2.1.

Primary Nerve Cell Preparation

Imaging dishes (MatTek) of 35 mm were coated with $0.2\mathrm{mg}/\mathrm{mL}$ Poly-L-Lysine (PLL) solution and incubated for 3 h. PLL solution was aspirated and washed three times using culture grade water at 10-min intervals. Dishes were dried in a sterile hood before plating cells. Primary hippocampal neurons were dissected from embryonic 17- to 18-day rats and plated onto coated 35-mm imaging cell culture dishes as previously described.¹⁰ Briefly, hippocampi were dissected into Hanks balanced salt solution containing 10 mM HEPES and 1% penicillin/streptomycin. Following trypsin treatment and trituration, dissociated hippocampal neurons were plated at low density into imaging dishes with plating media (2% b27, 1% Glutamax, 5% fetal bovine serum in Neurobasal). The following day, the plating media was replaced with maintenance media (2% b27 1% Glutamax in Neurobasal). Neurons were fed fresh maintenance media every two to three days until the day of the experiment. Subaxotomy experiments on hippocampal neurons were performed three to six days postdissection.

2.2.

iPSC-Derived Nerve Cell Preparation

Control iPSCs and neural progenitor cells (NPCs) derived from Craig Venter have been previously described.¹¹ Briefly, fibroblast biopsies were induced to a pluripotent state using the traditional Yamanka factors and expanded as iPSCs.¹² NPCs were made following published protocols by culturing iPSCs on a feeder layer of mouse stromal cells (PA6) for two weeks under conditions of dual SMAD inhibition.^13,14 For neuronal differentiation, NPCs were seeded onto 10-cm plates coated with poly-ornithine and laminin (Sigma) and expanded as previously described in NPC media [Dulbecco's modified eagle medium:F12 + Glutamax, and $20\mathrm{ng}/\mathrm{ml}$ Fibroblast growth factor (FGF) (Millipore)].^11,14,15 After NPCs reached confluence, the FGF was removed from the NPC media to promote neuronal differentiation. At three weeks, cells were dissociated using Accutase and Accumax (Innovative Cell Technologies) and resuspended in NPC media supplemented with 0.5 mM dibutyryl cyclic AMP (Sigma), $20\mathrm{ng}/\mathrm{ig}$ brain-derived neurotrophic factor, and $20\mathrm{ng}/ì\mathrm{g}$ glial cell line-derived neurotrophic factor (Peprotech). In order to isolate true axons from the bulk culture, differentiated NPCs were seeded at a density of 2 million onto a microfluidic culture system coated with $100\mu \mathrm{g}/\mathrm{ml}$ poly-L-ornithine and $5\mathrm{ug}/\mathrm{ml}$ laminin (Sigma). Microfluidic culture systems have been shown to efficiently separate axons from soma and dendrites.^16,17 A modified previously designed microfluidic system was used.¹⁸ Briefly, a central soma compartment ($3\mathrm{mm}\times 100\mu \mathrm{m}\times 39\mathrm{mm}$) is flanked by two open axon compartments ($3\mathrm{mm}\times 23\mathrm{mm}\times 10\mathrm{mm}$). The three compartments are connected by intermediate capillary channels ($10\mu \mathrm{m}\times 3\mu \mathrm{m}\times 450\mu \mathrm{m}$) that are thin enough to exclude soma and dendrites. After six to eight days, differentiated NPCs extend their axons through the capillary channels where laser subaxotomy is performed.

2.3.

Tubulin Transduction and Immunofluorescence Staining

To visualize real-time cytoskeletal repair following laser subaxotomy, hippocampal neurons were transduced with tubulin tagged with red fluorescent protein (RFP). Cultures approximately three- to four-days old were incubated overnight with $10\mu \mathrm{L}$ of CellLight tubulin RFP (C10614, Life Technologies) and subsequently prepared for imaging. In order to examine tubulin and actin, some neurons were fixed with 4% paraformaldahyde for 1 h at room temperature immediately after subaxotomy. After PBS wash, fixed cells were incubated in a blocking buffer for 30 min and incubated overnight with the primary tubulin antibody (T9026, Sigma-Aldrich). The primary antibody was washed with PBS three times before incubating cells in the secondary antibody (A-11005, Life Technologies) against the mouse tubulin for 30 min. Neurons were then incubated for 20 min in Phalloiden (A12379, Life Technologies) diluted to $40:760\mu \mathrm{L}$ in PBS to stain for f-actin.

2.4.

Laser Optical System

The optical system used in these studies has been described previously.^19,20 Briefly, an Nd:YVO₄ 532-nm laser (Spectra-Physics, Vanguard) delivering 12-ps pulses at a 76-MHz repetition rate was used to create subaxotomies. The amount of laser power and energy entering the microscope was controlled using a polarizer mounted in a rotary mount (PR50PP, Newport Corp.). A mechanical shutter (Vincent Associates) with a 30 ms duty cycle gated the laser beam resulting in a 30-ms burst of pulses that entered the microscope and exposed the specimen at the laser focal point. A dual-axis fast scanning mirror (FSM-200-01, Newport Corp) was used to steer the laser beam at an image plane conjugate to the back focal plane of the microscope objective. At the specimen plane, the laser beam followed a linear path defined by the user (Fig. 1).

Fig. 1

Hippocampus axon with growth cone. An example subaxotomy experiment which shows the neuron at prelaser (a) and postlaser (b). Axon thins [arrow in (b)] within 1 s after laser damage. Red line in (b) indicates the path of the laser beam.

The amount of laser power/energy needed to create subaxotomy was determined by targeting an axon at a low laser power and gradually increasing the power until visible thinning of the axon was detected within a second after laser exposure. The pulse energy threshold to damage hippocampal neuronal axons was 0.351 nJ (peak irradiance $1.27\times {10}^{11}\mathrm{W}/{\mathrm{cm}}^2$), while the pulse energy required to damage iPSC-derived neuronal axons was 0.268 nJ (peak irradiance of $0.97\times {10}^{11}\mathrm{W}/{\mathrm{cm}}^2$). Only one to two 30-ms bursts of laser energy were exposed to the nerve axon since it is only a few microns in width. The total subaxotomy energy for one to two 30-ms bursts of laser exposure was 801 to $1602\mu \mathrm{J}$ and 610 to $1220\mu \mathrm{J}$ for hippocampal and iPSC-derived neurons, respectively.

2.5.

Live Cell Imaging

The optical system utilized a Zeiss Axiovert 200M microscope with a 63X, phase NA 1.4 Plan-Apochromat oil immersion objective. Neurons were mounted on an X-Y stepper stage (Ludl Electronic Products). Phase and fluorescent images were acquired with a Hamamatsu Orca_AG deep-cooled $1344\times 1024\text{pixel}$, 12-bit digital CCD camera.

To perform live imaging of neuronal cells for extended periods of time, the temperature (37°C), humidity, and ${\mathrm{CO}}_2$ (5%) concentration of the environment was controlled using a microscope incubator (OkoLab, NA, Italy). Once environmental conditions stabilized, neurons were placed in the microscope incubator and allowed to acclimate for 5 min. Neurons with established growth cones with active filopodia and lamellipodia were followed for 5 min using phase contrast imaging to characterize preirradiation morphology and behavior.

Axons were identified based on thickness, length, and morphology. Ideally, axons were more than 1 mm in length with a clear, active growth cone and no secondary extensions sprouting from the primary process.²¹ If no such processes were present, the longest process extending from the soma without branch points was targeted for laser irradiation. Based on these criteria, axons as opposed to dendrites, which are shorter thick processes with multiple branch points, were selected for subaxotomy. Phase contrast and fluorescent time series imaging were performed immediately and up to 60 min after axonal damage. Imaging was stopped when it was apparent that the recovery process was complete and the cell had stabilized. Based on morphological changes of the growth cone, the axonal damage site, and the cell body, it was determined that the cell had either returned to its prelaser state or it did not recover from the laser damage. Possibly due to inherent cell variability (age, physiology, neuronal subtype, etc.), it was observed that different cells required different lengths of time to reestablish normal growth cone morphology and movement. Individual cells were followed for the period of time necessary to recover normal morphology, or until it was obvious that repair/recovery was no longer proceeding.

Images from pre- and postlaser exposure were compiled and analyzed using ImageJ. In order to quantitatively score growth cone recovery, a LabView program was written to analyze the magnitude of the movement within and around the growth cone before and during the recovery process. The algorithm utilizes a wavelet transform to analyze the changes in an image series over time. Iterative application of temporal high- and low-pass filters allows for separation of the areas of the growth cone and axon, which are moving with high and low frequencies, and determine a concrete measurement of the magnitude of the oscillations at these frequencies.^22,23 These magnitudes can then be plotted versus time as a relative measurement of growth cone activity before and after laser exposure. The measurements can also be used to characterize different classes of responses.

3.1.

Hippocampal Response

Phase contrast images of hippocampal neurons after laser subaxotomy showed a visible thinning of the axon followed by a repair process involving the growth cone and localized cytoskeletal remodeling (Figs. 2 and 3). Axons were scored qualitatively based on visible changes in morphology following laser exposure. A “++” response, which showed the highest level of recovery, was characterized by the return of the axon to its original thickness, with a recovery process involving either turning or retraction of the growth cone. A “+” response was a return to initial axon thickness without dynamic involvement of the growth cone. A “−” response was the degeneration of the axon and growth cone after laser damage. Completely transected neurons (axotomy) were noted and excluded from analysis. In 10 out of 23 axons imaged (Table 1, ++ response 47.82%), the growth cone retracted in the retrograde direction, extending filopodia toward the damage site (Figs. 2 and 3). The ++ response generally began 5 to 10 min after laser damage and had a duration of $17\pm 5\min$.

Fig. 2

Axonal growth cone dynamics of a representative rat-derived hippocampal neuron following subaxotomy. Still frame images from a representative imaging session show several stages of injury, repair, and recovery. (a) Growth cone before laser subaxotomy. (b) The axon is damaged as indicated by the white arrow causing a thinning of the axon. (c) and (d) Axon is slightly thinner at ablation site and lamellipodia-like structures appear to form at the base of the cell body (arrow). (e) Growth cone retracts and lamellipodia-like components appear to be moving toward the damage site. (f) The growth cone is reduced in size. (g) and (h) Neuron axon thickens and returns to prelaser morphology. (i) Lamellipodia progress forward and combine with the growth cone. See (Video 1 MOV, 6.46 MB) [URL: http://dx.doi.org/10.1117/1.NPh.2.1.015006.1], time series of laser induced damaged and recovery of hippocampus neuron.

Fig. 3

Hippocampus growth cone turning in response to axonal damage. (a) Hippocampus nerve cell before laser ablation. (b) Nerve cell thinned at ablation site (arrow). (c) and (d) Filopodial-like structure forming at damage site. (f) to (h) Growth cone turning and extending filopodia toward the damage site. (i) Repaired axon. See (Video 2 MOV, 3.78 MB) [URL: http://dx.doi.org/10.1117/1.NPh.2.1.015006.2, time series of laser induced damaged and recovery of hippocampus neuron.

Table 1

Hippocampal neuron recovery. A “++” was assigned to neurons that displayed extensive axonal cytoskeletal remodeling as well as active involvement of the growth cone, “+” was assigned when repair occurred, without involvement of the growth cone, and “−” was assigned to neurons that did not recover from laser damage.

In hippocampal neurons, 47.82% of the cells exhibited a ++ response, 43.48% a + response, and 8.7% of the neurons did not recover from the laser damage (Table 1). Of 23 damaged hippocampal neurons, 91.3% exhibited some level of repair and recovery, which involved growth cone and/or axonal remodeling.

Wavelet analysis of the growth cones using the LabView algorithm was used to identify and quantify the qualitative cell scoring of ++, +, and −. The activity trace for − cells, for example, shows a distinct drop-off as the growth cone activity decreases. Other responses, such as those seen in + or ++ cells, may decrease due to damage, but eventually return to near precut levels, or they may not show significant decreases at all (Fig. 4).

Fig. 4

High frequency (6 Hz) activity traces for different growth cones. “++” and “+” responses generally did not show a significant lasting change in activity and were similar. “−” responses were both characterized by a decrease in activity. An additional delayed response was observed in one case: activity decreased but did recover. Changes in growth cone morphology in the “−” time series indicated that no further increase in activity occurred after 800 s. In the above graph, laser subaxotomy occurred at 300 s. Values were normalized by division by the largest value in the time series.

Wavelet analysis was used to analyze the dynamics of the 23 hippocampal growth cones in this study. The low-frequency magnitude of the growth cone movement before subaxotomy was averaged, as was the magnitude for the last 100 s of the time series. The normalized recovery was calculated as follows:

Normalized recovery was plotted for each neuron (Fig. 5). Although the distribution for recovery is wide, the lowest two points on the plot represent the two—scored cells. Based on these values, it is possible to estimate a threshold in movement, which separates—scored growth cones from + and ++ growth cones. This threshold can be approximated between $-0.6$ and $-0.8$, and is shown as $-0.75$ in Fig. 5.

Fig. 5

The normalized change in movement for all growth cones, based on values before subaxotomy and values from the last 100 s of imaging. The two “−” cells fall below all other cells and appear to indicate a threshold for positive versus negative recovery, approximated by the line at $-0.75$. “++” cells are indicated as blue squares, “+” cells are indicated as green circles, and “−” cells are indicated as red diamonds.

Our live-imaging indicated large-scale changes in filopodia and lamellipodia dynamics along the axon, suggesting that cytoskeletal remodeling may be involved in repair after injury. Therefore, to examine the role of the cytoskeleton in the damage-repair process, two types of cytoskeletal elements, microtubules and actin, were examined. Microtubules are crucial for the axonal transportation system of a highly polarized neuron. Furthermore, actin is known to be a critical component of growth cone motility and dynamics, and is also essential in the formation of filopodia and lamellipodia.²⁴ Conventional immunofluorescence was used to determine if these two cytoskeletal components play a role in the changes observed during axonal repair. Hippocampus neurons were transiently transduced with RFP-tubulin. Fluorescently labeled RFP-tubulin was monitored before and after laser subaxotomy. Immediately after laser damage, the RFP-tubulin signal decreased in intensity at the damage site corresponding with the thinning observed in the phase contrast images. However, the RFP fluorescence recovered in intensity as axonal repair proceeded (Fig. 6). In addition, a visible retraction and accumulation of tubulin was observed at the damage site, consistent with the response seen in the phase contrast images. The results confirm that laser subaxotomy structurally damages microtubules, and that microtubule function and/or polymerization may be an important step in the axonal repair process in the system studied here.

Fig. 6

Hippocampal neuron transduced with red fluorescent protein tubulin before and after laser radiation: (a) before laser radiation and (b) neuron immediately after laser radiation. Location of subaxotomy is indicated by the arrow. (c) and (d) Loss of tubulin near damage site. (e) and (f) Retraction of tubulin toward the damage site. (g), (h), and (i) Tubulin retraction continues and fluorescence at the damage site recovers.

Cytoskeletal remodeling was also observed by fixing hippocampal neurons at different time points of recovery and staining for tubulin and actin. Thirteen cells were successfully damaged and stained. Immunofluorescent staining, while lacking the flexibility of transduction with fluorescent tagged actin and tubulin, provides stronger fluorescent signals and eliminates concerns of photobleaching and autofluorescence. The fixed immunofluorescent stains show a clear loss of tubulin at the laser ablation site, confirming the structural cytoskeletal damage caused by subaxotomy during live imaging analysis (Fig. 7, red channel). In hippocampal neurons fixed 6 min after laser subaxotomy, fluorescent images showed morphological changes in the neuron’s growth cone (Fig. 7, green channel). In addition, they showed visible accumulation of actin at the damage site consistent with the formation of lamellipodial-type structures during a ++ repair process. This suggests that formation of lamellipodia, and more specifically, changes in actin accumulation at the damage sites, may be an initial and important step in the axonal repair process.

Fig. 7

Actin (green) and tubulin (red) staining of hippocampal neurons. (a) Phase contrast image of neuron prior to laser ablation. Location of cut indicated by the arrow. (b) Neuron fixed 5 min after cut. The arrow indicates the laser damage. (c) Zoomed image at the damage site. (d) Phase contrast image of neuron prior to laser ablation. Location of laser cut indicated by the arrow. (e) Neuron fixed 6 min after cut. The arrow indicates the laser damage. (f) Zoomed image at the damage site. See (Video 3 MOV, 1.98 MB) [URL: http://dx.doi.org/10.1117/1.NPh.2.1.015006.3], time series of laser induced damaged and recovery of iPSC-derived neuron.

Fig. 8

Induced pluripotent stem cell (iPSC) derived neurons cultured in microfluidic compartmentalized chambers. Axonal marker SMI31 labels axonal compartment, while the somatodendritic marker Map2 labels only the cell body compartment. The nuclear marker DAPI clearly indicates which microfluidic well contains the cell body.

One reported example of damage following subdendrotomy was an observed localized swelling at the damage site followed by either recovery or further degeneration. Although our experiments were on axons, rather than dendrites, parallels may exist between the two.²¹ A proposed explanation for the swelling is that a change in membrane permeability following irradiation occurred, thus resulting in a change in osmotic equilibrium.⁷ Although irradiation likely changes plasma membrane properties in a number of ways, another explanation, based on the observed cytoskeletal dynamics, could be the occurrence of actin and/or tubulin accumulation as part of the remodeling process. The causes of axonal or dendritic swelling following irradiation may not be entirely attributed to either changes in osmotic balance or accumulation of cytoskeletal components.

3.2.

Human iPSC-Derived Neuronal Response

iPSC-derived human neurons were studied in order to determine whether they exhibit axonal repair properties similar to the rat hippocampus cells and the previous work on goldfish retinal ganglion neurons. iPSC-derived neurons were grown in compartmentalized chambers in which the axon could be separated from the cell body. As Fig. 8 shows, dendritic MAP2 staining remains in the cell body compartment, and axonal SM131 (phosphorylated neurofilament H) staining carries into the axonal compartment. Using this compartmentalized system, axons were easily identifiable from the rest of the cell. After laser subaxotomy a visible thinning was observed in the axon, similar to that observed in the rat hippocampal cells. The growth cones often retracted and appeared to shuttle membranous material in a retrograde direction toward the damage site, ultimately resulting in the formation of a varicosity at the site of laser damage (Fig. 9). Following this behavior, the iPSC-derived neurons reformed a growth cone with lamellipodial and filopodial extensions. The entire process of growth cone retraction, repair, and return to normal morphology took $40\pm 5\min$. This process was two to three times as long as the repair process in rat brain hippocampus cells.

Fig. 9

iPSC-derived neuron repair and regeneration. (a) Neuron prior to laser ablation. (b) Arrow indicates site of laser damage. (b) and (c) Thinning at the damage. (d) to (f) Growth cone retracts toward damage site (arrows). (g) Axon shows no visible thinning (arrow). (h) and (i) Growth cone is re-established. See (Video 3 MOV, 1.98 MB) [URL: http://dx.doi.org/10.1117/1.NPH.2.1.015006.3], time series of laser induced damaged and recovery of iPSC-derived neuron.

iPSC-derived neurons were also scored using a similar ranking system used previously for the rat hippocampal neurons. A ++ score was assigned to neurons that exhibited both axonal repair and growth cone reformation, + was assigned to neurons that exhibited axonal repair with partial or no reformation of growth cone, and − was assigned to neurons that did not recover from laser damage (Table 2). iPSC-derived neurons had a ++ response of 18.5%, a + response of 25.9%, and a − response of 55.6%. Thus, of 27 damaged iPSC-derived neurons, 44.4% exhibited some level of repair and recovery to predamage morphology, about half of the rate recorded for rat-derived hippocampal neurons.

Table 2

Induced pluripotent stem cell derived neuron recovery. A “++” score was assigned to neurons that exhibited both axonal repair and growth cone reformation, “+” was assigned to neurons that exhibited axonal repair with partial or no reformation of growth cone, and “−” was assigned to neurons that did not recover from laser damage.