Spinning Disk Anisotropy
Figure 1. Fluorescence Polarization and Anisotropy A. Polarized excitation is used to selectively excite dipole-aligned fluorophores.
B. Fluorophores bound or in high viscosity media diffuse or rotate more slowly - higher anisotropy
C. Rotational diffusion or resonance energy transfer reduces polarization – lower anisotropy
Anisotropy imaging can be used to elucidate dynamic molecular organization in living cell studies. For optimum speed and dynamic range, laser spinning disk confocal imaging is the best tool currently available. We have overcome a limitation in the standard CSU scan head to enable use for real time confocal imaging of fluorescence anisotropy.
The technique of fluorescence anisotropy is well-known in fluorescence spectroscopy and biochemical assays (Lakowicz 2006) and it has been used in epi-fluorescence microscopy in living cells (Lidke et al 2003, Sharma et al 2004, Goswami et al 2008) and even in super-resolution imaging (Gould et al 2008). However, when implemented in point scanning confocal instruments with live cell specimens it has suffered from photo-bleaching, low frame rates and temporal skew resulting from sequential point scanning.
The CSU (Ichihara et al 1996), manufactured by Yokogawa Electric Corporation, is the leading spinning disk confocal scanner on the market today and finds widespread application in live cell confocal imaging. The various CSU models provide scan rates up to 2000 frames per second, with good confocality (~1 um FWHM) and low background across the visible range (400-650 nm). Unfortunately, the CSU cannot be used in its native form for anisotropy studies because its excitation path degrades the polarization of laser light. We have overcome this limitation and we present here a spinning disk confocal anisotropy system (SpiDA) which exploits simultaneous dual camera detection of orthogonal emission polarizations to provide high definition spatio-temporal anisotropy imaging.
Figure 2. Key components of the spinning disk confocal anisotropy imaging system are highlighted. A. Shows a schematic of the CSU-P polarizer mounted on custom-designed carrier. The component is motorized to allow introduction into the optical path or removal for convenient switching of operational modes. B. TuCam high performance dual camera adapter with two iXon EMCCD cameras attached. TuCam is designed to deliver alignment stability and low differential distortion (<1%). These features ensure that calibration tables for sub-pixel image registration do not require frequent updates to deliver precise anisotropy computation.
We have worked with Dr Satyajit Mayor and his team at NCBS in Bangalore, to bring this solution to fruition and a Revolution XD systems now resides is in his laboratory, where it is heavily used for confocal anisotropy studies in living cells.
2. Anisotropy and Homo-FRET
Isotropic is derived from the Greek and means quite simply "equal in all directions". While anisotropic means not equal in all directions. Fluorescence anisotropy measurements are based on the principle of photo-selective excitation by polarized light.
Fluorescent molecules preferentially absorb photons whose electric vectors (polarization) are parallel to their absorption (transition) electric dipole. The dipole has a defined orientation with respect to the molecular axis. Thus, when polarized light is incident on a population of molecules it is absorbed preferentially according to orientation. Further, the resulting fluorescence emission is also aligned relative to the molecular axis and the relative angle between absorption and emission polarization determines anisotropy.
Fluorescence anisotropy, r is defined as follows:
r(t) = (Ip(t) – Is(t))/ (Ip(t) +2 Is(t))
where Ip(t) is the fluorescence intensity parallel to excitation and Is(t) is fluorescence intensity perpendicular to excitation (Lackowicz 2006).
As the notation indicates fluorescence anisotropy, r is a function of time and decays with a relaxation time, Trot. The relaxation time is a measure of rotational diffusion which occurs during the lifetime of the excited state (typically 1-10 ns). In fluids molecular rotation can take place in a few tens to hundreds of ps and as result little anisotropy is observed. The rotational diffusion rates of larger molecules, such as proteins, are of the same order as fluorescence lifetimes and therefore anisotropy is sensitive to factors affecting these rates.
Anisotropy can therefore be used as an indicator of the state of biological macro-molecules within the cell and its membranes, including molecular size, aggregation and binding state (Lidke et al 2003).
When excited fluorescence molecules (donors) come close enough to engage in dipole interactions (0.3-0.5 nm) with unexcited fluorescence molecules (acceptors), an effect known as resonant energy transfer (RET) can occur. Provided the excitation spectrum of the acceptor overlaps the emission spectrum of the donor, this can result in transfer of energy from donor to acceptor and is non-radiative (no photon is emitted). The newly excited molecule can now emit a photon, but its polarization will depend on its own orientation. When averaged over an ensemble, the result of RET is loss of polarization and reduction in anisotropy.
Figure 3. Calibration images are created by imaging FITC-loaded glycerol solutions of different concentrations. Higher glycerol concentration slows rotational diffusion rate to a greater extent, leading to higher anisotropy. A. Pseudo-color anisotropy image of 50% glycerol solution with mean isotropy of 0.09 ± 0.008 (SD). B. Pseudo-color anisotropy image of 70% glycerol solution with mean isotropy of 0.17 ± 0.009 (SD).
When the RET interactions happen between fluorescent molecules of the same type the process is known as "HomoFRET", and can be used to monitor molecular interactions and binding states. Varma and Mayor (1998) used anisotropy to monitor HomoFRET interactions in so-called lipid rafts and GPI-anchored proteins, which are located in the plasma membrane of living cells. These structures are important in key cellular processes and Mayor and co-workers have made substantial contributions to understanding these structures and their function using anisotropy imaging.
Figure 1 provides a visual summary of the principles of selective excitation and how rotational diffusion and HomoFRET affect polarization and anisotropy.
3. Polarization and the CSU
Polarization measurements in the Yokogawa CSU products shipped from the manufacturer show that laser polarization is degraded in the excitation optics of the instrument (Table 1). This was common in 5 different units tested and included CSU-10, 22 and X1 models. A low extinction ratio precludes use of an unmodified CSU for anisotropy imaging. In contrast the emission path was found to maintain polarization with high fidelity enabling confocal detection of the polarization state in all units tested to date.
|ALC + SMP/PM Fiber
|ALC + Fiber +CSUX1
|ALC + Fiber + CSU-P – see text
Table 1 - Polarization results from laser engine and SM/PM fiber (before the CSU-X1), standard CSU-X1scanner and the Andormodified CSUIX1 polarization solution (CSU-P). Standard CSUscanners degrade input laser polarization as this table shows.
We found that a good solution to the polarization problem is to integrate a high performance polarizer into the optical path of the CSU at a position which affects the excitation (laser) light, but not the detected fluorescence, whose polarization contains the useful information. A custom carrier was designed for the polarizer as shown in Figure 2A and this is mounted on a motorized drive for insertion and removal under computer control. In this way conventional intensity imaging can be re-selected without compromising sensitivity.
Figure 4. Anisotropy images from live cells after registration and pixel computation. Hot spots in anisotropy are indicative of tethered or bound fluorophores and reflect microstructure domains in the plasma membrane. Image courtesy of Mayor Lab, NCBS, Bangalore. Note GG8 cells are a CHO variant devoid of transferring receptor (Tfr) stably transfected with human Tfr and GFP-GPI (Sabharanjak et al 2002).
4. Confocal Dual Camera Anisotropy Imaging
The epi-illumination system described in Varma and Mayor (1998) was limited for dynamic studies because it used a single camera and sequential detection of p and s polarization images using a filter wheel. This approach introduces a time-skew between the s and p channels. To overcome this Mayor's group constructed a dual camera solution (Goswami et al 2009) and access to dynamic data was extended. Post acquisition image processing is required to achieve pixel alignment of the two images prior to the calculation of anisotropy.
However, a remaining problem with such a system is that emission from microscopic polarization domains is masked by out of focus fluorescence. Hence resolution, dynamic range and signal to background ratio of anisotropy are all compromised. The SpiDA system greatly reduces these effects by rejecting out of focus fluorescence, allowing anisotropy domains to be monitored in greater detail than ever before.
To achieve maximal temporal resolution and full field of view, SpiDA employs TuCam, our dual camera adapter. TuCam is a third generation adapter optimized for throughput, distortion and ease of alignment and enables simultaneous detection of p and s polarizations onto back-illuminated EMCCD cameras i.e. iXon3 897 - see figure 2B.
An image quality polarizing beam splitter is a critical component of the system: it must be flat and mounted with minimal stress. We utilize laser quality components with a surface flatness of <λ per inch and radius of curvature ≥ 30m. This minimizes distortions and lensing effects which can lead to focus error across the field of view. The beam splitter is mounted in a kinematic assembly to provide precision of adjustment and stability.
Mechanical stability is the most critical quality of an image splitter to ensure robust and repeatable measurement. Drift or creep leads to registration errors. Even though image registration to sub-pixel precision requires image processing in software, this process is driven by calibration whose temporal repeatability depends on the physical design. We have optimized image registration in our own iQ software, but an open source solution is also available in ImageJ (http://rsbweb.nih.gov/ij/).
The iXon cameras deliver high signal to noise anisotropy imaging with exposures in the 10-100 ms range. This provides a dynamic imaging tool for protein-protein interactions and microstructure modulation during events such as endocytosis and vesicle recycling. Figures 4 and 5 show example calibration data and live anisotropy images respectively.
Figure 5. Images showing anisotropy of cultured GG8 cells in single confocal optical sections at or close to the plasma membrane. A. Domains of GFP labelled lipo-proteins showing localized high anisotropy (mean r = 0.29) are clearly highlighted in yellow and red in the pseudo color mapping. These domains have been shown (Sharma et al 2004) to be associated with cholesterol. B. The same field in the specimen imaged after treatment with saponin, a natural "soapy" product of some plants (e.g. soapwort), which forms complexes with cholesterol in the plasma membrane. The impact of this on lipo-protein domain anisotropy is dramatic (mean r =0.17). Image courtesy of Mayor Lab, NCBS, Bangalore.
This technical note describes the Andor solution for real time confocal anisotropy imaging, which has been developed to overcome inherent constraints in the CSU scan head for polarized excitation of fluorescence. SpiDA utilizes a high performance image splitting dual camera adapter, TuCam and ultra-sensitive EMCCD cameras to deliver high contrast dynamic anisotropy data for live cell studies. We highlight cooperation with a scientific research group whose vision had not been realized previously. This process has now delivered a solution which is generally available to the wider research community as the SpiDA option for Revolution XD.
We take this opportunity to thank Dr Satyajit Mayor and his team at NCBS for their support, feedback and collaboration during this project.