Hank Hogan, Contributing Editor
As with the microscopes and detectors that measure their emissions, fluorescent proteins are constantly being improved. Three recent projects promise a new spatially confined imaging method, offer more reliability in what has been a mistrusted color, and provide clues about photoswitching mechanisms in fluorescent proteins.
Researchers from Riken Brain Science Institute in Wako, Japan, and from Catholic University of Leuven in Belgium do not have a problem with the photoswitchable fluorescent protein Dronpa. However, that has not kept them from seeking variants with distinctive characteristics. Using semirandom mutagenesis, they recently created two, Dronpa-2 and -3, that could be used for a new imaging modality.
Researcher Atsushi Miyawaki, leader of Riken’s Laboratory for Cell Function and Dynamics and director of its Advanced Technology Development group, said the new imaging technique will make use of two lasers, one at 405 and the other at 488 nm. The lasers will cause Dronpa-2 and -3 labels to oscillate between bright and dark states and, if properly aligned, they will do so only at the focal point. This approach could offer advantages over other spatially restricted methods. “While two-photon excitation microscopy uses a single objective, the proposed microscopy [technique] will have more freedom regarding the alignment of multiple lasers or objectives,” Miyawaki said.
Dronpa is a good fluorescent protein for reversible highlighting techniques because it has two states. In one it absorbs light at 503 nm and emits in the green. When hit with strong light at 488 nm, the protein converts into a nonfluorescent, or dark state, which switches back to the bright, emissive form after a small dose of 405-nm light. This switching behavior has made it possible to study fast protein dynamics in cells by turning fluorescent tags on at different times, but the conversion process from bright to dark does take intense light.
As detailed in the March 23 online version of Biophysical Journal, the researchers subjected the standard DNA-encoding Dronpa to an error-prone polymerase chain reaction that amplified the DNA, introducing mutations. After creating colonies of E. coli cells that expressed these mutants, they screened them for different photoswitching behavior using a custom-built image analyzing system equipped with two xenon lamps, a bright one with 300 W of power and a dimmer one with 75 W. They illuminated the colonies with 490- or 400-nm light from the 300-W lamp and used 490-nm light from the 75-W lamp to induce fluorescence, which they captured with a cooled CCD camera from Roper Scientific.
This approach yielded Dronpa-2, while site-directed random mutagenesis produced Dronpa-3. The two showed rapid photobleaching when the 490-nm light was present. They were not stable in the dark state because of a thermally and photochemically driven back reaction. “We see very efficient spontaneous recovery of the mutants,” Miyawaki said.
It takes two to tango or, in this case, fluoresce, as seen in these fluorescent images of human cancer HeLa cells expressing Dronpa-3 targeted to mitochondria (left) and to plasma membranes (right). A variant of the photoswitchable fluorescent protein Dronpa, Dronpa-3 has a bright (fluorescent) and a dark state, with the transition between the two driven by exposure to light of the proper wavelength, 488 and 405 nm, respectively. As a result, it takes both to make the fluorophore visible. On top is a scan with an argon-ion laser at 488 nm, in the middle is a scan with a 405-nm laser diode, and on the bottom is a scan with both. Reprinted with permission of Biophysical Journal.
The researchers showed that this behavior could be used in imaging by simultaneously illuminating Dronpa-3 targeted to mitochondria or to the plasma membrane of human HeLa cancer cells with lasers at 405 and 488 nm. Neither laser alone produced bright fluorescence, but the two together caused the fluorophore to cycle, leading to strong fluorescence.
More work is needed before the new imaging technique can be used, and it will require aligning the lasers independently using separate objectives. But if successful, the approach could provide researchers with another fluorescence imaging tool.
Switching at high speed
Dronpa is also the focus of research that seeks to uncover exactly how it goes from bright to dark and back to bright again. Because of the structural similarities between Dronpa and other photoswitchable fluorophores, such information could be useful.
“Understanding their photoswitching mechanism would be helpful to develop the best protein for a certain application by tuning its photoswitching efficiency,” said Johan Hofkens, a physical chemistry professor at Catholic University of Leuven. He added that better proteins could be used in subdiffraction-limited microscopy and as optical highlighters for protein tracking.
Hofkens was part of a group at the university that, along with researchers from the Riken Brain Science Institute, investigated how Dronpa changes from its dark back to its bright incarnation when prodded by violet light. The dark form is protonated and neutral, whereas the bright form is deprotonated and anionic. The spectral changes associated with photoswitching are similar to those related to pH changes. Because of these facts, the theory was that the back conversion involves efficient excited-state proton transfer from the dark to an intermediate form, which then evolves into the bright form.
To test this idea, the researchers turned to ultrafast transient absorption spectroscopy. The group at Catholic University developed a custom setup, with a light source consisting of a Ti:sapphire laser and optical parametric amplifiers, both from Spectra-Physics of Mountain View, Calif. The scientists used a CCD camera from Princeton Instruments of Trenton, N.J., for polychromatic detection and a Hamamatsu photomultiplier tube for a monochromatic detector. The setup allowed them to spot very small absorption changes — crucial when dealing with biological samples and small signals.
The researchers in Japan provided an ample supply of the protein. “It was necessary to use a flow cell to avoid back photoconversion, and very large sample volumes with a relatively high concentration were needed,” Hofkens said.
The spectroscopy results showed various time constants in the reaction, ranging from 2.0 up to 90 ps. One that measured 4 ps appeared only when back conversion was taking place, and it was close to the calculated value of 3.4 ps derived from the reaction’s quantum yield. It also was affected by the presence of deuterium oxide. The researchers therefore concluded that proton transfer is indeed involved in the photoswitching. “We were pleased to confirm our assumption by an independent measurement,” said postdoctoral researcher Cristina Flors. The work is detailed in the April 25 issue of the Journal of the American Chemical Society.
As for the future, both kinetic and structural information is needed to develop better photoswitchable proteins. Currently, different instruments and techniques are required to extract this total picture. “We are also exploring the possibilities of ultrafast infrared spectroscopy to simultaneously retrieve kinetic and structural information,” Hofkens said.
Can’t avoid the blues
Researchers using fluorescent proteins to highlight activity within a cell have had a case of the blues, and Robert E. Campbell wants to fix that. Campbell, Canada Research chair in bioanalytical chemistry and assistant professor of chemistry at the University of Alberta in Edmonton, is part of a research team that includes others from the university and from Howard Hughes Medical Institute at the University of California, San Diego. The researchers engineered improved versions of blue fluorescent proteins that should make multicolor fluorescent imaging more practical and robust. They also have developed variants on fluorescent proteins of other colors.
Current blue fluorescent proteins suffer from two drawbacks. They are relatively dim, which makes signals hard to spot, and they photobleach quickly, possibly within seconds, unless special precautions are taken. Thus, by the time researchers find a cell and bring it into focus, the fluorescence could have subsided substantially. As a result of these problems, blue fluorescent proteins have not been embraced. “Blue fluorescent proteins have been viewed with suspicion as far as their utility in multicolor imaging is concerned,” Campbell said.
He added that one of his own research goals involves developing useful fluorescent proteins that span the visible. With that done, pairs could be created for Förster resonance energy transfer (FRET) imaging. Better blue fluorescent proteins would help achieve this goal and also could benefit plant biologists because chlorophyll fluoresces in the red.
In developing an improved blue fluorescent protein, the researchers used random mutagenesis, creating bacterial colonies that expressed variants of the fluorescent protein. The researchers induced mutations into enhanced blue fluorescent protein, then screened the colonies for fluorescence using a 175-W xenon lamp passed through a 375- to 415-nm bandpass filter for a light source. They measured the resulting fluorescence with a CCD camera from QImaging of Surrey, British Columbia, Canada, fitted with a 440- to 480-nm bandpass filter. They captured the full emission spectra of those colonies identified as promising with a plate reader from the Tecan Group of Männedorf, Switzerland.
This image shows living HeLa cells transiently transfected with a blue fluorescent protein variant, EBFP2, fusion vectors. The cells were allowed to recover for 48 hours before imaging. Courtesy of Michael Davidson, Florida State University.
Azurite, a relatively new blue fluorescent protein, is 190 times as photostable and some 2.8 times as bright as enhanced blue fluorescent protein. By combining the mutations the researchers created in their variants with those in Azurite, they produced what they dubbed EBFP2 — 550 times as photostable as EBFP and four times as bright.
Campbell said that the researchers originally thought that the poor properties of blue fluorescent protein come from the chromophore structure. However, because the best one was a descendant of the original blue fluorescent proteins, that clearly is not the case. Also, he noted, much of the improvement in EBFP2 resulted from incorporation of the mutations found in Azurite. The work is published in the May 22 issue of Biochemistry and in the Dec. 15 Biochemical Journal.
Campbell sees possible further improvements of blue fluorescent protein variants and noted that his group will be working toward another goal. “We intend to continue to pursue the development of spectrally orthogonal FRET pairs that incorporate EBFP2 and mTFP1, another new fluorescent protein we recently described.