2.1.

SRS Microscope

A Ti:Sapphire laser with 80-MHz repetition rate and 140-fs pulse duration (Chameleon Vision; Coherent, Santa Clara, California) pumped an optical parametric oscillator (Chameleon Compact OPO; Angewandte Physik and Elektronik, Berlin, Germany), providing the pump field at frequency ${\omega }_p$ and the Stokes field at frequency ${\omega }_S$. An acousto-optic modulator (15180-1.06-LTDGAP; Gooch and Housego, Melbourne, Florida) was used to modulate the intensity of the Stokes beam at 7.1 MHz. The two beams were collinearly combined and directed into an upright laser-scanning microscope (FV300 + BX51WI; Olympus, Tokyo, Japan), and a $25\times$ water immersion objective (XLPlan N, $\mathrm{N.A.}=1.05$, Olympus) was used for laser focusing. The forward propagated laser beams were collected by an oil condenser (U-AAC, Olympus, $\mathrm{N.A.}=1.4$) and directed to a photodiode (S3994, Hamamatsu, Japan). For SRS imaging, we detected the stimulated Raman loss (SRL) signal, where the Stokes beam was blocked by a bandpass filter ($825/150\mathrm{nm}$, Chroma Technologies, Bellows Falls, Vermont). The signal in the pump beam was then amplified with a resonant amplifier^10,31 and extracted by a lock-in amplifier (HF2LI, Zurich Instrument, Switzerland). The wavelengths of the pump and Stokes beams were selected to 830 and 1087 nm, respectively, and the frequency difference of $2854{\mathrm{cm}}^{-1}$ resonated with the symmetric stretch of ${\mathrm{CH}}_2$.

The SRS signal originates from the interaction of the optical field with the electronic cloud oscillation of chemical bonds. Therefore, SRS microscopy enables us to probe molecule’s chemical bond in cells or tissues and provides us with information regarding their molecular composition.^32–35 As illustrated in Fig. 1(a), the pump field (${\omega }_p$) and the Stokes field (${\omega }_S$) excite the molecules from the ground state to the excited vibrational state by passing through a virtual state. When the frequency differences (${\omega }_p-{\omega }_S$) between the pump and Stoke fields match the vibrational frequency (${\mathrm{\Omega }}_{\mathrm{vib}}$) of a specific chemical bond, the Raman transition is efficiently enhanced by stimulated excitation. The coherent excitation of a molecular vibration results in one photon annihilated in the pump beam and, simultaneously, another photon created in the Stokes beam. These processes are called SRL and stimulated Raman gain, respectively.³⁰ Intensity transfer techniques are usually employed to extract the SRS signal from the intense excitation laser beams with shot-noise limited detection.^10,36 By modulating the intensity of the Stokes beam at the megahertz level, the pump beam experiences an intensity loss. The SRL signal can then be extracted and detected by using a lock-in amplifier [Fig. 1(b)].

Fig. 1

The principle of the stimulated Raman scattering (SRS) process and scheme of the SRS microscope. (a) Energy diagram of the SRS process where the pump photons coincide with the Stokes photons, exciting molecules from the ground state to an excited vibrational state and resulting in one photon being absorbed in the pump beam and one photon being created in the Stokes beam. When the energy difference between pump photons and Stokes photons matches the vibrational frequency (${\mathrm{\Omega }}_{vib}$) of a specific chemical bond, the amplification of Raman signals is achieved by stimulated excitation. (b) The experimental scheme of the SRS microscopy for in vivo imaging of live tadpoles. Anesthetized tadpoles were mounted with low temperature gelling agarose and imaged by a single-frequency SRS microscope. AOM: acousto-optic modulator, DM: dichromic mirror, GM: galvo mirror, LIA: lock-in amplifier, PD: photodiode, FL: filter.

2.2.

In Vivo Tadpole Imaging

Tadpoles around stage 40 were bought from Nasco (Fort Atkinson, Wisconsin) and maintained in a modified rearing solution (60-mM NaCl, 0.67-mM KCl, 0.34-mM $\mathrm{Ca}{({\mathrm{NO}}_3)}_2$, 0.83-mM ${\mathrm{MgSO}}_4$, 10-mM HEPES pH 7.4, $40\text{-}\mathrm{mg}/\mathrm{l}$ gentamycin) with 0.001% phenylthiocarbamide (Sigma-Aldrich, Missouri) added to prevent melanocyte pigmentation. Tadpoles were raised to about stage 49 and were selected for SRS imaging; all staging was carried out according to a previous report.³⁷

Before imaging, the tadpoles were anesthetized with 0.02% tricane methanesulfonate (Sigma-Aldrich, Missouri) and then transferred to a glass-bottom Petri dish. Low gelling temperature liquid agarose (2%, Sigma-Aldrich, Missouri) was kept in a 37°C water bath, and drops of agarose were poured into the Petri dish to cover the anesthetized tadpole. The appropriate rearing solution with 0.02% tricane methanesulfonate was added into the Petri dish in an attempt to maintain the tadpoles in their optimal physical condition for SRS imaging. Three-dimensional (3-D) imaging with $1\text{-}\mu \mathrm{m}$ depth was then performed to encompass the entire morphology of the spinal cord and optic nerve in Fig. 2, the Schwann-like cell in Fig. 3, the node of Ranvier in Fig. 4, and the cranial nerve VII in Fig. 5. The 3-D stacked images were processed by ImageJ (NIH, Bethesda, Maryland). The optical power of pump and Stokes beams on samples are 10 and 40 mW, respectively. As most of the excitation power is carried by a longer wavelength, the photo damage can be minimized to live animals.³⁸

Fig. 2

The transparent characteristic of X. laevis tadpoles ensures label-free, noninvasive SRS imaging of the nervous system. (a) X. laevis tadpoles at stage 45; the neural system can be directly seen by the naked eye. (b) The optic tectum of a stage 49 tadpole was optically imaged; and the red rectangle area was imaged by SRS microscopy shown in (c). (c) SRS imaging showing the membrane of single neurons in the optic tectum. (d) The spinal cord of a stage 49 tadpole was optically imaged; and the red rectangle area was imaged by SRS microscopy shown in (e). (e) SRS imaging of myelin sheath in the spinal cord. (f) SRS imaging of myelin sheath in the optic nerve. (g) Overlaid SRS (red) and two photon fluorescence imaging (green) showing the myelin membrane wrapping on a single axon.

Fig. 3

In vivo SRS imaging of myelination in a Schwann cell. A Schwann-like cell in a branch of the trochlear nerve interacting with two axons was imaged by the SRS microscope. From 6 to 9 h, it can be seen that a new myelin segment presented in the process of the Schwann cell (indicated by the arrow).

Fig. 4

In vivo SRS imaging of the formation of a node of Ranvier. (a) A node of Ranvier in a single axon in the trochlear nerve. Under initial observation and through to 6 h, the morphology of the node of Ranvier showed little change. (b) At 0 to 20 min, the cell membranes of two adjacent Schwann cells can be seen contacting each other to form a contractive junction. At 40 min, the two myelin segments can be visualized as having retracted and formed a gap (red arrow). At 90 min, the gap is clear. After 90 min, it can be seen that the gap finally developed into a relatively mature node. I: internode, P: paranode, and N: node.

Fig. 5

In vivo SRS imaging of myelin sheath degradation of a peripheral nerve after transection. (a) A branch of cranial nerve VII imaged before cutting. (b) The myelin sheath following disruption, losing its parallel structure 5 min after the cut. (c and d) 1 and 2 h after the cut, the damaged myelin can be seen to have slightly recovered. (e) Macrophages then migrated to the damaged region (red asterisk). (f) Myelin degradation starting at the distal end. (g) Myelin debris forming at both distal and proximal ends (red arrows). (h) Macrophages clearing up the myelin debris (red asterisk), and some myelin sheath is surviving following degradation (green arrows). For a better view of SRS images in the left panel, we enlarged the green rectangles about 50% to 100% in the right panel.

All animal manipulations were conducted in strict accordance with the guidelines and regulations set forth by the University of Science and Technology of China (USTC) and Purdue University Animal Resources Center and University Animal Care and Use Committee. Furthermore, the protocol was approved by the Committee on the Ethics of Animal Experiments of the USTC (Permit Number: USTCACUC1102012).

3.1.

SRS Imaging of the Nervous System in Live Tadpoles

We applied SRS microscopy to image myelin and the node of Ranvier in X. laevis tadpoles in vivo. Figure 2(a) shows that the brain and nerves of stage 45 tadpoles were more optically refractive than their skin and tissue and that they could be directly observed by the naked eye. After being anesthetized and mounted into agarose, the tadpoles were imaged by a laser scanning SRS microscope [Fig. 1(b)]. Also, as shown in Fig. 2(c), we were able to clearly resolve the cellular architectures in the optic tectum in which the strong SRS signal may be produced by abundant ${\mathrm{CH}}_2$ bonds of phospholipids in the cell membrane and CH vibrations of proteins and polysaccharides in the extracellular matrix. Myelin sheaths are highly compacted cell membranes with their major component being phosphonolipids. Therefore, using SRS allowed us to image myelin sheaths in the optic nerve and the spinal cord by ${\mathrm{CH}}_2$ symmetric stretching. As shown in Fig. 2(e), the 3-D overlapped SRS image delineated lateral myelin sheaths wrapping on axons in the spinal cord of a stage 49 tadpole. As they contain much less ${\mathrm{CH}}_2$ bonds than myelin, axons were not detectable due to their comparative signal level with background. We electroporated several retinal ganglion cells of a tadpole at stage 50 with the fluorescent protein mCherry to label axons in the optic nerve. SRS imaging showed two myelin segments in the optic nerve [Fig. 2(f)], and the two photon fluorescence (TPF) imaging overlaid with SRS imaging showed that one labeled axon was ensheathed by myelin [Fig. 2(g), dash line]. Therefore, using colocalization of SRS and TPF, we were able to demonstrate that the SRS signal was being elicited from the myelin and not from the axons.

3.2.

In Vivo Imaging of Myelination in a Schwann Cell

Myelination is an essential developmental process in which multiple layers of the myelin membrane from glial cells gradually wrap on axons, devoting the functionality of myelin to neural signal transduction.^39–41 By using SRS microscopy, we observed the appearance of a thin myelin sheath in the process of one Schwann cell. As shown in Fig. 3, this Schwann-like cell was observed in a small branch of the trochlear nerve. Furthermore, the nucleus and the nucleolus were clearly visualized by 3-D overlapping of SRS images. There were two axons along the process of the cell: one axon had a clear node of Ranvier, whereas the other axon was partly wrapped on the process. At the beginning of imaging, the cytoplasm of the cell produced a strong SRS signal that gradually decreased over time, and, from 4 to 9 h, some small granules showed up. These granules might be the organelles like endoplasmic reticulum, Golgi apparatus, and endosomes, which respond to proteins and lipids synthesis, sorting, and transportation during myelination.^42–44 At 6 to 9 h, we were able to observe a weak signal in the cell process, which represented a newly synthesized myelin segment (indicated by an arrow). The decrease in the SRS signal produced by ${\mathrm{CH}}_2$ in the cytoplasm might have resulted from lipid consumption of the multilayered cell membrane winding the axon. We believe that there were three axons in this axon bundle and that the cell imaged here was a Schwann cell. More specifically, we think that the cell process was wrapping on the third axon to form a myelin segment.

As there are several limitations of labeling and imaging techniques, some of which we have already discussed, it has been difficult to visualize the dynamic process of myelination in living animals. The observation of in vivo time-lapse imaging in transgenic zebrafish has indicated that the spacing of oligodendrocytes along axons is regulated by a repulsive cell–cell interaction of oligodendrocytes;⁴⁵ moreover, the myelin formation by oligodendrocytes is restricted to about 5 h in the zebrafish.⁴⁶ Here, by using label-free SRS microscopy and transparent X. laevis tadpole, we have observed the formation of one myelin segment in the process of Schwann cells without exogenous fluorescent protein incorporation.

3.3.

In Vivo Imaging to Monitor the Development of a Single Node of Ranvier

During myelination, the gap between accompanying myelin segments develops into a node of Ranvier. The maturation of a node of Ranvier is essential for the saltatory conduction of action potentials along myelinated axons. Due to their crucial role in signal transduction, the molecular mechanisms underlying the formation of nodes of Ranvier have been extensively studied,^47–50 and some hypotheses and models have been suggested and discussed.^51–53 However, the inability to observe nodes of Ranvier in real time has greatly limited our knowledge on the mechanisms of their dynamic formation. By using a SRS microscope, we were able to observe the developmental behavior of a node of Ranvier in a stage 49 tadpole in real time.

For imaging, we selected the trochlear nerve in order to visualize individual nodes of Ranvier on a single axon or two bundles of axons. We chose the trochlear nerve since it is the smallest of all cranial nerves, contains a few numbers of axons, and can be easily imaged as it exists on the superficial layer of the X. laevis brain. As shown in Fig. 4(a), a 3-D stacked SRS image clearly presents a typical myelin structure in which two adjacent myelin segments are separated by a node of Ranvier. In the internodal area, the SRS signal from the ${\mathrm{CH}}_2$ vibration delineates the parallel multiple layers of myelin membranes around the axon. At the paranodal region, the diameter of the myelin sheath can be visually confirmed as decreasing. Furthermore, no SRS signal is detectable from the nodal region, as the node only consists of axon with no myelin membrane. We monitored the node of Ranvier over 6 h and observed no obvious changes during imaging, except for a small reshaping of the myelin sheath. The length of the node was found to be consistent over time. Considering these findings, we believe that the lack of change over time indicates that this node of Ranvier was mature. In Fig. 4(b), the SRS image at 0 min shows two contiguous myelin segments with an atypical node of Ranvier. This node was determined to be atypical because it did not have a clear node comprising a naked axon without myelin ensheathment. At 0 and 20 min, a clear conjunction can be visualized in the middle of the axon where the two segments of myelin sheath closely contact, rather than separate with each other. At 40 min, the two myelin segments initiated separation in the opposite direction, with the conjunction stretching and two heminodes beginning to emerge. At 90 min, the stretching induced a gap between the heminodes and after 320 min, the two myelin segments had completely separated. Compared with our observation of the mature node of Ranvier in Fig. 4(a), we think that the node depicted in Fig. 4(b) was immature, and the imaging series reveals a new node formation.

Until now, the understanding of how nodes of Ranvier form was that nodes act as a barrier to prevent the movement of flanking paranodal domains into the nodal area.⁵³ Here, based on in vivo observation, we found that the paranodal domains indeed invaded into nodal domains; however, paranodal domains subsequently retracted from the nodal area, leaving the naked axonal plasma. Based on these findings, we propose the following hypothesis of node formation in the periphery nervous system: during myelination, the membrane of Schwann cells gradually wraps along an axon, and the pressure of this winding drives the membrane to spread along the axon. After the edge of the membrane of two adjacent myelin segments contact together, the closely contiguous myelin segments begin to retract in oppositional directions, leaving the naked axonal plasma between myelin membranes. The full functionality of the node then relies on the anchoring and clustering of sodium and potassium ion channel proteins.

3.4.

In Vivo Imaging of Myelin Degradation after Nerve Transection

Demyelination is a process of myelin protein and membrane destruction around neuronal axons. The damaged myelin sheath impairs the conduction of a neural signal, which can cause deficiencies in sensation, movement, cognition, or other functions depending on which nerves are involved.^54,55 Demyelination has been extensively studied, and many animal models have been generated to understand the basic cellular and molecular mechanisms that underlie this process.⁵⁶ This line of research is very important since once we determine these mechanisms, we might be able to devise a way to promote remyelination.^5,57 In the current study, we used SRS microscopy to observe myelin degradation in X. laevis tadpoles.

A branch of the nerve derived from the cranial nerve VII was selected and transected by a needle. The degradation process of the myelin sheath was then monitored by 3-D in vivo imaging. As shown in Fig. 5(a), before the transection, the parallel myelin membranes around axons compacted and interlaced with each other, constructing a bundle of nerves. At 5 min after damage, the myelin sheaths around the axons were found to be disrupted due to the stretch induced by the mechanical force. In Fig. 5(b), the nerve looked loose and blurry with no obvious parallel myelin structure. Some remaining myelin membrane looked swollen (yellow arrows) or twisted (red arrows), and there were several vacuoles [blue arrows in Figs. 5(b)–5(e)] that we believed to be the cross-section of the transected end of the damaged axons. After 1 to 3 h, some typical and parallel myelin membranes appeared [green arrows in Figs. 5(d) and 5(e)], which had recovered from the mechanical damage. Interestingly, the damaged axon tips showed different morphology between the distal and proximal ends. At the proximal end, the injured myelin showed some vacuole structures [Fig. 5(e), blue arrow]; however, these vacuoles were not observed at the distal end. Instead, the distal end displayed several open tubes, which tended to close [Fig. 5(e), blue arrowhead]. Moreover, we observed a macrophage that had migrated to the distal end [Fig. 5(e), red asterisk], and this area of the damaged nerve began to degrade at 4 h [Fig. 5(f)]. At the distal end of the nerve, the recovered parallel myelin membranes then became diffused and blurred, and some myelin debris was generated at the damaged area [Fig. 5(f), red arrows]. Over the next 15 h, more myelin debris appeared [Fig. 5(g), red arrows] and at 17 h, more macrophages [Fig. 5(h), red asterisks] migrated to the damaged nerve to clear up the myelin debris. At this time point, the distal end had totally degraded without any myelin structure left, whereas the proximal end exhibited some intact myelin sheaths that had survived the degradation [Fig. 5(h), green arrows].

X. laevis tadpole has a robust capability to remyelinate, both in the central and the peripheral nervous systems; therefore, this animal has been used as a model for investigating demyelinating diseases.⁵⁸ Studying remyelination mechanisms in lower animals will provide a better understanding of why remyelination fails in higher animals. Furthermore, these studies might help devise potential methods to promote remyelination.⁵⁹ Compared with mice and rats, X. laevis are easy to maintain/breed, and the aquatic characteristics of the tadpole makes it a suitable model for drug screening.^60–62 Label-free imaging of the myelin sheath by SRS microscopy provides a flexible and highly efficient tool to assess the properties and functionality of the myelin sheath. Therefore, the combination of X. laevis tadpoles and SRS microscopy may provide us with an excellent drug-screening platform for therapies aimed to target demyelinating diseases.