Setting up ‘Sensitized Emission’ based FRET experiments on Zeiss LSM780 confocal microscope

[Credits: Manoj V Mathew, Centre for Cellular and Molecular Platforms (CCAMP), Bangalore, India.
Experiments were conducted at the Central Imaging and Flow Cytometry Facility (CIFF), National Center for Biological Sciences (NCBS), Bangalore, India.]

Optical microscopes are resolution limited. However, there are ways to obtain sub-resolution information or measure sub-resolution distances in resolution limited images. One such technique is FRET.

FRET stands for Förster Resonance Energy Transfer. It involves two sets of molecules-one called 'Donor' and the other called 'Acceptor'. When a 'Donor-Acceptor' pair of molecules is in close proximity to each other the 'Donor' molecule can transfer its energy to the 'Acceptor' molecule. 

This happens non-radiatively via a dipole-dipole interaction mechanism. A 'Donor' molecule gets excited by excitation with its excitation wavelength. It transfers its energy to an 'Acceptor' molecule in its close proximity. The 'Donor' molecule then relaxes to the ground state without emitting a photon. The energy transfer results in the 'Acceptor'  molecule getting excited to its excited state and subsequently relaxing to its ground state radiatively (spontaneously emitting a photon).

These interactions are non-linearly (to the 6th power) dependent on the distance between the 'Donor' and the 'Acceptor'. They are significant only at very short distances between the 'Donor' and 'Acceptor', usually of the order of 10nm. By looking at FRET signals in diffraction-limited fluorescence images one can determine regions where the 'Donor-Acceptor' pairs are in very close proximity (<10nm) and measure the distance to an accuracy of the order of 1nm.

For FRET to occur there are some conditions the 'Donor' and 'Acceptor' molecules need to meet. For example, the 'Donor' emission spectrum should overlap with the absorption spectrum of the 'Acceptor'. One widely used FRET pair is ECFP and EYFP which satisfies this condition. Their spectra can be seen here.

There are many applications of FRET. One could study protein folding for example. If a FRET pair is tagged to the two ends of a protein, when it folds, the FRET pair gets closer to each other and exhibit stronger FRET signals. 

There are multiple ways to detect FRET. The following are some of them:

  1. Sensitized Emission: The 'Donor' is excited. The 'Donor' transfers energy to the 'Acceptor'. Acceptor fluorescence increases. 'Acceptor' fluorescence is measured.
  2. Acceptor Photobleaching: In a region of the sample, the Acceptor molecules are photobleached. This prevents the non-radiative transfer of energy from the Donor to the Acceptor. This results in an increase in 'Donor' fluorescence emission as more of the 'Donor' molecules in the region decays to the ground state radiatively. 'Donor' fluorescence before and after 'Acceptor' photobleaching is measured.
  3. Polarization Anisotropy: This technique is more useful for Homo-FRET ('Donor' and 'Acceptor' are the same species) measurements. Polarization anisotropy decreases with increasing FRET. Polarization anisotropy is measured.
  4. FRET by FLIM: This is probably the most sensitive FRET measurement technique. 'Donor' fluorescence lifetime decreases with increase in FRET. Donor fluorescence lifetime is measured most commonly using a Time-Correlated Single Photon Counting (TCSPC) based system.

For more details on the technique refer to the link

The idea of 'Sensitised Emission' would seem straightforward: Excite the 'Donor' and measure the fluorescence of the 'Acceptor'. However, there are few things we need to keep in mind. The very fact that we need the Donor's emission spectra to overlap with the absorption spectra of the Acceptor also means that there is inevitably some amount of:

  1. Cross Excitation: Some 'Acceptor' molecules will get excited directly by the 'Donor' excitation wavelength.
  2. Bleed Through: Some amount of 'Donor' emission will bleed into the 'Acceptor' channel.

This means that not all 'Acceptor' fluorescence is a result of FRET.  There are few other factors as well that can bias the results.

To have a more unbiased data from 'Sensitized Emission' based FRET experiments one needs to have a measure of these biasing factor. The following is the basic protocol:

a. Take three sets of images with the following sample sets:

  1. Sample with 'Donor' molecules only
  2. Sample with 'Acceptor' molecules only
  3. Sample with both 'Donor' and 'Acceptor' fluorophores

It is also possible to use only one sample if one can find regions within the sample with 1) only 'Donor',  2) only 'Acceptor' and 3) both 'Donor' and 'Acceptor'.

b. For each of these three sets take images using the following channels:

  1. Donor Channel: Excitation with 'Donor' excitation wavelength and collection in the 'Donor' emission spectrum.
  2. Acceptor Channel: Excitation with 'Acceptor' excitation wavelength and collection in the 'Acceptor' emission spectrum.
  3. FRET Channel: Excitation with 'Donor' excitation wavelength and collection in the 'Acceptor' emission spectrum.

In this write-up, I will describe how to use the Carl Zeiss LSM 780 confocal microscope to set up a 'Sensitized Emission' based FRET experiment using the 'ZEN' software.

Step 1: Setup the tracks and channels 

Setup three tracks with the following channels for sequential imaging with the following channels:

  1. Donor Channel
  2. Acceptor Channel
  3. FRET Channel

This is shown in Figure 1907/1.

Figure 1907/1: Setting up the three tracks 1) Donor Channel, 2) Acceptor Channel and 3) FRET channel

I used a fixed thin section of convallaria as the sample. While this is definitely not a good sample to demonstrate FRET, it is certainly enough for us to learn the software features. Convallaria is autofluorescent for excitations with most visible laser excitation sources. I used 488nm and 561nm excitation sources and spectral detector settings commonly used for Alexa 488 dye and Alexa 568 dye for the 'Donor' as well as 'Acceptor' and 'FRET' channels respectively. 

The channel configurations for the three tracks are shown in Figure 1907/2.

Figure 1907/2: Channel configurations for the three tracks- Left) Donor Channel, Middle) Acceptor Channel and Right) FRET channel

The following need to be kept in mind when configuring the channels:

  1. Use the same primary dichroic mirror for all channels
  2. Adjust the spectral detection bands to minimize bleed-through in the Donor and Acceptor channels.
  3. Use exactly the same spectral detection bands for both Acceptor and FRET channels

In addition, make sure that the Laser Power and PMT gain for each channel are set appropriately so as to avoid saturation of images. This is very important in any intensity quantification experiments. Saturation can be checked using the 'Range Indicator' feature in the software. It is also good to have the sequence mode as 'Change track every Frame' rather than 'Change track every Line'

Step 2: Acquire Images

Acquire three sets of images. Each set is imaged with one of the three samples using all the three tracks ( 1.only 'Donor',  2. only 'Acceptor' and 3. both 'Donor' and 'Acceptor'). This results in three image sets with 3 channels each (one each track).

In my case, I used a single sample and identified two regions 1) that had only 'Donor' and 2)only 'Acceptor' fluorescence. The whole field of view was used as the one with both 'Donor' and 'Acceptor'. For this, I first took an overview image with the track configuration described above. Figure 1907/3 shows the overview image.

Figure 1907/3: Overview image- Green: Donor Channel, Red: Acceptor Channel, and Blue: FRET channel

From the overview image, I identified regions with only green or red fluorescence. Using the 'Regions' feature the required region with only 'Donor' was marked for imaging. The 'Regions' setting is shown in Figure 1907/4.

Figure 1907/4: Regions setting

Make sure to uncheck bleach and check acquisition.

Check all the tracks and click 'Snap'. This will acquire an image of the ROI with only 'Donor' fluorophore in all the three channels. Save this image as 'Donor Only'

Similarly, use the 'Regions' feature to select a region with only 'Acceptor' fluorophores. Snap the image with all the three tracks and save the image as 'Acceptor Only'. 

The ROI images acquired using all the three tracks are shown in Figure 1907/5.

Figure 1907/5: ROI image with Donor Only (Left) and Acceptor Only (Right)

Now snap an image using all the three tracks of an ROI that has both 'Donor' and'Acceptor' fluorophores. I used the whole field of view. Call this image 'Donor and Acceptor'. In this case, the image will be same as the overview image I took. Now we have acquired three images each with three channels (one in each track). This is shown in Figure 1907/6.

Figure 1907/6: The three acquired images

Step 3: Set Donor and Acceptor parameters

Select the on the 'Donor Only' image and click on the 'FRET' tab on the side panel. This is shown in Figure 1907/7.

Figure 1907/7: FRET tab on the Visualization side panel

Note that the 'FRET' tab appears automatically under the following circumstances:

  1. Acquisition using 3 tracks (For 'Sensitized Emission')
  2. Acquisition using 2 tracks with one having a bleaching with time series (for 'Acceptor Photobleaching')

In the FRET GUI click on 'Parameters'. This is shown in Figure 1907/8.

Figure 1907/8: Setting Donor parameters

Do the following: 

  1. Select the Donor, Acceptor and FRET channels from the track assignments
  2. Click 'Donor' tab. This will calculate the Donor parameters Fd/Dd and Ad/Fd
  3. Select 'Donor' from the bottom drop down menu and click save. Save the parameters as 'Donor Parameters'

Similarly, select the 'Acceptor Only' image and click on the 'FRET' tab on the side panel. In the FRET GUI click on 'Parameters'. This is shown in Figure 1907/9.

Figure 1907/9: Setting Acceptor parameters

Do the following: 

  1. Select the Donor, Acceptor and FRET channels from the Track assignments
  2. Click 'Acceptor' tab. This will calculate the Acceptor parameters Fa/Aa and Da/Aa and Da/Fa.
  3. Select 'Acceptor' from the bottom drop down menu and click save. Save the parameters as 'Acceptor Parameters'

Kindly note that for the Convallaria sample I used has broad autofluorescence. The 'Donor' only and 'Acceptor' only regions may not exactly be that way. I selected them based on how they appeared in the overview image.

You can adjust the thresholds in the 'Thresholds' tab manually or select a background ROI in the image which will help the software calculate the thresholds. It would be advisable to define a background ROI (described below). The 'Thresholds' tab is shown in Figure 1907/10.

Figure 1907/10: 'Thresholds' Tab

For most applications set the 'Settings' tab to as shown in Figure 1907/11.

Figure 1907/11: 'Settings' Tab

Step 3: Analyze

Select the 'Donor and Acceptor' image and click on the 'Parameters' tab on the FRET GUI. Select 'Donor' from the drop-down menu. Load the 'Donor Parameters' file which was saved earlier. Then select 'Acceptor' from the drop-down menu and load 'Acceptor Parameters' file.

Subsequently, select the 'FRET' tab in the FRET GUI.  This is shown in Figure 1907/12. 

Figure 1907/12: 'FRET' tab

Select the Analysis method from the dropdown menu. There are three options available:

  1. Fc (Youvan)-"This method assumes that the signal recorded in the FRET channel is the sum of real FRET signal overlaid by donor crosstalk and acceptor signal induced by direct (donor) excitation. There is no correction for donor and acceptor concentration levels and as a result, the FRET values tend to be higher for areas with higher intensities". (Source: Zeiss LSM 880 ZEN 2 Manual)
  2. FRETN (Gordon) -"This method calculates a corrected FRET value and divides by concentration values for donor and acceptor. This method attempts to compensate for variances in fluorochrome concentrations but overdoes it. As a result cells with higher molecular concentrations report lower FRET values."(Source: Zeiss LSM 880 ZEN 2 Manual)
  3. N-FRET (Xia)-"This method is similar to the Gordon method with the difference that for concentration compensation the square root of donor and acceptor concentration is used. The resulting image is properly corrected for variances in the fluorochrome concentration." (Source: Zeiss LSM 880 ZEN 2 Manual)

Here I have used the N-FRET (Xia) method. 

Once a method is selected use the ROI tools in the FRET GUI to select a background ROI (a region with no signal). Check the 'Background' and 'Enabled' checkboxes for the background ROI. This helps the software to calculate the thresholds. 

Click the 'Analyze' button. This sted processes and analyzes the 'Donor and Acceptor' image to generate the FRET image. The FRET image will be displayed as shown in Figure 1907/13. 

Figure 1907/13: FRET image and FRET analysis results

Use the ROI tools in the FRET GUI to select an ROI on the image. The FRET analysis data for this ROI will be displayed below the image.

Kindly note again that I have used a highly autofluorescent sample which displays strong cross-excitation and bleedthrough. For the same reason, the calculated parameters might not be accurate enough as well. So please do not read too much into the FRET image and the results that are shown here. 

Performing FRAP experiments using Olympus FV3000 confocal microscope

[Credits:   Manoj V Mathew, Centre for Cellular and Molecular Platforms (CCAMP), Bangalore, India.
Experiments were conducted at the Central Imaging and Flow Cytometry Facility (CIFF), National Center for Biological Sciences (NCBS), Bangalore, India.]

FRAP stands for Fluorescence Recovery After Photobleaching. This is a very simple technique to study dynamics like diffusion rates in biological samples at the microscopic level. A better way to study dynamics would be at the single-molecule level. For such studies, FCS (Fluorescence Correlation Spectroscopy) might be a better tool. However, FCS is not easy, requires specialized hardware and some complex analysis modules. FRAP, on the other hand, can be easily performed on most confocal fluorescence microscopes. For a better understanding of the differences between FRAP and FCS kindly see (Link).

FRAP involves photobleaching a Region of Interest (ROI) in the biological sample and studying the recovery of fluorescence in that region over time. Since photobleached molecules cannot recover the ability to fluoresce, the only way recovery can happen is by the diffusion of non-bleached molecules from outside ROI. The average intensity of the ROI can be measured as a function of time and plotted as a graph. This can then be fitted onto standard diffusion models to compute parameters like diffusion rates. 

In this write-up, I will describe how to use the Olympus FV3000 confocal microscope for performing FRAP experiments.

The main interface of the FV3000 software is shown in Figure 1708/1.

Figure 1708/1: The main interface of the Olympus FV3000 confocal microscope software

Step 1: Image the sample

Use the software interface to generate an image of the sample. Adjust the laser power and PMT gain such that the image is not saturated but covers most of the dynamic range of the system. This can be ensured by using the HiLo mode in the image visualization window. I used a fixed thin section of convallaria as the sample. While this is definitely not a good sample to demonstrate FRAP, it is certainly enough for us to learn the software features. Convallaria is autofluorescent for excitations with most of the available visible laser excitation sources. I used 488nm excitation source and spectral detector settings commonly used for Alexa 488 dye for detection. 

In general, it is always recommended to use the 10% AOTF dampening (Laser ND filter set to 10%) to protect the detectors against excessive light exposure. This is especially important when using GaAsP detectors. However, when performing FRAP we need laser powers higher than 10%. Hence imaging should also be performed with the 10% AOTF dampening factor removed (Laser ND filter set to 'None'). In this case be extremely careful with the AOTF laser power setting and PMT gain setting to make sure you donot saturate the detectors.

Step 2: Enable the 'LSM Stimulation' and 'Synchronization' tabs

For performing FRAP experiments you need to enable the 'LSM Stimulation' and 'Synchronization' tabs in the software. Ths can be done by going to 'Tools Window' in the main menu of the software. This is shown in Figure 1708/2.

    Figure 1708/2: Enabling LSM Stimulation and Synchronization tabs from the Tools Window.

Once enabled they will appear within one of the tab containers in the interface as shown in Figure 1708/3.

Figure 1708/3: 'LSM Stimulation' and 'Synchronization' tabs added

Step 3: Set the 'LSM Stimulation' tab

Go to the 'LSM Stimulation' tab and set the photobleaching parameters. The 'LSM Stiumation' settings are shown in Figure 1708/4.

Figure 1708/4: 'LSM Stiumation' settings

The 'LSM Stimulation' tab provides the following:

  1. ROI tools: A number of tools to select one or more ROIs on the acquired image or the Live view
  2. Bleaching Dwell Time: The bleaching dwell time can be set different from the imaging dwell time. The image dwell time is entered in the 'LSM Imaging' tab. It might be useful to set a higher bleaching dwell time for effective bleaching.
  3. Bleaching Duration: You can set the bleaching duration to 'continuous'. In this case, the beaching continues till the user stops it. You can also enter a total bleach duration. For FRAP experiments it might be better not to use the continuous mode.
  4.  Bleaching Laser selection and powers: Bleaching is usually most effective with the same laser wavelength as the excitation wavelength of the fluorophore.  You would certainly need much more power for bleaching than what you use for imaging. But increasing the laser power much beyond the value where the fluorophore saturates is counterproductive. This is because at saturation, fluorophores achieve the maximum possible excitation-emission cycle rate. For photobleaching, most people simply blast the sample with maximum available laser power. This is not ideal. It would be great to perform a fluorophore saturation experiment first. Then set the bleaching laser power to about 10% more than what would saturate the fluorophore. After that play with the bleaching dwell time and total bleach duration for optimized bleaching. (Perform a few pilot bleach tests to optimize the parameters before commencing the actual experiment).

Step 4: Select one or more ROIs for bleaching and analysis

Once the "LSM Stimulation' parameters are set, select one or more ROIs on the image using the ROI tools in the 'LSM Stimulation' tab. Please note that selecting ROIs using the ROI tools on the visualization tab will not help.

I used the circle ROI tool to select a circular ROI. This is shown in Figure 1708/5.

Figure 1708/5: Selection of a single ROI on the image (labeled 1S)

Step 5: Set the Time Series

Set the Time Series from the 'Series' tab as shown in Figure 1708/6. 

Figure 1708/6: Setting the Time Series

The number of cycles and interval depend on how fast the fluorescence recovery is. If the recovery is fast, set the system to image as fast as possible ('Freerun'). It would also help to explore other ways of improving imaging speed like:

  1. Zooming into as small a regions as possible for imaging
  2. Decreasing imaging dwell time
  3. Decreasing the number of pixels (sometimes at the cost of resolution)

Please note that the total number of cycles set in the 'Time Series' also includes the pre-bleach cycles.

Step 6: Set the 'Synchronization' tab

Set the pre-bleach and post-bleach imaging cycles in the 'Synchronization' tab. This is shown in Figure 1708/7.

Figure 1708/7: Synchronization settings

Do the following:

  1. Base Method: Select 'LSM Imaging'
  2. Rest in stimulation: Select 'OFF'
  3. Stimulation Wait: Enter the number of required pre-bleach cycles. 


No: Post-Bleach cycles= Time Series Cycles - No: Pre-Bleach cycles

In my case: Time Series Cycles=30, Pre-Bleach Cycles=5, so Post-Bleach cycles=30-5=25.

Step 7: Run the photo-bleaching protocol

If only photo-bleaching is required and FRAP analysis is not required then go to the 'Normal' tab in the 'Aquire'' container and press 'Stimulation Start' as shown in Figure 1708/8.

Figure 1708/8: Photo-Bleaching only by pressing 'Stimulation Start' in 'Acquire' tab

Photobleaching will be performed on the selected ROIs as per the ROIs selected and settings in the 'LSM Stimulation' tab. A time series will not be run.

If FRAP analysis is required then, go to the 'Sync' tab in the 'Acquire' container and press 'Synchronization Start' as shown in Figure 1708/9.

Figure 1708/9: Sync Tab

The system will now run the time series: acquire the pre-bleach images, bleach the ROI/s and then acquire the post-bleach images as per the settings in the 'LSM Stimulation', 'Time Series' and 'Synchronization' tabs. The whole set will be saved as a time series. 

Note: Don't forget the mention the target folder for saving the 'Synchronization' files. This location is not mapped to be same as LSM acquisition ('Normal')  folder location.

The resultant image after bleaching and recovery is shown in Figure 1708/10.

Figure 1708/10: Resultant image post bleaching and recovery

Note: Since I used a fixed sample there is hardly any recovery (about 25 seconds post bleaching).

Step 8: Load the time lapse data into CellSens software for analysis

Open the acquired time series (with pre-beach and post-bleach) data in the CellSens software. This can either be done by opening the CellSens software and loading the file from the target folder or by right-clicking on the time lapse data on the acquisition software and clicking 'Export to CellSens'.

Once the time series data is opened in CellSens, we need to define two more ROIs in addition to the bleach ROI which is already defined and indicated in the time series data. The two additional ROIs are:

  1. Background: ROI in a region where there is no signal. This ROI would be used for background correction.
  2. Photobleaching Correction: ROI in a region which has signal but has not been bleached. This ROI will be used for correcting any bleaching that occurs during the image acquisition process after bleaching.

Note that the bleach ROI is already embedded in the time series data. But if you want to modify that or create a fresh one you can.

Make sure you select the appropriate ROI tools. This is shown in Figure 1708/11. 

Figure 1708/11: CellSens ROI selection tools for FRAP Analysis

Once the three ROIs (Bleach, Background, and Photo Bleaching Correction) are defined the time series data will look as shown in Figure 1708/12.

Figure 1708/12: Time series data with the 3 ROIs, Bleach Region (1S-Red), Background (ROI1-Blue) and Photobleaching Correction Region (ROI-2-Majenta) defined

Step 8: Plot the Intensity Profile

Click on the measurement tab on the top panel of CellSens. At the bottom of the tab, you will find tools for FRET and FRAP analysis. This is shown in Figure 1708/13.

Figure 1708/13: Tools for FRET and FRAP analysis

Click on the 'Intensity Profile' option. It presents a GUI shown in Figure 1708/14. Select the Time Series file location, Bleach ROI, and Background ROI and then press 'Execute'.

Figure 1708/14: Intensity profile GUI

This will display a background corrected intensity profile over time (both pre-bleach and post-bleach). This intensity profile as obatianed in my experiment shown in Figure 1708/15.

Figure 1708/15: Intensity profile over time

Note that you can choose to display the average, integral, maximum or minimum intensity on the intensity profile GUI. For the fixed sample I performed FRAP on, after bleaching while the fluorescence intensity drops very close to zero there is understandably no recovery of fluorescence over time.

Step 9: Perform FRAP analysis

You can now open the FRAP analysis GUI by clicking on the button on the top right corner of the intensity profile (Figure 1708/15) or use the option in the measurement tab (Figure 1708/13).

The FRAP analysis GUI is shown in Figure 1708/16. 

Figure 1708/15: FRAP Analysis GUI

As shown in Figure 1708/16  enter the following in the FRAP analysis GUI.

  1. Time Series file location
  2. Bleach ROI
  3. Single or double exponential fit. If the diffusion has only one component, use the 'Single Exponential Fit' and if the diffusion has two components select the 'DoubleExponential Fit'. For more complex diffusion scenario consider using some other fitting algorithm.
  4. Time series cycles over which the recovery data is fitted.
  5. Background ROI
  6. Photobleaching Correction ROI

The data fitting results are shown in the same GUI window. It also shows the raw and fitted data graphs. By pressing 'Execute' you can export the fitted data and results.


Confocal Microscopy II

In this write-up I will discuss some advanced topics related to confocal microscopy. 

Why do we need to use lasers in a confocal microscope?

Lasers have a lot of disadvantages when compared to using an incoherent light source like a mercury arc lamp.

  1. They are expensive
  2. They lase at a single wavelength, unlike an arc lamp which produces like 10 discrete spectral lines. 
  3. For multiple excitation wavelengths, we need multiple lasers and then make their beams colinear in a laser combiner. The lasers, the combiner, and associated electronics and controller could cost almost 25 to 30% of a confocal microscope's cost.
  4. They have more complex electrical and heat dissipation requirements
  5. Finally, the very nature of coherence is not good for imaging. Coherence creates unwanted interference patterns (speckles) resulting in non-uniform illumination.

So why lasers in a confocal microscope? Why not use in-coherent light sources like arc-lamps as excitation sources in confocal microscopy?

The answer lies in Figure 12, from our discussion on Köhler Illumination. Let's take a closer look at this figure again as shown in Figure 1567/1.


Figure 1567/1: Light from an extended source is not focused to a diffraction limited spot. An image of the source is created at the sample plane.

Incoherent light sources like a tungsten lamp or an arc lamp are not point sources of light. They are extended sources. If we consider a single point on the source, let's say at the center of it, it creates a diverging beam of light. This can very well be collimated by a collimating lens and then projected onto the back focal plane of the objective (see Figure 1567/1). The objective, in turn, will create a single focused diffraction limited spot on the sample plane. This is what we ideally need in a confocal microscope. 

But now all the other points on the extended light source which also create collimated beams of light after passing through the first lens.  But these individual collimated beams are not parallel to each other. See Figure 1567/1. After passing through the objective these collimated beams get focused at different spots, essentially creating an image of the source on the sample plane. So we do not have a single diffraction limited spot on the sample plane. In fact, a demagnified image of the filament/arc gap is created on the sample plane. The image is demagnified since the objective focal length is shorter than the focal length of the collimating lens and light is traveling from collimating lens towards the objective. 

In a point scanning imaging system like a confocal microscope, the resolution depends on how small the excitation spot is. Scanning an image of the light source in a raster pattern does not certainly help in achieving any resolution at all. 

So how do lasers help?

See Figure 1567/2. A laser by virtue of being coherent produces a single collimated beam of light. This single collimated beam of light is focused to a single diffraction limited spot. 

Figure 1567/2: A Laser produces a single collimated beam of light which is focused to a single diffraction limited spot.

For the objective, it would look like light coming into it from a single infinitesimally small point source at an infinite distance away from it. The objective is in-fact trying to create an image of this single point source, which ends up being a diffraction limited spot. It is in this sense that a laser is a point source of light. 

As discussed in my earlier write-up on confocal microscopy, modern confocal microscopes do not directly beam the laser light onto the objective back focal plane. Instead, a laser combiner combines laser beams from different laser sources into a single collinear beam of different wavelengths. This colinear beam is then launched into an optical fiber using a focusing lens. This is shown in Figure 1567/3. Light out of the fiber is diverging and is collimated by a collimating lens. This collimated beam is further passed through the scanning mirrors and expanded by the telescope formed by the scan lens and tube lens (microscope) combination. An expanded and collimated beam of about 1cm is projected onto the back focal plane of the objective. This helps slightly overfill the back aperture of the objective to engage the full numerical aperture of the objective.

Figure 1567/3: Excitation light path of a modern confocal microscope

It may, however, be noted that the collimating lens, scan lens, tube lens and objective combination is projecting an image of the tip of the core of the optical fiber onto the sample plane. A demagnified image of the tip of the core of the fiber is projected onto the sample plane.

Let us assume a magnification of 60X for the objective, tube lens combination, and a magnification of about 1.5 for the scan lens, collimating lens combination. So a total magnification of about 90X. Assuming a resolution limit of about 250nm and considering the fact that the image is demagnified,  a diffraction limited spot is only formed if the fiber diameter is less than 250x90=22.5micrometers.

This is the reason why modern confocal microscopes use the so-called single-mode fibers for launching the excitation light. Single mode fibers should ideally have a core diameter so small that they propagate only one mode of the laser (transverse modes are not propagated). The core diameters are usually less than 15 micrometers. This would qualify them to be single mode for optical communication applications (where this term is most commonly used) as they use much longer wavelengths there like 1.5 um. For wavelengths used in confocal microscopy (350nm to 800nm) these fibers may not single mode in its literal sense. However, they satisfy our core diameter criteria for achieving diffraction limited spot on the sample plane. 

There is an added advantage to launching excitation light into an optical fiber. If a diode laser which may not have a proper Gaussian beam profile, the near single mode optical fiber will cut-off a large number of higher order modes and the fiber output would be near Gaussian.

So why not launch the light from an extended source into an optical fiber and clean it up?

The extended source will create an image of the source at the fiber input and hardly any light will get coupled in! So we definitely need lasers. A laser with a slightly bad beam profile (slightly incoherent) would work but not something that is totally incoherent.

So what about the issue of coherent light creating speckles?

The point scanning confocal microscope is only detecting the intensity of emission light from the focal spot. The PMT or other single pixel photodetectors used in confocal microscopy donot measure the spatial distribution of intensity. So this is not an issue.

What is de-scanning?

Look at the block diagram of a confocal microscope as shown in Figure 33. The scan optics (scanning mirrors + associated lenses) is placed between the primary dichroic mirror and the objective. This means that in addition to the excitation light fluorescence emission passes through the scan optics as well. A lot the precious emission light is lost here. 

So why can't we place the scan optics upstream in the excitation path before the primary dichroic mirror and avoid emission light from passing through the san optics?

If what we need to do is only scan the focused laser beam across the sample plane this will certainly work. The problem arises when we pass the emitted light through the pinhole.

Lets us understand this with Figure 1567/4. The figure shows the block diagram of a non-descanned detection based scenario for a confocal microscope. Here the scan optics is placed between the excitation path, upstream of the dichroic mirror. The fluorescence emission does not pass through the scanning mirrors. 

Figure 1567/4: Scan optics in non-descanned configuration with both X and Y scan mirrors set at 0 degrees.

Figure 1567/4 shows one situation where the scan mirrors are not deflecting the light. Both the X and Y mirrors are set to 0 degrees deflection. In this scenario, the laser beam would ideally be focused exactly on the center of the field of view of the sample by the objective. The emitted light retraces the path of the excitation light and passes straight through the primary dichroic mirror and will strike the pinhole exactly at its center. The out of focus light will be blocked and the in-focus light will be passed through the pinhole. 

Now let us consider a situation where the X scanning mirror is deflecting the light towards the right at a certain angle. This situation is shown in Figure 1567/5.

Figure 1567/5: Scan optics in non-descanned configuration with Y scan mirror set at 0 degrees and X mirror deflected to the right by a certain angle.

Figure 1567/5 shows a situation where one of the scan mirrors is deflecting the light. The Y mirror is set to 0 degrees and X mirror is deflected off-axis to the right by a certain angle. In this scenario, the laser beam should be focused to a point to the right of the center of the field of view of the sample. The emitted light retraces the path and is deflected off-axis as well and after passing through the primary dichroic mirror does not hit the center of the pinhole. This means the pinhole is simply blocking most of the light, including the in-focus ones.

In a non-descanned configuration, as the excitation beam is scanned in a raster pattern,  the emission light gets scanned across the pinhole in a raster pattern as well. Light from the entire field of view is blocked from reaching the detector, except for the fluorescence emission coming from exactly the center of the field of view.

One way out of this problem would be to install a second set of scan optics in the detection path, before the pinhole. This second scan optics (descan optics) will have to be precisely synchronized with the first one in such a way that it cancels the deflection of the emitted light. This way the emitted light will remain stationary at the pinhole while the excitation light is being scanned across the sample plane. So the scan optics will scan the excitation light on the sample plane and the descan optics will descan and  keep the emitted light beam stationary at the pinhole.

It turns out that we donot need to two sets of scan optics. A single set of scan optics can do both scanning and descanning if it is placed in the path that is common for both excitation and emitted light (between the primary dichroic mirror and the objective). This is what modern confocal microscopes do. 

Lets us assume that the X scanner is deflecting the excitation light by +10 degrees. When the emitted light returns it retraces the path and is traveling in the opposite direction. So as it passes through the scan mirrors it is deflected -10 degrees. The net deflection would be +10 + -10 = 0 degrees. This is true for any deflection angle for both the X and Y axis scan mirrors. The net deflection of the emitted light in the descanned configuration is 0 degrees. The emitted light remains stationary at the pinhole.

A single set of scan optics hence performs both scanning and descanning.

It must be noted that the scan mirrors are continuously moving. There is a time delay between the excitation light leaving the primary dichroic mirror and the emission light arriving at the dichroic mirror. The mirrors would have moved slightly during this time. Note that the fluorescence lifetime is in nanoseconds and light travels at the speed of light. So compared to the speed of motion of the mirrors (even for the high-speed scanners) the fluorescence emission is arriving almost instantaneously after the excitation has left the primary dichroic mirror. The mirrors are practically at the same position. 



Confocal Microscopy

Fluorescence microscopes are excellent tools to generate contrast in images of biological samples. However, they suffer from a major problem when you are imaging samples even slightly thicker than the z resolution limit. The problem is the out-of-focus blur. This happens because the system is unable to capture light from precisely the focal plane of the objective. Some out of focus light (light from planes below and above the focal plane) inadvertently creeps in and introduces noise. This results in blurred images.

There are few ways to overcome this problem and get sharper, in-focus images. This helps the microscope to perform precise optical sectioning. The most widely used and probably the gold-standard in optical sectioning is confocal microscopy. 

How does the confocal microscope reduce out-of-focus-blur?

A very basic understanding can be had from the description below.

Let us take a laser that emits at the excitation wavelength of the fluorophore of interest. The laser beam is reflected by a dichroic mirror towards an objective. See Figure 29. The beam is a made to fill the back aperture of the objective. This creates a focused, diffraction-limited spot of the excitation light at the focal point of the objective. If there are fluorophores in this focal volume, they will get excited and will emit fluorescence isotropically. A cone of this emitted light (that travels in the backward direction) is collected by the objective. You can see that we are using the epi-illuminaiton configuration here. The objective sends this fluorescence emission light out in a collimated beam shaft (remember Figure 2). Here the focused laser spot creates something similar to a point source of light at the focal point of the objective. This means the beam comes out collimated and is directed towards the dichroic mirror.

Figure 29: Measurement of fluorescence intensity excited using a diffracted limited laser spot. Excitation light is shown in violet and emission light is shown in green.

The dichroic mirror is designed to reflect the excitation wavelength (from the laser) and transmit the emission wavelength (from the fluorophores in the sample). The dichroic mirror hence transmits this collimated beam shaft of emitted light. A tube lens further down focuses this light to its focal point. If we now place here an emission filter and a photodetector we can measure the intensity of emitted light from the focal spot of the objective. Note that the objective-tubelens combination is acting like an infinite tube length microscope and is projecting a magnified image of the focal spot of the objective onto the detector plane. 

The detector measures the intensity of the light emitted from the focal spot. However, we have out of focus light which also gets detected by the detector.

The solution to this problem is a pinhole. The pinhole should be kept exactly at the focal point of the tube lens. This is very important. The detector is placed after the pinhole. The detector used in most common implementations of a confocal microscope is a Photo Multiplier Tube (PMT).  This is shown in Figure 30.

Let us now consider what happens to the light coming from the focal spot (Figure 30). If the light comes from a point source placed at the focal spot (refer Figure 2) it comes out collimated through the back aperture of the objective. Now a collimated beam of light enters the tube lens. The tube lens focuses the light to its focal point. The pinhole is placed exactly here. Almost all of the emission light that originates from the focal spot passes through the pinhole and is detected by the PMT.

Figure 30: Confocal Microscope-Light Coming from the focal plane of the sample. Violet shows excitation light and green shows emission light.

Now let us consider the out of focus light. First, the light that comes from a point at a distance shorter than the focal length of the objective. Look at Figure 8 from left to right. If the light comes from a point source placed at a distance shorter than the focal length, the light exits the lens slightly diverging. This scenario for a confocal microscope is shown in Figure 31.

Note that when we use high NA objectives the out of focus light it is not produced over large distances away from the focal point but probably a couple of micrometers above and below the focal point. This is the range of distances over which the objective collects fluorescence light. Beyond these distances, the intensity of excitation is very low to excite any detectable fluorescence. Hence the divergence of the beam as the light exits the lens is not very large.

This divergence, however, is significant enough to have an influence on how the beam exits the tube lens as is seen in Figure 31. Lets us consider Figure 6. Look at this figure from right to left. For that matter, you will see the same thing in Figure 4 and Figure 5 when looking at them from right to left. (Remember lenses are reciprocal systems, a ray diagram is true in both directions).

When a diverging beam enters a lens, it gets focused at a distance that is longer than the focal length of the lens. Now this means the light is not focused all the way down when it hits the pinhole (Figure 31). The pinhole hence blocks the light coming from a point on the sample that is at a distance shorter than the focal length of the objective. 

Figure 31: Confocal Microscope-Light Coming from a plane above the focal plane of the sample. Violet shows excitation light and green shows emission light.

Lets us now consider the situation when the objective collects light from a point that is at a distance longer than the focal length of the objective. Look at Figure 6 again but this time look at it from left to right. You will see the same thing in Figure 4 and Figure 5 as well when looking at them from left to right. If the light comes from a point source placed at a distance longer than the focal length of the objective, the light exits the lens converging. Again, in a confocal microscope, this light is collected only from very short distances away from the focal point. So the light out of the objective is only slightly converging. This scenario is shown in Figure 32.

Now a converging beam is passing through the tube lens. Lets us consider Figure 8. Look at this figure from right to left. When a converging beam enters a lens, it gets focused at a point that is at a distance shorter than the focal length of the lens. Now this means the converging light in the confocal microscope is focused before the pinhole by the tube lens. After being focused as light propagates further it diverges and the pinhole blocks this light as well.

Figure 32: Confocal Microscope-Light Coming from a plane below the focal plane of the sample. Violet shows excitation light and green shows emission light.

So the combination of the objective, tube lens, and the pinhole blocks most of the out of focus light allowing only in-focus light to be detected by the PMT.

We have however collected intensity information from only a single diffraction limited spot. This does not create an image. To create an image, we need to scan the laser beam in a raster pattern and collect intensity information from many points across the XY plane. This creates a 2D array of intensity information stored in the memory of the computer. This array is then converted into an image.

How is the point scanning confocal microscope system configured?

The block diagram of one of most common implementations of a point scanning confocal microscope is shown in Figure 33. The diagram gives a more detailed description of the components involved in making a confocal microscope work. 

Figure 33: Block Diagram of a Point Scanning Confocal Microscope

The excitation light for a point scanning confocal microscope comes from a laser combiner. The laser combiner has a bank of lasers with different wavelengths. The different wavelengths are required to excite fluorophores with different excitation wavelengths.  These wavelengths are combined into a single collinear beam using a set of mirrors and dichroic mirrors. This combined laser beam is passed through an Acousto Optic Tunable Filter (AOTF). AOTF is controlled electronically through the software that runs the confocal microscope. The AOTF serves two functions:

  1. Select a wavelength or a set of wavelengths for excitation
  2. Control the intensity of each wavelength independently

The light out of the AOTF is launched into a single mode optical fiber. This arrangement helps the laser combiner to be separated from the scan head and imparts flexibility in placing the laser combiner with respect to the scan head. The other end of the optical fiber is connected to the scan head.  The excitation light diverges out as it exits the optical fiber. A collimating lens is used to collimate this light. The collimated output is directed towards the dichroic mirror.

The dichroic mirror reflects the excitation light towards the scanning mirrors. The scanning mirrors scan the excitation wavelength/s in X and Y dimensions in a raster pattern. The scanning mirrors are controlled electronically through the sofware of the confocal microscope.

Before directing towards the objective, the light excitation light out of the scanning mirrors is passed through a set of two lenses called the scan lens and the tube lens. The tube lens sits in the body of the microscope and the scan lens inside the scan head. These two lenses have the following functions:

  1. Together they act like a telescope to expand the laser beam from about few mm to about 1 cm. This is to ensure that the back aperture of the objective is overfilled. This, in turn, ensures that the full Numerical Aperture of the objective is available and maximum possible resolution is attained.
  2. Relay the image of the scanning mirrors to the back aperture of the objective. This ensures a proper scanning of the excitation laser spot across the defined ROI on the sample plane.

As described before the objective focuses the excitation light onto the sample. The fluorescence emission (in the backward direction) is collected by the objective. The fluorescence emission traces back the path taken by the excitation light all the way till the dichroic mirror. The dichroic mirror transmits the emission light. A tube lens (of the scan head) focuses the emission light onto the pin hole. The light out of the pinhole is filtered by an emission filter and directed towards the PMT.

The PMT produces a current that is proportional to the number of emission photons detected by it. A Trans Impedance Amplifier (TIA) is used to convert this current into a voltage and to amplify the voltage to appropriate levels. An Analog to Digital Converter (ADC) is used to convert the analog voltage signals into digital signals. Computers can only read digital signals. The intensity information in digital format is read by the computer and it converts this information into an image.

How does a confocal microscope generate an image?

Widefield fluorescence microscopes use imaging detectors like CCD or sCOMOS cameras for generating images. Cameras have an array of detectors in them which corresponds to the pixels we see in the images they generate. If a camera is specified as 5 Mega Pixel, then the camera chip has 5x10^6 independent detectors. When the lens system of the widefiled microscope projects an image onto the detector plane, such a camera can generate a digital image with 5x10^6 pixels.

However confocal microscopes use single pixel detectors like PMTs. Single pixel means they donot have an array of detectors but only a single detector. So how are these detectors generate images?

The fact is in a confocal microscope the image is not generated by the detector. It is generated by a computer that has a certain amount of memory and processing power. The image is constructed by the computer using a set of intensity information sequentially gathered from the PMT and sequentially stored in the memory of the computer. 

The laser spot is scanned in the raster pattern and spot moves across the XY plane. As the laser spot steps from one spot to next, the output of the PMT changes every set time period. This time period is the Dwell Time-the time it takes the scanning system to move the laser spot from one pixel to the next on the sample plane. The ADC is programmed to take a reading of the PMT voltage (through TIA) every time period. So every successive time period (dwell time) the ADC spits out a digital value. This value corresponds to the intensity of fluorescence emission from the position of the laser spot at that particular time point.

Before you start imaging you need to tell the confocal software how many pixels you need in your image. If you say you need 512x512 pixels, the computer allocates a memory location in the form of an array of size 512x512. Each memory location has an address ([0,0] to [511,511]). Let us assume that an ROI is scanned. The scanning system positions the laser on the first spot (let's say on the top left of the ROI). The ADC reads the PMT output and generates a digital value. This corresponds to the intensity of fluorescence from the first spot. The computer stores this value in the array at memory location [0,0]. The laser spot is then moved right to the next spot. The ADC reads the value and this is stored in memory location [0,1], then the next one onto [0,2] and so on till memory location [0,511] is filled. This completes scanning of the first row. Now the laser spot is moved to the next row and the process repeats. Starting from [1,0] the memory locations are progressively filled till [1,511].

So as the laser beam progressively moves across the sample plane the memory locations get progressively filled up. This happens until the laser spot reaches the last point in the ROI on its bottom right. This fills the last cell in the array with address [511,511] and created a completely filled memory array. This array of intensity information is called a bit map. This bitmap is converted into an image in a given format like JPEG, PNG, TIF etc. or custom formats proprietary to confocal microscope manufacturers. This image can be displayed on a computer monitor or stored.

This whole process requires precise synchronization between the scanning system and the ADC. This is done by the control electronics which works under the command of the confocal software. Modern confocal microscopes donot step the laser spot across the XY plane. The laser spot instead is moved continuously in a raster pattern. The ADC simply reads periodically (every dwell time) from the PMT.

How does a confocal microscope generate 3D images?

To generate a 3D image, a 3D array is defined with the number of pixels in X, Y and Z directions specified. After the image of one plane is scanned, the relative position between the objective and the sample is changed by the Z step size. This is done by moving the objective or moving the sample placed on a Z stage. Z step size is calculated from the number of pixels defined in the Z direction and thickness of sample needed to be imaged. This process is repeated for multiple planes and a raster scan performed and an image generated for each plane. The number of image planes equals the number of pixels in the Z direction. The end result is a 3D array of intensity information. This array can be volume rendered or projected onto a single plane for visualization. 


Basic Fluorescence Microscopy

Biological samples are quite complex structures with a lot of stuff in them. Looking for something very specific (for example looking at neurons inside C.elegans larvae) could be almost impossible many a times. Bright Field microscopy cannot reveal this.

Fluorescence microscopy comes in very handy in such situations. Here a component of interest is specifically labeled with a fluorescent molecule called a fluorophore (such as green fluorescent protein (GFP), DAPI etc:). Then by observing the fluorescence of the label the component of interest can be observed. Fluorescence Microscopy collects light only from the component of interest and not from other structures surrounding it (by filtering in the fluorescence wavelengths and filtering our all others).

Fluorescence is a quantum mechanical process that can be explained using the Jablonski Diagram shown in Figure 27.

Figure 27: Jablonski Diagram

The excitation process excites the ground state (S0) electron of the fluorophore to its lowest excited singlet state (S1). Excitation process happens very quickly in about few femtoseconds (fs). Since the S1 state is vibrationally broadened (has a number of vibrational states) excitation can occur over a range of wavelengths. This gives rise to the absorption spectrum of the fluorophore. Once the electron is in one of the higher vibrational levels of S1, it non-radiatively relaxes to the lowest vibrational level of the singlet excited state. This non-radiative decay happens in few picoseconds (ps).

The electron waits at the bottom of the singlet excited state for about few nanoseconds.  Fluorescence occurs when the electron returns to the ground state with the emission of a photon. The ground state is vibrationally broadened as well. The electron can return to any of these vibrational levels before non-radiatively decaying to the lowest vibrational level of the ground state.  This allows for a range of emitted photon energies (wavelengths), observed as the fluorescence emission spectrum.

Since the electron has lost some of its energy in non-radiative processes in the excited and ground states, the emitted photon has energy lower or in other words wavelength longer than the excitation photon. The difference between the peak excitation wavelength and the peak emission wavelength is called the Stoke's shift. 

The total time for the whole process (time for excitation+non-radiative decay+emission) is the fluorescnce lifetime. This is fs+ps+ns which is essentially ns. So, the fluorescence lifetime ends up being the time taken for the electron to decay from the bottom of the excited state to the ground state in the fluorescence emission process.

The basic ray diagram of a fluorescence microscope is given in Figure 28.

Figure 28: Fluorescence Microscope-Ray Diagram

Fluorescence Microscopy is all about separating the emission wavelength from all other wavelendths inlcuding the exciation light and any stray light. This is achieved with the help of the following optical components:

  1. Excitation Filter
  2. Dichroic Mirror
  3. Emission Filter

An epi-fluorescence configuration where the objective also acts as the condenser is used. The light from a broad-band intense light source like a Mercury Arc-Lamp is passed through the excitation filter. This filter transmits the excitation wavelength (band) towards the dichroic mirror. The dichroic mirror reflects the light onto the objective. After interacting with the sample the excitation light excites florescence in regions of the sample that has the specific fluorophore.  

Fluorescence is mostly isotropic. The fluorescence light that emitted in the backward direction is collected by the objective and relayed to the dichroic mirror. The dichroic mirror is designed to transmit the emission wavelength and it, in turn, relays the light to the emission filter. The dichroic mirror behaves differently for the excitation and emission wavelengths. It reflects the excitation wavelength and transmits the emission wavelength. The emission filter, filters out any residual excitation or stray light and transmits the emission wavelength towards the detector. 

The excitation light is way stronger than the emission light. The excitation filter should be designed to handle the entire power of the illuminaiton source usually in 100s of mWatts, of which few 10s of mWatts at the excitation band of the fluorophore is transmitted by the excitation filter . The fluorescence emission is usually in micro-watts, so the emission filter needs a very high transmission at the emission bandwidth and a sharp roll off outside this bandwidth.

In commercial fluorescence microscopes the three components are mounted in a single cube called the Filter Cube. This makes it convenient to change between fluorophores by changing the filter cube as a whole and not worry about individual components.

Configuring the fluorescence microscope in the epi-illumination configuration has a number of advantages:

  1. Objective doubles as the condenser. Alignment is easier. 
  2. Resolution depends on both Objective and Condenser NAs. Since an Objective usually has a high NA and since it is doubling as the condenser the resulting resolution is high.
  3. Easier to separate the excitation and emission wavelengths. Most of the intense excitation light is transmitted through the sample and only a small fraction is scattered back. Hence separating excitation and emission wavelengths in the backward direction (epi-configuration) is easier.

Like in case of bright field imaging we need to setup the epi-fluorescence microscope to achieve Köhler Illumination. There is a small problem here. The Aperture Diaphragm needs to be placed at the back focal plane of the condenser. Since the objective doubles as the condenser, placing it there means both the illumination and collection NAs are affected. Reducing the Aperture Diaphragm to decrease the NA of the condenser will lead to decrease in NA of the objective as it collects the emitted light. To overcome this problem Köhler Illumination in epi-illumination uses a configuration as shown in Figure 28.

Figure 28: Köhler Illumination in Epi-Illumination Configuration. Green is excitation path and red is emission path.

Figure 28 shows the schematic of an epi-fluorescence microscope that uses an arc lamp as the excitation source. Unlike in the case of Köhler Illumination for bright field imaging, the Aperture Diaphragm comes first in the light path and then the Field Diaphragm.

Consider a single point at the center of arc-gap. Light from this point is collimated by the Collector Lens and then focused by Lens B. Focal point of Lens B creates an image of that point on the arc-gap. As we discussed for Köhler Illumination in transmitted illumination configuration  the focal plane of Lens B is a conjugate plane of the light source. We can place the aperture diaphragm here. Lens C is placed at its focal distance from Aperture Diaphragm. This hence produces a collimated beam of light that is directed into the Field Lens.

The Field Aperture is kept at the focal distance away from the Field Lens. This creates an image of the Filed Aperture on the sample plane. The Field Lens focuses the collimated beam of light reflecting through the dichroic mirror onto the back focal plane of the objective. This creates an image of the arc-gap point at the back focal plane of the objective. The objective then collimates this light into a beam pencil to impinge on the sample plane. Like what we discussed earlier the different points on the light source create beam pencils at different angles that intersect at the focal point of the objective (acting as condenser). This creates the cone of light, condensing it on the sample plane.

Note that we have essentially moved the position of the Aperture Diaphragm from the back focal plane of the objective to a point before it. We can now control the NA of illumination without affecting the NA of light collection by the objective.