Broadly speaking, photoswitchable probes may be divided into two broad categories: those that may be genetically expressed (e.g., photoactivatable fluorescent proteins, PA-FPs [5]) and those synthetic dyes that may be exogenously introduced via antibodies or small-molecule labeling strategies [6]. Each class has its own strengths and weaknesses: PA-FPs are small (a few nm in size), may be targeted with molecular precision and are thus highly specific, and may possess a high (>1,000) contrast ratio between bright and dark states, helpful in isolating a single fluorescent molecule from its neighbors and acquiring high-density (and thus high resolution) images. Unfortunately, PA-FPs are in general dimmer than most synthetic dyes, limiting localization precision and challenging the detection of activated single-molecule signal above cellular autofluorescence and residual faint background emission from the potentially large number of inactive molecules. They are also available only in limited colors, and the brightness and contrast ratio of the green-emitting PA-FPs is generally lower than the red-emitting PA-FPs (although the palette of PA-FPs continues to expand [7] and the performance of green PA-FPs continues to improve [8]). Switchable synthetic dyes can be very bright and are available in many colors [9], significant advantages over PA-FPs. However, the coupling method (e.g., to antibodies) can introduce significant uncertainty in their position, thus defeating the brightness advantage and effectively limiting resolution. Exogenously introduced dyes often label their target less specifically than PA-FPs, leading to an increased level of background in the reconstructed super-resolution image and sometimes introducing uncertainty as to whether localizations are real or spurious. Finally, many exogenous dyes require an additional chemical cocktail (typically reducing agents or oxygen scavengers) and/or high excitation intensities, limiting application in live cells. In this chapter, we focus on instrumentation particularly suited to the somewhat dimmer PA-FPs.

Initial PALM experiments were conducted in 2D [2, 10] near the coverslip surface, where optical aberrations and scattering are minimized, and the full numerical aperture (NA) of oil objectives may be used to maximize signal collection and aid in the localization precision of dim PA-FPs. An additional advantage of 2D imaging near the coverslip is that total internal reflection (TIR [11]) may be used to limit excitation to the focal plane and further aid in single-molecule detection. As it forms the basis for 3D experiments, we cover 2D PALM instrumentation in Subheading 3.1.

While suited to membrane [12] and focal adhesion [13] studies in live cells, the obvious drawback of TIRF is that the vast majority of the cell is inaccessible. The simplest way around this problem is to move the excitation out of TIRF, thus employing epi- or wide-field imaging to illuminate the entire cell. This imaging modality also creates several problems for 3D super-resolution imaging: (a) in addition to x and y, a molecule’s z coordinate must be localized with subdiffractive precision, and the resulting information used in assembling a 3D PALM dataset; (b) axial drift must be measured and/or controlled; and (c) background photoactivation and subsequent fluorescence of probe molecules at axial locations other than the focal (imaging) plane must be minimized, lest their localizations be wasted and the extraneous background signal so created impede localization in the focal plane. We discuss technical solutions to (a–b) in Subheading 3.2 and (c) in Subheading 3.3. We also refer to the reader to [14, 15], as this chapter draws heavily from these sources.