A simple slice culture system for the imaging of nerve development in embryonic mouse



Newborn neurons elaborate an axon that undertakes a complicated journey to find its ultimate target in the brain or periphery. Although major progress in the study of this process has been made by analysis of dissociated neurons in vitro, one would like to observe and manipulate axonal outgrowth and pathfinding as it occurs in situ, as fasciculated nerves growing within the tissue itself. Here, we present a simple technique to do this, through cultivation of embryonic mouse slices expressing enhanced green fluorescent protein (EGFP) specifically in newborn neurons. This system allows for imaging of outgrowth of peripheral nerves into structures such as the developing limb. We demonstrate a reproduction of normal innervation patterns by spinal nerves derived from spinal cord motor neurons and sensory neurons of the dorsal root ganglia. The slices can be manipulated pharmacologically as well as genetically, by crossing the EGFP-expressing line with lines containing targeted mutations in genes of interest. Developmental Dynamics 236:3514–3523, 2007. © 2007 Wiley-Liss, Inc.


Directly after their birth in the developing nervous system, many neuronal cell types begin the long process of innervation by elaborating a process that will ultimately become an action potential-generating axon. It is of fundamental importance for the development of a properly functioning organism that the axon arrives at the correct target for the subsequent establishment of a functional synapse. The molecular pathways responsible for axon guidance have for the most part been elucidated for dissociated neurons growing on artificial substrates, but the natural environment in which an axon grows is much more complicated. Spinal nerves, for example, are known to elongate into the limb bud in closely apposed fascicles of sensory and motor axons, and the adhesive interactions binding these axons together could temper the pathfinding decisions that the growth cones make (Honig et al.,1998). In addition, Schwann cells provide a stabilizing function to the nerve bundle, as mouse mutants lacking this lineage show major defasciculation of nerve bundles (Garratt et al.,2000). Thus, the conclusions made from in vitro experimentation may not necessarily apply to the physiological situation. It is, therefore, important to manipulate nerve outgrowth in situ and observe the effects of such manipulations in real time, using similar techniques to those used so successfully in in vitro experiments.

Although organotypic slice culture systems are commonly used in preparations of the brain and spinal cord for the examination of synaptic development and function (reviewed in Gahwiler et al.,1997), their use for the study of axonal growth has been comparatively limited. In the brain, co-culture of different functionally connected regions has been used to study pathway formation (reviewed in Bolz et al.,1993). For example, axonal outgrowth of dopaminergic midbrain neurons to the striatum (Holmes et al.,1995), optic nerve growth at the optic chiasm (Chung et al.,2000; Lin and Chan,2003), layer V and VI efferent projections from the visual cortex (Novak and Bolz,1993), and axonal development in layer 2/3 of the visual cortex (Uesaka et al.,2005) have all been studied using ex vivo culture systems. In the periphery, a transverse preparation of rat spinal cord was used to study axonal regeneration after injury (Lee et al.,2002). Transverse sections from chick embryos have been prepared in which spinal nerve outgrowth could be followed over time (Hotary et al.,1996; Hotary and Tosney,1996). Although these preparations helped to elucidate crucial molecular mechanisms in pathfinding decisions, their technical disadvantage is that the outgrowing nerves need to be labeled either post hoc with an immunohistochemical marker or a lipophilic fluorescent dye (e.g., 1,1′, di-octadecyl-3,3,3′,3′,-tetramethylindo- carbocyanine perchlorate [DiI]) to assess outgrowth, or labeled with a fluorescent dye when the slices are prepared in order for imaging to be performed.

Because of the relative ease with which the mouse can be genetically manipulated, an easier solution would be to use a transgenic line that expresses a fluorescent marker in the appropriate cells at the appropriate times. Using a mouse line that expresses the enhanced green fluorescent protein (EGFP) in all newborn, postmitotic neurons in the central and peripheral nervous system (Tucker et al.,2001), we describe a simple system that allows for the imaging of early peripheral nerve development in slice cultures prepared from embryos. We have prepared transverse slices at various axial levels of the body, but we focus upon the innervation of the forelimb by motor neurons of the spinal cord and sensory neurons of the dorsal root ganglion (DRG). Using epifluorescent microscopy, strong signal from newborn neurons of the spinal cord and DRG can be followed along the entire length of the outgrowing spinal nerves derived from these two populations. We could perform imaging of the outgrowth and pathfinding decisions made by spinal nerves during innervation of the forelimb. This enables us to not only image and quantify outgrowth at distinct embryonic stages, but also allows for the investigation of the genes involved in these processes through a combination of genetically manipulated mouse strains and pharmacological manipulation of the slices themselves.


General Approach and Design Characteristics

To visualize growth of nerves into the periphery in real time, slices were prepared from a transgenic mouse line (tauGFP) expressing EGFP from the locus (Mapt) encoding tau, a microtubule-binding protein (Tucker et al.,2001). During prenatal development, the tau gene is specifically and strongly expressed in postmitotic neurons shortly after their birth. The EGFP cDNA was inserted into exon 1 of the tau gene, in frame with the endogenous initiation codon, resulting in a fusion protein between the first 31 amino acids of the tau protein and EGFP that distributes freely in the cytosol. The first fluorescent signal is observed from 9.0 days post coitum (dpc) in the trigeminal ganglion. At 24 hr later, it appears throughout the peripheral and central developing nervous systems (Fig. 1). The entire length of developing axons can be clearly visualized and imaged over time. A previously reported method of slice culture used a “guillotine” to slice embryos in a transverse manner (Tucker et al.,2001). In this method, a steel grate wrapped serially with tungsten wire was projected with a spring onto embryos lying beneath on a piece of agarose. Although this method had the advantage of allowing 4–6 embryos to be prepared at one time, it resulted in only one or two slices in a litter of 6–8 embryos that were both properly cut and that would show robust growth over time. This finding made it essentially impossible to work with homozygous lethal mouse mutants, in which only 25% of the embryos in a litter are homozygous mutant in the gene of interest. To circumvent this problem, and to improve the quality of the cultures in general, we developed a new method that allows for robust and reproducible growth from practically every slice. This method is based on the individual slicing of embryos with a Vibratome, followed by culturing on Millicell inserts in culture medium.

Figure 1.

Expression of enhanced green fluorescent protein (EGFP) in the developing nervous system of tauGFP mice, revealed by epifluorescence. A: Sagittal view of the uncut embryo. Brachial plexus (asterisk) of a 10.75 days post coitum (dpc) embryo. The dorsal root ganglia (DRGs) are indicated (C5–Th1) from which the spinal nerves contributing to the forelimb (outline in white dots) are derived. The rostral (R) –caudal (C) and dorsal (D) –ventral (V) axes are indicated. B–D: Acute transverse slices from embryos at the axial levels C5–Th1. B: At 10.5 dpc. The ends of the outgrowing spinal nerves are indicated (asterisk) on the right side of the slice only, as are the DRG (D) and the motor neurons (M) of the ventral spinal cord. C: At 11.0 dpc. D: At 11.5 dpc. On the right side of the slice, the nascent suprascapular nerve (arrow), the paravertebral (1) and prevertebral (2) sympathetic ganglia are indicated. The D-V axis is indicated for B–D. Scale bars = 1 mm in A, 500 μm in B–D.

Preparation of Slice Cultures

Embryos at 10.25–10.75 dpc (Theiler stage 16–17) were embedded into low melting point (LMP) agarose as follows: Whole embryos were placed on their side on bacterial plates on ice. Extra phosphate buffered saline (PBS) was removed with sterile strips of Whatmann (3MM) chromatography paper, and the paper was also used to position the embryos so that they were lying straight in the rostral–caudal axis at the axial level where slices were collected. Embryos were fixed in this position with 5 drops of molten agarose from a 7-ml plastic Pasteur pipette (VWR) placed on the ventral side of the embryo and later completely embedded in agarose applied to the dorsal side. After the agarose solidified, the trunk of the embryo (approximately the axial levels between cervical vertebra 1 to lumbar vertebra 2) was isolated by cutting with a razor blade, covered successively on the second lateral side with agarose, and mounted with superglue on the chuck of the microtome, such that the caudal aspect of the embryo was sitting upon the chuck, with the dorsal aspect facing the blade. The 350–400 μm slices were cut in Ca++-Mg++-free Hanks' balanced saline solution (HBSS) at 4°C, using a HM 650 Microtome with a vibrating blade (Microm). Each slice was transferred with a Pasteur pipette into 4–5 ml of HBSS in a six-well plate kept on ice. At the axial level containing the developing forelimb, two to three slices per embryo could be obtained. Agarose was then carefully removed around the slices with fine forceps (Dumont). Four to five slices were transferred onto a 3-cm Millicell insert (Millipore) that had been rinsed with culture medium. The membranes were placed into 10-cm Petri dishes with 6 ml of culture medium and placed in an incubator at 37° C and 5% CO2. Half of the culture medium was changed at least once a day (every 8–12 hr).

Imaging of Peripheral Nerve Outgrowth

The embryonic forelimb bud first appears as a distinct evagination of the lateral ectodermal tissue between the fifth cervical (C5) and the first thoracic (Th1) vertebrae at 9.5 dpc, as the neural tube is closing. At the same time, neural crest-derived migratory cells form condensations on either side of the neural tube that will give rise to DRGs. As soon as sensory neurons are born in the DRG, the postmitotic neurons elaborate a bifurcated process, one end of which grows in the direction of the dorsal neural tube, the other end of which proceeds ventrally to reach the developing limb bud. At 10.5 dpc, newborn motor neurons situated in the mantle zone of the ventral spinal cord have initiated an axon that projects ventrally, to join the DRG-derived axons in the brachial plexus between the ventral spinal cord and the forelimb bud. The brachial plexus is formed by spinal nerves, arising from the fifth cervical to the first thoracic segment of the spinal cord (Fig. 1A). They form three trunks that separate and refasciculate to establish finally the nerves innervating the forelimb.

All prepared slices showed a common tissue morphology, regardless of the developmental stage at which they were prepared (Fig. 1). The neural tube had strong green signal in the lateroventral half, derived mostly from newborn motor neurons, with commissural connections between the left and right sides. The DRGs located adjacent to the upper lateral half of the neural tube showed a strong speckled signal corresponding to individual sensory neurons. Axons derived from spinal cord motor neurons left the neural tube ventrally in the direction of the forelimb. Processes left the DRGs also in the direction of the forelimb and met the spinal cord-derived axons to form the brachial plexus. Transverse slices of 10.5 dpc embryos usually contained one or two intact spinal nerves. At this time point, the spinal nerves had entered the forelimb bud and in many slices, the initial division into trunks and subsequently into dorsal and ventral cords had occurred (Fig. 1B). Slices made of 11.0 dpc embryos showed a clear separation of these dorsal and ventral cords (Fig. 1C). In slices of 11.5 dpc embryos, these two nerve cords proceeded to extend along the length of the forelimb (Fig. 1D), and a laterally projecting ramus from the proximal portion of the spinal nerve could also be observed (Fig. 1D, arrow).

To reproduce the endogenous developmental pattern of limb innervation ex vivo, we cultured slices from 10.25 or 10.5 dpc embryos and took images at regular time intervals (Fig. 2A). The first images were obtained after a 2- to 5-hr acclimation period in the incubator. We found that the most critical parameter for maintenance of short-term growth was temperature. Short drops in temperature of even 5° C reduced outgrowth tremendously. The embryos were placed upon ice during dissection, and the slices were cut in a bath that had been chilled to 4° C. It was critical, therefore, that the slices were brought back to a warm incubator and allowed to recover before the imaging could begin. Although one could in principle begin earlier, normally we waited for at least 2 hr before starting to observe nerve behavior.

Figure 2.

Timed imaging series and ultrastructural analysis of nerve outgrowth in a tauGFP embryo. A: Epifluorescent images of a transverse slice prepared from a 10.25 days post coitum (dpc) tauGFP embryo. The dorsal side of the embryo is at the top of each figure. Hours spent in culture is indicated in the bottom left of each image. B–D: Toluidine-blue stained semi-thin transverse section of a slice prepared from a 10.5 dpc tauGFP embryo and cultivated for 24 hr. B: Arrow indicates nerve deep within the forelimb slice. C: Higher magnification of B. Asterisk indicates nerve bundle. Arrowhead indicates blood vessel in the vicinity of the nerve bundle. D: Higher magnification of B. Arrows indicate periderm cells. Asterisk indicates underlying epidermis. E–G: Transmission electron microscopy (TEM) upon the nerve in the transverse section in B. E: A lower-magnification view of the nerve bundle. F: Several axons within the nerve. E,F: Presumptive Schwann cells (S) directly adjacent to the nerve bundle can be seen. G: View of a single axon. Scale bars = 500 μm in A, 50 μm in B, 20 μm in C,D, 2 μm in E, 500 nm in F,G.

At the beginning of the image series, sensory projections derived from DRGs have already met motor axons from the spinal cord and formed together the nascent brachial plexus (Fig. 2A; 5 hr). In the three following images, a separation of the trunks into dorsal and ventral cords can be clearly observed, accompanied by a defasciculation at the ends of the cords (Fig. 2A; 12–18 hr). The following two images demonstrate a continued outgrowth of the dorsal and ventral cords and the emergence of a proximal ramus, the nascent suprascapular nerve (Fig. 2A; 25 hr, arrow), that grew toward the lateral edge of the slice. The final two images demonstrate branching of the ventral and dorsal cords into main nerve branches and a more pronounced growth of the ventral cord, while the dorsal cord reached the edge of the forelimb bud (Fig. 2A; 33 and 39 hr). After 48 hr, the axons started to grow along and down the length of the forelimb. The tissue does not physically degenerate over the course of the next 12 hr, but the continued growth of the axons along the edges of the limb does not appear to reflect a physiologically normal growth pattern. Therefore, we restricted our culture period to 2 days.

Examination of slices after the culture period revealed that nerves grew deep within the tissue, up to 150 μm from the surface of the section, as revealed by a toluidine blue staining of semi-thin transverse sections (Fig. 2B,C). Nerves were sometimes found to grow in the vicinity of blood vessels (Fig. 2C, arrowhead). Ultrastructural analysis by transmission electron microscopy (TEM) demonstrated the characteristic structure of a developing nerve bundle (e.g., Xue and Honig,1999). Individual axons showed a regular spacing of microtubules and intermediate filaments within the axon (Fig. 2G), and the axons were closely apposed to one another (Fig. 2E,F). The nerve bundles were partially surrounded and contacted directly by polyribosome-rich cells, with extensive cytoplasmic processes that are very likely to be Schwann cells (Fig. 2E,F).

To observe this outgrowth process at a finer resolution, we focused upon the ends of the outgrowing nerve bundles. Outgrowth was measured in slices prepared from 10.5 dpc embryos, in which spinal nerves had already formed the nascent brachial plexus and started to branch into dorsal and ventral cords (Fig. 3). All time series (n > 100) displayed a continuous outgrowth of the dorsal and ventral cords (Fig. 3A–D). Motor and sensory axons grew from the neural tube as fasciculated spinal nerves (e.g., derived in Fig. 3A from two different axial levels), converged into three trunks, defasciculated, and divided into the dorsal and ventral cords (Fig. 3A–D; 4 hr). At the beginning of the series, the axons of the future dorsal and ventral cords showed a decrease in fasciculation in combination with ongoing outgrowth (Fig. 3A–D; 12 hr). After the decision to divide into ventral and dorsal cords was made, the more proximal part of the trunks became thinner, and dorsal rami emerged from the proximal portions of the nerves to grow out to the lateral myotome (Fig. 3A–D; 22 hr). Examination at higher magnification (Fig. 3G) allowed the identification of the ends of individual axons with a morphology similar to that reported for growth cones in vivo (e.g., Hollyday and Morgan-Carr,1995), but the depth of nerve growth in the tissue prevents adequate imaging of these structures with epifluorescence microscopy. However, this application would be ideal for two-photon microscopy.

Figure 3.

Spinal nerve outgrowth in slices prepared from 10.5 days post coitum (dpc) tauGFP embryos. A–D: Slices were prepared from separate embryos and photographed with epifluorescence after 4, 12, and 22 hr in culture. The spinal cord and dorsal root ganglia (DRGs) are located to the left of each figure. E: Nerve outgrowth was measured at the indicated time points. Each line represents a nerve bundle in a different slice. F: The velocity of nerve outgrowth (μm/hr) was calculated for four spinal nerves from the measurements in E. Identical colors/numbers represent the same nerves in E and F. G: Higher magnification image of individual axon ends (arrowheads). Scale bars = 200 μm in A–D, 100 μm in G.

Almost all nerve branches showed a continuous outgrowth during the culture period. Although one might assume that nerves at the same developmental stage display similar outgrowth rates, there was no apparent direct correlation between various slices. Some nerves grew very slowly, some began slowly and sped up, while a third group grew at a moderate pace throughout the culture period (Fig. 3E). This finding can be seen in the measurements of nerve velocity over the culture period (Fig. 3F). There were nerves that showed a relatively stable outgrowth (Fig. 3E,F; traces 1 and 4), and others with a variation of more than 30 μm/hr in their outgrowth rates (Fig. 3E,F; traces 2 and 3). The cause of this diversity could lie in the intrinsic developmental pattern of different spinal cord segments that are sending nerve fibers to the forelimb, and this is currently under investigation.

To summarize, we were able to reproduce spinal nerve outgrowth in slice cultures that duplicated outgrowth as it occurs in vivo, although ex vivo growth occurred over a longer time frame (approximately 30% decrease in nerve outgrowth rates compared with development in vivo).

Development of Non-neural Tissues Within Cultured Slices

To show that the slices retained a developmentally normal morphology over the course of the culture, we examined them with scanning electron microscopy (SEM) after 24 hr in culture (Fig. 4A,B). The proximal half of the forelimbs, where the tissue had been cut (Fig. 4A, arrows), had not been overgrown with epidermal cells and remained free of debris. The apical epidermal ridge could clearly be seen on the distal end of the forelimb (Fig. 4A, arrowhead), the ventricles of the developing heart were clearly separated, and the spinal cord and DRGs retained their proper relationship to each other. In one-third of the slices, the dorsal portion of the neural tube opened up after culture, as shown (Fig. 4A). At higher magnification, the distal forelimb shows the characteristic morphology of the periderm cells that develop between 10.5 and 11.5 dpc (Nakamura and Yasuda,1979), which have a flattened, longitudinally oriented, polygonal shape (Fig. 4B, asterisks). Developing periderm cells can also be observed in toluidine blue-stained semi-thin transverse sections as a single layer of thin, elongated cells covering the epidermal layer beneath (Fig. 2D). In summary, aside from some cell growth out of the dorsal neural tube (Fig. 2A; 33 and 39 hr) and a split in the dorsal-most aspect of the neural tube in a proportion of the slices (Fig. 4A), the tissue appeared morphologically normal after the culture.

Figure 4.

Development of non-neural tissues within embryonic slices prepared from 10.5 days post coitum (dpc) tauGFP embryos and cultivated for 24 hr. A: Scanning electron microscopy (SEM) image of an entire slice. The dorsal aspect of the slice is at the top of the figure. Arrows indicate where the forelimb was cut by the vibratome. Arrowhead indicates the apical ectodermal ridge of one forelimb. B: SEM image of the distal end of the forelimb in a slice. The proximal side of the forelimb is to the left. Asterisks indicate periderm cells. C,D: Chondrogenesis was examined using an Alcian blue stain. Dots outline the forelimb in each case. C: Left: A slice from a 10.5 dpc embryo fixed immediately after cutting. Right: A slice cultivated for 24 hr. D: Forelimb from a 11.5 dpc embryo. E: Myogenesis was examined using an antibody recognizing MyoD1. Left: A 11.5 dpc embryo. Right: A slice cultivated for 24 hr. The asterisk indicates somitic myotome lateral to the DRGs. The lateral (arrows) and medial (arrowhead) myogenic compartments of the forelimb are indicated. Scale bars = 200 μm in A,C,D,E(left), 20 μm in B, 500 μm in E(right).

Between 10.5 and 11.5 dpc, several important changes occur in the development of the limb bud itself. At 11.5 dpc the first signs of cartilage formation can be observed in the forelimb in vivo, as revealed by Alcian blue stainings (Martin,1990). As can be seen in slices prepared at 10.5 dpc, no Alcian blue staining was visible in the forelimb (Fig. 4C, left). We compared the extent of staining of forelimbs from 11.5 dpc embryos with slices of 10.5 dpc embryos that had been cultivated for 24 hr. In both cultivated slices (Fig. 4C, right) and the embryo (Fig. 4D), a blue staining along the proximal central course of the forelimb could be observed, indicating that chondrogenesis had begun appropriately in the cultivated slices.

In the time period between 10.5 and 11.5 dpc, the first signs of myogenesis can also be observed within the forelimb, as evidenced by the expression of the myogenic transcription factors MyoD1 and myogenin (Sassoon et al.,1989). We compared the spatiotemporal pattern of MyoD1 expression in forelimbs from 11.5 dpc embryos with slices of 10.5 dpc embryos that had been cultivated for 24 hr, using an antibody recognizing MyoD1 (Fig. 4E). In both cases, large numbers of myogenic cells could be identified in the somitic myotome lateral to the DRGs, the lateral and medial myogenic compartments of the forelimb, and nascent cardiac muscle in the developing heart.

In summary, three of the major tissue components of the developing forelimb, the peridermal cells of the epidermis, the chondrocytes of the developing humerus, and the myogenic precursors of forelimb muscles, all develop in an appropriate time frame in slices prepared from 10.5 dpc embryos. Although the development of these tissues appears to be normal, the forelimb clearly does not achieve the same size as it does in vivo (cf. embryo and cultivated slices in Fig. 4C–E), indicating that overall proliferation rates are reduced in the ex situ environment.

Quantification of Cell Death

To examine the viability of neurons and non-neuronal tissue in the slices over the culture period, we used fluorescent-activated cell sorting (FACS) upon dissociated slices at progressive time points. Using FACS analysis we could determine with GFP signal the ratio between neurons (GFP-positive) and all other cell types (GFP-negative), and by adding the dye propidium iodide (PI) to the cell suspension, we could determine the ratio between damaged or dead cells (PI-positive) and viable cells (PI-negative; Fig. 5A). We examined slices directly after the cutting procedure and after 24 and 48 hr in culture. At all three time points, we observed a very low amount of cell death of both neurons and non-neuronal tissue (Fig. 5B), by comparing the number of PI-positive and PI-negative cells within each population. There was no significant difference in cell death between neurons and non-neuronal tissue.

Figure 5.

Analysis of cell death and apoptosis in slices prepared from 10.5 days post coitum (dpc) embryos. A: Quantification of cell death by fluorescent-activated cell sorting (FACS) analysis. Slices were trypsinized and dissociated into single-cell suspensions after the indicated culture times. Propidium iodide (PI) was added to assess the number of viable cells, and the cells were then analyzed for red (PI) and green (green fluorescent protein [GFP]) fluorescence by FACS. A: Representative FACS plots of slices cultured from tauGFP mouse lines. Forward (FSC) vs. side-scatter (SSC) is indicated on the left. The population of whole cells analyzed for red and green fluorescence is indicated by the gate. Green (GFP) vs. red (PI) fluorescence is indicated on the right. All cells to the right of the vertical line in this plot were judged to be GFP-positive (GFP+) and, therefore, neurons. All cells above the horizontal line in this plot were judged to be PI-positive (PI+) and, therefore, dead. B: The percentage of PI+ cells was calculated from the fraction of cells in the upper left (GFP−, non-neuronal) and upper right (GFP+, neuronal) quadrants of the scatter plot shown on the right (GFP vs PI) in A, compared with the two respective PI-negative populations (SEM, n = 3). C,D: Apoptosis was examined using an antibody recognizing activated caspase-3 (left). The DAPI staining of cell nuclei is shown on the right. C: 11.5 dpc embryo. D: A slice prepared from a 10.5 dpc tauGFP embryo and cultivated for 24 hr. Scale bars = 100 μm in C,D.

To compare the spatial pattern of apoptosis occurring within the cultured slice with uncultured embryos, we used an antibody recognizing the cleaved, activated form of caspase-3. In 11.5 dpc embryos, almost all of the positive cells were found in the DRGs, but in the forelimbs of 11.5 dpc embryos, we saw very little staining (Fig. 5C), as reported previously (Fernandez-Teran et al.,2006). Examination of slices prepared from a 10.5 dpc embryo and cultivated for 24 hr displayed in contrast a higher amount of caspase-3–positive cells. Small numbers of activated caspase-3–positive cells could be found scattered throughout the forelimb, and a higher number of cells appeared in the DRG and the dorsomedial portion of the spinal cord (Fig. 5D). Quantification of the activated caspase-3–positive cells showed a close agreement with the results delivered by FACS (data not shown).

Imaging of Nerve Outgrowth in the Targeted Sema3A Mouse Mutant

The use of the mouse for this slice culture system offers the opportunity to examine the effect of mutations in specific genes in the development of peripheral nerves. As an example, we have investigated the role in peripheral nervous system development of the gene encoding Semaphorin 3A (Sema3A). Sema3A is a secreted protein within the semaphorin gene family that acts as a chemorepulsive cue throughout the developing central and peripheral nervous system (Pasterkamp and Kolodkin,2003). Sema3A binds to the receptor complex neuropilin-1/plexinA to induce a signaling cascade leading to cytoskeletal rearrangements and growth cone collapse. In the developing forelimb, Sema3A is expressed in tissues surrounding the outgrowing spinal nerves (Wright et al.,1995). Elimination of either Sema3A (Taniguchi et al.,1997; White and Behar,2000) or its receptors neuropilin-1 (Kitsukawa et al.,1997) and plexin-A4 (Suto et al.,2005; Yaron et al.,2005) leads to promiscuous outgrowth of spinal nerves into normally nonpermissive regions.

To perform timed imaging of embryonic slices derived from mice lacking Sema3A (Behar et al.,1996), we crossed the Sema3A knockout mouse to the tauGFP line. As reported previously (White and Behar,2000), at 10.5 dpc, we could clearly observe premature outgrowth and a strong defasciculation of spinal nerves in Sema3A−/− embryos, compared with GFP-expressing littermates that were wild-type or heterozygous for the Sema3A mutation (Fig. 6). At the beginning of the imaging series, slices from Sema3A−/− embryos showed an approximately 12-hr advance in spinal nerve outgrowth compared with wild-type or Sema3A+/− littermates (Fig. 6; 9 hr). In the Sema3A−/− embryos, not only had the spinal nerves prematurely formed trunks and fasciculated into dorsal and ventral cords, but also the ventral cord had branched and the main myotomal rami had already developed. As the spinal nerves in the Sema3A+/+ slices began their outgrowth (Fig. 6; 17 hr, 25 hr), the nerves in the Sema3A−/− slices elongated much more slowly (Fig. 6; 13 hr, 17 hr, 25 hr). After 25 hr of culturing, the Sema3A+/+ and Sema3A−/− slices displayed an equivalent progress in nerve outgrowth (Fig. 6; 37 hr). Comparing the Sema3A−/− slice at the beginning and end of culture revealed that the nerve branches at the end of culture did not appear as defasciculated as at the beginning of culture, resembling more the Sema3A+/+ slice (Fig. 6; 37 hr). This could represent the beginning of the “self-correction” process that is known to occur between initial outgrowth and birth (White and Behar,2000).

Figure 6.

Spinal nerve outgrowth in slices prepared from a litter of 10.5 days post coitum (dpc) Sema3A mutant embryos containing the tauGFP construct. Upper panel: Sema3A+/+, lower panel: Sema3A−/−. The dorsal side of the embryo is at the top of each figure. Hours spent in culture is indicated in the bottom right of each image. Asterisks indicate the ends of the outgrowing spinal nerves, only on the right side of the slice. Scale bar = 500 μm.


The method described here allows one to prepare and culture embryonic mouse slices successfully on membranes for up to 48 hr and to follow the outgrowth of spinal nerves into the forelimb bud over time. We did not observe any change in the morphology of the tissue that could impair normal nerve outgrowth into the periphery. In almost all prepared slices of the tauGFP mouse line, the outgrowing axons showed normal patterns of growth and no innervation of normally avoided tissue regions in the forelimb. FACS and apoptosis analysis further demonstrated a low mortality rate of both neurons and non-neuronal cells, even after 48-hr in culture. We, therefore, conclude that this slice-culture system allows for a faithful reproduction of the initial steps of innervation of the forelimb by neurons of the spinal cord and DRG and, thus, provides a useful model system to investigate spinal nerve development in situ.

Although the gross patterns of nerve development appeared normal, it remains to be seen if specific subpopulations of DRG and motor neurons are responding to short- or medium-range cues in the slices in the appropriate manner. To definitively answer this question will require a combination of anterograde and retrograde labeling of neuronal soma and nerve ends, respectively, followed by immunohistological analysis of the labeled neurons to determine their subpopulation identity. In addition, we recorded variable rates of growth from the individual nerves, varying not only between nerves themselves but also varying for a single nerve. With respect to variation between nerves, this could be partially explained by the fact that the outgrowth of ventrally oriented nerve cords was faster than the dorsally located cords. Both branches reached the edge of the slice at approximately the same time, although the ventral cords had a much further distance to grow. This finding is similar to the observation that cranial ganglia-derived nerves display an inherent difference in their outgrowth rates that correlates with the distance they must travel to reach their target, in that nerves that need to cover a greater distance grow faster (Davies,1989). With respect to variation in growth rates of individual nerves, the “pausing” of nerves during outgrowth is a well-documented phenomenon. In the periphery, spinal nerves have been seen to gather in the brachial plexus and wait for up to 24 hr before they continue to grow into the fore- and hindlimbs (Tosney and Landmesser,1985), whereas in the developing cortex, thalamic efferent axons pause from days to weeks in the transient subplate structure before they continue to grow to their targets in the cortical plate lying above (Allendoerfer and Shatz,1994). However, the periods of very slow growth that we observed all occurred after the nerves had left the brachial plexus, possibly indicating new, previously unreported “waiting periods.” The cause of this behavior is not clear, but this is a topic of active investigation.

The validity of this ex situ model is confirmed by analysis of slices prepared from mice lacking the gene encoding the chemorepulsive secreted protein Sema3A. We were able to reproduce in the slice culture system several key features found in the mutant embryo. First, we observed a precocious outgrowth of spinal nerves (Huber et al.,2005). Second, we observed aberrant projections derived directly from the DRG and aberrant branches from the spinal nerves themselves (Taniguchi et al.,1997; White and Behar,2000). Third, we recorded the beginning of the “self-correction” procedure that takes place after initial outgrowth in the Sema3A mutant, resulting in a pattern of peripheral projections that at birth appear relatively normal (Behar et al.,1996; White and Behar,2000). The plethora of targeted mutations in genes responsible for controlling nerve outgrowth could allow for a widespread use of this model to examine the effects of these mutations in real time, thereby complementing the static immunohistochemical analysis that is currently the norm. In addition, the slices are amenable to pharmacological and microsurgical manipulation, either through the application of beads containing growth factors or function-blocking antibodies (Tucker et al.,2001), through bath application of such substances, or through heterochronic transplantation experiments, as has been done so successfully in the chick (Wang and Scott,2000).

Although we have used transverse sections of the embryo at the level of the forelimb bud as a model system, this slice culture method is applicable to a wide variety of alternatives. For example, to obtain more insight into the longitudinal interaction of spinal nerves in the brachial plexus, and not to be restricted to the imaging of one or two axial levels, we have also prepared sagittal slices at the level of the forelimb. In this preparation, all spinal nerves contributing to the brachial plexus (C5–Th1) could be observed at the same time (I. Brachmann, unpublished observation). Likewise, the outgrowth of intercostal nerves growing toward the axial muscles could be imaged with such a slice preparation. In the central nervous system, growth of axons from olfactory epithelium toward the olfactory bulb (Zaghetto et al.,2007), thalamocortical tract formation (M. Shakèd, personal communication), and perforant pathway formation from the entorhinal cortex to the hippocampus (Hechler et al.,2006) could all be successfully imaged over time in slice cultures prepared from the developing forebrain. Ideally, such cultures would be applicable to two-photon confocal microscopy, as this would allow not only for a visualization of nerves located deeper within the slice, but also a three-dimensional reconstruction of nerve growth over time.


Transgenic Mice

All experiments were conducted according to the guidelines of the state of Baden-Württemberg, Germany. TauGFP mice were generated as described (Tucker et al.,2001). This line had been backcrossed more than 10 generations to the C57BL/6 background and maintained as a homozygous mutation. The mouse line lacking Sema3A (Behar et al.,1996) was crossed into the tauGFP mouse line. As the Sema3A knockout mouse is perinatal lethal as a homozygote, mice homozygous for tauGFP and heterozygous for Sema3A were mated to obtain homozygous-deficient Sema3A embryos. Offspring of both lines were genotyped by polymerase chain reaction (primer information available upon request).

Vibratome Slicing and Culture of Slices

Noon of the day that the vaginal plug was observed was designated as embryonic day 0.5 dpc. Timed pregnant mice were killed by cervical dislocation, and embryos at developmental stages 10.25– 10.75 dpc were removed from the uterus, cleared of the amnion and chorion, and kept in ice-cold PBS (without Ca2+/Mg2+). Embryos were embedded (described above) in 4% LMP agarose (Ultrapure, Invitrogen) in PBS kept at approximately 40°C on a heating plate. The agarose block was affixed to the chuck of the microtome using a few drops of Loctite 406 glue (Henkel corporation). Slices were cut using a HM 650V Microtome (Microm International GmbH) with the following settings: thickness, 350–400 μm; frequency, 50; amplitude, 1.2; velocity, 8; single stroke mode. Slices were removed from the slicing chamber using Pasteur pipettes from which the narrow end was cut off, generating an even glass tube with an inner diameter of 5 mm, and collected in six-well plates. Slices were then transferred to 30-mm Millicell inserts, (Millipore, PICMORG50) and cultured in an incubator at 37°C and 5% CO2.


TEM was performed as described (Finotto et al.,1999).


Slices were fixed overnight at 4°C in 2.5% glutaraldehyde/0.1 M PIPES pH 7.4, washed 3 times in 0.15 M PIPES pH 7.4 at 4° C, treated for 1 hr at room temperature with 1% OsO4, washed 3 times with 0.15 M PIPES pH 7.4, and finally dehydrated in an ascending ethanol series. The slices were transferred in a small steel basket to a CPD 030 critical point dryer (BAL-TEC AG) that was one-third filled with 100% ethanol, and the cap was sealed airtight. Once a temperature of 10°C and a pressure of 50 Bar was reached, the chamber was repeatedly emptied of liquid and filled with CO2 (approximately 6–8 times) without uncovering the sample so that all ethanol was removed. The chamber was then completely filled with liquid CO2 and heated to a temperature of 40°C and a pressure of 80 Bar. The slices were then sputter-coated with a 5-nm carbon layer in a high vacuum coating system (BAL-TEC MED 020). For SEM, a LEO 1530 field emission scanning electron microscope with a Schottky cathode was used (LEO Elektronmikroskopie GmbH, Oberkochen, Germany).

Alcian Blue Staining

Slices and dissected embryos were fixed in Bouin's solution for 2 hr, rinsed with a solution of 70% ethanol/0.1% NH4OH for 12–24 hr in five to eight changes, placed in 5% acetic acid for 2 hr, stained with 0.05% Alcian blue 8GX in 5% acetic acid, rinsed for 2 hr in 5% acetic acid, dehydrated in methanol, and finally cleared in 1:2 benzyl alcohol/benzylbenzoate.

Immunohistochemical Analysis

Slices were fixed in 4% paraformaldehyde at 4°C for 2 hr; transferred to 10, 20, and 30% sucrose in 1× PBS; and then mounted on dry ice in tissue freezing medium (Jung, Leica, Nussloch, Germany). Twelve-micrometer sections were cut on a cryostat (Leica CM3050 S) and blocked for 1 hr at room temperature (blocking buffer: 0.5% Triton X-100, 1% bovine serum albumin, and 5% native goat serum in 1× PBS). Primary antibodies (Ab) were diluted in blocking buffer as follows and incubated overnight at 4°C: anti-MyoD1 (mouse monoclonal clone 5.8A; BD Pharmingen, Erembodegem-Dorp, Belgium) 1:1,000, and anti-cleaved caspase-3 (Asp175, clone 5A1, rabbit mAb; Cell Signaling Technology, Frankfurt, Germany) 1:200. After washing slides 4 times with 1× PBS, the secondary Abs (goat anti-rabbit AlexaFluor 488 or 555 and goat anti-mouse AlexaFluor 488 or 555; Invitrogen, Karlsruhe, Germany) were applied for 1 hr at room temperature, followed by 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI; 2.5 μg/ml) staining, washed 3 times in PBS, and mounted for microscopy (Aqua PolyMount, Polysciences Europe, Eppelheim, Germany).


Solution reagents were purchased from Invitrogen, unless otherwise indicated. Dissection buffer: 1× Ca++-Mg++-free HBSS, 10 mM HEPES buffer pH 7.3, 500 U/ml penicillin/streptomycin. Culture medium was DMEM, 25% 1× HBSS, 25% fetal bovine serum (heat-inactivated, Invitrogen), 0.5% glucose (Sigma), 1 mM L-glutamine, 2.5 mM HEPES, pH 7.3. We have also used a serum-free medium (F-12, Sigma) that included insulin, progesterone, sodium selenite, and transferrin (recipe in Hotary et al.,1996) and found that peripheral nerves were also able to grow in a similar pattern and with similar outgrowth rates as with serum-containing medium. Therefore, a serum-free medium could be used when one wishes to avoid the potential influence of serum-based growth factors.

FACS Analysis

Directly after the Vibratome slicing procedure, the slices were trypsinized (0.25% Trypsin in PBS [Invitrogen], 0.5 mM EDTA pH 8.0) for 10 min in a 37°C waterbath. The trypsin was washed out twice with warm PBS, and the tissue was dissociated by pipetting up and down with a Gilson pipette (200 μl). FACS analysis was performed on a FACSCalibur device (Becton Dickinson, Biosciences, San Jose, CA). Just before FACS analysis, PI (250 ng/ml final concentration) was added to allow quantification of dead cells in the red channel. A proper compensation for the GFP-based fluorescent signal was first determined using cells not labeled with PI. The 50,000 events were counted for each condition, with a sort rate of < 1,000 events per second. Nontransgenic embryos were used as a negative control to estimate background (less than 1%).

Microscope Setup

Imaging was performed with an up- right epifluorescent microscope (BX61WI, Olympus Germany, Hamburg) using ×4 (numerical aperture [NA] 0.1), ×10 (NA 0.3), and ×20 (NA 0.5) objectives. Filter: excitement: 470–490 nm, dichroic mirror: 505 nm, emission: 510–550 nm. Images were taken with an F-view II 12-Bit monochromatic CCD camera (Peltier-cooled, 1,376 × 1,032 Pixel, Soft Imaging System, Muenster, Germany). Slices were kept in the incubator next to the microscope setup at 37°C and 5% CO2. Outgrowth measurement was performed with the program analySIS (Soft Imaging System, Muenster, Germany) and image editing with Adobe Photoshop.


The authors acknowledge the original source for the idea to perform slice culture upon mouse embryos (Hotary et al.,1996); the receipt of the Sema3A knockout mice from Reha Erzurumlu; Karin Gorgas, Hanno Svoboda for critique of the manuscript; Joachim Spatz (Department of Biophysical Chemistry, University of Heidelberg and Max Planck Institute for Metals Research, Stuttgart, Germany) for the use of the scanning electron microscope; Barbara Brühl for semi-thin sections and TEM; Stefan Wölfl for kind use of the FACS device; C. Peter Bengtson and Claudia Mandl for assistance with the microtome; Bernice Pawletta, Christina Spassova, and Dmitry Rusanov for excellent technical assistance; and Jochen Wittbrodt for the use of a fluorescence-equipped binocular microscope.