Fluorescence and Raman microscopy

Roel Arts
Biomedical applications have been a driving force in microscopy innovations since the inception of the microscope. In order to better understand the complex biochemical processes that occur in any kind of living material, better imaging techniques are always needed. The two questions that any kind of biomedical microscopy experiment aims to answer are: "what's in there and where is it?" The more information that can be obtained from a sample, the better.

Broadly speaking, there are two major ways to obtain specific biochemical information. One strategy is to somehow label the molecules of interest and then look for these labels. This is what happens in fluorescence microscopy: a fluorophore—a molecule that glows when illuminated— or "dye" is made to bind to specific molecules. After adding the dye to the sample and washing it, the only fluorophores left under the microscope are the ones that are bound to the molecule of interest. In addition, these fluorophores all have specific colors they emit and specific colors that cause them to glow ("excitation wavelength"). By controlling the color of excitation light and filtering the emitted light, the position of the fluorophores—and hence the molecules of interest—can be obtained.

Another method for obtaining specific molecular information is a more direct one. A large number of spectroscopic techniques are available for identifying molecules. Most of these techniques rely on measuring the vibrational energy levels of a molecule. By measuring the vibrational energy levels of a molecule, a "fingerprint" of it is obtained. One technique for measuring these vibrational levels makes use of Raman scattering, which is a nonlinear optical process.

Raman scattering begins when an incoming photon gets absorbed by the molecule, which brings it to a so-called "virtual" energy level. Normally, this molecule would then emit a photon with exactly the same frequency (Rayleigh scattering). However, it was shown in the 1920s that this is not always the case; sometimes, a slightly lower-frequency (lower-energy or "Stokes shifted") photon is emitted. The difference in energy between incoming and outgoing photon just so happens to match the energy difference between two of the molecule's states; the energy of the photon enabled the molecule to reach a vibrational energy level. If the molecule happened to already be in a higher vibrational energy level when absorbing the photon, it can also fall back to a lower vibrational energy level than before. In this case, the molecule will emit a photon with a higher energy than the one it absorbed ("anti-Stokes shifted"). So by shining some monochromatic light on a sample, measuring the newly-generated colors, and subtracting the original color, the energy differences between vibrational levels of molecules in the sample can be measured. This is called Raman spectroscopy.

The big downside to this technique is that most of the incoming light will not be Raman-scattered. Most molecules have a small absorption cross-section for Raman scattering, which means that the number of photons generated by this process will be low and the signal will hence be weak and hard to detect.

For this reason, fluorescent labeling and Raman microscopy are difficult to combine; the (much) larger signal emitted by fluorescent labels will effectively drown out the Raman signal ("cross-talk"). A recent publication by the Biophysical Engineering Group (BPE) from the University of Twente claims that they have managed to combine the two techniques. This was achieved by making sure that the colors generated by Raman scattering, and the colors emitted by the fluorophores are far enough apart. They used semiconductor quantum dots (QDs) as the labeling agent rather than standard molecular dyes. QDs have been described as "artificial atoms" because they are designed to confine a small number of electrons spatially, which causes them to occupy a discrete set of energy levels, like in an atom. Tailoring these energy levels and hence their emission spectrum, can be done by carefully designing symmetry as well as dimensions of the quantum dots. BPE's letter in Nano Letters demonstrates two experiments in which QDs are used for two hybrid fluorescence/Raman microscopy experiments. Despite the QDs high luminescent yield, they were able to effectively perform resonant- and non-resonant Raman microscopy using a QD-stained sample.

Combined resonance Raman (RR) / one-photon excitation (OPE) imaging was demonstrated using neutrophils (white blood cells). In RR imaging, the sensitivity is improved compared to non-resonance Raman (NR) scattering because the energy of the incoming beam is chosen to make certain vibrational transitions more likely; this will increase the amount of light that gets Raman scattered. OPE is just the "regular" fluorescence process; one photon gets absorbed and a lower-energy photon gets emitted.

A fluorescent image was obtained by labeling specific parts of the cells with 15-20 nm diameter QDs. In addition, the researchers looked for the RR spectrum of flavocytochrome b558, which has a well-known RR spectrum. The sample was illuminated with 413.1 nm UV radiation, which resulted in the QDs emitting 605 nm radiation and the RR spectrum of flavocytochrome b558 being emitted in the 420-445 nm range. These two emitters were well separated and could be imaged independently of each other. The RR spectrum was not found to be affected by the presence of the quantum dots, but the QDs were shown to bleach (become inactive) after a certain amount of illumination, which makes it necessary to obtain the OPE signal before measuring the RR signal.

Combined non-resonance Raman / two-photon excitation (TPE) imaging was done on macrophages, another type of white blood cell. TPE differs from OPE in that the emitted photons have a higher energy than the absorbed photons. In this experiment, a slightly different QD type was used. The sample was illuminated with 647.1 nm radiation resulting in the QD's emitting 585 nm radiation. The nonresonant Raman (NR) spectrum was then measured in the 660-730 nm region (Stokes shifts of 300-1800 cm-1, "fingerprint region") as well as 770-870 nm (2500-4000 cm-1, high frequency region). These spectral regions allowed the researchers to identify certain proteins and lipids.

Although again no cross-talk was observed, it was also observed that the cells were damaged after illumination, presumably due to localized heating due to the presence of the quantum dots. This unexpected complication limits the amount of excitation power that can be delivered to the sample without wrecking all those interesting white blood cells.

The significance of this work is not that it shows new fundamental advances, but rather that it demonstrates the significance of being able to tailor the luminescent properties of QDs. In this case, it is the application that is important; by using a QD labeling agent, areas of interest within a sample can be identified, after which Raman-scattering can be used to get a detailed view of the molecules and molecular processes involved. Because of the highly-tunable optical response of quantum dots, this technique can be extended to any kind of sample where QD labeling is viable.