Designed and performed the study: MR SD. Conceptual input: MR FK UP SD. Wrote the manuscript draft: SD. Contributed to the manuscript draft: MR. Critically evaluated the results and finalized the manuscript: MR FK UP SD.
The authors have declared that no competing interests exist.
Second and Third Harmonic Generation (SHG and THG) microscopy is based on optical effects which are induced by specific inherent physical properties of a specimen. As a multi-photon laser scanning approach which is not based on fluorescence it combines the advantages of a label-free technique with restriction of signal generation to the focal plane, thus allowing high resolution 3D reconstruction of image volumes without out-of-focus background several hundred micrometers deep into the tissue. While in mammalian soft tissues SHG is mostly restricted to collagen fibers and striated muscle myosin, THG is induced at a large variety of structures, since it is generated at interfaces such as refraction index changes within the focal volume of the excitation laser. Besides, colorants such as hemoglobin can cause resonance enhancement, leading to intense THG signals. We applied SHG and THG microscopy to murine (
Light microscopy is the method of choice for visualizing cells in their biological or physiological context. The classic light microscopy techniques such as Köhler illumination and phase contrast can be applied without prior labeling, allowing the observation of unperturbed cells or tissues. However, these techniques are not suitable for thick tissues and they do not allow high resolution three-dimensional reconstruction of image volumes, since the microscopic image also contains data from above and below the focal plane. This limitation can be circumvented by restricting image generation to photons originating from the focal plane as achieved by confocal laser scanning microscopy or by two-photon excited fluorescence microscopy. Both approaches require a fluorescent structure, and thus except for a few autofluorescent structures staining with structure specific fluorescent dyes such as DNA stains or fluorescent antibodies is necessary. Images from adjacent focal planes then can be combined to generate a three-dimensional image volume. For staining of living cells or tissues, however, it often remains unclear how far the staining perturbs physiologic functions of cells or tissues. For example, the expression of GFP or GFP fusion proteins was linked to induction of apoptosis
Second and Third Harmonic Generation (SHG and THG) microscopy
In THG, the energy of three simultaneously incoming photons is combined to emit one photon with one third the wavelength. THG is more complex to realize because detection of the signal in the visible range above 400 nm requires an excitation wavelength of above 1200 nm, a wavelength range not provided by standard two-photon microscopes. Detection in the visible range is not an absolute requirement dictated by physical theory, but rather by the optical elements of standard microscopes which have limited transmission for UV light. In addition, Rayleigh scattering with accompanying light loss in tissues increases with the 4th power with shorter wavelengths
While SHG and THG are well-established phenomena in the optical sciences, the biomedical community is less aware of their potential. To evaluate the capabilities of these label-free techniques, we applied THG and SHG microscopy to an established tissue model. The rodent cremaster muscle can be prepared from the scrotum as a thin, light permeable sheet and thus became widely-used for in vivo physiological research with conventional bright field microscopy (e.g.
Tissue extraction was performed according to German legislation for the protection of animals. Male C57BL/6 mice at the age of 10 – 12 weeks were purchased from Charles River (Sulzfeld, Germany). Animals were housed under conventional conditions with free access to food and water. The Institutional Board of the Walter-Brendel-Zentrum has approved the tissue extraction after euthanization of the animals (approval ID K09), according to the local law "Bayerisches Tierschutzgesetz, Paragraph 6 Abs. 1 Satz 2 Nr. 4". The animals were euthanized by an intraperitoneal pentobarbital overdose (Narcoren, Merial, Germany). Subsequently, the right or left cremaster muscle was exposed through a ventral incision of the scrotum. The muscle was opened ventrally, and epididymis and testicle were detached, subsequently the cremaster was explanted. Throughout the procedure as well as after surgical preparation during microscopy, the muscle was immersed in buffered saline or 0.9% sodium chloride solution.
Structures and cells in isolated cremaster muscles were studied without fixation or other further treatment, except for those counterstained for DNA. For DRAQ5 staining, freshly excised, unfixed tissue was incubated for 15 – 60 min at room temperature in 0.9% NaCl with a 1∶1000 dilution of the 5 mM stock solution (Biostatus Ltd, Shepshed Leicestershire, UK). Microscopic observation was in 1∶10 diluted staining solution. For TO-PRO-3 (Invitrogen, Karlsruhe, Germany) staining, excised tissue was fixed for 10 min with buffered 4% paraformaldehyde (Microcos, Garching, Germany). Staining was for 1 hour in PBS with 2 µM TO-PRO-3 and 0.5% Triton X-100 (Sigma). TO-PRO-3-stained tissue was mounted in PermaFluor (Beckman Coulter, Fullerton, CA) on glass slides.
For imaging of unstained, unfixed cremaster preparations at room temperature, the muscle was placed on a standard microscopic glass slide or on a round cover slip and covered with sodium chloride solution (0.9%) solution. Round cover slips were mounted with vacuum grease (Baysilone-Paste mittel-viskös, GE Bayer Silicones GmbH & Co. KG, Leverkusen) in a plastic Petri dish with a hole in the bottom. On the sides, the edges of the muscle were weighed down to avoid movement of the tissue. Such cremaster muscle preparations have a thickness of 160 – 200 µm. For observation of leukocyte motility the tissue was placed in a custom-made heating device, submerged in Hank's Balanced Salt Solution and maintained at physiological temperature.
For the microscopy of thigh muscles, a longitudinal 10 mm incision through the skin of the right hind limb of a sacrificed mouse was made, beginning at the inguinal crease. The skin was removed to expose the musculus aductor magnus. The muscle was either directly immersed with saline, or excised and placed in a Petri dish filled with saline.
Microscopic observations were performed using a commercially available system, a TriMScope (LaVision BioTec, Bielefeld, Germany) built around an Olympus BX 51 microscope (Olympus, Hamburg, Germany) and equipped with a tunable Ultra II Titanium:Sapphire Laser (Coherent, Dieburg, Germany) and an Optical Parametric Oscillator (Chameleon OPO; APE, Berlin, Germany; typical pulse width 200 fs, repetition rate 80 MHz) which is pumped by the Ti:Sa. The OPO generated 1270 nm or 1275 nm light with 640–700 mW output. The unattenuated intensity at the sample (i.e. after the objective) was ∼250 mW. Scanning cremaster tissue with full intensity and high resolution (400x400 µm with 1680×1680 pixels and 200 lines per second) did not produce any visible damage, except if the area of observation contained light absorbing dirt which would induce plasma formation. An Olympus XLUMPlanFl 20×/0.95W objective was used (working distance 2 mm). The following detection channels were used: backward (epi) detection blue (417 – 477 nm), red (604 – 644 nm) and far red (645 – 695 nm), forward detection blue and longpass 488 nm. 700 nm short pass filters blocked out excitation light. Light collection in forward direction was performed by an Olympus WI-UCD condenser, NA 0.8. Photomultiplier tubes (PMTs) were Hamamatsu H6780-01 for the blue channels and H6780-20 for the others. Where mentioned (see main text) more sensitive gallium arsenide phosphide detectors (Hamamatsu H7422-40) were used for backward detection.
To characterize the performance of the system we measured the full width half maximum (FWHM) of signals from subresolution particles embedded in 1% agarose with 1275 nm excitation. Two-photon excitation fluorescence of deep red beads (size 175 nm, exc 633, em. 660, Invitrogen P7220) provided an FWHM of 1.1 µm laterally and 6.2 µm axially. This compares to theoretical values of (0.51*λ) / ((√2)*NA) = 484 nm laterally and (0.64* λ) / (n-√(n2-NA2)) = 1.7 µm axially
Images were processed in ImageJ, Fiji and/or Imaris. ImageJ (W.S. Rasband, U. S. National Institutes of Health, Bethesda, Md, USA,
Striated muscle and collagen fibers were inducers of SHG and THG (
A) The relative SHG signal intensity ratio collagen/myosin is higher in backward SHG (green) than in forward SHG (red). B) While in striated muscle SHG visualizes myosin and thus A-bands of the sarcomeres, THG signals are from the interjacent I-bands as well as from the muscle fiber border (b) and muscle cell nuclei (n). C) Forward THG signal of a muscle fiber running parallel to the optical axis, single optical section. The fiber border is delineated by a surrounding THG signal. The small bubble at the upper fiber border may belong to a cell nucleus. No SHG signal was detected in this case. D) Thick collagen fiber producing a double signal in THG (left, cyan in overlay) which is filled by the SHG signal (center, red in overlay). Scale bars 20 µm.
For some collagen fibers, we observed a double signal with a hollow core of 1 – 1.5 µm (
We visualized red blood cells (RBCs) in vessels of explanted, unfixed cremaster muscles, thus without flow, with forward detectors (
A) Projection of optical sections. Without flow, red blood cells (RBCs) sink to the bottom of the vessel, forming a tightly packed mass. B) Visualization of typical RBC shape. C,D) Leukocytes adhering to the inner vessel wall. E–G) Blood vessel with strong THG signal from the vessel wall. Vessel diameter is between 40 and 45 µm. E shows a 3D rendering with a 2x magnification in the inset. A section perpendicular to the vessel axis is shown in F and the same section is magnified in G. The arrows point to the prominent THG signal from wall sections parallel to the optical axis, the arrowhead points to the RBCs at the vessel bottom. All scale bars are 20 µm.
Many blood vessels within the tissue were identifiable by THG not only by their RBC content but also by strong signals originating from the vessel walls (
In the tissue, outside of blood vessels, we observed THG signals reminiscent of leukocytes. DNA staining after fixation confirmed that they were nucleated cells, with the shapes of some nuclei reminiscent of multi-lobed granulocyte nuclei (
A) Fixed cremaster muscle, optical section about 50 µm from the surface. THG (left, cyan) revealed shapes reminiscent of migrating leukocytes deep in the tissue. DNA counterstaining (right, red) confirmed that these shapes are nucleated cells. In some cases, nuclei were lobed, a morphology typical for mature neutrophil granulocytes (Gr). Note that some cells identified by DNA stain are not visible in THG. B) Projections of THG (cyan) and SHG (red) in a time series of a freshly explanted cremaster, recorded in forward direction. While most leukocytes were stationary one moved ∼25 µm in 385 seconds (arrowheads). Image on the right shows a yz-section (view from the side, at end of the time series) illustrating that the moving cell is inside the tissue, not at the surface. The respective movie is shown in
A,B) Single optical forward THG sections from unstained, explanted cremaster. A, Section near the surface, nuclei most likely belong to macrophages. B, In deeper regions, nuclei of muscle fibers (m) can be observed, in addition to leukocytes. C,D) Nuclei in the cremaster after DNA-staining with Draq5. Chromocenters colocalize in THG channels and by DNA staining. C, Single optical section. D, Projections of five sections spaced 3 µm each. Not all DNA-stained-nuclei are visible in THG. Scale bar 20 µm for all images.
THG also visualized the nuclear chromatin structure in unfixed tissue (
When excited with 266 nm, DNA shows autofluorescence with the maximum around 350 nm and extending up to around 425 nm
To our knowledge, this is the first visualization of unlabeled chromocenters or any heterochromatin structure in living cells in 3D. The distribution of chromocenters changes during mouse cell differentiation to patterns characteristic for each cell type (see
The strongest THG signals we observed in the mouse cremaster muscle in forward as well as in backward direction were produced by peripheral nerve fibers (
A) Unfixed, explanted muscle, projection of several optical sections covering 50 µm depth. Peripheral nerve (forward THG, cyan) surrounded by collagen fibers (forward SHG in red, backward SHG in green). Some blood vessels and small structures also produce THG signals. B–E) Single channel images of A near the nerve junction. Forward THG and SHG images have a much better signal to noise ratio than backward images. While forward and backward SHG signal differ substantially in some image details, backward THG (not included in A) does not contain any information in addition to the crisper forward THG image. F,G) Fixed cremaster with fluorescent DNA counterstain (red) and forward THG signal visualizing nerves. Several cell nuclei are associated with the nerve fibers. Scale bars are 20 µm, bar in F valid also for G.
As a potential source for the strong THG signal in nerve fibers, the packed membranes of the myelin sheath come to mind. Indeed, it was demonstrated earlier this year that nerve fibers in the mouse CNS also produce a strong THG signal and that this signal colocalizes with specific myelin staining
For tissues substantially thicker than the mouse cremaster (150 – 200 µm), detection of transmitted forward generated SHG and THG signals is not an option. Instead, only backward detected signal can be obtained. While backward detection is hampered by a lower signal-to-noise ratio than forward detection, we did obtain backward detected SHG and THG signals through the whole depth of the cremaster. To estimate the depth until which backward signals can be recorded from thicker muscle tissue, we imaged mouse thigh muscles. A layer of connective tissue with a thickness of 100 – 250 µm was detected on top of striated muscle fibers (
The first 100 µm consist of connective tissue, followed by a dense meshwork of collagen fibers (150 and 200 µm) and finally striated muscle fibers with interspersed collagen. For this display, at each depth a different x,y position from the original 200×200 µm stack was selected. Each focal plane was adjusted for brightness individually to compensate for intensity differences. Scale bar is 20 µm.
The backward detected signals at a given distance from the surface may actually improve with increasing tissue thickness, since an increasing amount of the forward generated signal will be backscattered in deeper tissue layers. This assumption is supported by Légaré and colleagues showing that backscattered signal increased in intensity when a thin tissue section was put on top of a strongly scattering medium
As others
With THG, visualized structures did not usually show relative intensity differences for forward and backward detection. Backward THG generally appeared like a weak forward THG signal with decreased signal-to-noise ratio (
A number of studies have established SHG as a tool to image collagen fibers and striated muscle myosin. More recently it becomes clear that THG may have an even broader potential for label-free three-dimensional microscopy of tissues. To our knowledge, this study is the first to demonstrate three-dimensional movement for unlabeled leukocytes within tissues (
While we have shown THG microscopy to provide resolution at a single cell level, it may be challenging to apply it to cells cultured directly on glass or plastic, since the medium/substrate interface generates a strong THG signal, raising the background for intracellular observations (data not shown).
Local imaging conditions are important also in other ways. The strength of SHG and THG signals depends heavily on the relative orientation of the potentially harmonophoric structure to the incoming polarized laser beam and on the optical conditions in the surrounding sample, such as absorption, scattering and internal refraction index mismatches in the beam path. For example, we observed shadowing by blood filled vessels on areas underneath. DNA staining revealed some cell nuclei that were not visualized by THG while others were visible (
Two reasons argue for detection of THG signals in the visible range instead of generating UV-THG, e.g. with wavelengths available on Ti:Sa lasers (≤1060 nm): Transmission of the THG signal through typical glass optics is better in the visible range and the strong absorption and stronger scattering of UV light by biological tissues is avoided. Previous studies used a home-built Cr-Fosterite lasers emitting light at 1230 nm (e.g.
(AVI)
(MP4)
(AVI)