Cytomechanics of neurite outgrowth from chick brain neurons

Axonal development by peripheral neurons has been shown to be intimately dependent on mechanical tension. Axons of chick sensory neurons and PC12 cells can be experimentally initiated by applying tension to the margin of the cell body (Bray, 1984; Zheng et al., 1991; Lamoureux et al., 1997), and tension also mediates spontaneous initiation of axons by cultured chick sympathetic neurons (Smith, 1994). After initiation, axonal elongation occurs visibly over the course of seconds and minutes both from experimentally applied tension (Bray, 1984; Zheng et al., 1991) and from tension produced by the growth cone (Lamoureux et al., 1989). When axonal elongation is experimentally stimulated by ‘towing’ with a needle, the axon can elongate at physiological and far-above-physiological rates for many hours (Bray, 1984; Zheng et al., 1991). Most significantly, in our view, the rate of neurite elongation by chick sensory neurons and PC12 cells is a simple linear function of the applied force above some minimum threshold tension (usually around 100-200 μdynes) under a variety of culture conditions and at all elongation rates (Zheng et al., 1991; Lamoureux et al., 1992, 1997). That is, axonal elongation bears a robust mechanical equivalence to the elongation of a Newtonian fluid-mechanical element, i.e. a ‘dashpot’ like the piston on screen doors that prevents slamming. Thus, tension apparently integrates the complex chemistry underlying axonal elongation to produce a simple fluid-like relationship between a tension input and a growth output.

In addition to this fluid-like growth behavior, chick sensory neurons and PC12 cells show clear evidence of solid, elastic behaviors below the tension threshold required for growth. That is, neurites of these cells behave like springs at low tension levels (Dennerll et al., 1988; Lamoureux et al., 1992). If the neurite is rapidly lengthened by plucking with a needle, the increase in force obeys Hook’s law. At longer time scales (and low tension), the neurites behave like fluid-damped springs, i.e. behave like viscoelastic solids (Dennerll et al., 1989). And neurites of both cultured cell types support a static ‘rest tension’ that does not dissipate with time; i.e. the ‘spring’ of the neurite is normally slightly stretched.

Van Essen (1997) has proposed that significant aspects of brain morphogenesis are driven by tension, based in part on the reasonable but untested assumption that central neurons behave similarly to peripheral neurons in response to mechanical forces. We report here a study of the responses of embryonic chick forebrain neurons to experimentally applied tension. Brain neurons show fluid-like growth responses to tension, entirely similar to that of peripheral neurons from chick embryos. However, brain neurons show only transient elastic behavior to low tension levels, in contrast to chick sensory neurons. We also report that these brain neurons are unusually reliable and convenient to culture.

Neurons from the cerebral hemispheres of seven- or eight-day-old chick embryos were cultured essentially by the method of Sensenbrenner and colleagues (Sensenbrenner et al., 1978; Pettman et al., 1979) with some minor modifications. The principal modification is growth of neurons in L-15 medium supplemented with 0.6% glucose, 2 mM glutamine, 100 i.u./ml penicillin, 100 μg/ml streptomycin, 100 ng/ml 7S nerve growth factor (Harlan, IN), and 10% FCS. The use of L-15 medium (Gibco Labs, NY) permits cell growth and micromanipulation without a CO₂-supplemented atmosphere. Cells were grown at low density (1,500-2,000 cells/cm²) on 60 mm tissue culture dishes pretreated with poly-L-lysine (0.1% in phosphate buffered saline for 30 minutes to 1 hour and rinsed three times with sterile water).

Force was applied to various regions of chick forebrain neurons with calibrated glass needles as described by Zheng et al. (1991) and Lamoureux et al. (1992). Briefly, two needles were mounted in a micromanipulator; one needle was calibrated for its bending constant and used as a pulling needle applied to the cell, while the other needle was used as an unloaded reference for bending of the towing needle and for possible drift of the micromanipulator system. The bending constants of the pulling needles were between 3 and 7 μdyne/μm, and needles were pre-treated with concanavalin A (10 mg/ml in phosphate-buffered saline) to aid attachment of the cell to the needle. All applications of force were recorded by videotape at 24× time lapse for subsequent analysis of neurite length and needle bending, i.e. magnitude of forces applied.

For neurite intiation, chick forebrain neurons without extant neurites were attached to calibrated experimental needles at the cell margins. The response of the cell to the application of various tension levels was recorded on videotape and analyzed for tension and neurite length as above.

For determination of the relationship between applied tensions and elongation rate, a calibrated needle was attached to the growth cone of a spontaneous neurite, and the neurite was pulled in 30-60 minute ‘steps’ of constant force (Zheng et al., 1991). That is, a tension magnitude was chosen, beginning at 10-20 μdyne, and this tension was held constant for 30-60 minutes by moving the micromanipulator to maintain the appropriate deflection of the calibrated needle. Subsequently, the same technique was used to apply 30-60 minute periods of higher tension to the neurite, each level typically 20-50 μdyne higher than the previous value. Neurite lengthening was measured from the videotape record of the experiment. For each period of constant force, a rate of elongation was calculated from the new length of neurite acquired during the period divided by its duration.

Some tension-induced outgrowths were examined by immunofluorescence microscopy for the presence of microtubules. Following initiation or elongation by tension, the distal ends of such neurites were micromanipulated from the needle back onto the culture substrate. A diamond-tipped ‘objective’ was used to mark the experimental cell by circling the dish beneath. Immunofluorescent staining of microtubules was carried out by a method similar to that of Thompson et al. (1984): medium was carefully removed, and the culture was permeabilized in 0.5% Triton X-100 in the microtubulestabilizing buffer described by Thompson et al. (1984), then fixed in formaldehyde solution in this same stabilization buffer, all at 37°C, followed by extraction in methanol at −20°C. The neurites were then incubated with primary antibody to tubulin (kindly provided by Dr David Asai), rinsed, then incubated with fluorescein-labeled secondary antibody. The cells are observed through an Odyssey confocal microscope (Noran Instruments, Middleton, WI) equipped with fluorescent optics.

The presence or absence of polyribosomes in axon-like and dendrite-like neurites extending from chick sensory neurons after 5 days in cultures was determined by transmission electron microscopy. Fixation, dehydration, embedding and sectioning of cells were carried out by standard methods previously described (Joshi et al., 1986). Sections stained with uranyl acetate and lead citrate were viewed with a Phillips CM-10 transmission electron microscope.

We cultured brain neurons from the telencephalon of 7- to 8-day-old chick embryos by the method of Sensenbrenner and colleagues (Sensenbrenner et al., 1978; Pettmann et al., 1979; Louis et al., 1981) with small modifications as described in Materials and Methods. Our principal modification is the use of supplemented L15 culture medium, which does not require a CO₂ atmosphere for pH control. We find that culture of the forebrain neurons is unusually simple, inexpensive, and reliable, even when compared to the relatively straightforward culture of chick sensory neurons (Bray, 1991). The two hemi-spheres of the forebrain are easily identified in the intact embryo within the V-shaped notch formed by the two large eyes and the beak. The hemispheres can be removed without recourse to a dissecting microscope. Dissociation by trypsin treatment of both hemispheres from a single embryo produces a large number of cells, enough for thirty to forty 60 mm culture dishes. Smaller numbers of dissociated cells can be obtained without trypsin by incubation in Ca²⁺, Mg²⁺-free Hanks’ balanced saline for 30 minutes and mechanical trituration. Culture on polylysine- or collagen-treated surfaces appears to be required for neurite outgrowth. On plain tissue culture plastic, or laminin-treated surfaces, the cells rapidly form clumps, most of which do not attach to the surface. As previously reported by Sensenbrenner et al., we had most success with polylysine-treated surfaces.

Fig. 1 shows the developmental course of these cultures over approx. 5 days. These neurons undergo a stereotyped developmental sequence of neurite outgrowth that resembles that observed for rat hippocampal neurons in culture (Dotti et al., 1988). Within hours of plating, the margins of cells become highly motile and display the characteristic lamellipodial activity described for many types of cultured neurons prior to neurite outgrowth. By 24 hours in culture, most cells have developed short ‘minor’ processes that varied between 10 and 30 μm in length, with a few cells developing a single, long neurite of uniform caliber (Fig. 1A). By 2-3 days in culture, 25-30% of all cells had developed a single long outgrowth in addition to several shorter, ‘minor processes’ extending from the cell body (Fig. 1B). After 4-5 days in culture, 10-20% of all cells show a single, long, axon-like projection and several shorter, tapered processes that resemble dendrites (Fig. 1C).

The majority of cells, however, do not develop to this stage (stage 4 of Dotti et al., 1988), and a substantial fraction of cells never even develop a single long neurite (Fig. 1B). In those cells that do develop axon-like and dendrite-like processes, ultrastructural examination of the two types of processes showed two well-described cytoplasmic differences characteristic of axonal/dendritic polarity (Deitch and Banker, 1993; Craig and Banker, 1994). In all five neurons examined, the cytoplasm of the dendrite-like processes contained abundant polysomes throughout their length and a lower density of microtubules and neurofilaments than the axon-like processes (Fig. 2A). In contrast, the long axon-like processes of forebrain neurons (Fig. 2B) showed a complete absence of polysomes in regions 50-100 μm from the soma and a similar density of microtubules and neurofilaments to that observed previously in the axon-like neurites of chick sensory neurons (Baas et al., 1987). However, compartmentation of MAP-2 into the somatodendritic compartment, a commonly used marker for axonal/dendritic polarity in rat neurons (Craig and Banker, 1994), did not occur in chick forebrain neurons even after 9 days in culture (immunofluorescence data not shown). In rat hippocampal neurons, differential compartmentation of MAP2 is a relatively late event, requiring approximately 7 days for complete compartmentation (Caceres et al., 1986) while axonal elongations can be identified within 2-3 days after plating at which time ribosomal compartmentation has occurred (Deitch and Banker, 1993).

As prevously described for chick sensory neurons (Zheng et al., 1991), forebrain neurons in culture were able to initiate and elongate neurites in response to mechanical tension. Eight hours after plating, initiation of neurites from cultured chick forebrain neurons was induced by experimentally applying tension to random locations along the margin with calibrated glass needles (Figs 3 and 4). In approximately half the cases, a uniform caliber cytoplasmic process formed with its distal tip attached to the needle (Fig. 3). Initiation in which the pulling needle was distal to the neurite required quite small forces. In 14 experiments, the mean initiating tension was 31±6 (s.e.m.) μdynes, with the minimum tension of 5 μdynes and a maximum of 80 μdynes. This is substantially less than the forces typically required for experimentally induced neurite initiation from chick sensory neurons (100-200 μdynes; Zheng et al., 1991) or PC12 cells (300-1,000 μdynes; Lamoureux et al., 1997). Within the experimentally initiated forebrain neurites, immuno-cytochemistry using fluorescently labeled antibodies against β-tubulin demonstrated the presence of intact microtubule arrays (Fig. 3D). The intensity of staining within experimentally initiated neurites was entirely similar to that of spontaneously initiated (growth cone-mediated) neurites. The remaining half of experimentally initiated neurites elongated with the needle remaining attached to the cell body and the distal tip of the elongation remaining attached to the dish (Fig. 4). As shown in Fig. 4, a growth cone was observed at the distal tip attached to the dish, which showed normal motile activity. Curiously, such ‘proximal initiations’ (needle on the cell-body side of the neurite) required much higher forces than distal initiations (needle at the distal end of the outgrowth as shown in Fig. 3). The mean initiating force in 14 ‘proximal initiations’ was 127±26 μdynes, with a minimum of 35 μdynes and a maximum of 330 μdynes.

We determined the quantitative relationship between neurite elongation and mechanical tension by a method similar to that described by Zheng et al. (1991) for peripheral neurons from embryonic chicks. In one experimental series, experimentally initiated neurites from cells plated 8 hours previously (as in Fig. 3) were tethered by their distal end to a calibrated glass needle and subjected to a step function protocol of towing. Each tension step was maintained at constant magnitude for 30-60 minutes by appropriate adjustments of the micromanipulator, and a sequence of 3-6 steps was applied to single neurite, each step being 20-50 μdynes greater in magnitude than the previous step. As previously observed for chick sensory neurons, the neurites did not equilibrate their length to a given force; rather they elongated continuously in response to experimentally applied tension. Indeed, the elongation rate was a linear function of applied tension (Fig. 5a). A similar protocol of applying steps of tension to spontaneously initiated (growth cone-mediated) neurites after 24 hours in culture also produced a linear relationship between elongation rate and applied tension (Fig. 5b). Fig. 6 shows one such experiment in which a spontaneously initiated neurite was towed and then subsequently processed for immunocytochemistry for microtubules, demonstrating that the tension-induced length of neurite had a normal array and density of neuritic microtubules. Quantitatively, the tension-growth relationship of forebrain neurons showed two similarities to that of chick sensory neurons and one important difference that were consistent for both spontaneously- and experimentally-initiated neurites: the first similarity is that the linearity of the relationships was robust: in 10 of 11 such experiments the correlation coefficient between tension and elongation rate was 0.9 or greater. Second, the frequency distribution of the tension sensitivities of neurite elongation (Fig. 7, the values of the slopes of the lines, i.e. the elongation rate per unit tension) is quite similar to that of chick sensory neurons (compare to Fig. 3; Zheng et al., 1991). For both peripheral and central neurons the sensitivity ranged between 0.5 and 5 μm per hour per μdyne of applied force, with the majority of cells having a tension sensitivity around 1 μm/hour per μdyne. However, a significant difference between peripheral neurons and these central neurons was the absence of a clear positive threshold tension required for elongation, which was typically between 100 and 200 μdynes for sensory neurons. Rather, as shown in Fig. 5, most cells showed a zerogrowth-rate intercept very close to zero tension. In one case, the analysis of the data suggested a small elongation rate under compression (left-most graph, Fig. 5b). As suggested by these analytical results, we observed that chick forebrain neurons elongated continuously at even the smallest applied tensions we could measure, i.e. the neurites elongated with any pull on the neurite.

The absence of a positive threshold tension for neurite elongation and the low net tensions at which neurite elongations were observed suggested that forebrain neurons show less elastic behavior than chick sensory neurons. We found it impossible to analyze elastic behavior of forebrain neurons by the ‘plucking’ technique previously used for chick sensory neurons and PC12 cells (Dennerll et al., 1988; Lamoureux et al., 1992) because of the growth habit of cultured forebrain neurons. Sensory neurons and PC12 cells are attached to the culture surface only at the cell body and growth cone, allowing the unattached neurite to be plucked like a guitar string. In contrast, chick forebrain neurites are attached all along their length on surfaces that support cell adhesion and neurite outgrowth such as polylysine-treated and collagen-treated culture dishes. (On plain tissue culture plastic, and plastic treated with laminin or fibronectin, the cells adhere poorly, form floating clumps, and generally fail to extend neurites.) Nor could these elastic measurements be made on forebrain neurites peeled from their dish attachment: they elongated significantly during such micromanipulations and formed flaccid, roughly sinusoidal elongations totally unsuitable for plucking. Consequently, we analyzed the tendency of such flaccid, sinusoidal processes to contract and shorten. We previously found that chick sensory neurites rapidly take up slack when the needle attached to the growth cone is moved toward the cell body; i.e. after the micromanipiulator is ‘backed up.’ In response to such sudden declines of static tension, chick sensory neurites contracted to rapidly take up the sinusoids that form initially, and in about 2/3 of the cases the neurites strongly contracted to acheive tension levels substantially greater than the initial tension prior to slackening of the neurite (Dennerll et al., 1989). This robust contractile/elastic behavior was not observed in chick forebrain neurons. Rather, as shown in Fig. 8, slackening of forebrain neurites after peeling the neurite from the dish was followed by complex behavior with weak recovery of tension accompanied by neurite elongation at the recovered tension. First, the sinusoids clearly straightened out but the tension recovery was substantially less than expected for a Hookean spring with no evidence for active contraction (tension overshoots compared to initial value). In the experiment shown in Fig. 8, for example, a neurite of 250 μm at an initial tension of 120 μdynes (following detachment and a 30 minute period to reach mechanical equilibrium) was subjected to four small reductions in length (‘back ups’) over the course of about an hour. The net reduction in length totalled approx. 50 μm or 20% of the initial length, but tension was reduced to approx. half the initial value. Upon tension recovery, the neurite began to lengthen. For example, between the second and third slackening shown in Fig. 8 (15-30 minutes) the needle was stabilized at the peak of the spontaneous tension recovery (lower trace), causing the neurite to lengthen (upper trace). Following the third slackening (30-50 minutes), tension again initially recovers but rapidly declines without intervention and the neurite again lengthens. In general, it was not possible to stabilize neurite length for even several minutes at the smallest positive tension we can confidently measure with our needles (approx. 10 μdynes). This is an easy experimental manipulation for both chick sensory neurons and PC12 cells.

The principal result of our investigation is that brain neurons of chick embryos respond to mechanical tension with growth responses qualititively similar to that previously reported for peripheral neurons from the same source. Microtubule-rich cytoplasmic processes can be initiated de novo by applying tension to the margin of the cell body. Like peripheral neurons, brain neurons also show a simple, robust, Newtonian-fluid-like relationship between neurite elongation rate and tension magnitudes. Thus, we suggest that increases in axonal, and possibly, dendritic length that accompany the growth of the brain following synapse formation (Van Essen, 1997) occur by the same tensile ‘towed growth’ mechanism long postulated to allow peripheral axons to accommodate skeletal (overall) growth of animals (Weiss, 1941).

Indeed, our data suggest that the response to tension by chick forebrain neurons is more purely growth-related than with peripheral neurons. These brain neurons initiate and elongate neurites at substantially lower net tensions than are required for chick sensory neurons, routinely growing in response to tensions <100 μdynes. This sensitivity to low tensions is not the result of greater intrinsic growth sensitivity to tension; both sensory and forebrain neurons increase their growth rate at roughly 1 μm/hour for each additional μdyne of tension (Fig. 7). Rather, chick forebrain neurons in culture show substantially less elastic behavior at low tensions than do chick sensory neurons. In sharp contrast to chick sensory neurons or PC12 neurites, chick forebrain neurons show no clear evidence of a minimum tension required for growth; forebrain neurons elongated with the smallest pulling force we could confidently measure with our needles. In addition, forebrain neurons showed substantially weaker recovery of tension following slackening than chick sensory neurons. Unlike sensory neurons, chick forebrain neurons showed no capacity to actively generate tension in the neurite shaft following slackening. Although chick forebrain neurons showed clear evidence of short-term elastic recovery following slackening, these neurites were unable to maintain a given force over longer periods (20-30 minutes) without lengthening and reduction of tension. That is, forebrain neurites generally behave as viscoelastic fluids; they show some immediate recovery of tension in response to slackening (elastic behavior) but behave as a fluid over longer time scales. In contrast, neurites of peripheral neurons behave principally as viscoelastic solids at low tensions, showing fluid behavior only in providing a damped approach to an otherwise spring-like length-tension equilibrium. These differences in elastic behaviors between forebrain and sensory neurons is unlikely to be due to microtubule or neurofilament differences because both cell types show similar cytoskeletal arrays in axon-like processes. Further, the elastic properties of neurons are due principally to the actin cytoskeleton of the neurite shaft (Dennerll et al., 1988), which is poorly understood in terms of density, arrangement and functional differences. The mechanical behavior of forebrain neurites would appear to be unusual in that many decades of work on cytomechanics has shown that a wide variety of cell types respond to small forces or deformations as viscoelastic solids (e.g. Chambers and Fell, 1931; Hiramoto, 1963; Harris et al., 1981; Pasternak and Elson, 1985; Elgsaeter et al., 1986; Albrecht-Buehler, 1987; Hudspeth, 1989; Harold, 1990; Ingber and Folkman, 1988). Thus, cultured chick forebrain neurons show an exaggerated fluidlike growth response to tension, even when compared to sensory neurons from the same source.

The attractive proposal of Van Essen (1997) for brain morphogenesis is based primarily on the assumption that neuronal processes are under sustained tension. This entails that the tissue show macroscopic elasticity for extended periods. Van Essen also invokes other elastic properties, e.g. the mechanical compliance of different brain regions should depend upon local anisotropies in cellular architecture. Although our results do not provide direct support for Van Essen’s hypothesis, neither do they constitute strong negative evidence against the hypothesis. First, we do not wish to minimize our clear observations of transient elastic behavior in forebrain neurites, measured on a time scale of minutes. The failure to demonstrate sustained tension over a longer time scale might reflect insensitivity of our needle methodology, which applies minimum forces in the 10 μdyne range and might be too crude for these sensitive neurons. That is, forebrain neurons may have a substantially lower growth threshold than sensory neurons, which above their threshold also show little evidence of long-term elastic behavior. A difference in growth threshold seems plausible given the substantially greater ‘delicacy’ of forebrain tissue compared to dorsal root ganglion tissue, the source of chick sensory neurons. It also seems plausible that under normal in vivo conditions, chick cerebral neurons might have higher growth thresholds and generate substantial tensile forces that do not occur in the particular in vitro conditions reported here. In this regard, we note that the intact embryonic cerebral hemispheres show clear evidence of macroscopic elasticity, which would be unlikely if none of their constituent neurons were under tension. Additional experiments using different sources of central neurons and possibly lower tensions will be needed for a more incisive test of the role of tension in brain morphogenesis.

This work was supported by a grant from the NSF (IBN 96--03460). We thank David Van Essen for stimulating conversations concerning the role of tension in brain morphogenesis, for sharing his work prior to publication, and for reading and commenting on the manuscript.