1.

Introduction

Spinal cord injury (SCI) is a severe medical condition resulting in motor and sensory deficits, with dramatic functional and socioeconomic impacts.¹ The pathophysiology of SCI includes axonal severing and loss of neuronal fibers, with failure to sustain a regenerative response that can reconnect long descending motor tracts and ascending sensory fibers. Multiple long-term factors contribute to regeneration failure. After injury, the axons originating in descending cortical and subcortical tracts exhibit dieback over one or more segments, but do not degenerate to the cell body. Additionally, glial scar formation results in a chemical and physical barrier to reconnection of descending motor axons below the lesion.^2–4 We have recently demonstrated the capacity of poly(lactide-co-glycolide) (PLG) bridges implanted into a C5 cervical hemisection injury to promote regeneration of the descending corticospinal tract (CST). In that study, CST fibers expressing a green fluorescent protein (GFP) reporter exhibited growth through the bridge and into the distal spinal cord,⁵ supporting implantation of PLG bridges as a substrate for axonal regeneration.⁶ Implantation of this type of scaffolding biomaterial is applicable to SCI models, in which a surgical incision has been made to remove a segment or more of spinal cord, and potentially to traumatic clinical injuries, in which there has been penetration of the gray and white matter of the spinal cord. Current clinical applications of PLG include advanced drug delivery systems, dental sciences, and biodegradable bone grafts.⁷

Evaluation of axonal damage, myelination, and axonal regeneration is critical to investigate regeneration dynamics after SCI or in other implantation paradigms. Accordingly, we sought to develop a tool to identify neuronal fiber growth in a mouse model of SCI, using confocal and two-photon imaging on live animals. This imaging approach demonstrates proof of concept for in vivo imaging within the spinal cord lesion and regenerative fibers with the potential to evaluate remyelination on in situ spinal cord section. In vivo imaging studies of axon trajectories and neuronal fibers after SCI enable unambiguous identification of spared, injured, and regenerating neuronal fibers, providing a strong stage to assess therapies aimed at spurring neuronal fiber growth.⁸ We used Thy1-enhanced green fluorescent protein (eGFP) mice, which have previously been employed in other neurobiological studies and are commercially available.^9,10 These mice express eGFP in neuronal areas, including the spinal cord, enabling visualization of axons of multiple origins. We used a modified upright two-photon microscope equipped for live animal imaging, capable of visualizing fluorescent emission originating from eGFP. We used the same microscope to visualize third-harmonic emission signals arising from myelin on spinal cord sections.^11,12 Myelin is a compact fatty structure that wraps axons to enhance the speed of signal transmission.¹³ Myelin can be detected via third-harmonic emission signals in the intact spinal cord; in contrast, myelin identifying dyes can only be applied in fixed tissue sections and disrupt its lipid structure.¹² We visualized fluorescence inside the intact spinal cord rostral and caudal to the lesion site, as well as that inside the PLG bridge arising from regrown neuronal fibers. We then sectioned the spinal cord and visualized the eGFP fluorescence signal through the PLG bridge, and we evaluated the third-harmonic signal to investigate possible fibers remyelination. Third-harmonic generation (THG) is a suitable method to detect myelin inside the central nervous system (CNS).^12,14 THG can be detected from myelin in a label-free imaging modality that provides micrometer resolution imaging of myelin in the mouse spinal cord.^14,12 Using this method, we tried to assess axonal myelination adjacent to and inside the PLG bridge. We were able to visualize third-harmonic signal rostral to the bridge entrance, but not inside the bridge. However, our qualitative results show it is possible using reporter mice to visualize a time-lapse regrowth of spinal cord fibers through a PLG bridge.

2.

Methods

2.3.

Microscopy and Live Imaging

Live imaging was performed at day 0, immediately before SCI and bridge implantation to establish a baseline, and 8 weeks post-SCI and PLG bridge implantation. Animals were anesthetized using isoflurane (2%) and kept on a warming pad to maintain temperature at 37°C for all procedures; live imaging was completed within a 1-h time limit after the animals were anesthetized [Fig. 2(b)]. During imaging, the vertebral column was stabilized by clamping forceps placed on the C4 and C6 vertebrae and attached to a fixation apparatus to prevent movement of the cord. Images were acquired using an commercial upright microscope (Olympus BX module, Center Valley, Pennsylvania) coupled to a tunable femtosecond pulsed Ti:sapphire laser (Mai Tai HP with pulse-width compensator unit DeepSee, Spectra Physics, Santa Clara, California) working at 80 MHz. The setup was equipped with two digital photon counting PMTs (H7422P, Hammatsu, Japan) in the reflective path and a wide area PMT (R7600P-300, Hamamatsu, Japan) in the transmission path of excitation. A photon time tagging unit capable of lifetime imaging (FLIM-Box, ISS, Urbana-Champaign, Illinois) was used to collect the signals from the PMTs, enabling simultaneous detection and imaging. The imaging and data analysis were performed using the lab-made software (SimFCS, LFD). The third-harmonic images were imaged at 1038 nm (THG at $\sim 346\mathrm{nm}$) and collected with a narrow bandpass filter combination of glass filters (BG39 and UG11, Edmund Optics, Barrington, New Jersey) centered at $360\pm 25\mathrm{nm}$. The single-photon confocal imaging was done using microscope’s internal 473-nm diode laser. Images were collected by raster scanning at fixed pixel dwell time of $20\mu \mathrm{s}/\text{pixel}$ and image size of $256\times 256$. The scanning was controlled by SimFCS for THG and Olympus Fluoview for confocal imaging. A schematic representation of the microscope and data collection is shown in Fig. 1.

Fig. 1

Schematic representation for the microscope setup. The ultrafast laser system is coupled to the auxiliary input of the commercial upright laser scanning microscope system using an AOM (acoustic optical modulator) for power control and a beam expander to tune the beam size (telescope). The light is launched into an elevated scanning unit using a periscope-mirror set (M1, M2, and M3). The internal single-photon lasers are combined to the auxiliary laser at the DM1 dichroic mirror. The excitation laser goes through the XY galvano-scanner mirror unit and is launched through the microscope frame BX61 to the sample. The emission is collected using three separate geometries: (a) the nondescanned detection path (NDD), which is split from DM3 dichroic mirror, (b) the confocal unit with variable pinhole size from DM2 dichroic mirror, and (c) the transmission path detection using DIVER module, which uses combinational filters. The DIVER unit has an in-built filter wheel that can tune into visible and ultraviolet filters. Conventional samples can be mounted on stage 1, the commercial stage, and a second animal mounting stage with heating pad can be mounted on top of DIVER module (stage 2). The red bold line shows the multiphoton, the blue line shows single-photon excitation, green line shows the emission path, yellow shows the ocular path, the dashed-orange line shows the signal lines, and the black line shows the fiber.

3.

Results

For the live imaging of the cord, animals were maintained under anesthesia for the duration of the experiment and positioned under the microscope as shown [Fig. 2(b)]. An initial (0 day) live imaging session on the Thy1-eGFP animals was performed immediately prior to SCI and bridge implantation to establish the capacity to detect in vivo neuronal fibers. Using a 488-nm (blue) laser, we imaged the eGFP neuronal fibers through cord up to $140\text{-}\mu \mathrm{m}$ deep inside the tissue [Fig. 2(d)]. We were able to observe very distinct neuronal fibers inside the cord, expressing eGFP.

Fig. 2

(a) Schematic representation of PLG bridge implanted at C5 vertebral segment, (b) Thy-eGFP animal is positioned under microscope clamped on C4-C6 with exposed C5 segment, (c) picture showing C5 hemisection and bridges placed inside the black hole, and (d) circle imaging of the spinal cord prebridge implant showing neuronal fibers at the C5 level after bone removal.

After the initial imaging session, we performed a C5 hemisection and bridge implantation as described in Sec. 2.2 [Figs. 2(a) and 2(c)]. For bridge implantation, the dimensions of the biomaterial bridge were adapted to the lesion size to ensure apposition within the lesion margins of the hemisection. A second live imaging session was performed 8 weeks post-SCI and bridge implantation, with anesthesia and acquisition time limited to 1 h for data collection. We used reflectance microscopy to detect the bridge and green fluorescence (eGFP) to image the neuronal fibers. We imaged rostral to and within the PLG bridge to establish feasibility to identify regrown fibers entering the bridge and traversing through it. Figure 3(a) shows a schematic representation of the bridge placed inside the spinal cord at C5 level. Numerous green fluorescent neuronal fibers were observed to accumulate in the rostral parenchyma [Fig. 3(b)], where they were positioned immediately adjacent to the entrance to the PLG bridge. Although few fibers were found to enter the bridge channels at this timepoint in the Thy1 model, we were able to observe some [Figs. 3(c) and 3(d)]. These fibers reached distances of up to $\sim 500\mu \mathrm{m}$ from the rostral side of the injury edge, progressing toward the caudal margin.

Fig. 3

(a) Schematic representation of a C5 cross section after bridge implant, (b) neuronal fibers (green) at the rostral edge of the PLG bridge entrance, (c) left panel: reflection microscopy showing the bridge surface on and right panel: the arrows show the neuronal fiber growing into the bridge, and (d) representative images of two more animals we tested with variable results.

Following the imaging session, animals were resutured and maintained for an additional 4 weeks. At 12 weeks post-SCI and bridge implantation, animals were euthanized as described in Sec. 2 to confirm the observation of regenerated fibers within the bridge in histological sections and to test THG imaging for the visualization of myelinated fibers within the bridge. C2-T1 spinal cord segments were cryosectioned horizontally at $160\mu \mathrm{m}$ to enable comparison of injured and uninjured spinal parenchyma, as well as regions rostral to, within, and caudal to the implanted bridge. Cryosections were mounted on glass slides as described in Sec. 2.

PLG biomaterial bridges exhibit a degradation profile.¹⁶ Using these procedures, we were able to observe the degrading bridges inside the tissue at C5 level [Fig. 4(a)] and histologically verify the presence of regenerating neuronal fibers. We observed increased green fluorescence (eGFP) arising from the area proximal to the bridges in both rostral and caudal area [Figs. 4(a) and 4(b)]. We speculate the presence of high fluorescence at bridge boundaries is due to the regenerating neuronal fibers accumulating at the bridge entrance. In Fig. 4(a), the green fluorescence is evidently present across the whole injury, suggesting that many more neuronal fibers have crossed the bridge in the 4 weeks after the live imaging session. This timecourse is in agreement with our previous publications in which we observed the neuronal fibers starting to enter the bridge and crossing through it in 4 to 12 weeks postinjury.⁵

Fig. 4

(a) Imaging of the bridge area showing the fibers (eGFP) are passing through it, in gray is the degrading bridge and (b) $Z$ stack imaging of the spinal cord after sectioning, in gray is the bridge and the fluorescent neuronal fibers are in green.

Additionally, we tested the detection of THG signal within the bridge, as described in Sec. 2. THG has been demonstrated to be capable of imaging compact myelin in Schwann cells and in the CNS ex vivo.^12,14 We tested this possibility by applying the same method to study the remyelination of the spinal cord after injury. Figure 5(a) illustrates detection of GFP+ axons in the rostral spinal parenchyma adjacent to the bridge using region reflectance microscopy. Figure 5(b) demonstrates successful detection of THG signal associated with these axons. However, while we were able to detect THG from neuronal fibers in the rostral parenchyma and fluorescent eGFP+ fibers throughout the bridge structure [Fig. 5(c)], we were unable to detect any THG signal within the bridge [Fig. 5(d)]. Taken together, these data suggest that the myelination is a process that does not directly follow the neuronal fibers regeneration and could occur at a later time.

Fig. 5

(a) Reflecting microscopy showing axons on the rostral entrance of the bridge, (b) THG originating from myelin surrounding the neuronal fibers in panel (a), (c) imaging showing the neuronal fibers around and through the bridge and in gray is the bridge itself, and (d) THG is absent in the bridge area while signal is detected from the surrounding areas.

4.

Discussion

Evaluation of axonal damage, myelination, and axonal regeneration is critical to investigate the dynamics of regeneration after SCI. In vivo studies of neuronal fibers after SCI would improve knowledge of regenerating fibers growing into and through a permissive environment, such as the PLG bridge. This approach is also important as a strategy for improving understanding of axonal regeneration and sprouting in the spared parenchyma of the spinal cord after injury. Visualizing the regrowth of the neuronal fibers would enable us to investigate the direction of regeneration and eventually investigate any physical barrier slowing the regeneration process.

For many fiber tracts, reconstructing a circuit through regeneration or regenerative sprouting may be only the first step in establishing functional connectivity. In this regard, myelination of these new fibers will in many cases be a determining factor. Assessing the level of axonal myelination in vivo is a challenging task because the third-harmonic signal is easily absorbed from the tissues and along the microscope pathway, limiting the detection signal. However, it is possible to obtain myelination data from ex vivo fibers. Here, we demonstrate the feasibility of combining live imaging of neuronal regrowth at two separate time-points, before and after the PLG scaffold implantation, with ex vivo determination of myelination status of these fibers. Myelination in the spinal cord after PLG bridges implant has been demonstrated previously to be very low.²⁷ When PLG bridges were implanted without trophic factors or sonic hedgehog only 18% of fibers exhibited colabeling for neurofilament and myelin basic protein 8 weeks postimplant. Although we were unable to detect any signal arising from THG in the PLG bridge area, we cannot exclude that the absence of myelination is due to the properties of the bridge. For example, the signal arising from the myelin could be quenched given the low proportion of myelinated fibers at this time point. Additionally, THG has only been used to visualize mature myelinated structures, in which compact myelin has formed;^11,12 thus, it is also possible that we were unable to visualize THG signal due to the immature state of the myelin at this time point. Critically, three-photon imaging and THG microscopy have the potential to go deeper than two-photon imaging schemes, owing to greater penetration depth due to using a longer wavelength. Three-photon imaging and THG microscopy also render higher resolution images because of the spatial excitation photon density in a third-order nonlinear process. However, one main hurdle in increasing depth of imaging is the wavefront distortion suffered by the excitation beam through scattering in the layers of penetrated tissue thinning excitation photon density. This can be corrected by adaptive optics schemes²⁸ and other photon density optimizing techniques to focus the light at one point using objective lens optimization.²⁹ Future development of these systems will lead to an improvement into the live imaging of the spinal cord and to detect myelination in an in vivo setting.