Written by Jesse Aaron
In a previous blog post, we discussed probe selection for single molecule localization microscopy (SMLM) techniques such as PALM and STORM. This class of techniques is powerful in that it can achieve some of the highest spatial resolutions (in all three dimensions) of all current super-resolution microscopy techniques, up to 10-20nm (1–3). Expanding PALM/STORM to imaging more than one type of molecule within a cell opens up exciting new possibilities for exploring the biomolecular interactions that underpin nearly all biological functions. This blog post continues earlier discussions to give some practical guidance toward both probe selection and imaging considerations when performing SMLM across multiple channels.
In general, more care needs to be taken when selecting probes for multi-color SMLM, as compared to conventional multi-color imaging. Aside from the obvious requirement that each fluorescent tag should be spectrally well-separated from each other, there are further considerations to be made. This stems from the fact that in multi-color SMLM, the image acquisition process in one color channel can affect the results in another channel in different ways.
Multi-Color Imaging and Photochromism
To help illustrate this point, we’ll consider a hypothetical example. Let’s say a researcher is interested in assessing the interaction between actin and actinin in a cultured cardiomyocyte, using the PALM approach. She chooses to utilize photo-activatable GFP to tag the actinin, and mEos to label the actin. At first glance, these dyes appear to be an appropriate combination – the peak emission wavelength of each probe’s “on” state is reasonably well-separated (517 vs. 584nm).
For simplicity’s sake, let’s assume that each color channel is acquired sequentially. If the researcher acquires the PA-GFP channel first, we run into a problem: The non-photoconverted mEos is excited and emits at the same peak wavelengths as the PA-GFP! Thus color-discrimination between actin and actinin becomes impossible in this scenario.
But what if she activates and acquires the mEos channel first? This approach solves the problem of color discrimination – the PA-GFP will not be visible in the activated mEos channel. However, as she activates the mEos probe with low 405nm excitation, the PA-GFP will also become activated under these conditions. Recall that PA-GFP is activated irreversibly – that is, once it is switched to the “on” state, it cannot be returned to a temporary dark state, but only be permanently photobleached. As a consequence, when she subsequently begins acquisition of the actinin image, nearly all the signal is already present, with no hope of imaging the well-separated single molecules necessary for PALM! Clearly, PA-GFP and mEos will make a poor pair for multi-color SMLM.
A better choice is to replace PA-GFP with the reversibly photoswitchable Dronpa. In this way, the researcher can first activate and image the mEos-actin molecules. Upon switching to the Dronpa-actinin channel, she finds that most of the Dronpa has also been activated, as expected. But, by illumination at 488nm, she can drive nearly all the Dronpa fluorophores back into the dark state. Then, by carefully re-applying 405nm activation, she can acquire single-molecule images to create a high-quality PALM reconstruction, as demonstrated by Shroff and colleagues (4). This example illustrates that in addition to a probe’s emission properties, the consequences of a fluorophore’s photochromic properties are equally important.
Acquisition Order in Multi-Color SMLM Experiments
As the previous example hints, the order of color channel acquisition can be important in SMLM, along with the effects due to photochromism. Let’s consider another example in which a researcher wants to use a dSTORM approach to image interactions between microtubules and mitochondria in an epithelial cell line. He uses immunofluorescence staining to label tubulin with Alexa Fluor 647, and TOM-20 (a mitochondrial marker) with Cy3B. As before, these dyes are reasonably well separated in their excitation and emission profiles. Further, both dyes can be also reversibly photoswitched, avoiding the issues described in the previous example. However, recall that reversibly switchable SMLM dyes have a property termed recovery rate. This refers to the fraction of total dye molecules that continue to photoswitch from one blinking event to the next. Due to permanent photobleaching, the recovery rate will always be less than one; because of this, the number of molecules that can be detected continuously decreases under the high-intensity illumination required for dSTORM.
How does this relate to the order of channel acquisition? Let’s assume the researcher acquires the Cy3B (TOM-20) channel first. He illuminates the sample with high power 561nm laser light, and collects single molecule images of TOM-20. But the Alexa Fluor 647 is also absorbing some of this light at the same time. In fact, Alexa Fluor 647 will absorb 561nm light at 8% of its maximal value. While that may not seem consequential, dSTORM often utilizes illumination intensities of >10kW/cm2. Multiplied over 30-60 minutes of acquisition, the cumulative effect can be drastic. By the time the researcher begins acquisition of the microtubule image, he finds that only a small portion of the Alexa Fluor 647 signal is left, and the resulting image is poor. Clearly, a better approach is to image the more red-shifted dye first. Cy3B will absorb 647nm excitation at only a fraction of a percent of its maximal value. Thus, the chances of the TOM-20 signal succumbing to permanent photobleaching before acquisition is considerably less. In this way, high quality images can be acquired in both color channels.
To aid users of the Advanced Imaging Center (AIC) at Janelia, we have developed a simple flow chart, shown in Figure 1 below, to aid proper selection of probes for multi-color PALM/STORM. In addition, Table 1 following this gives some example dye combinations that are compatible with multicolor PALM/STORM imaging. While not comprehensive, these examples will give AIC users a number of options to consider.
Researchers are continually developing new approaches for multiplexed PALM/STORM. Recently, researchers have proposed a conceptually simple method that allows imaging multiple cellular targets with a single fluorophore (5,6). They showed that a sample can be immuno-stained with suitable dSTORM dye (such as Alexa Fluor 647), and imaged until nearly all fluorophores have been driven into a permanent dark state due to photobleaching. Any remaining active fluorophores can also be quenched by addition of a reducing agent such as NaBH4. The sample can then be washed and re-labeled using the same fluorophore conjugated to an antibody raised against a new target structure, and imaged again. While theoretically straightforward, the major technical hurdle in this approach is channel image registration. Manipulating the sample between imaging rounds can cause great difficulty in maintaining its original position on the microscope at the nanoscale. Thus, incorporating fiducial markers or careful image correlation to re-align images afterward is crucial. Indeed, even if samples are not manipulated between color-channel acquisitions, alignment fiducials are nearly always needed due to the nanoscale chromatic aberrations that are present in even the most well-corrected microscope systems.
Both probe choice and imaging sequence must be carefully considered in any multi-color SMLM experiment. Merely relying on probes that are spectrally distinct from each other is not sufficient for generating high quality data. We highly encourage potential AIC users to contact us prior to proposal submission to discuss PALM/STORM probe selection or any other technical matters related to your project.