
> the blue light will excite the GFP and inturn the green  
> fluorescence from the GFP will be observed by an absorbing glass.
Observed in what (see below)?
> Problem
> The limitation of homologous lambda recombineering is that it is a  
> very very rare event and the chance of occurance of recombination  
> is 0-1% (frequency is 1 in 10e6). Even if I have only  one positive  
> recombinant, the chance of finding it amongst the non-gfp cells  
> (negative background) is very low therefore for this purpose I want  
> to optimise FACS so that I will be able to screen and detect for  
> the positive recombinant.
> Another limitation is that of the signal intensity of the GFP, the  
> recombinant will have only one copy of the gfp gene integrated into  
> the chromosome and I am really not sure / do not know about the  
> intensity if the fluorescence with one copy of the gene.. Can't I  
> get  around FACS if the signal intensity is low?
> I understand that it works quite well with eukaryotic cells but I  
> want to know how to optimize the instrument with bacterial cells. I  
> have already tried using FACS once, the controls worked fine but I  
> faced a couple of problems when I have actually performed lambda  
> recombineering which I feel will be worth mentioning.
> 1. the minimum number of cells (in terms of volume and the number  
> of cells) that can be injected into the machine at one time.	
> Experimental error and the chance of missing the fluorescing cell  
> when too many cells are injected at once are the concerns that I have.
> 2. the huge amount of negative background (non-fluorescing cells).
> 3. autofluorescence from the non-fluorescing cells and also  
> interference that might affect the fluroscence.
> I would like to know if the model 'BD FACSCalibur System' can be  
> used for this pupose and if so how to optimize 'BD FACSCalibur  
> System' for this purpose (recommended settings) and will it be able  
> to detect a single fluorescing cell (recombinant) amongst many non- 
> fluorescing cells? How can I modulate (light source and other  
> technical aspects) the instrument model for the current purpose?
Here's the good news:

Detection of rare events at frequencies on the order of 1 in 10e6 is  
extremely technically demanding, but has been accomplished in BD  
benchtop flow cytometers:

Gross H-J, Verwer B, Houck D, Recktenwald D: Detection of rare cells  
at a frequency of one per million by flow cytometry. Cytometry	
14:519-26, 1993

Gross HJ, Verwer B, Houck D, Hoffman RA, Recktenwald D: Model study  
detecting breast cancer cells in peripheral blood mononuclear cells  
at frequencies as low as 10(-7). Proc Natl Acad Sci U S A 92:537-41,  

And here's the bad news:

The rare events in question in the papers cited above were antibody- 
labeled mammalian cells, presumably bearing at least tens of  
thousands of molecules of antibody/cell, and both positive and	
negative gating were necessary to discriminate the labeled cells from  
others in the sample.

What you're up against is not how many copies of the GFP gene there  
are per cell, but how many copies of the protein. If you were looking  
for rare recombinant bacteria bearing a few thousand copies of GFP,  
you might have a chance. If you are trying to detect a recombinant  
bearing a single GFP molecule, by flow cytometry or otherwise, you've  
got a really serious problem. GFP is not going to be the only  
fluorescent material in the bacteria, so your "negative" background  
is not going to be as negative as you think.

The extinction coefficient of an improved GFP at 488 nm is on the  
order of 20,000 M^-1cm^-1, with quantum efficiency near 80%. Like  
most pro- and eukaryotic cells, E. coli contains flavin nucleotides  
(FMN and FAD) and flavoproteins, which can absorb light or be excited  
at 488 nm and which emit in the same spectral region as GFP; these  
are a major source of autofluorescence. Although extinction  
coefficients and quantum efficiencies of FMN, FAD, and various	
flavoproteins differ, it would probably be reasonable to assume an  
average extinction coefficient of 5,000 M^-1cm^-1 at 488 nm and an  
average quantum efficiency of 10% for the total flavin fluorescence.  
The fluorescent lifetimes of GFP and flavins probably do not differ  
by more than a factor of two, suggesting that the fluorescence	
measured in a given interval from a single molecule of GFP will be no  
more than would be measured from about 64 molecules of flavin.	
Evidence suggests there are several hundred thousand molecules of  
flavins in a single E. coli K12; however, even if there were only one  
tenth as many, excitation of a single organism containing a single  
molecule of GFP would produce 500 times as much fluorescence from  
flavin nucleotides and flavoproteins as from GFP.  Although imaging  
apparatus using cooled intensified CCDs and exposure times of  
milliseconds can detect a single GFP molecule in the absence of  
interference from other fluorescent substances, detecting a single  
GFP against that level of flavin background fluorescence (and/or any  
other similar autofluorescence) is essentially impossible, even using  
such instrumentation.

 From the known extinction coefficient at 488 nm and quantum  
efficiency of fluorescein, we would expect it to be at least 30 times  
as fluorescent as flavin nucleotides. It has been reported that, in a  
typical BD Biosciences benchtop flow cytometer (FACScan, FACSCalibur,  
etc.), it takes the emission from about 80 molecules of fluorescein  
to produce a single photoelectron at the cathode of the green  
fluorescence detector PMT. It would take about 2,400 molecules of  
flavin, or a few dozen of GFP, to produce the same signal. So, if  
there were no autofluorescence at all, it would be difficult to  
detect fewer than about 100 molecules of GFP reliably in a single E.  
coli using a FACSCalibur. A slow flow system could probably detect  
single molecules in single [non-autofluorescent] bacteria, but the  
typical processing rate - fewer than 100 organisms/second - would  
mean that several hours of observation would be required to find a  
single recombinant, even if there were some way of eliminating flavin  
fluorescence completely.

But back to good news:

My original question to you was, basically, How are you "screening"  
the recombinant by observation? If you can *see* the recombinants  
against the background of autofluorescence expected from E. coli,  
there have to be at least a few thousand GFP molecules in each	
recombinant, and you ought to be able to detect the recombinants in a  
FACSCalibur, if - and probably only if - you follow the entire	
obsessive-compulsive ritual described in the papers by Gross et al.  
If recombinants are such rare events, it is probably unlikely that  
you will detect many with more than a single copy of the GFP gene.  
Although you would might expect a recombinant with two copies of the  
gene to produce roughly twice as much GFP as one with a single copy  
of the gene, whether or not you could distinguish them by GFP content  
would depend on the width of the GFP fluorescence distribution, which  
would be noticeably broadened by photoelectron statistics if there  
were only a few thousand molecules of protein/organism. Also, since  
bacteria typically initiate multiple replication forks, you might  
expect to find different numbers of GFP gene copies - not all of  
which might be active - in organisms originally containing only a  
single copy that had initiated DNA replication but had not yet divided.

It won't be easy, but you might be able to do it.

Good luck,


More information about the Cytometry mailing list