A revolution in specific staining was initiated with the introduction of antibody-mediated immunohistochemical stainings and DNA hybridization-based stainings. These allow a single protein or DNA sequence to be highlighted within a huge background of chemically similar molecules that could not be differentiated with the classical chemical dyes. The latest important progress in this context has been made with fluorescent proteins. These are introduced to stain structures, molecules and cells or are used in sensors for metabolites, ions and other species in living cells and organisms to provide a window for the observation of the machinery of life – in live mode. Nevertheless, all these staining procedures involve more or less invasive interventions, and require dead or fixed preparations in many cases. Additionally, the fluorescent protein method depends on the expression of foreign proteins in the target cells that often significantly interferes with the sensitively regulated homeostasis in cells. Therefore, methods are explored that allow imaging of bulky samples without any staining.
Fig. 1: Left: Overlay image of two recordings from sandwich sauce: Lipids are shown in red, water in green, 25x objective, scale bar 25 μm. Right: Study of localization and grade of accumulation of lipids in different types of macrophages: Lipids are displayed in red, actin in green, 63x objective. Courtesy: Alba Alfonso Garcia and Jeff Suhalim, Potma Lab, University of California, USA. Cells: Dr. Adelheid Kratzer, Department of Medicine Division of Pulmonary and Critial Care, University of Colorado.
Rayleigh and Raman scattering
Light can interact with matter not only by absorbing photons followed by the molecule assuming a higher excited state, but also by assuming what is called a "virtual state". In that case, the photon does not have the quantum of energy that would fit any of the state differences in the given molecule. For that reason, the released photon is not Stokes-shifted, as there is no thermal substate series in a virtual state which could cause the emission shifted to the red (in fluorescence, the excited state will first rapidly relax to the lowest vibrational substate of the excited state, releasing thermal energy which leaves the emitted photon with less energy – a redder tone). Under normal scattering conditions, the incident and the scattered photon therefore have the same energy, which is the same color, and cannot carry any specific information from the interacting molecule. This standard case is called “Rayleigh scattering” (Figure 2: Y).
Nonetheless, there is a chance that the emission does not cover the full span to the lowest vibrational state of the molecule, but, for example, the energy only returns to the first thermal state. In that case, the emitted photon only lacks the energy difference of that thermal state to the ground state. The emission is red-shifted. This phenomenon is known as Raman scattering (here, Stokes scattering, see below for an alternative, Figure 2: S). As the energy differences of these substates are very specific for molecular structures, the energy difference between incident and released photons reveals details of the compound that was interacting with the light. Raman spectroscopy has therefore become an extremely useful tool in analytical chemistry. Alternatively, the molecule might just happen to reside in the vibrational state (which to some extent happens even at ambient temperature). In this case, the virtual state upon illumination is higher compared to the example discussed above. As the molecule will typically assume the lowest energy after emission, the scattered photon has a higher energy than the incident photon. This is called Anti-Stokes scattering (Figure 2: A). Both processes occur very rarely; the Stokes process some 10–3-10–4 times less than the Rayleigh process, and the Anti-Stokes process even one or two orders below that. For imaging, the efficiency of spontaneous Raman scattering is by far too low to generate a sufficient signal within a reasonable time.
Amplification of the Anti-Stokes Raman Signal
In order to generate useful intensities of Raman photons, it is necessary to artificially increase the population of the first vibrational state. When this is achieved, many photons will emit at Anti-Stokes wavelength, and a measurable signal is generated.
To achieve this goal, an intense infrared beam (pump beam) is initially focused into the sample by using a standard pulsed high power IR laser. This beam will promote many virtual states in an ensemble of molecules. To avoid their return to the zero vibrational level, a second beam is applied (probe beam), which stimulates the return from the virtual state, much like a STED beam that returns excited states to the ground state. If the second beam is tuned to exactly match the first vibrational state (meaning that the energy must be the difference of pump beam and vibrational energy), then the molecules will not assume the zero level in the ground state. As a consequence, many molecules in the focus have been promoted to the vibrational state. The second laser is a tunable IR laser, and modern systems use a compound device that provides both wavelengths coaxially.
These two processes occur more or less simultaneously. Both wavelengths need to illuminate the sample. Consequently, as the pump beam is still on, the molecules can assume a higher virtual state, starting off from the first vibrational ground state. They will relax from this virtual state to the zero substate of the ground state and consequently emit photons with higher energy than the pump beam. The difference is exactly the energy of the first thermal state. All we need to do is collect this scattered light and separate it from pump and probe beam. It is not necessary to fit emission bandpass filters to the anti-Stokes wavelength, as the specific signal is produced by populating a large number of thermal states that are specific for the compound in question (Figure 3). The specific signal is the difference between pump and probe, which has to be tuned to select the desired compound.
Things we can see now
As the thermal energies correlate very specifically with chemical bonds, it is possible to localize different bond types in the sample. The contrast is based on the specific thermal excitation and the signal will scale with the concentration of the compound. CH-CH bonds are highly concentrated in lipid bilayers and other non-polar structures. Therefore, it is possible to image the lipid areas against the polar (mostly aqueous) background – without any fluorescence staining procedure. This has proven to be very successful with many types of samples. Whole organisms, like drosophila embryos, adult insects and tissue samples, cells and even living yeast cells become visible without any staining procedure. In combination with a fast scanning system, e.g., the resonant scanner in the Leica TCS SP8 CARS, secrets inside cells can be monitored under physiological conditions. Food samples can be examined for lipid-aqueous distribution of components and a huge range of applications is opened up in polymer research and industry.
Fluorescence microscopy assumed a pivotal role in cell biology once it was possible to stain cell components selectively by fluorescing dyes. One of the first explorers of targeted stainings, Paul Ehrlich, had the idea that something that stains specifically should also kill specifically – which was associated with the term "magic bullet", the essential idea of chemotherapy. His group discovered Salvarsan, a tailored drug against syphilis – though not specific enough not to cause substantial side effects. Screening many fluorescent dyes led to a long list of stainings which are used in histology, including dyes like DAPI or hematoxylin and eosin.
