Methods

Instrument

The microscope utilized was based on the original STORM setup described in (STORM1). In this particular setup though, the components were selected to achieve super resolution while maintaining the lowest cost possible. Images of the setup are contained in a Powerpoint presentation file with their respective tags. These images show the full setup in acquisition configuration, top, front and side views and details of the laser mounting, the stage and the light paths.

Microscope frame

The frame was custom built using a cage system by Thorlabs. Support was provided at the alignment mirrors, a post near the mirrors, the main stage and the barrier filter/emission tube lens mount, as seen on the picture below. Additionally, an enclosure made with Thorlabs anodized beams and black foam was built to fit the whole setup.

Trans illumination

A near infrared LED was fitted with a 25 mm plano convex lens to serve as a condenser above the objective of the microscope. It was powered via a power supply (120 mA, 2.1 V).

Laser path

We used a 532 nm, 5 mW diode laser (Z-BOLT) and a HeNe 5 mW, 632.8 nm laser (Melles Griot) as our light sources. The former was mounted on an adaptor plate and directly into the base plate of the scope. The latter was mounted in a V support and both were pointed at a pair of mirrors for alignment (either one or the other). From here, the beam was expanded approximately 7 times by a Galilean telescope fitted with a -30mm plano-concave lens and a 250mm plano-convex lens separated by around 220mm. A 200 mm plano-convex lens (Thorlabs) was used to focus the laser beam on the back focal plane of the objective via a dichroic mirror (LWP-45-Runp-532-Tunp-670-PW-1025c, CVI Melles Griot for the first bead and cell images; z543/647rpc, Chroma Technologies for the last images). The objective was a 100X, 1.25 NA oil immersion (Nikon). For the cell imaging, the beam expander was removed so as to concentrate the power of the laser in a smaller area and achieve the required photoswitching.

Camera path

The light coming from the sample was filtered partially by the dichroic mirror mentioned above. After that, an emission bandpass filter (ET700/75m, Chroma Technologies) eliminated most of the remaining off-wavelength radiation. Finally, the light was focused on a CMOS camera detector (Orca Flash 2.8, Hamamatsu Photonics) by another 200 mm plano-convex lens (Thorlabs).

Imaging

For the bead experiments we ran a series of different acquisition routines, changing both the preparations and the exposure times. All of them had 100 frames in length using the full area of the camera and gain of 255. Finally, the sequences were captured using the Stream acquisition option in Metamorph (Molecular Devices).

For the cell imaging, we used the same stream acquisition feature of the software, but in this case, we captured between 3000 and 8000 frames at 500 ms acquisition time and 255 gain.

Samples for Imaging

Bead Samples

To characterize the optical set up with the 532nm laser, images of sub-diffraction fluorescent beads (110nm with Nile Red Dye) were mounted on a glass slide at a 1:5^4 dilution. 1um silica beads were added at a 1:5^5 dilution, and the beads were mounted using agar. To test the optical system with the 632nm laser, PS-Speck™ Microscope Point Source Kit Component D (deep red) beads were imaged. The beads were vortexed, and 7.5ul of 1X bead solutions was added to a glass slide along with 7.5ul of 1um beads at a 1:10^5 dilution. The slides were incubated at room temperature overnight to dry the slide and then mounted using 7.5ul of mounting medium with an anti-fade agent (MOWIOL with DABCO).To find the focal plane, the 1um silica beads were imaged in bright field. The 110nm beads were then imaged with fluorescence microscopy. Sequences of 100 images were taken with 100ms, 60ms, 30ms, and 10ms exposure times with a gain of 255. The beads were imaged with an approximate 7X beam expander in the optical system. The field of view was changed after each image sequence acquisition to minimize the effects of photobleaching.

Cell Samples

Samples of BSC-1 cells (derived from green monkey kidney cells) fixed with paraformaldehyde were obtained from the Xiaowei Zhuang laboratory. For the images of michrotubules, the cells were treated with microtubule-targeting antibodies conjugated with Alexa647. The images of mitochondria were taken with the BSC-1 cells with mitochondria-targeting antibodies with both Cy3 and Alexa647. These fixed cells were stored in PBS. Just prior to imaging, the PBS was removed and 700ul of an imaging buffer (200mM Tris pH8, 10mM NaCl, 10% glucose) as well as 7ul of 100% beta-mercaptoethylamine and 7ul of 0.5mg/ml glucose oxidase were added to the sample. The cell samples were imaged with an exposure time of 500ms and gain set to 255 with the 632nm laser without the beam expander in the optical setup.

Image Processing

The images were processed using QuickPalm, an open source ImageJ plugin for particle detection and compiling STORM images (http://code.google.com/p/quickpalm/). The following input parameters affect how the software detects acceptable particles to include in the resulting STORM image: (1) FWHM: maximum full-width-half-max (FWHM) of the particle tolerated; (2) Threshold: The fraction of the maximum intensity of the image above which the signal must be to be a candidate for a particle; (3) SNR: The minimum signal to noise ratio tolerated; and (4) Symmetry: the expected symmetry of the particle. The weighted centroids of the particles are calculated and are stored as position data. To reconstruct the image, the positions of all the acceptable particles are plotted, resulting in the STORM image.