Localisation Microscopy

Stochastically switched single molecule localisation provides a class of far-field SRM techniques that essentially realise super-resolved imaging of a (densely) fluorescently labelled sample by acquiring a series of wide-field intensity images while arranging that only a small fraction of the fluorophores is emitting light (i.e. “switched on”) during each image acquisition - such that they are spatially separated by more than the width of the microscope PSF.  By acquiring many such images, all the fluorophores can, in principle, be localised and the superposition of all the maps of individual fluorescing molecules can provide a super-resolved image. The lateral resolution – given by the precision of the fluorophore localisation – is that of the wide-field microscope divided by the square root of the average number of detected fluorescence photons per molecule. Using dye fluorophores, resolutions of a few tens of nm are typically realised. 3-D super-resolved variants of PALM¹ and STORM² have also been demonstrated that provide axial localisation of the single molecule emitters to tens of nm, thus realising 3-D SRM.

Stochastically-switched single molecule localisation was first implemented using samples labelled with photoactivatable fluorescent proteins (PA-FP)^{3, 4}, i.e. genetically expressed proteins that only become fluorescent after irradiation with light at a specific wavelength that is different from the wavelength of light used for excitation of fluorescence⁵. By weakly irradiating the sample with radiation at the activation wavelength, a small number of (sparsely distributed) PA-FP are “switched on” and imaged to record their localisation and then they are “switched off” by strong irradiation at the excitation wavelength to cause photobleaching. Then further irradiation at the activating wavelength can switch on a further sparse subpopulation of the PA-FP, which can again be localised and photobleached.  This sequential switching on and off thus provides super-resolved images and is known as photoactivated localization microscopy (PALM). Initially PALM required hours for the sequential photoswitching/imaging acquisitions but the development of asynchronous photoswitching and imaging of reversibly switchable fluorescent proteins,^{6, 7} has significantly improved imaging speed to a few minutes. 

An alternative approach, known as stochastic optical reconstruction microscopy (STORM) is to employ photoswitchable dye labels (initially a specific pair of cyanine dyes⁸) that can be reversibly switched between bright (fluorescing) and dark (non-fluorescing) states with irradiation at appropriate wavelengths, rather than using photobleaching to switch off the fluorescence. STORM was subsequently simplified in terms of its implementation by utilising a wide range of conventional dye labels and chemically-induced switching of fluorophores under continuous excitation, which is described as d-STORM⁹.

A third, related, approach to SRM using single molecule localisation utilises fluorophores whose fluorescence emission switches on and off spontaneously and can be described as super-resolution optical fluctuation imaging (SOFI). Here the samples are labelled with fluorophores that “blink” spontaneously. The separation of the multiple fluorophores emitting within each PSF to permit their single molecule localisation is realised through an analysis of their random fluctuations. This statistical approach was first demonstrated with quantum dots^{10, 11} and later extended to dyes^{12, 13} and fluorescence proteins¹⁴.

Because they are inherently single molecule localisation techniques, PALM, STORM and related techniques benefit significantly from the low background associated with TIRF microscopy or the use of highly inclined illumination. Typically, the excitation light for TIRF illumination is delivered via a single mode optical fibre and the relatively low coupling efficiency means that relatively high power lasers are required to realise STORM or PALM and SRM systems are consequently expensive. We have recently demonstrated low-cost STORM using multimode optical fibres for delivery and used a multimode diode laser to provide excitation for STORM with or without TIRF. Combined with open source software for STORM data acquisition and analysis, we propose that this “easySTORM” approach to SRM¹⁵ can be widely implemented for little more cost that a standard fluorescence microscope. We are currently working to extend this to 3-D localisation.