Written by Jesse Aaron
Background
The last 10 years has seen an explosion of techniques aimed at creating biological fluorescence images with unsurpassed clarity via super-resolution microscopy. Among these various techniques that circumvent the diffraction barrier, there is a group of methods that can be termed single-molecule localization microscopy (SMLM). This term encompasses the potpourri of acronyms such as PALM, fPALM, STORM, dSTORM, GSDIM, and others. The first theoretical proposal for SMLM was suggested by Eric Betzig in 1995 [1], but wasn’t experimentally demonstrated until more than 10 years later [2], closely followed by other similar implementations [3,4], primarily as a result of advances in camera technology. Despite their differences, all SMLM techniques are fundamentally governed by a single general principle:
If a biological sample can be imaged such that only a very small portion of its fluorescently labeled molecules are visible at any given time, then an image of nearly all such molecules can be made with much higher resolution than diffraction would allow.
But why? The first answer to this question has actually been known for some time. If an image can be made of a single molecule using a conventional microscope, it will take the form of a “blob” that described by that microscope’s point spread function (PSF). The size of the PSF is ultimately determined by diffraction, and is orders of magnitude larger than the molecule itself. But, the location of the PSF center gives the most likely position of that molecule. Clever use of a curve-fitting algorithm can reveal the molecule’s position with accuracy in the 10s of nanometers, or even less – much closer to the actual size of the molecule. This is termed “localization”.
But this trick only works well if the PSFs in the image are not overlapped with each other; otherwise, it creates a hopelessly impossible curve-fitting task! This leads us to the second reason why SMLM works. One may ask: “How can we image single molecules that are well-separated in space if there are hundreds of thousands or even millions of fluorescently tagged molecules in a relatively small space?” The critical advance in SMLM was the development of ways to view all of these molecules a few at a time, over many sequential images. And that depends on inventing new types of fluorescent labels that appear in the image only occasionally. This blog introduces the conceptual basis, and gives pointers for selecting these fluorescent tags for use in SMLM.
Photochromism and “blinking”
Photochromism is a general term that refers to the ability of certain molecules to change their optical properties when exposed to light, and it forms a basis for how SMLM works. Broadly speaking, there are two types of photochromism that are useful in this context: (1) Photoswitching, and (2) Photoactivation. Photoswitching refers to a reversible change in either the fluorescence color or intensity in response to light. Photoactivation refers to an irreversible change to a molecule to produce the same effect(s). In either case, a general schema can be described. A fluorescent molecule starts (or is initially forced into) a state that isn’t detectable by the microscope. We’ll call this the “off” state. Upon absorption of light (very often in the UV range), the molecule now becomes visible (the “on” state) and emits photons that strike the detector. Then, one of two things can happen. Either the molecule returns to the previous “off” state, or it enters a permanent dark state, termed photobleaching. Key to balancing each of these steps is a very careful application of the UV light. Typically, it is applied at very low intensity – such that only a small number of random molecules get turned “on”, followed by a return to the dark state (or permanent photobleaching). In the correct implementation, it appears as a stochastic “blinking” in each image of a several single molecules, which can then be subjected to the localization process described above. Very often, 104 images are needed to capture a “blink” from nearly every molecule in the sample.
Although many probes change from a dark, non-emissive, state to a bright/emissive state, it is important to note that other probes change color, from a shorter wavelength state, to a longer wavelength state. Such molecules are often termed photoconvertible, and can change in a reversible (photoswitching) or irreversible (photoactivation) manner. The end result will be the same, but users should be sure that the microscope is only detecting molecules that are undergoing conversion to the second color. In a special case of photoactivation, a molecule can contain photolabile moieties that are degraded via absorption of light, which then renders the molecular visible. This is termed caged photoactivation, as it is typically irreversible. Figure 1 illustrates these various mechanisms to control single molecule fluorescence in SMLM.
Considerations when Selecting a Fluorescent Probe
The mechanisms outlined in Figure 1 encompass a wide variety of molecules, including fluorescent proteins (FPs) or small organic dyes. FPs have the advantage of genetically encoded specificity, while organic dyes will usually exhibit higher brightness, and are thus able to be localized more accurately. Most dyes also have an additional requirement that they can only blink in a specific imaging buffer, although this is not universally the case. However, it is important to note that FPs and dyes can be used in tandem for multi-color experiments, provided they are spectrally well separated from each other. A more detailed blog entry will follow that discusses the factors that need to be considered for multicolor SMLM.
However, the picture gets murkier as we look closer, as there are several considerations to make when selecting an optimal probe. For example, a probe’s on/off duty cycle refers to the fraction of time a fluorophore spends in the “on” state. A probe with lower duty cycle implies that fewer molecules will be in the “on” state at any given time — this can be good since that reduces the chance that any two PSFs will overlap in a single image. But it can also be inconvenient, as a very low duty cycle probe requires more images to capture all the fluorophores in the sample. Furthermore, each fluorophore can differ in its blinking rate. Surprisingly for some fluorophores, the blinking rate depends on the excitation intensity, while its overall brightness does not. Thus, higher laser powers will result in the same number of photons emitted per blink, but compressed into a shorter time. This allows faster image acquisition rates, since ideally the blinking rate should match the image acquisition rate. However, for some dyes, increasing the excitation power can also reduce the recovery rate – that is, the number of times a photoswitchable molecule can cycle between “on” and “off” states before permanent photobleaching sets in. Unfortunately, many SMLM probes are still not fully characterized.
More generally, it is useful to consider the differences in the final image when using a photoswitchable vs. photoactivatable fluorophore. Unlike a photoactivatable probe, a photoswitchable tag can blink (and thus be localized) many times before it is permanently photobleached. This can serve to better delineate a structure via more photons, but can also cause ambiguity when trying to ascertain relative molecular abundance. For instance, it can be very challenging to discern whether a single molecule has blinked four times, or two very closely-spaced molecules each blinked twice. Needless to say, there are numerous factors that can be taken into account when selecting SMLM probes. We encourage readers to refer to [5-6] for two particularly useful reports that delve into the myriad of FPs and dyes that can be used in SMLM.
In an effort to aid users of the Advanced Imaging Center (AIC) at Janelia Research Campus, we have compiled a short list of fluorophores (both fluorescent proteins and dyes) that have commonly been used in SMLM experiments, as shown in Table 1. Excitation/emission wavelengths for photoconvertible probes are shown for the red-shifted state. While certainly not comprehensive, each fluorophore in this list is compatible with use in the AIC’s iPALM – arguably the highest overall resolution SMLM system in the world, and capable of imaging up to three color channels. Probes with the same color scheme in the excitation/emission columns will use the same laser and filter configurations. As such, they cannot be used together in multi-color experiments. Brightness is calculated as the product of each probe’s extinction coefficient and quantum yield, and expressed in units of mM-1-cm-1. Under the Class column, PA-FP refers to photo-activatable fluorescent protein; PC-FP refers to photo-convertible fluorescent protein; and PS-Dye refers to organic small molecule fluorophores that undergo photoswitching.
We highly recommend that all prospective users of the AIC contact us to discuss appropriate probe selection for use in the iPALM system, as well as any other technical considerations.
References
1. Betzig E., Proposed method for molecular optical imaging. Opt. Lett. 20:237-239 (1995) 2. Betzig E., et al., Imaging Intracellular Fluorescent Proteins at Nanometer Resolution. Science 313(5793):1642-1645 (2006) 3. Rust M.J., et al., Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM). Nat. Meth. 3:793-796 (2006) 4. Hess S.T., et al., Ultra-High Resolution Imaging by Fluorescence Photoactivation Localization Microscopy. Biophys. J. 91(11):4258-4272 (2006) 5. Dempsey G.T., et al. Evaluation of fluorophores for optimal performance in localization-based super-resolution imaging. Nat. Meth. 8(12):1027-1036 (2011) 6. Shcherbakova, DM, et al. Photocontrollable Fluorescent Proteins for Superresolution Imaging. Annu. Rev. Biophys. 43:303-29 (2014)