Single-molecule fluorescence spectroscopy and imaging

Observability of individual fluorescent molecules under a microscope has many applications in biology.^1,2 In particular, by grafting fluorophores to biomolecules, they can be tracked in real time in live cells for monitoring of their conformational changes and interactions with other molecules. In other words, single-molecule imaging allows scientists to read the textbook of life pretty much as they will end up writing it, i.e., as a series of molecular events involving one or a few partners engaged in a spatiotemporal ballet. Observing single molecules is challenging because very few photons are emitted and even fewer are detected. There is, therefore, a strong incentive to improve detector performance to increase their temporal resolution and throughput.

Figure 1. Operational principle of our photon-counting camera. A fluorescence photon (with energy hν) is collected by the imaging optics and interacts with the photocathode, creating a photo-electron (e^- ). This photo-electron is amplified by a microchannel-plate stack, which thus generates an electron cloud. Measurement of the delay between the charge pulse at the back of the MCP and the laser pulse provides the nanotime information (τ). A position-sensitive anode determines the position (X, Y) of the electron cloud. A clock provides coarse timing information (T). The four coordinates are asynchronously sent to a computer for storage and processing.

Single-molecule observations are usually performed in either wide-field or confocal geometry, using electron-multiplying CCD cameras or single-photon point detectors, respectively.³ Wide-field imaging with a camera allows looking at many molecules that are immobilized on a surface or in live cells, but this is limited by the camera's noise and finite frame rate. Confocal microscopy, on the other hand, only illuminates a single microscopic spot in the molecular sample. Light emitted from this spot is collected by a point detector such as a single-photon-counting avalanche diode (SPAD). This approach has excellent temporal resolution but suffers from very low throughput.

We have collaboratively explored several strategies to take advantage of the best of both worlds, i.e., the wide-field imaging capability and large parallelism of cameras and the high temporal resolution and noise-free detection of single-photon-counting detectors.

In collaboration with the group of Oswald Siegmund (Space Sciences Laboratory of the University of California at Berkeley), we developed a photon-counting camera (see Figure 1) based on a large-area photocathode combined with a microchannel plate to amplify each photo-electron.⁴ A position-sensitive anode measures the location of each impact, while timing electronics provide high-temporal-resolution information for each photon. Our first-generation detector has allowed us to detect, track, and perform high-temporal-resolution studies of individual quantum dots, a novel type of single fluorescent probes with a potentially great future in biology.⁵ This photon-counting camera is also an excellent fluorescence-lifetime imaging device when used in conjunction with pulsed laser excitation,^6,7 potentially outperforming standard approaches based on confocal or time-gated, camera-based imaging.

Figure 2. New single-photon-counting avalanche-diode (SPAD) arrays used in our single-molecule studies. (A) Custom CMOS technology allows arraying SPADs in a tight package (left) but is limited to a relatively small number of devices (eight: red rectangle in right inset). (B) Standard CMOS technology allows construction of very dense arrays of SPADs (32×32, left), incorporating quenching and counting electronics (single-cell detail, right) at the expense of lower sensitivity. VLQC: Variable-load quenching circuit.

Jointly with the group of Katsushi Arisaka (Department of Physics and Astronomy, University of California at Los Angeles), we explored the use of hybrid photon detectors developed by Hamamatsu for single-molecule confocal detection. We were the first to emphasize that they lack the afterpulsing phenomenon that prevents SPAD use for short-timescale autocorrelation analysis.⁸ A new multipixel version of this detector seems a potentially very interesting device for high-throughput, single-molecule spectroscopy.

Our most recent work is done in collaboration with the group of Sergio Cova, Massimo Ghioni, and Franco Zappa (Politecnico di Milano, Milano, Italy), who have developed advanced prototypes of SPAD arrays (see Figure 2). SPAD arrays manufactured in a custom CMOS technology exhibit identical pixel performance as the single-pixel devices, but provide higher throughput because of the large number of pixels per array.⁹ We recently demonstrated parallel detection of single molecules in an 8×1 linear geometry.¹⁰ Standard CMOS technology has lower sensitivity (it is not optimized for photon detection), but allows tight integration of quenching and counting electronics on the same chip as the SPADs, resulting in very dense arrays of SPADs.¹¹ We obtained extremely encouraging preliminary results with a 32×32pixel² prototype, which promises to enhance the throughput of single-molecule spectroscopy by three orders of magnitude.¹²

The future of single-molecule imaging and spectroscopy calls for higher throughput and better temporal resolution. Although none of the different technologies we have explored can meet the requirements of all single-molecule approaches, each one has specific strengths that makes them well suited to fulfill these goals for one or more applications. Upcoming advances in detector sensitivity, array size, and data-processing algorithms will help make these goals a practical reality in the near future, to which we will actively contribute.

We thank our many colleagues in the Weiss group, our long-time collaborators on the photon-counting camera project at the University of California at Berkeley (Oswald Siegmund, Anton Tremsin, and John Vallerga), Katsushi Arisaka's group in the Physics and Astronomy Department at the University of California at Los Angeles, and Motohiro Suyama's team at Hamamatsu, as well as the group of Sergio Cova (Massimo Ghioni, Franco Zappa, Ivan Rech, and their talented collaborators and students) for very enjoyable collaborations. This work was funded by grants EB006353 and GM084327 of the National Institutes of Health and National Science Foundation grant DBI-0552099.