Online fingerprinting on Zeiss 510 META basic questions

[Anders Lunde]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Dear Confocal List,

Im a PhD student trying to do some colocalization anlysis on multiple
transcription factors in mouse neural tissue using uimmunofluorescence. To
get as clean data as possible I want to utilize the "online fingerprinting"
capabillity of our Zeiss 510 META confocal microscope.

I understand that the first step is to generate reference samples for each
fluorophore. So far so good. But what I am confused about is that in most
tutorials etc. it seems that online fingerprinting is explained in samples
where you only use one laser-line to excite all your fluorophores. For
example: I plan on doing online fingerprinting with Hoechst, Alexa 488,
Cy3, and Cy5. There is really no practical way to use the same laser to
exite both Hoeckst and Cy5. So optimally I would like to use 4 different
lasers in sequential scanning for the 4 different fluorophores. Is this
compatible with online fingerprinting? In my mind I guess this means that
for all reference samples I have to obtain reference spectra of each
fluorophore with all 4 lasers? So that means 4*4=16 reference spectra (+
background and autofluorencese references)? Moreover, I want to do this on
Z-stack samples, but that should be possible, right?


Another question: While it is important to use the same dichroic mirrors
and settings for the reference samples and real sample imaging, this will
not be possible if I want to image 4 fluorophores, because (as far as I
remember) the dichroic mirror is only optimized for 3 wavelengths so I have
to switch when going from Hoechst to, say, Cy5. However, this issue might
be solved if it turns out that I only need to generate 4 reference spectra,
and not each for each laser line (4*4=16), due to me misunderstanding the
concept properly.

Lastly, while it is important to use the same dichroic mirrors and other
settings for the reference samples and real sample imaging, can digital
offset and detector gain be adjusted when imaging real samples? If not it
might be difficult to get the grey values withing optimal ranges as there
is some natural variability between samples.

Best regards.

[Julio Vazquez]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hi Anders,

From our experience with a 510 META a few years ago, I would say that the benefit of online fingerprinting would be primarily for separating two labels that are too close spectrally to be separated with the conventional emission filters. In your case, I have a strong feeling that you will be better off collecting your four channels the normal way on the standard detectors. These will probably give you a better signal to noise, and therefore better data. You can always verify the registration of the channels with some fluorescent beads. In terms of image quality (noise), normal PMTs will probably be better. The META with online fingerprinting might have a benefit with regard to channel registration if you could image your four channels at once (in a single track), but if you need to multi-track, then the benefit is gone.

More specifically to your point, we haven't done unmixing with multiple laser lines. You could either do it in a single track, four channels at once ( so you would need to collect spectra for each dye with all lasers ON), but you would  need to block the laser lines and collect rather narrow bands (for example, if you have a 543 laser, your "green" channel will be pretty narrow). You seem to say you don't have a four line dichroic, so this means you would need to do two tracks at a minimum. How you combine them would depend on your choice of dichroics. You could maybe do DAPI/Cy3 and 488/Cy5, so that laser lines and emission bands don't overlap too much, and you would collect four spectra, one for each dye, under the conditions that you will be imaging them (for example, spectrum of Cy3 with 405 and 543 laser ON, if you will be imaging DAPI and Cy3 at the same time). I don't remember how the unmixing would work in multi track mode. But again, I see no obvious reason why you wouldn't just image your samples in the normal (emission filter/standard P<MT) mode. The imaging will be much simpler, and the data quality probably better.


Julio Vazquez
Fred Hutchinson Cancer Research Center
Seattle, WA 98109

http://www.fhcrc.org/en.html

--



On Jul 31, 2014, at 2:47 PM, Anders Lunde wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> Dear Confocal List,
>
> Im a PhD student trying to do some colocalization anlysis on multiple
> transcription factors in mouse neural tissue using uimmunofluorescence. To
> get as clean data as possible I want to utilize the "online fingerprinting"
> capabillity of our Zeiss 510 META confocal microscope.
>
> I understand that the first step is to generate reference samples for each
> fluorophore. So far so good. But what I am confused about is that in most
> tutorials etc. it seems that online fingerprinting is explained in samples
> where you only use one laser-line to excite all your fluorophores. For
> example: I plan on doing online fingerprinting with Hoechst, Alexa 488,
> Cy3, and Cy5. There is really no practical way to use the same laser to
> exite both Hoeckst and Cy5. So optimally I would like to use 4 different
> lasers in sequential scanning for the 4 different fluorophores. Is this
> compatible with online fingerprinting? In my mind I guess this means that
> for all reference samples I have to obtain reference spectra of each
> fluorophore with all 4 lasers? So that means 4*4=16 reference spectra (+
> background and autofluorencese references)? Moreover, I want to do this on
> Z-stack samples, but that should be possible, right?
>
>
> Another question: While it is important to use the same dichroic mirrors
> and settings for the reference samples and real sample imaging, this will
> not be possible if I want to image 4 fluorophores, because (as far as I
> remember) the dichroic mirror is only optimized for 3 wavelengths so I have
> to switch when going from Hoechst to, say, Cy5. However, this issue might
> be solved if it turns out that I only need to generate 4 reference spectra,
> and not each for each laser line (4*4=16), due to me misunderstanding the
> concept properly.
>
> Lastly, while it is important to use the same dichroic mirrors and other
> settings for the reference samples and real sample imaging, can digital
> offset and detector gain be adjusted when imaging real samples? If not it
> might be difficult to get the grey values withing optimal ranges as there
> is some natural variability between samples.
>
> Best regards.

[Anders Lunde]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Dear Julio,

Thank you for your answer. As you say the imaging is running relatively ok
in normal mode. But I have had some problems with high background,
autofluorescence, and possibly excessive bleed through due to large
difference in concentration of fluorophores, and I want to examine
posibillities for eliminating this. I also use a ImageJ plugin to analyze
the final images, and the plugin does much better with a better signal to
noise ratio (which I hope I can get by removing some background
autofluorescece).

By blocking laser lines, do you mean that the bins (10nm) that are directly
illuminated by a laser should be turned off?

And yes, I mean that I dont have a 4 line dichroic mirror.


On Fri, Aug 1, 2014 at 12:56 AM, Julio Vazquez <[hidden email]> wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> Hi Anders,
>
> From our experience with a 510 META a few years ago, I would say that the
> benefit of online fingerprinting would be primarily for separating two
> labels that are too close spectrally to be separated with the conventional
> emission filters. In your case, I have a strong feeling that you will be
> better off collecting your four channels the normal way on the standard
> detectors. These will probably give you a better signal to noise, and
> therefore better data. You can always verify the registration of the
> channels with some fluorescent beads. In terms of image quality (noise),
> normal PMTs will probably be better. The META with online fingerprinting
> might have a benefit with regard to channel registration if you could image
> your four channels at once (in a single track), but if you need to
> multi-track, then the benefit is gone.
>
> More specifically to your point, we haven't done unmixing with multiple
> laser lines. You could either do it in a single track, four channels at
> once ( so you would need to collect spectra for each dye with all lasers
> ON), but you would  need to block the laser lines and collect rather narrow
> bands (for example, if you have a 543 laser, your "green" channel will be
> pretty narrow). You seem to say you don't have a four line dichroic, so
> this means you would need to do two tracks at a minimum. How you combine
> them would depend on your choice of dichroics. You could maybe do DAPI/Cy3
> and 488/Cy5, so that laser lines and emission bands don't overlap too much,
> and you would collect four spectra, one for each dye, under the conditions
> that you will be imaging them (for example, spectrum of Cy3 with 405 and
> 543 laser ON, if you will be imaging DAPI and Cy3 at the same time). I
> don't remember how the unmixing would work in multi track mode. But again,
> I see no obvious reason why you wouldn't just image your samples in the
> normal (emission filter/standard P<MT) mode. The imaging will be much
> simpler, and the data quality probably better.
>
>
> Julio Vazquez
> Fred Hutchinson Cancer Research Center
> Seattle, WA 98109
>
> http://www.fhcrc.org/en.html
>
> --
>
>
>
> On Jul 31, 2014, at 2:47 PM, Anders Lunde wrote:
>
> > *****
> > To join, leave or search the confocal microscopy listserv, go to:
> > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> > Post images on http://www.imgur.com and include the link in your
> posting.
> > *****
> >
> > Dear Confocal List,
> >
> > Im a PhD student trying to do some colocalization anlysis on multiple
> > transcription factors in mouse neural tissue using uimmunofluorescence.
> To
> > get as clean data as possible I want to utilize the "online
> fingerprinting"
> > capabillity of our Zeiss 510 META confocal microscope.
> >
> > I understand that the first step is to generate reference samples for
> each
> > fluorophore. So far so good. But what I am confused about is that in most
> > tutorials etc. it seems that online fingerprinting is explained in
> samples
> > where you only use one laser-line to excite all your fluorophores. For
> > example: I plan on doing online fingerprinting with Hoechst, Alexa 488,
> > Cy3, and Cy5. There is really no practical way to use the same laser to
> > exite both Hoeckst and Cy5. So optimally I would like to use 4 different
> > lasers in sequential scanning for the 4 different fluorophores. Is this
> > compatible with online fingerprinting? In my mind I guess this means that
> > for all reference samples I have to obtain reference spectra of each
> > fluorophore with all 4 lasers? So that means 4*4=16 reference spectra (+
> > background and autofluorencese references)? Moreover, I want to do this
> on
> > Z-stack samples, but that should be possible, right?
> >
> >
> > Another question: While it is important to use the same dichroic mirrors
> > and settings for the reference samples and real sample imaging, this will
> > not be possible if I want to image 4 fluorophores, because (as far as I
> > remember) the dichroic mirror is only optimized for 3 wavelengths so I
> have
> > to switch when going from Hoechst to, say, Cy5. However, this issue might
> > be solved if it turns out that I only need to generate 4 reference
> spectra,
> > and not each for each laser line (4*4=16), due to me misunderstanding the
> > concept properly.
> >
> > Lastly, while it is important to use the same dichroic mirrors and other
> > settings for the reference samples and real sample imaging, can digital
> > offset and detector gain be adjusted when imaging real samples? If not it
> > might be difficult to get the grey values withing optimal ranges as there
> > is some natural variability between samples.
> >
> > Best regards.
>

[Julio Vazquez]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Yes, you get fairly strong reflections over the parts of the spectrum that overlap with the laser lines (at least we did with our confocal). However, if I remember correctly, in the case where you use multiple excitation lines in the same track, these will be collected (you can select the band for acquisition, but can not turn OFF individual bins during image acquisition, although you can tell software to ignore them later on). This is why it is preferable to do the online fingerprinting with a single excitation line, so you don't have to worry about this. If you decide to do it with four tracks, i.e. one per channel, then you would need four spectra (one for each dye acquired with its own specific laser line), and you may also need to collect spectra for the autofluorescence/background, so maybe four more spectra (the background may be different for different excitations). Now my memories of the 510 and linear unmixing are a bit rusty, so I can't be of much help for the actual procedure of collecting four channels in multi-track mode with online fingerprinting. For instance, I don't remember if the Zeiss software has a direct way of handling multi-track unmixing, but I am guessing that there should be no special issue with treating each dye separately. For example, you can use 488 excitation and extract the Alexa 488 signal from the background, and repeat this for all other dyes. Then you rebuild your four channel image from the extracted channels. However, it may be possible to do that a bit more directly. For that, depending on where you are located, you may be able to get a hold of one of Zeiss' applications specialists, and they could guide you through the procedure, or you could contact their support in NY if you're in the US (1-800-509-3905). They also  have some info here:

http://zeiss-campus.magnet.fsu.edu/articles/spectralimaging/index.html


Julio




On Jul 31, 2014, at 4:12 PM, Anders Lunde wrote:

> Dear Julio,
>
> Thank you for your answer. As you say the imaging is running relatively ok
> in normal mode. But I have had some problems with high background,
> autofluorescence, and possibly excessive bleed through due to large
> difference in concentration of fluorophores, and I want to examine
> posibillities for eliminating this. I also use a ImageJ plugin to analyze
> the final images, and the plugin does much better with a better signal to
> noise ratio (which I hope I can get by removing some background
> autofluorescece).
>
> By blocking laser lines, do you mean that the bins (10nm) that are directly
> illuminated by a laser should be turned off?
>
> And yes, I mean that I dont have a 4 line dichroic mirror.

[Wendy Salmon]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Dear Anders,

This is a common type of sample challenge I see in my core. Based on my experience, I think your time will be better spent optimizing your sample preparation rather than figuring out the linear unmixing. As you are experiencing and Julio states, unmixing is often more challenging than it sounds. Additionally, by splitting the signal into the small bins for the spectrum you make a huge sacrifice in SNR.

What type of sample optimization have you done? Spending a week on sample optimization can save you weeks of instrument and processing time as well as make conclusions more obvious.

In case you haven't already tried them, here are a few adjustments (in order) that help my users who have similar problems:

Autofluorescence:
There are a lot of previous discussions on autofluorescence in the list archive so I won't go into details, but the two main things to consider are fixation --fresh paraformaldehyde versus formalin is transformative-- and clearing or quenching agents

Bleedthrough:
Ideally your sample will have similar intensities in each channel. If you are staining with antibodies or dyes, there are a couple strategies to achieve this:
- First, for channels that are bright, play the "fluorescence limbo"--see how low you can go.
- If that is not possible (such as if the bright channel is a fluorescent protein and the dim channel is low abundance), you can try amplifying a dim antibody with TSA amplification (note: not good for quantifying intensity, which is a bag of worms unto itself).

Finally, if adjusting the staining conditions and autofluorescence isn't sufficient, you can measure the bleedthrough in standard imaging (using single-channel samples) and correct for it in the post processing. For more details on this, check the archives or let us know.

Best,
Wendy

Wendy Salmon
Light Microscopy Specialist
Whitehead Institute for Biomedical Research
W.M. Keck Imaging Facility
9 Cambridge Center, Rm 447
Cambridge, MA 02142
------------------------------

Date: Fri, 1 Aug 2014 01:12:13 +0200
From: Anders Lunde < [hidden email] >
Subject: Re: Online fingerprinting on Zeiss 510 META basic questions

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy 
Post images on http://www.imgur.com and include the link in your posting.
*****

Dear Julio,

Thank you for your answer. As you say the imaging is running relatively ok
in normal mode. But I have had some problems with high background,
autofluorescence, and possibly excessive bleed through due to large
difference in concentration of fluorophores, and I want to examine
posibillities for eliminating this. I also use a ImageJ plugin to analyze
the final images, and the plugin does much better with a better signal to
noise ratio (which I hope I can get by removing some background
autofluorescece).

By blocking laser lines, do you mean that the bins (10nm) that are directly
illuminated by a laser should be turned off?

And yes, I mean that I dont have a 4 line dichroic mirror.


On Fri, Aug 1 , 2014 at 12:56 AM, Julio Vazquez < [hidden email] > wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy 
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> Hi Anders,
>
> From our experience with a 510 META a few years ago, I would say that the
> benefit of online fingerprinting would be primarily for separating two
> labels that are too close spectrally to be separated with the conventional
> emission filters. In your case, I have a strong feeling that you will be
> better off collecting your four channels the normal way on the standard
> detectors. These will probably give you a better signal to noise, and
> therefore better data. You can always verify the registration of the
> channels with some fluorescent beads. In terms of image quality (noise),
> normal PMTs will probably be better. The META with online fingerprinting
> might have a benefit with regard to channel registration if you could image
> your four channels at once (in a single track), but if you need to
> multi-track, then the benefit is gone.
>
> More specifically to your point, we haven't done unmixing with multiple
> laser lines. You could either do it in a single track, four channels at
> once ( so you would need to collect spectra for each dye with all lasers
> ON), but you would need to block the laser lines and collect rather narrow
> bands (for example, if you have a 543 laser, your "green" channel will be
> pretty narrow). You seem to say you don't have a four line dichroic, so
> this means you would need to do two tracks at a minimum. How you combine
> them would depend on your choice of dichroics. You could maybe do DAPI/Cy3
> and 488/Cy5, so that laser lines and emission bands don't overlap too much,
> and you would collect four spectra, one for each dye, under the conditions
> that you will be imaging them (for example, spectrum of Cy3 with 405 and
> 543 laser ON, if you will be imaging DAPI and Cy3 at the same time). I
> don't remember how the unmixing would work in multi track mode. But again,
> I see no obvious reason why you wouldn't just image your samples in the
> normal (emission filter/standard P<MT) mode. The imaging will be much
> simpler, and the data quality probably better.
>
>
> Julio Vazquez
> Fred Hutchinson Cancer Research Center
> Seattle, WA 98109
>
> http://www.fhcrc.org/en.html