Third Harmonic Generation Microscopy
Label-Free 3D-Tissue Imaging and Blood Flow Characterization
- Third Harmonic Generation Microscopy - Label-free 3D-tissue Imaging and Blood Flow Characterization
- Fig. 1: Mouse cremaster muscle arteriole visualized with THG. Among other structures, two parallel lines from bottom left to top right are seen (asterisks). They are part of the arteriolar vessel wall. Smooth muscle cells (s) on the outside and endothelial cell nuclei (e) on the inside can be recognized. A nerve fiber (n) cuts through the focal plane, striated muscle fibers (f) are parallel to it. This is a clipping from the article title image where in addition smooth muscle cell actin (antibody staining, Cy3, orange) and nuclear staining (To-Pro-3, red) is shown. After staining with several washing steps, no red blood cells remained in the vessel. Single optical section, scale bar in title image 20 µm.
- Fig. 2: Intravital THG imaging of flowing red blood cells. Arrow indicates flow direction, the dotted line outlines the vessel wall. The scanning process added additional lines from right to left. At a moderate flow speed (a), this causes cells in the main vessel to appear elongated. At high flow speed (b), the cells cause streaks elongated in the same direction. In the smaller vessel towards the left, flow speed of the cells is approximately as fast as the addition of scan lines to the image; cells cause extended smears. Scale bar 10 µm.
- Fig. 3: THG line scanning. (a) Scheme (from ) showing a red blood cell flowing along the scan line from left to right at four time points T1 – T4. At the bottom an xt-representation with the resulting 4 scan lines is shown. (b) xt-representation from a line scan in mouse cremaster arteriole with 30 µm diameter. Flow of cells was from right to left. (c) Blood flow velocity in that arteriole over time. Pulse curve and the heart rate are clearly discernable.
Third Harmonic Generation (THG) microscopy is a non-fluorescent multi-photon technique that combines the advantages of label-free imaging with restriction of signal generation to the focal spot of the scanning laser. It allows three-dimensional imaging of refraction index mismatches and of hemoglobin. We applied it to image mouse tissues and to characterize blood vessels and blood flow velocity. In small arteries, THG line scanning reveals the pulse velocity curve and hence the heart rate.
Third Harmonic Generation
A physicist may see Third Harmonic Generation (THG) as a well described multi-photon effect. Biomedical researchers are, however, still mostly unaware of the possibility to use THG as a signal generating process in light microscopy. This is most likely due to the rarity of the required excitation source. A conventional tunable Titanium Sapphire laser with wavelengths up to about 1000 nm as found in most multi-photon microscopes (MPMs) in life science labs is not sufficient. It does meet one requirement, providing a pulsed laser beam for high photon density, but not another, an extended wavelength range.
THG is a process in which the energy of three incoming photons is combined to generate one outgoing photon with a wavelength of exactly one third of the excitation beam. All the photons' energy is converted. None is deposited in the specimen, as during a fluorescence excitation and emission cycle. Thus, to create a THG signal in the visible range above 400 nm wavelengths, an excitation source with a wavelength over 1200 nm is required. While a signal also is generated with shorter excitation, this UV signal will be largely absorbed by glass lenses.
In a life science lab, a suitable light source usually would be an Optical Parametric Oscillator (OPO) pumped by a Titanium Sapphire laser. Recently, lasers tunable from around 700 to 1300 nm became an alternative choice. As a rule of thumb, an MPM designed for the excitation of red fluorescent proteins will also provide the capability to perform THG microscopy.
Therefore, quite a few newer MPMs will ‘accidentally' also allow for THG imaging.
A microscope designed for THG or Second Harmonic Generation (SHG) should be equipped with forward detectors. Unlike fluorescence which is emitted equally in all spatial directions, the bulk of those signals is generated parallel to the excitation beam, in forward direction . When forward detectors are unavailable or cannot be used due to spatial constrictions in live animal microscopy, signal intensity from transparent samples may be significantly increased at the normal backward detector by placing a mirror under the sample. The reflection of a large part of the forward-generated photons towards the detector can lead to a several-fold intensity increase .
THG and SHG in the Life Sciences
So what good is THG microscopy and what is the difference to SHG microscopy? In SHG, the energy of two photons is combined to create one with half the wavelength of the excitation beam. Therefore, most multi-photon microscopists will have observed such a signal e.g. at 400 nm when exciting their fluorochromes at 800 nm. SHG, however, is limited to non-centrosymmetric, dense structures. In mammalian soft tissues, that essentially equals collagen fibers and striated muscle myosin. In plants, starch granules and cellulose are SHG sources, among others . Microtubules have been reported to induce SHG, but apparently so weak that in tissue imaging, we never detected them.
THG, on the other hand, is induced by several mechanisms. One is an interface in the specimen, such as a refractive index mismatch. As a result, structures are highlighted that also would come up in phase contrast or interference contrast microscopy. THG, however, is induced only at the focal point of the excitation laser, enabling 3D-imaging of tissues.
Another THG mechanism is resonance enhancement. When a chromatic material absorbs the wavelength of the produced THG signal, contrary to intuition, this causes an increase in signal intensity by resonance processes. The strong THG signal of red blood cells was explained by this effect with hemoglobin .
THG Microscopy of Tissues
I first became fascinated by THG microscopy after hearing talks from CK Sun's group at the Focus on Microscopy meeting 2008, which incidentally took place while we were waiting for the delivery of an MPM with an OPO. We began using the mouse cremaster muscle as a sample tissue (fig. 1), a widely used animal model in physiology research. Thanks to the long excitation wavelengths of 1275 nm, reducing scattering and absorption, full penetration of the 150 - 200 µm of that tissue is easily achieved. THG signals were sufficiently strong for informative imaging of blood vessel walls and blood cells, muscle fiber sarcomeres, nerve fibers and nuclei of some cell types. We also observed migration of leukocytes through the tissue .
The strong THG signals created by red blood cells spurred the desire to use it for characterization of the microcirculation also in living animals . The microcirculation, i.e. capillaries, small veins called venules, and small arteries, the arterioles, is the site of inflammation as well as the regulation of blood flow and blood pressure. With THG imaging, we visualized flowing blood cells (fig. 2), showing for example the cell-free layer at the edge of the lumen of blood vessels . We also could perform velocity measurements if the vessel was oriented parallel to the scanning direction of our scanner.
THG Line Scanning
The more versatile approach for velocity measurements, however, proved to be line scanning where a single line is scanned at high speeds, e.g. 1000 - 2000 Hz. Introduced to two-photon excitation fluorescence microscopy over 15 years ago  and even earlier to confocal fluorescence microscopy , we are, to our knowledge, the first ones to have applied it to a label-free technique . If an erythrocyte flows along the line, its signal moves progressively. In an xt-representation of the line, this causes a diagonal streak (fig. 3a, b). From the length of the scan line and the number of lines per second, the velocity of the erythrocyte can be calculated.
Tyson N. Kim et al. developed an algorithm called ‘line-scan particle image velocimetry' to evaluate line scanning data with fluorescent labels . Fortunately for us, they also made it available on the internet for free download. With an adapted version, we monitored velo-city changes over time, allowing for example to visualize the velocity pulse curve in arterioles (fig. 3c).
In the mouse ear, intravital THG imaging and line scanning led to label-free and non-invasive determination of blood flow velocity, velocity pulse curve, heart rate, velocity profiles over the blood vessel width, the shear rate and, via detection of the vessel wall, blood vessel diameter . Thus, potential physiological perturbations by invasive procedures and/or the administration of labels can be avoided.
In vessels wider than ~25 µm, THG signals from the central axis were weaker than from more peripheral regions. Most likely the 425 nm signal gets absorbed by hemoglobin between the site of signal generation and the objective. This effect limited observation of axial flow to vessels of about 50 µm in diameter. There is, however, hope that this will substantially improve in the future. The full width half maximum of the point spread function at 1275 nm of our system was about twice as wide in each direction as it should be. For the three photon process THG, diffraction limited optics might therefore produce an about 60 times brighter signal, and thus accordingly increase penetration into blood vessels . It is to be expected that future objectives specifically designed for up to 1300 nm excitation will perform much better, providing diffraction limited optics, and maybe some newer objectives are already today.
Another concern is excitation power. THG signal intensity increases with the third power of excitation intensity. Our system transmits up to around 200 mW behind the objective at 1275 nm and thus much more than other commercial systems we could test. In many settings we did use that maximal power and sometimes we would have used more if available. While such a value around 800 nm would easily melt the tissue, the low absorption between 1200 and 1300 nm prevented damage except when a small area was scanned at slow speeds. That is, as long as no black dust specks or melanin spots in the skin are present which heavily absorb and cause immediate destruction of their environment. Diffraction limited optics and adaptive optics  may greatly help produce the same signal strength with much lower excitation powers.
THG and Fluorescence
THG is well compatible with two-photon excited fluorescence, thus providing additional information without the need for additional labels. We found that our typical THG excitation wavelength 1275 nm also excites some far red fluorochromes such as the nucleic acid stain To-Pro-3 (title image). Other dyes such as Cy3 (title image) or GFP can be recorded sequentially. In time-laps observations of processes visualized with GFP or other dyes, THG measurements of hemodynamic parameters before and after the observation allow to measure blood flow and heart rate without the need of additional labels.
I thank my coauthors [2,5,6], B. Balluff and K. Wachal for their collaboration in developing the described techniques and/or for their help with the primary data for the figures shown.
 Moreaux L. et al.: Biophys J 80: 1568-1574 (2001)
 Rehberg M. et al.: J Biomed Opt 15: 026017 (2010)
 Cheng P-C and Sun CK. In: Pawley JB, editor. Handbook of biological confocal microscopy. 3rd ed. Springer Science+Business Media LLC 703-721 (2006)
 Chang CF. et al.: J Biophotonics 3: 678-685 (2010)
 Rehberg M. et al.: PloS ONE 6: e28237 (2011)
 Dietzel S. et al.: PloS ONE 9: e99615 (2014)
 Kleinfeld D. et al.: Proc Nat Acad Sci U S A 95: 15741-15746 (1998)
 Dirnagl U. et al.: J Microsc 165: 147-157 (1992)
 Kim TN. et al.: PloS ONE 7: e38590 (2012)
 Thayil A. et al.: J Biomed Opt 16: 046018 (2011)
Dr. Steffen Dietzel
Walter Brendel Centre of Experimental Medicine (WBex)