Senior Project : Bi-Plane FPALM


1 September 2009

Science demands an ever-smaller view of our world to realize and understand the microscopic parts which compose the macroscopic whole. Light microscopes reach an inherent barrier in resolving very small (~10-7m) details due to the wave nature and interference patterns of light. This barrier, quantified by Ernst Abbe in 1873, is termed the diffraction limit, and determines the closest proximity to which two separate light sources may be resolved.  Fluorescence microscopy, with techniques1,2,3,4 which work around the diffraction barrier, earns its name by using natural and synthetic fluorescent dyes which stochastically turn on and off. If the right amount of fluorescent molecules becomes activated then the diffraction-limited spots are spatially separated enough to be accurately localized. The localization precision of the resolution is determined by the background fluorescence, the magnification of the sample on the camera chip and the number of photons collected, and turns out to be much smaller than the diffraction limit. These techniques work because the fluorescent dyes absorb energy in the form of laser light and emit a second color light, which is optically separated from the laser light and therefore detectable. A sparse subset of these dyes will fluoresce under the activation energy of a laser and then photobleach a short time later. A very sensitive high-speed camera images the fluorescence and custom written re-construction and rendering software completes the image. Of the several techniques which have emerged in the field of fluorescence microscopy, many have also been adapted to image in the axial (z) dimension, as well as the lateral (x, y) dimensions5,6,7,8,9,10. Among the techniques to image below the diffraction limit, FPALM4 will become the focus of this project; specifically, an adaptation to the original technique in order to image all three spatial dimensions. Termed Bi-plane FPALM7, this 3D technique simultaneously images two focal planes spatially separated on the same CCD chip. Because of the spreading of light emitted by a point source (fluorescent molecule or bead), the images in each focal plane will contain spots which are either slightly away from the focal plane (appear larger, blurrier), or closer to the focal plane (appear smaller, brighter). By analyzing both focal planes simultaneously and knowing accurately the point spread function11, a fitting algorithm is able to determine the axial position of the emitted light relative to the two focal planes. Analysis of the gathered images is processed in Matlab on custom-written software which localizes the imaged blinking fluorescence both radially and axially with 30nm and 75nm localization precision7 respectively.

The goals of this project are:

1) To continue to develop the Biplane fitting algorithm as a link between measured spot sizes and axial positions so as to optimize – as much as possible – both speed and localization precision;

2) To test the developed algorithm against a known geometry for precision and accuracy; and

3) To image the structure of a cell membrane and/or the membrane protein hemagglutinin (HA) in order to determine the topography of cell membranes.

The three parts of this project are intertwined inasmuch as the results from imaging the known structure offer feedback to the accuracy and precision of the fitting algorithm. Not until the algorithm produces acceptable fits axially will the cell membrane images carry any weight; however, with an axial localization precision equal to- or better than published results7 the topography of cell membranes will offer valuable insight for additional research in cell membrane structure and the process of viral infection.

Specifically, the biplane fitting algorithm will be developed through the repeated imaging of fluorescent beads at a number of axial positions. With the use of a piezo-driven stage (Mad City Labs), the samples can be moved precisely with less than 10nm precision over several microns. With images obtained, the two focal planes will be analyzed and compared, and a regression defined to fit the axial position of the fluorescent source as a function of ratio of spot sizes in each focal plane (rA/rB). This regression will be specific to the exact experimental setup, magnification arrangement and emission fluorescence color. If any of these parameters are changed, the calibration process will have to be repeated. Once defined, this fitting regression will appropriately describe axial positions for an effective range of about 1 micron. The localization precision of the regression fits will be described by imaging a planar arrangement (fluorescent beads on a coverslide) and calculating the standard deviation of the fitted positions against the “known” plane. With an acceptable (~100nm or better) axial localization precision developed, a biological sample will then be prepared by culturing cells to express fluorescent dyes in HA protein on the membrane. A successful biplane image of a cell membrane will show structure and detail at length scales smaller than those achieved with traditional light microscopy. The HA protein in particular is important in the process by which viruses like influenza infect cells. Recently, the H1N1 strain of Swine Flu has gained worldwide notoriety as a fast-spreading virus. Though its effects are not yet systemic or pandemic, fears about possible mutations have pushed this virus to the front pages. A better understanding of how viruses like H1N1 infect cells will lead to better treatments and precautions against this and other viruses in the future.

            Several other considerations must be addressed in the matter of this project. A typical FPALM setup includes the electron multiplying charge coupled device (EMCCD) for imaging ($25-40k), microscope ($15-20k) High-NA objective ($7k), lasers ($10-25k), optics, lenses, filters ($8k), and computer and Matlab software ($2-3k)12. From scratch, this setup would cost $60-100k to establish; however, due to current research at the University of Maine, two FPALM setups have been established and will be used for this project. Furthermore, to convert FPALM to image biplane, merely an additional 50/50 beamsplitter and mirror are used in the detection pathway. Additionally, the photoactivatable dyes and cells to be used for imaging will be donated to this cause. This project will pose little to no economic burden to the student to image a cell thanks to the ongoing research in the Department of Physics.

            Environmentally, this technique poses little to no risk. The green fluorescent proteins which are used to fix cells do not harm the cells; in fact, mice and pigs have both been developed to express this protein in every cell in their body, all the while remaining healthy and alive. Additional dyes and fluorescent bead solutions may be used, although these materials will be properly handled with latex gloves. Furthermore, to ensure the safety of the experimenter, formal training for proper use of the lasers is imperative; sufficient laser training in compliance with the University has been completed.

The work that this project will result in involves the membrane protein HA which plays an integral role in viral infection to host cells. A three dimensional picture of how this protein functions will no doubt illuminate the subject, which is recently spotlighted in the spread of H1N1 Swine Flu. Scientific research in the area of viral infection will carry deep social and political ramifications and may ultimately lead to a deeper understanding of the biological functions and arrangements of viruses, or to better vaccination attempts.

For now, BP FPALM is still under development, and is not ready for manufactured production. In fact, BP FPALM may never be manufactured commercially, but instead remain a learned technique which requires the expertise and knowhow of a skilled experimentalist.

Figure 1.

Typical BiPlane Setup. BPFPALM requires modifications to the beampath only after L2. The Addition of a 50/50 Beamsplitter (N) and M4 create an additional and longer reflected pathlengththan the transmitted path, resulting in an offset focal plane.


References and Related Papers