All tags Flow Cyt Color Compensation in Flow Cytometry webinar

Color Compensation in Flow Cytometry webinar

Watch this on-demand webinar presented by Graham Pockley and Ian Dimmick 

Review hardware configurations, instrument settings and consideration of dyes. 

Webinar Topics:

  • Choice of instrument configuration
  • Making sure where your photons are going
  • Choice of fluorochromes
  • Are your controls doing their job?

About the Presenters:

Graham Pockley

Graham Pockley is currently the Associate Director of the John van Geest Cancer Research Centre at Nottingham Trent University. His research is focused on immunoregulatory mechanisms in health and disease. Graham has used and been a supporter of flow cytometry throughout his career.

Ian Dimmick

Ian Dimmick began his career in a clinical setting, and has worked in various research posts using flow cytometry as a diagnostic and research tool, primarily for HIV, leukemia, lymphoma and a broad range of immunological testing. Ian currently manages the flow cytometry core facility at Newcastle University. 

Webinar Transcript:

Hello, and welcome to Abcam's webinar on color compensation in flow cytometry. Today's guest speakers are Graham Pockley and Ian Dimmick. Graham's research areas are focused on immunoregulatory mechanisms in health and disease, and he is currently the Associate Director of the John van Geest Cancer Research Centre at Nottingham Trent University. He has used and been a supporter of flow cytometry throughout his career.

Ian began his career in a clinical setting, and has worked in various research posts using flow cytometry as a diagnostic and research tool primarily for HIV, leukemia, lymphoma and a broad range of immunological testing. Ian currently manages the flow cytometry core facility at Newcastle University. 

Joining Ian and Graham today will be Ken, Supply Manager based at Abcam's Boston office. Before we start, just a quick reminder that any questions you have during the webinar can be submitted via the Q&A panel on the right hand side of your screen. Also, when you log-off from the webinar you will be directed to a webpage where a copy of the presentation files can be downloaded. I will now handover to Graham who will start this webinar.

GP:    Thank you, Lucy, and welcome everybody to today's webinar on color compensation. As Lucy said, I've spent quite a large amount of my research career working in flow cytometry, and I think if there's one aspect to flow cytometry that still causes concerns, it surrounds color compensation and an understanding of how one should design appropriate experiments. With the increasing number of lasers and detectors in the next generation instruments, the problems are only going to get more difficult to resolve. So before I turn the main part of the webinar over to Ian, we'll probably just go through the key elements of the webinar that we're going to go through today. So, really, what we want to cover today is, why do multicolor experiments? I think maybe we all have our own views on that, but it's important to consider why we do them, and is it necessary, and to what extent is the information going to provide insight into our particular experimental questions? There's a large range of fluorochromes available these days, so which of those should I use in these experiments? What is spectral overlap and why do we need compensation? A key element is what is the procedure for compensation? ​

Really, it's always good to see problems in real-life, and so Ian's going to take us through some examples of spill over events, and how you can calculate the spill over. As with any experimental procedure, it's crucial to have the correct controls in the experiment and so he's going to go through those elements of experimental design as well. Without further ado, I'd just like to handover to Ian and just once again remind you that if you do have any questions, please don't necessarily wait until the end, if you can just get those to us as we go through, then we can start to try and answer those as we go through the webinar. Ian?

ID:    Thanks very much, Graham. The first question: Why do multicolor experiments? So it's pivotal to all of our research that we actually accumulate as much data as possible from experiments. The next slide will effectively show you that if we use, for example, six antisera within a single tube, then you will see that you will get a maximum of 18 phenotypes out of this. So you are limited to the actual phenotypes depicted by the antisera per tube. 

If, however, you then put all six antisera, depicted here by three antisera per tube into the single tube, then you can see these dark areas are the feeder types that you would never ever see by keeping the antisera separate within two tubes. So if you then take on board that all you need to do is have perhaps a little bit more in the way of a more expensive instrument, and put a little bit more thought into the actual experiment set up and protocol itself, then you can actually increase the data that you're getting out of your experiment, and considerably so. These are 36 discrete phenotypes, as opposed to the 18 phenotypes and that's a two-fold increase. However, what is not shown here are the non-discrete phenotypes, so the doubling effect can be enhanced when you start looking at the very subtle changes in the phenotypes that you can see between the discrete phenotypes. So it's quite an exponential rise in data that you're going to get. 

This then begs the question, which fluorochrome do I use? Well, this is probably a very, very good pivotal question, and this is where you shouldn't just log-in to a catalogue on a website and just choose whichever ones fit into the laser that you have. You need to consider a considerable amount of variables that you're going to come across when you actually create the experiment. Fluorochromes all basically work on the same principle, in that we excite the fluorochrome by a laser and this laser should be roughly in the area of the maximum extinction coefficient of the fluorochrome. Once the electron goes into the higher orbital which is quite unstable, it will then go back to the stable orbital. Within this process there's a certain amount of heat dissipated and this is consequential, in that the excitation wavelength and the omission wavelength of the photon falling back down to the more stable configuration, will always be a longer wavelength than the excitation. This is the basic state of play for all fluorochromes that you will excite at a lower wavelength, and you will admit at a higher wavelength. 

This very, very useful tool is the spectrum viewer, and the advantage of this tool is that it will give you some degree of capacity of looking at which fluorochromes will fit onto your particular instrument. However, please remember this is very graphical and very often you can be misled by this tool sometimes. So be very, very careful when using this tool, don't exclude fluorochromes on the basis of this tool, but it is a very, very useful tool in its basic form.

The fluorochrome itself is excited in terms of the extinction coefficient, as I've already discussed, and it's excited at a specific absorbed maximum wavelength, and this is the optimum. With few examples, the excitation of the fluorochrome is never at the absorbance maxima, it's far away from that. Then the emission is depicted by the quantum fluorescence, the degree of emission is depicted by the quantum fluorescence of that particular fluorochrome. The value that we are given for the quantum fluorescence is a value that's measured over the entire spectral emission value. You never measure the entire spectral emission of a fluorochrome, apart from occasionally when you're using a very broad, long pass filter, but that is not on your instrument; but, that again, is not always the norm. So the theoretical value that you get that should depict the fluorochrome of extinction coefficient multiplied by the quantum fluorescence, is virtually never reached but it will give you the extinction coefficient and the quantum fluorescence. This value that usually for the quantum fluorescence goes between 0.5 to less than 1, just below less than 1, is very much a theoretical value. But, again, it's very much like the spectral viewer, it will actually give you an indication of what you are dealing with, but it will not be a finite value that will give you an absolute value.

These are some very useful values; these stain indexes are basically the function of the fluorochrome that has been placed onto a cell, it is then run on a flow cytometer and you look at the negative content of that staining, as opposed to the positive content of the staining. This stain index is a very, very good value in terms of the brightness of that fluorochrome. The brightness of the fluorochrome with respect to the autofluorescence of the cells that you're dealing with. It should be remembered that all these values down here rely on the quantum fluorescence, the extinction coefficient, the autofluorescence of the cells that you're dealing with, the power of the lasers and the filter setup of your instrument. So these values should not be interpreted as a finite value that you can achieve on your particular instrument, it's instrument-dependent, it's cell-dependent and it's laser power dependent also.

GP:    Ian, just a quick question whilst we're going through. So, certainly, the stain index value is where the value is very much dependent on the instrument, but would you expect the order of the fluorochrome to be the same irrespective?

ID:    Yes, the order should be very similar to what you see in front of you. Yes. This gets us to the statement that we've all heard people say before, is that you reserve the ‘brightest’ fluorochrome, for the antigens that have the lowest cellular concentration. Also, reserve the less bright fluorochromes for the antigens that contain the highest cellular concentration of your antigen, and this is just a common sense approach.

The stain index, you can see here, is depicted by the signal from the cell which is depicted by this positive peak, but it's also depicted by the width of the autofluorescence. So the stain index is this distance from the negative to the positive, and divided by the width or standard deviation that can be achieved by the negative part of the cell. The standard deviation is very important, because, as you can see, if we take this narrow peak the distance from the positive to this narrow peak is far greater than if you have a highly autofluorescent cell, such as Mesenchymal stem cell or embryonic stem cells, as opposed to, say, lymphocytes, so the stain index also will vary from cell type to cell type, as well as from instrument to instrument.

The type of fluorochromes can be broadly put into three distinct categories: the first is the non-tandem fluorochromes which are very easy to deal with, very straightforward; and the second type of fluorochromes are the tandem fluorochromes. These tandem fluorochromes are absolutely fine, they don't fall to pieces, they've acquired a bad reputation, but if you treat them properly they're absolutely fine, there's not a problem with them. The third are the Qdots; very, very useful third fluorochrome type, but they have to be used very intelligently.

This is an example of the non-tandem fluorochrome, and this is whereby we are looking at this particular dye here is it's a violet laser dye and it’s Brilliant Violet 421. This dye here is a V450 dye and you can see that this Brilliant Violet 421 as opposed to the V450 dye here, is actually much brighter than this dye here; and you can see that on this overlay here. So that is a very, very useful first step in that if you have a low concentration of antigen, then you would probably go for this dye as opposed to this dye here.

The other advantage when you're looking and ascertaining which fluorochrome you need to use in your experiment, is how much compensation do you need from, for example, the next PMT along whereby the red tail of most of these fluorophores will actually spread into? As you can see, for this Violet 421 dye, then the compensation from 525 nm to 450 where this dye is being detected is 14%; so it's coming into this detector 14%. If you then look at the V450, the compensation on this is actually 30.4%, so not only do you have a brighter dye here, you have a dye that has less spectral overlap into the next PMT detector. So now we have two reasons for looking very objectively at the Brilliant Violet 421 dye, as opposed to the V450 dye. However, it's got to be said that both of these dyes can be equally well-used in any experiment as long as you are aware of the spectral overlaps. As long as you are aware that maybe you need to put a higher concentration antigen onto the V450, as opposed to using the V421 for a lower density antigen; and be aware that the compensation is going to be slightly higher. So you need to make sure that the voltages that you apply to this experiment account for this increase in the compensation.

The tandems themselves: here we have an example of basically why tandems are produced, it was to give extra dyes from single laser instruments, initially. Here we have phycoerythrin that is excited by 488 nm, and it will emit at 580 nm. If we then look at Cy5; Cy5 dye is excited by the 633 nm laser, and it emits at 675 nm. Now, here we're using two lasers: this laser, the blue laser, is showing an emission of 560 with phycoerythrin and this red laser is showing an emission with Cy5 at 675 nm. However, if you put both of these molecules together, you can create a blue laser excitable phycoerythrin molecule that transfers its energy at 580 nm, which is close enough to this 633 nm excitation spectra for Cy5. You will then get a molecule, as you can see here, that is excited at 488 nm, it then emits at 675 nm thereby giving you a third option for the excitation wavelength and emission wavelength off the blue laser.​​

This slide kind of depicts what you've got to do to cause problems with a tandem, and this particular tandem dye you can see this is from a FACSCalibur. If you create this particular dye, and this is - if I remember correctly - this is PeCy5; if you create this tandem, then you need 20.98% compensation after the sample has been prepared for 20 min at room temperature, and this is quite acceptable from the excitation to the emission. Here, if you leave the sample out at 24 hr at 4°C, then the breakdown of the tandem molecules will mean that the excitation or donor electron molecule starts to leak a little bit, because we have a breakdown of the bonding and the compensation then goes up to 22.75%, which is quite acceptable. However, if you leave this sample out for 24 hours at room temperature, then the compensation, because of the leakage of the donor molecule, will go up to 34.59%, and this is obviously unacceptable. So this shows two things: you've got to really try hard to get a change in compensation if you treat the samples correctly, but also if you leave it out at room temperature, then the dye does breakdown quite significantly. So you should treat a tandem dye with respect.

The Qdots are exceptionally good fluorophores and they have many advantages, but I think the major advantage is that the Qdot can almost exclusively and uniquely be excited by any wavelength that is below its emission wavelength. So you have the extinction coefficient curve for the Qdot, and this particular Qdot here, which is Qdot 605, can be excited by any lasers that are below this emission. However, as you can see, the extinction coefficient reduces and the effectiveness of the emission starts to reduce the degree of quantum fluorescence and excitation, and reduces as you get closer to the emission spectra. Secondly, the Qdot itself has a very uniform emission spectra, and it doesn't have the typical red tail that would normally come out like this; it has a very uniform emission. This means that usually the compensation values are very much less when you're dealing with Qdots, than when you are dealing with normal fluorochromes. The structure of the Qdot here, the emission spectra relies upon the actual size of molecules that make up this semi-conductor core; the larger the molecule, then the higher the emission spectra that will be created.

Also, the structure on the actual antibody differs. Here we have an antibody that is conjugated with a fluorochrome, and this is the excitation wavelength, this is the emission wavelength with the red tail going out here. Here the Qdot is actually central and antibodies are bound onto the Qdot, so it's a complete contradiction of this graphic here. Also, the absorption is all the way along here. It doesn't have a specific extinction coefficient maxima, it is all the way along here. The emission, you can see if you compare this emission here with this emission here, where there's a long red tail here, just underlines what I was saying in terms of the actual spectra.

But what is spectral overlap and how do we see it, and why is compensation needed? We need to take a step back now to actually visualize what happens in front of us when we're trying to create an experiment. So the spectral overlap itself is mimicked within the instrument, you have a filter in front of your photomultiplier tube that is going to actually measure the emission from your fluorochrome. Here we're going to measure the emission of 520 nm, plus or minus 10 nm for this particular fluorochrome, and this is usually put in front of a detector that you're going to look at FITC. This is the emission spectra of FITC. The whole quantum fluorescence value is the area underneath this curve, however, we're only going to measure this part of the emission spectra, because this is the bounds of the filter that we have put in front of the detector, so we will measure the area underneath that defined curve. 

This is a phycoerythrin filter and there's also a phycoerythrin detector at the other side of this filter, and, as you can see, the phycoerythrin detector will in actual fact measure this spillover due to the red tail of the emission of the FITC molecule. So you will see this effect within an instrument if you do not use any sort of color compensation. So this is obviously erroneous and we need to get rid of this. This is phycoerythrin. Now, as you can see here, there's a tiny amount of phycoerythrin goes into the FITC detector and this is the phycoerythrin emission spectra here that's going to be measured, and this is not going to be measured. 

This is another filter that will actually measure the tail end and fourth of these fluorochromes. There's the emission filter there, so here now we need two compensation values here for FITC, and for phycoerythrin to measure this particular fluorochrome correctly. This is the spillover value and this is the spillover value here. 

In the real world when we're doing multicolor work, this is what it looks like. Now, for each of these fluorochromes and for each of these detectors, you do need to do exactly the same basic thing time and time again. It's no more complex than doing the same basic compensation for each of these individual detectors. There may be some instances where you need to subtract four or five fluorochromes from a PMT, or perhaps just one or two as is the case here, but the mechanism for doing that is the same for every instance. So what does it look like on your dot plots? Here's the spectral overlap here in the phycoerythrin, here's our FITC molecule, and here's our FITC signal: the negative cells and the positive cells. Here is the erroneous FITC signal within the PE detector, and this is what it looks like. This is actually showing a true representation of what's happening up here, we can see the FITC signal and we can see also the PE signal here. What happens is, when we take this compensation away and this spectral overlap is here and here, then we get this negative cell here. We get the FITC cell here. We look at the mean value for the negative against the positive for phycoerythrin. We look at the mean value of this negative against the positive. For any dual-stained samples we look at the percentage PE and FITC that places this cell within this quadrant proportional to how much FITC and phycoerythrin is on this cell.

So here we have the FITC and the spill over, and if this is not subtracted this is where the FITC cell will stay. Once it's subtracted the cell will come back down onto the X-axis. So if we subtract this, then this is what we get: the signal goes away from the PE detector and we get the mean values for the negative and the positive to be the same.

Compensation shouldn't have any surprises, and this is a very easy way of mapping out what your instrument is. This is quite a busy graphic, but basically what we have this is our five laser LSL2, and here you have the filter values depicted in terms of height and values on each of these graphics. So you can see here that these heights, certainly here are very similar. So if you run two fluorochromes and they are captured by each of these filters, then the compensation is going to be high. However, if we go to this particular set up here, this is off the UV laser, then the fluorochrome that you run here, something like Alexa 350 and whatever fluorochrome or Qdot you run here, there'll be very little spectral overlap because of the difference in filter value here. So the more PMTs usually the closer the filters, and therefore the higher compensation values. None of those things are a problem, but you should plan out your instrument like this to ensure that you will not have any surprises when you come to putting certain fluorochromes within certain photomultiplier tubes with certain filters in front of them. 

So compensation is fluorescent intensity independent, and this is something that people keep asking: why, explain this to me? Well, it's quite easy, this is the intensity of our signal here, and this is a green signal, let's say, FITC. The compensation is the spillover within this yellow detector, and this represents, let's say, 10%, and this is 10% of this value here that is in this green detector. If the fluorescent intensity in this case goes down, then this is still 10% of this value here. So it is completely independent of the intensity as to the spectral overlap. If you find that you have differing spectral overlap percentages that you need to apply, because of differing fluorescent intensities, then it's probably because the compensation hasn't been set up correctly in the first place, or you've changed fluorochrome manufacturers.

However, if the sensitivity on this green detector increases, whether it be by you increase the PMT voltage or whether somebody's come in and realigned your instrument to make this more sensitive, then what will happen is that by increasing this PMT voltage here, this compensation value here will decrease. So, again, the scenario we can put to bed is that you can actually decrease compensation by increasing the sensitivity or PMT voltages on another detector, and that's a very important take-home message. Increasing PMTs does not always necessarily mean increasing your color compensation values. Again, what do I subtract from what? Well, here you've got a green fluorochrome and it's in the yellow detector, the only thing you can do is you can subtract green. If here you have a red fluorochrome, then within the yellow detector the only thing you can do is subtract red. Remember, you cannot subtract it against itself, so it must be green minus yellow for the yellow, it must be yellow minus green for this green and here we've got far red minus yellow for this yellow signal here, and here for the red signal far red minus red. So just keep that diagram in your head.

We've already talked about PMT voltages and how they affect compensation. Think of your PMT voltages as your first compensation control, because if you have very discrepant, or if you don't put the thought into the fluorochromes you use, and you have very discrepant PMT voltages, it will significantly affect your compensation values. Also, the available lasers, make sure, firstly, you have the appropriate lasers for the fluorochromes you are using. The fluorochromes themselves obviously play a major part in this, and also the instrument for filter configuration which I've already talked about. Whereby if you have filters that are very, very close together, i.e. the taking into consideration the tolerance on the filters and the position of the filters, you will have more or less compensation independent of fluorochromes. So a lot of consideration has to go into the instrument filter configuration.

The procedure for compensation, again, it's a very simple procedure, it's working with basic rules all the time and it's not working with anything complex, but you must apply these basic rules very sensibly. So if we take an example of a three-color experiment that contains the CD3-FITC, the CD8-PE and the CD19-PerCPCy5.5, then the first thing to do is check for spectral overlap by running each fluorochrome separately and apply compensation, if necessary. So the first tube you run is FITC only, the second tube you run is PE only and this is what we get, we could see immediately what was needing to be compensated. Another thing to remember is that it is very, very useful to have all combinations on this three-color example here, then we have every fluorochrome against the other, so you can immediately see what is required in terms of compensation. In a three-color example this is fairly straightforward. If you are doing 13 colors, then it becomes almost imperative that you look at the whole thing.

So look for the plots with PE on the axis, is there spectral overlap and there obviously is spectral overlap. Here is the PE axis here, and here is our spectral overlap for PE and then they need adjusting, and they adjust like this. Again, it's adjusting the mean of the negative with the mean of the positive on each occasion for each of the single fluorochromes, and then when you run your triple stain sample, then you get nice separation of each of the fluorophores.

Spillover examples: if you are running a multi-laser, multi-parameter instrument, a very, very easy way of seeing whether you're going to run into problems later on is to look at the spillover of the fluorochromes. We can often do this by putting the antibody with the particular fluorochrome we want to look at in our experiment onto an antibody capture bead. Here, if we look to Qdot 605 and here we can see this is the 488 laser with the 610, with a tolerance of 20 nm, so there's a 600 to 630 nm which nicely detects the 605 Qdot. You can see here the Qdot is nice and bright, you can see here it's less bright and here it's less bright, so these are very easily compensatable. Don't be misled by this, this is actually the autofluorescence of the antibody capture bead. Remember also that you've got to base these measurements on something. What the red peaks are, this is an experiment that we were setting up for looking at lymphocytes, peripheral blood lymphocytes, and we set all of the voltages on these unstained lymphocytes. So you need to have appropriate voltages before you do any of this sort of antibody fluorophore testing. 

Here we see an example of phycoerythrin, so here this is the blue laser, the phycoerythrin channel here, which is fine. You can see, as I showed in the Excel graphs, that 57525 and 488610, these filters are very close together, and you can see the values for the mean fluorescent intensity for phycoerythrin in the 57525 channel, and in the 488610 channel with the voltages as they have been set. These values are very close together and you will need significant spectral overlap to get rid of this phycoerythrin signal out of this 610. Once we go further away from the 488575 phycoerythrin peak emission, you can see that the spectral overlap is very easily taken into account. For the rest of the lasers, this is a UV laser, this is the violet laser, this is the red laser, phycoerythrin does not really enter into any of these detectors at all. So you know what you're dealing with. 

Pacific blue, again, you can see here for the violet laser at 450 nm emission there's a nice fluorescence here. At 525 emission here, then it is much less, so you can take this away by normal spectral overlap, and you can see for Pacific blue it's not actually excited by any of the other lasers at all. This is very, very useful information for you because when you're designing the multicolor experiment, then you need to know if any of these fluorophores are excited by additional lasers. Because you need to take that into consideration when you're combining fluorophores with the potential at looking at weak dual positive results from your multicolor experiment.

This is PeCy5 and this is one of the great examples that with PeCy5 then, in actual fact, with these basic voltages here, firstly, this is the signal, it's off-scale, even at 5.4 log decades it's off-scale. This is all rounded up against this Y-axis, so the first thing is this voltage here you could quite easily drop down. It's very bright in these detectors here, so if you drop this voltage then you would have problems with high compensation, and certainly in this detector here. But also the Cy5 molecule off the red laser is independently excited from the PE, and this can also cause you problems: one, in terms of a large dye vector, so that it would be very difficult to look at weak positives that were perhaps APC positive and PeCy5 positive; it would be unworkable. So this underlines that this fluorophore is actually excited by more than one laser, and it's the second photon accepter part of the PeCy5 molecule, a Cy5 molecule that's been independently excited by that laser.

GP:    Could I just remind everybody that please do keep sending your questions in, and we'll try and answer those either at the end of the session or offline. We are running a few minutes late; we will continue with the webinar, and please do stay with us, it's important that we get the message, the critical message across this afternoon. So we will be going slightly beyond the one hour, but, as I say, please stay with us. Thanks Ian.

ID:    So now we're on to the relatively easy part of the color compensation, which is the spillover calculation procedure and the controls that you should use. Here we can see that for PeCy5.5 off the blue laser, a 710 emission band pass filter, in front, we have this nice negative, this nice positive here. The compensation can be performed quite easily if you have a very bright signal like this, and lots of cells that are positive, and you have a nice negative population.

Here you can see we go from this uncompensated dot plot here, to this dot plot. We know that these antigens are mutually exclusive, and so it's a matter of aligning the positive mean with the negative, and the positive mean here with the negative. That doesn't happen always, and what we then need to do is after setting up your experiments on your basic cells, you then have to run fluorochromes attached to antibody capture beads, so that you get a good positive, bright signal from each fluorochrome. For these particular beads they have anti-kappa on the surface and they attach the antibody via the kappa light chain, and the fluorochrome is exhibited here, and then you have a beautifully bright bead with a specific fluorochrome on its surface. 

So the calculation of this spillover and the fluorochrome combination rules are: weak antigen use a bright fluorochrome, differentiate between compensation and data manipulation, as in the PeCy5 scenario. If the fluorochrome is excited by more than one laser you need to evaluate it, if it is usable in your multicolor experiment. Always evaluate percentages of antigens by single color controls, this is very important. If you're setting up a ten-color experiment, then run your single color controls for each of those ten antigens. Make sure you know roughly what those ten antigen positive values are, and then when you run your multicolor experiment make sure that once you've added all ten colors together, you're getting more or less those same percentages. The lower the compensation value within your experiment, the more stable your experiment will be.

Here, we want to differentiate between compensation and data manipulation. This is Cy3 off a green laser, and it's absolutely fine. This is PE off a blue laser and a green laser; this is absolutely correct. PE off a green laser and the blue laser with this emission and pass filter in front of it, which is specific and is the same for the PE on the green and the blue laser are the same values for the PE emission, will show you this and it's not a problem. So we've got CD19 here and I think CD3 here, and so this is correct, you shouldn't compensate that, that is correct. If you compensated that back over there, then you would get a high dye vector and its data manipulation, it's not compensation. 


Here is a very simple way of finding out what is excited by different lasers. Here we see a blue laser with chrome orange, and here we see a violet laser with chrome orange. You can see that the blue laser doesn't excite the chrome orange at all. If we look at V500 against a blue laser, it's excited by the blue laser and with the violet laser it's excited again by the violet laser. With V500 off the blue laser here, you can see that it's 31% to compensate that. Although this is excited by the blue laser, you can in this instance manipulate the data, so that it doesn't appear on the blue laser. It has the same mean value and there's not a wide dye vector so it is useable, but it is excited by the blue laser. Some dyes that have dual laser excitation will give you a very broad dye vector and it becomes unusable, because any dual positives will not be seen, any weak dual positives will not be seen.

Here is a nice example of a nine-color sample, all antigens are nicely expressed and this looks very, very nice. I'm not going to go through all the detail of this, but then we have things like this. This was a ten-color example, and here we wanted a gate on CD3. This was quite a complex experiment, but you can see there's no CD3 expression. This is a CD3 Qdot, and we titrated everything out and what happened was that when we put all the antibodies together, this happened and it wasn't well-expressed. So sometimes you need to change the concentrations once you've put all the antibodies together. Also, this, we weren't happy with at all and it wasn't giving us the results we expected, and this was an issue with the fluorochromes that we chose. So if we increase the Qdot concentration, then we get a nice expression of CD3 on these particular cells. Then if we change the fluorochrome we get what we expect, and the fluorochrome, when from CCR7PE to CCR7P-Cy7 and we changed this fluorochrome here to CD4 Alexa 700 to accommodate that, and give us a much more stable experiment. So sometimes you do need a little bit of fine-tuning. 

After each multicolor experiment, this is the single multi-colored difference in percentage at each of the fluorochromes, i.e. each of the antigens, and you must check that they're in concordance, you don't have any large errors.

So compensation; if you have high compensation values you will increase the instability of the experiment, and the instrument fluorescent intensity, i.e. the ability for the intensity within those photomultiplier tubes, the monitoring becomes critical. Low compensation, then the instrument fluorescent intensity monitoring becomes less critical.

Lastly, controls. Our gold standard is the negative cells that are clearly distinguishable from the positive population of stained sample, and this is an excellent standard. The next is negative cells. Now, unfortunately, you're looking at instrument noise and autofluorescence, you're not actually looking at the effect of the antibody. This is not a good control and, unfortunately, sometimes we have to use it, but it's not a good control and you're not getting the non-specific binding of the antibody in the actual sample. Isotypic controls are very often used, and they should be the same protein concentration as your antibody that you're using, and that's extremely important. They've got to be the same immunoglobulin isotype, and the same fluorochrome as the antibody, obviously. A bad isotype control is a different protein concentration that's going to be very, very misleading, they've a different immunoglobulin isotype and a different fluorochrome even we've seen, or manufacturer. Isoclonic controls, they are very, very good and very difficult to get correct, and I just want to leave that for today.

When to use and when not to use? Picking correct isotype controls fluorescein and protein ratio, and protein concentration considerations and isotype controls are very important. Here we see a test antibody, non-specific binding, good positive/negative differentiation. No antibody, just on negative cells. The cells will be very much to the left and they won't have any non-specific binding of the antibody. The isotype control, depending upon what the protein concentration is, it can be anywhere. 

Fluorescence minus one is more appropriate for multicolor experimentation, but, again, you must remember that for fluorescent minus one you have not got to have a high dye vector, because very often you are looking at dual positives. Here you can see the fluorescent minus one in this antigen that we're detecting here has being missed out of this tube. We then run a tube with the antibody in for that specific antigen, and this is the result and you can see that the compensation is perfect. If the compensation is not perfect, and if this compensation is out, then it will be erroneous. The same old story, the means of the positive to the negative and no matter how many colors you do, then this is the actual modus operandi. Thank you.

GP:    Thanks, Ian. I think there's plenty for everybody to consider there. I think everybody will have some key take-home messages, and we can come back to them during the question session in a few seconds. But I would just like to turn the floor over to Ken, who is going to give some brief information.


KH:    Thanks Ian and Graham, and thanks everyone for attending the webinar. I'd like to take a few minutes to draw your attention to some new services and products that Abcam offers to help you in your research. With Abcam's flow cytometry multicolor selector tool, designing your multicolor flow experiments has never been simpler. This tool allows you to quickly and easily search and compare across Abcam's wide range of competitively priced flow antibodies, go to to try it out. Our multicolor selector tool allows you to search and compare over 2,500 conjugated primary antibodies validated in flow cytometry. The search includes a broad range of over 20 fluorophores, including six tandem dyes. Whether you're looking for a well-known CD antigen or a novel marker, all search results are grouped by clone number making it easy to identify the right product for you. Please visit our blog to watch a brief tutorial on the new tool.

Abcam provides much more than just primary antibodies, we've a growing catalogue of proteins, kits and dye conjugates to advance your flow research. In addition, Abcam has a leading collection of flow protocols, posters and literature on current research methods. Discover more of Abcam's flow resources at the link shown. I'll now talk about some of these new products in more detail, and I'll just note that at the end of the webinar we will be offering a discount code valid for 25% off a number of these products.

We offer a wide range of biotin binding proteins, including avidins, streptavidins and deglycosolated avidins. Given their robust and stable binding to biotin molecules, these are useful tools in protein detection studies. You can choose to purchase these products conjugated to either a fluorophore or to an enzyme.

Flow cytometry is useful to quickly and reliably analyze apoptosis and cell viability. In this slide we’ve outlined some of the products available from Abcam to detect and quantify apoptosis markers using flow cytometry. For example, phosphatidylserine exposure, a hallmark of late apoptosis, can be measured with classic annexin V-FITC staining, or with our new kinetic apoptosis kit. This new kit contains a reversible probe that only binds to phosphatidylserine residues exposed to the outer membrane, making it an ideal tool to measure rescuable apoptosis. To learn more about this kit and other methods you can use to detect apoptosis with flow cytometry, please visit the link shown.

DRAQ5 is a robust DNA dye to help you stain both live and fixed cells, it's useful in a number of platforms including flow cytometry, immunocytochemistry and microarray technology. Abcam's extended collection of fluorescent dyes also includes DRAQ7 which selectively labels dead impermeabilized cells and CyTRAK Orange, which can be used to detect rare cell events.

Can't find the conjugated antibody you're looking for? EasyLink antibody conjugation kits offer a rapid and easy way to label your antibodies. The range includes the choice of 18 labels that are available in a number of convenient sizes to give you flexibility in your experimental design. The chemistry relies on covalent attachment of the label to free amine groups, which results in a perfectly stable bond. Our protocol requires only 30 sec of hands-on time, making this a rapid and cost-effective solution to getting the conjugated primary antibody you need. Find out more on the EasyLink range at the link shown. 

We are pleased to let you know that Abcam now offers Alexa Fluor® conjugated secondary antibodies. Currently available are Alexa Fluor® 488 and 647. From spring 2013 we will also be offering Alexa Fluor® 555 and 594. Why should you choose one of our Alexa Fluor® conjugated secondary antibodies? All of our Alexa secondaries have been extensively tested in-house to guarantee bright staining and low background. We offer a large selection of pre-adsorbed antibodies, which ensure low species cross-reactivity and recognition of endogenous immunoglobulin. If you'd like to use these products in ICC you can do at least 250 experiments at a 1/200 dilution. Find out more at

Abcam scientific support team are always available to help you with any questions or issues you may have. Support is provided in a range of languages including Chinese, Japanese, Spanish, French and German. For those of you in North, Central or South America, please contact our US team. For those in Europe, please contact our UK team. For those in China and Asia-Pacific, please contact our Hong Kong team, and for those in Japan please contact our Japan team.

Finally, I'd like to highlight a couple of upcoming meetings we're holding. The first is on Allergy and Asthma, and this conference will be held in Bruges, Belgium in May. It will feature key scientists from this growing field. More information, including the full speaker list, as well as information on presenting talks or posters can be found on the meeting website. 

The second meeting is entitled, Inflammasomes in Health and Disease. It'll be held at Harvard University in Cambridge, Massachusetts and it covers a broad range of topics, including an overview of the molecular mechanisms of inflammasome activation, the roles of IL1, IL18 and paraptosis in immune defences, and studies of the role of the inflammasome in infectious and sterile inflammatory disease. Check out the website for more details, including a list of the confirmed speakers and their talk titles.

We hope you enjoyed today's webinar. A couple of upcoming webinars include, Fluorescent Western Blotting: Introduction and Application, which will be held on February 28th and IHC-ICC Staining Techniques Using Single and Multiple Labels, to be held on March 6th. I'd now like to turn the talk back to our presenters who will answer any questions you've submitted during the webinar.

GP:    Thanks, Ken, that's great. Okay, so we've got a few minutes to go through some of the questions that you've sent in through the interface. I think everybody will pick up on some key take-home messages from today's session, and the three that I jotted down, really, was the concept that the intensity of the particular flow of chrome doesn't necessarily directly influence the amount of compensation that you will need in your particular experiment. It's also key to understand if you're using a panel of fluorochromes, to what extent that panel is excited by multiple lasers, because that will seriously impact on the decision-making process when it comes to putting your panel together. Another interesting and key factor that Ian raised, was the fact that if you titrate your antibodies in a single color experiment, you need to make sure that when you put that concentration of antibody into a multicolor experiment, that you're getting the same values because you might need to change the dilution of antibody which you use.

We have some of the questions here, and one question is a query about whether PeCy5 is the best and brightest fluorophore, and APCH7 is the least? Well, again, what needs to be done is you need to take ownership of your individual experiment, because in some experiments a particular combination or a particular fluorochrome would be better than others. So I think there's no - unfortunately, there's no hard and fast answer to that, but that decision should be based on a full understanding of your instrument configuration, and the experiment that you're trying to undertake. Okay, Ian?

ID:    The second question was: Would it mean any compensation values over 35 per cent is unacceptable, an example between FL9 and FL10? No, the finite values for compensation you can't really put a limit onto what is acceptable and what is unacceptable, because you've got to have a look at your fluorochromes, you've got to have a look at what you're trying to achieve, you've got to have a look at the filters on your instrument. But what I would say is, keep the compensation values to as low as possible. What I personally never do is to take compensation over the 80 or 90%, because I think that becomes unstable. It also means that if you go over 100%, then, quite frankly, your fluorochrome you are measuring in one PMT and you're trying to bring it back through another PMT that you should have originally measured it in. So I would say we never use compensation over 100%, and we just try and get everything as low as possible. But, again, it's incumbent on you to actually make sure that you have that as low as possible, and put some thought into your experiment so that you keep everything as low as possible. But 35% compensation is fine.

GP:    One of the other questions: How could we know when we prepare a good compensation? I think a picture says a thousand words, and I think certainly some of the experimental, the actual experimental data that Ian showed illustrated the point quite clearly. It's a case of having the correct controls in there and knowing what your individually stained cells look like, and making sure that when you have your combination of antibodies you essentially get the same data, so with regards to the percent positive. Also, generally speaking, that you have clear differentiation in your single stained cells between the two channels which you're looking at. Broadly speaking, things basically work in squares, and if you have a square type imagery then that would generally indicate that you've got the correct compensation settings. But, again, it comes back to having the crucial controls in all of the experiments, and understanding how your instrument works.

ID:    So the next question is: Is it true that brighter fluorochromes always have lower spectral overlap than dimmer fluorochromes? No, it's not, but what perhaps this question eludes to is that if you are using dimmer fluorochromes to set your compensation, then that gives you a greater margin for error. The brighter the fluorochrome, the more accurate you can be in terms of adjusting your negative population mean to your positive population mean. The brighter and the larger the numbers you're dealing with, will enable you to get that much more correct at the end of the day, than if you're using a dimmer fluorochrome, so that's not correct.

GP:    Another question is: Will the tandem dye that its molecule is supposed to be bigger, hamper the dye to permeate into the cells? So, essentially, the tandem dyes are clearly two fluorophores sort of conjugated covalently linked together, and the question is will that hamper their internalization into cells? Well, the answer to that is no it won't, because the size of these molecules is not going to be limited in that kind of experiment.

Slightly off this question is another issue, really, which relates to whether tandem dyes are more suitable or less suitable for intracellular experiments? So does the process of fixation and permeabilization have any effect on the functionality of these tandem dyes in these experiments? There are certain of the tandem dyes which don't perform particularly well in intracellular experiments, and an example would be PeCy7. But, I mean, for most of the other dyes it's, again, multifactorial, it very much depends on the cell which you're analyzing in the experimental setup. So, again, it's a case of understanding the system and doing some preliminary experiments to ensure that the experimental setup is fit for purpose.

Another question along those lines is compensation or the compensation requirements, is that influenced by the presence of fixed cells? Well, no, it shouldn't be, because the compensation is relating to the fluorescent characteristics, not the status of the cells. Although, clearly if you have an experimental setup, as Ian eluded to in one of his slides, that if you store cells and don't store them correctly, then you may get some degradation in the tandem dyes, and you may start to see compensation issues. But, again, if the experiment is done properly and the cells are stored appropriately, then you shouldn't really have any problems with that.

ID:    It's not surprising that we have some questions about this, and one question was: Why do we have many problems with tandem dyes, and how to solve the problems? The solution is easy, treat them with care. I tend not to want to start putting any sort of fixative on any experiments that I've got tandem dyes on, but there's a broad range of tandem dyes. You go from PerCPCy5.5 which almost acts like a fluorochrome in its own right, it's very, very stable indeed. Then perhaps on the other end of the scale you've got the one that we have a few issues with is PeCy7, but even PeCy7, if it's treat correctly, if you don't leave it out on the bench for too long, if you analyze the samples fairly quickly, i.e. within a couple of hours of preparing the sample, then there's not really a problem. So I would say tandem dyes have acquired this not very nice reputation, but it's unjustified. A lot of the tandem dyes are absolutely fine, you very often have to use them, but also always think of the alternatives. 

There are people that still are not thinking out of the box, and just going for the same old dyes all the time. Think in terms of Alexa 532, which is an excellent dye if you have a green laser. If you have a UV laser think in terms of Alexa 350, it's an excellent dye, yet, nobody seems to use it commonly, and it's a very, very good phenotyping dye. So think out of the box, there's a lot of options out there that you can go for, and sometimes you can get away from using tandem dyes. But also consider what sort of excitation your dyes have, i.e. single or dual lasers, as Graham said earlier, because the dye vector, if you're looking for dual positive, weakly dual positive cells is a big consideration. That is the major consideration, rather than trying to avoid tandems I think in multicolor work.

GP:    Then a question is regarding the controls: Do you use isotype control or fluorescence minus one to define the negative population? Again, to a certain extent, it really depends on your experimental setup. Personally, we tend to use both, because they provide different types of information. But just to reiterate what Ian was saying earlier on, is that if you are using isotype controls, then you need to make sure that you're using an isotype control which is at the same concentration as the antibody which you're using. Preferably one which is being purchased from the same supplier as your primary antibody, because you need to be sure that the conjugation system is exactly the same and so you're comparing like with like. The fluorescence minus one is also a good test, because it also allows you to see to what effect the sort of antigens, or the antibodies around the one of interest are influencing your target population. This is especially the case if you're looking at weakly-expressed antigens or the induction of weakly-expressed antigens, you need to be sure that you're not actually picking up a signal that should have been compensated out due to your multicolor panel.

So, again, for a lot of these questions there aren't any black and white answers, but it does come down to understanding the limitations of the experiment, and making sure that all of the controls that you need to be 100% confident of the interpretation of the results are in place.

ID:    A question about what are mass slides, on the three-color example, why the CD3 and CD8 positive cells were set nearly off-scale? Well, it doesn't really matter where you set them as long as you know where your negatives are, and the demarcation between the positives and negatives are, is good. So, yeah, I mean, that's sort of fairly arbitrary. If aesthetically you prefer them lower down, then that's fine, but remember if you place those cells lower down, if there's any weak positives that are just emerging from your negatives, you may lose those. So the actual PMT settings should be based on the actual maximum expression of your antigens, so that you can see absolutely all positive events. Now, if this case they were off-scale and you were only interested in the very bright events, you could place them anywhere you wanted to, because it wouldn't affect the results. But if you were looking at dim CD3 or dim CD8 events, then be very careful when you actually reduce your PMT voltages, because you may lose some of those dim events.

GP:    Thanks Ian. I'll just turn the session back to Lucy, but just on behalf of Ian and myself, I'd just like to thank you for attending the webinar today. Unfortunately, we don't have time to answer all of the questions which have come in, but we have received them and we will address those offline as soon as possible. Lucy?

Thanks. So I would like to extend our thank you to Ian, Graham and Ken for presenting this webinar, and on behalf of Abcam I would like to thank you all for attending as well. As Graham mentioned, for those whose questions were not answered, we will be contacting you shortly with a response. This webinar is also available for download as a PDF file, so when you log-off from the webinar you will be redirected to a webpage where this can be found, along with information about the special webinar promotions. Also, if you have any questions about what's been covered today, or if you have any scientific enquiry, please don't hesitate to contact our scientific support team at and they'll be very happy to help you. Finally, we hope you found this webinar useful and informative, and we hope to welcome you to another webinar in the future. Thank you again for listening, and good luck with your research!

Sign up