Responses of growth cones to changes in osmolality of the surrounding medium

Most walled cells generate a hydrostatic pressure, or turgor, greater than that of the atmosphere. Pressures within plant and microbial cells range to about 1.0 MPa (9.87 atmospheres; Cram, 1976; Zimmermann and Steudle, 1978; Money, 1990). This pressure results from the influx of water, caused by the difference in solute concentration between the cytoplasm and the external milieu; the cell wall is subjected to tension as the plasma membrane is pushed against its inner surface. The pressurized fluid within the cell constitutes a hydroskeleton and is thought to provide the driving force for cellular expansion (Harold, 1990).

For vertebrate cells, which are bathed in isotonic fluids and lack a rigid exoskeleton, the importance of osmotically generated pressure is less evident. Shapes are maintained in large part by the matrix of protein filaments forming the cytoskeleton, and the driving force for cell extensions such as cilia, microspikes and lamellipodia is usually attributed to the assembly of microtubules or actin filaments. However, it is conceivable that the development of tension in a membrane-associated cortex could maintain a small positive internal pressure and that local weakening of the cortex, like the loosening of the cell wall, would allow expansion of the plasma membrane (Harris, 1973).

Indeed, positive hydrostatic pressures within animal cells have been measured using a variety of different techniques including micromanometry (Kao and Chambers, 1954), cell compression (Yoneda, 1986) and membrane deformation with a suction pipette (Rand and Burton, 1964), and by comparison of the volumes of flaccid and turgid cells (Bereiter-Hahn and Strohmeier, 1987). Estimates range widely from 10⁻⁶ to 0.2 MPa (Bereiter-Hahn and Strohmeier, 1987). Furthermore, Oster and Perelson (1987) have suggested that the growth of microspikes and lamellipodia could be driven by local osmotic effects, such as a local increase in the number of osmotically active particles associated with actin polymerization or the expansion of a gelled network of actin filaments due to Ca²⁺-induced solation.

We addressed these questions by examining the responses of nerve cells growing in tissue culture to changes in the osmolality of the surrounding medium. Clearly, the swelling or shrinking of a cell under such conditions does not constitute evidence that expansion is usually driven by water movement. However, we can ask whether any changes that occur resemble in time course and morphology the events of normal axonal growth, and whether the applied osmotic changes are of a magnitude likely to occur under physiological conditions.

The standard tissue culture medium used in these experiments comprised equal volumes of C⁺ medium and DPBS. These have the following compositions. C⁺ medium (Bray, 1991) consists of L15 medium (Sigma) supplemented with 10% fetal bovine serum, 0.6% glucose, 2mM glutamine, 0.lmgml⁻¹ streptomycin, 100i.u.ml⁻¹ penicillin, 0.25μgml⁻¹ amphotericin, 0.5% hydroxypropyl methylcellulose (Methocel E4M Premium, Dow Chemical Company, Midland, Michigan) and 1μ1 ml⁻¹ of a preparation of mouse submaxillary gland nerve growth factor diluted to a level that produced a healthy outgrowth of axons after 24 h. DPBS contains CaCl₂-2H₂O (0.133g1⁻¹), MgCl₂-6H₂O (0.1gl⁻¹), KC1 (0.2gl⁻¹), KH₂PO₄ (0.2gl⁻¹), NaCl (8.0gl^–1), Na₂HPO₄ (1.15gl⁻¹).

All media used in the osmotic shift experiments were made by mixing equal volumes of C⁺ medium with DPBS modified by the addition of osmolytes or dilution with water and supplemented with nerve growth factor and antibiotics at the same level as in C⁺ medium. In this way the final concentrations of nutrients, serum factors and growth factors remained unchanged during the osmotic shift. Increased osmolality was obtained by the addition of sucrose, mannitol or polyethylene glycol (PEG-400) each made up at 0.4 M final concentration in DPBS and sterilized by filtration through a 0.2 urn filter and mixed with DPBS in suitable proportions. The osmolalities of media samples were determined by vapor pressure deficit osmometry as described below.

Media used in long-term growth studies had a similar composition to that just described except for one series of experiments in which the osmolality was reduced to a third of that of normal media. In this case media were made by mixing one part of C⁺ medium with two parts of DPBS diluted to the appropriate amount by addition of water, supplemented with nerve growth factor and antibiotics.

The methods of tissue culture used in these studies resemble those used previously (Bray and Chapman, 1985). Briefly, sensory or sympathetic ganglia were dissected aseptically from the lumbar region of chick embryos at the 9th or 10th day of incubation. Ganglia to be used for the observation of growth cone behavior were then placed onto a 22 mm × 22 mm glass coverslip positioned over a 10 mm diameter circular hole drilled in the bottom of a 35 mm plastic tissue culture dish and held in place with petroleum jelly and tape (Bray, 1991). Prior to use, the dish plus coverslip assembly was sterilized by u.v. irradiation and the coverslip was then treated for 20 min with 0.05 mg ml⁻¹ poly-L-lysine (Sigma, M_r40000) dissolved in 20 mM Na₂B₂O₇, pH 8.5, and rinsed repeatedly with water before use.

Cultures m 4 ml of medium were maintained in a humidified incubator at 37 °C for periods of 12–24 h. They were then placed onto the stage of a Nikon Diaphot microscope and maintained at 37 °C by means of a 1500 W ceramic element room heater regulated by an electronic temperature regulator (Dyna Sense, Scientific Instruments, Skokie, IL). Most observations were made using a ×40 phase-contrast objective and an MTI videocamera (70 series, Newvicon tube). Changes in growth cone morphology were observed using an additional intermediate zoom lens with a magnification of ×2.25.

Images were collected using an ICOS image analysis system (Quantex Corporation, Sunnyvale CA) and stored on 44 Mbyte disk cartridges (SyQuest, Fremont, CA). Images were analyzed either directly or following application of a spatial filter to enhance contrast. Lengths of filopodia and growth increments of axonal tips were measured using the ICOS length measurement program calibrated with reference to an image of a stage micrometer at the appropriate magnification. Lamellipodial areas, in numbers of pixels, were measured using the ICOS ‘arbarea’ program and converted to μm² with reference to a micrometer slide.

Shifts in osmolality were achieved by removing 2 ml of the total of 4 ml of medium in the culture dish and replacing this with 2 ml of prewarmed medium of the appropriate osmolality. Following observation of changes, culture medium was removed and stored frozen for subsequent determination of osmolality.

Axonal growth rates were measured from video images or photographic prints of cultures through a ×40 objective, captured at 20 min intervals. The positions of all clearly isolated growth cones were marked (omitting those at the tips of large axon bundles, which experience shows often travel as a cohort at atypically rapid rates). Growth cone positions from one image were transferred to the next, either electronically or by use of transparent overlays, and the increments of axonal length in the intervening 20 min were measured.

The long-term survival and growth of ganglia in media of different osmolality were assessed on cultures growing in 24-well tissue culture dishes (Corning Glassworks, Corning, NY), pretreated with polylysine as described above. Samples of media prepared with a range of osmolalities were allocated in 2 ml samples to the wells in a randomized sequence, each osmolality being present in at least 4 wells. After incubation for 24 h, the ganglia were assessed for neurite outgrowth using a low-power (×10) objective and Hoffman modulation contrast optics. They were scored on an arbitrary scale from 0 (no outgrowth) to 6 (dense outgrowth of long axons) by an observer (D.B.) with no knowledge of their osmolality. Subsequently the osmolality of culture fluid from each well was measured by a second observer (N.P.M.) with no knowledge of the growth scores. The results were then collated and analysed statistically.

The osmolality of culture solutions and tissue extracts were measured with a 5100C Vapor Pressure Deficit Osmometer (Wescor, Inc., Logan, UT). Samples (10μd) were pipetted onto 6 mm diameter discs placed previously into the sample chamber of the osmometer. The osmometer was frequently recalibrated over the range 100 to 1000 mmol kg⁻¹ (or mosM) using NaCl standards (Wescor, Inc.) during the experiments. In some cases, osmolality measurements were converted to osmotic pressure (Π; measured in MPa) using the formula: Π=RTc, where c=osmolality in mol kg⁻¹ and RT=2.446kgMPamol⁻¹ at 21°C.

Whole brains and dorsal root ganglia were dissected from chick embryos and washed in Hanks’ balanced salt solution (Sigma Chem. Co., St Louis, MO). The tissue samples were blotted dry on filter paper, frozen at -20 °C and later thawed by transfer to room temperature. This slow freeze-thaw was repeated three times. Finally, the samples were vortexed to produce a homogeneous viscous paste, whose osmolality was determined by osmometry.

The instrument is based upon the original design of Zimmermann et al. (1969) and is described in detail by Money (1990). The pressure probe consists of a micropipette (tip diameter ∼1μm) filled with a low-viscosity silicone oil (200 fluid, 2 cs viscosity, Dow Coming Corp., Midland, MI); this fits into a plexiglass block traversed by a T-shaped bore, which is also filled with oil. Two arms of the bore communicate, respectively, with a pressure transducer (model XT-190-300G; Kulite Semiconductor Products, Inc., Leonia, NJ) and a control rod positioned with a micrometer screw. The assembled apparatus is mounted on a micromanipulator (Brinkmann Instruments, Inc., Westbury, NY).

When the oil pressure within the probe is set lower than that of the cell (with the control rod), cytoplasm is forced into the tip of the pipette and when the probe pressure exceeds that of the cell, oil is ipjected into it. Once the micropipette is inserted into the cell body, the interface between the oil and cytoplasm is adjusted to the pipette tip with the control rod. The pressure within the probe then matches that of the intact cell; a small increase in probe pressure results in the injection of oil into the cell.

The most direct way to assess the osmolality of cytoplasmic constituents is to transfer broken cell preparations to a vapor pressure osmometer, a method applied previously to plant and microbial cells (Markhart and Linn, 1985; Woods and Duniway, 1986). Samples of brain and dorsal root ganglia dissected from 12-day chick embryos were compacted by gentle centrifugation and lysed by repeated freeze-thawing. Analysis by vapor pressure deficit osmometry then gave values of 272.2± 1.7 mosM (22 samples) for brain and 271.7±0.3mosM (3 samples) for dorsal root ganglia. The combined value from these two tissues, which is 272.2±1.4mosM, may be compared with the osmolality of 279.4±4.8mosM (26 samples) for samples of nominally isotonic tissue culture media taken from the experiments described below.

Single classification analysis of variance of these two populations demonstrated that there was no significant difference between the osmolalities of cell extracts and culture medium (F[l,49]=1.99, P>0.1). Statistical analysis also showed that the least significant difference between the two populations is 10.2 mosM. In other words, we have a 95 % degree of confidence that if the cytoplasmic osmolality does differ from that of the culture medium, then this difference will be less than 10 mosM.

The micropipette-based pressure probe, developed by Zimmermann et al. (1969), has recently been used to measure intracellular hydrostatic pressure in microorganisms (Money, 1990). The instrument offers some novel opportunities for the study of the water relations of animal cells. As described in Materials and methods, the tip of a pipette filled with silicone oil is introduced into the cell and the movement of cytoplasm into the pipette, or of oil into the cytoplasm, is balanced by a micrometer screw. Impalement of the cell bodies of dorsal root ganglion cells in culture failed to show a significant positive hydrostatic pressure. Droplets of silicone oil were injected into the cells by raising the probe pressure to <0.01 MPa. Therefore, these neurons can have a positive pressure no larger than 0.01 MPa, corresponding to a difference of osmolality across the plasma membrane of no greater than 4mosM.

The morphology of control cultures of sympathetic and sensory neurons growing under the conditions employed in this study is very similar to that described previously (Bray and Chapman, 1985). Growth cones possess highly irregular contours composed of filopodia and lamellipodia. These vary from minute to minute, although individual growth cones usually have a characteristic size and display of lamellipodia and filopodia, which are preserved for long periods provided no major changes occur in external conditions.

We observed that an increase in osmolality of culture medium through addition of sucrose, mannitol or polyethylene glycol caused a distinct and reproducible set of morphological changes in the growth cone. The first evidence of a response, seen 2–3 min after changing the medium, was an increase in length of filopodia (Figs 1 and 2). This accelerated elongation continued for an additional 3–5 min during which filopodia increased not only in length but also in number (Fig. 3, Table 1). Elongation rates up to 10 μm min⁻¹ were seen during this period. Responses were more pronounced and more rapid at the highest concentration of osmolytes employed (an excess of 100mosM over the normal culture medium). However, the increase in filopodial length was still detectable following an increase of as little as 20mosM. Twenty minutes following the change in medium, filopodia shrank and the growth cone adopted a stunted appearance, usually associated with a cessation of growth (see below). At lower concentrations the growth cone re-established its normal appearance and resumed growth.

At higher concentrations of osmolytes (50–100 mosM above basal osmolality) a pronounced reduction in the area of lamellipodia was observed (Fig. 2; Table 1). For most growth cones, a reduction in lamellipodial area also generates an increase in filopodial length, since it moves the ‘baseline’ from which measurements are made. It should be emphasized, however, that the response of filopodia described above was principally due to the increased elongation of filopodial tips over the culture substratum and not the retraction of lamellipodia.

A decrease in culture fluid osmolality, effected by the addition of diluted basal medium to the culture, had far less effect upon growth cone morphology. Hypo-osmotic responses were not the converse of hyperosmotic responses. Only the most extreme hypo-osmotic shock, that of a change to 50 % basal medium osmolality by addition of an equal volume of water to the culture fluid, caused a transient swelling of lamellipodia. This change was transient and reversed within a few minutes. No systematic increase or decrease in the length or number of filopodia was observed following a shift to hypotonic conditions (Fig. 4, Table 1).

It is difficult to measure the rate of axonal growth over periods of less than 15 min. At an average rate of progression of 40μmh⁻¹, a growth cone will advance hardly more than its own length; its progress will be largely obscured by the periodic fluctuations in form. Furthermore, there is substantial variation in growth rate between different cultures and from one growth cone to another in the same culture. The procedure we adopted (Materials and methods) was to allow cultures to equilibrate for 1 h on the microscope and then to take images at successive intervals of 20 min. Frames were analyzed by recording the advance (or retreat) of each growth cone in the field for each of the 20 min intervals. Mean values of the distance travelled then gave an index of growth.

Increases in osmolality caused a partial or complete arrest of growth (Table 2). At the highest concentrations of osmolytes employed (lOOmosM in excess over culture medium) growth was arrested within the first 20 min of observation. Growth remained arrested for a further 60 min and perhaps longer. Lower concentrations gave more variable results, which also appeared to depend on the osmolyte employed. For example, polyethylene glycol at 50 mM caused a more profound inhibition of growth than 50 mM sucrose (both solutes deviate from ideality, generating osmotic pressures of 0.09 MPa, equivalent to an osmolality of 37 mosM; Money, 1989).

Reductions in osmotic strength through addition of water to the culture medium stimulated neurite outgrowth. Statistically significant changes were measured with reductions to 50% and 80% of the basal medium osmolality (Table 2). The most dramatic effect was seen following a shift to 50% basal medium (final osmolality 136mosM). As early as 20 min following the hypotonic shift, individual neurites were observed to be growing at elevated rates. Individual axons had growth rates of 100 – 120 umh^{– 1} at this stage and appeared to be thinner than others in the culture (Fig. 5). It was our impression that these rapidly growing axons emerged from existing neurites or growth cones as branches, but studies with isolated neurons would be needed to confirm this.

The hypotonic growth response was observed in 12 separate experiments. In general the increased growth rate reached a maximum 40 – 60 min following the shift, as illustrated in Fig. 6. No cultures were monitored continuously for periods greater than 2h. All experiments were performed in such a way that the osmotic shift was achieved solely by changes in ions, with amino acids, vitamins, serum, NGF and Methocel remaining unchanged in concentration. In two experiments, the concentration of calcium was also maintained constant before and after osmotic shift with no effect on the phenomenon.

A series of experiments was performed to examine the long-term growth of neurons in cultures of different osmolality. Sympathetic ganglia from 9-day chick embryos were plated in media of different osmotic strength and assayed for the production of a halo of axons 24 h later. Ganglia were scored on an empirical scale (Materials and methods) that reflected both the length and the density of neurites. Considerable scatter was observed but the general trend was clear (Fig. 7). Apparently normal growth was found in media with osmolalities as low as 100mosM: less than half the value of normal ‘isotonic’ media. Cultures were less able to sustain increases in osmolality and some inhibition of growth was seen at osmolalities of 350mosM using PEG-400 and sucrose as osmolytes.

The apparent health of ganglia in low ionic strength culture media led us to examine cultures after longerperiods of time. Even after 5 days - the longest period examined - ganglia growing in 50% basal medium possessed long dense axonal outgrowths. Long straight axons were seen in such cultures, interspersed with a reticulum of fine, apparently single, axons (Fig. 8). Even after 5 days of culture in 50% basal medium, many ganglia had active growth cones at their periphery (Fig. 9).

Cultures in low ionic strength media seemed to be remarkably free of the non-neuronal cells present in large numbers in control outgrowths. Presumably, the low osmolality inhibits either the replication or the migration, or both, of non-neuronal cells from the ganglia.

Our estimates of cytoplasmic osmolality obtained from broken cell preparations were not significantly different, within the limits of resolution of the osmometer, from those for the media in which the neurons were grown. Assuming that cellular osmolality does not change when ganglia are explanted into tissue culture, then these estimates place an upper limit on the possible difference in osmolality across the plasma membrane. This value is about 10 mosM, corresponding to a hydrostatic pressure of 0.025 MPa (or 0.25 atmospheres). Consistent with this estimate, our measurements using the pressure probe indicate that, if an internal hydrostatic pressure does exist, it must be less than about 0.01 MPa. It may be noted also that experiments in which the processes of cultured neurons are severed fail to show the rapid expulsion of cytoplasmic constituents seen in organisms with high internal turgor pressures (Shaw and Bray, 1977).

It follows that if the formation of filopodia and lamellipodia from the growth cone is indeed driven by an osmotically generated pressure then the growth of these processes should be responsive to changes of osmotic pressure in the 0.01 – 0.03 MPa range. A decrease of 10mosM in the surrounding medium should, at least transiently, double the effective driving force for growth. An increase of 10 mosM should abolish it. Therefore, our finding that the formation of these two forms of actin-rich cell surface extension were not inhibited by an increase or decrease in the osmolality of the surrounding medium by amounts in excess of 20mosM is prima facie evidence against an osmotic driving mechanism. By comparison, the elongation of hyphae of the aquatic fungus Achlya bisexualis, which support pressures of 0.8–1.2 MPa, is arrested by increases in culture fluid osmolality of 100mosM (only ∼ 25% of the difference in osmolality between the cytoplasmic and culture medium; Money, unpublished data).

The case is especially strong for filopodia, which were observed in this study to increase in length following an increase in osmolality. This response, which was seen following hyperosmotic changes as small as 20mosM, is the opposite to that expected if filopodial growth were driven by turgor. Thus, if the elongation of filopodia were dependent on an osmotic-based expansion of the plasma membrane at the filopodial tip - as posited for lamellipodia and microspikes by Oster and Perelson (1987) - then it should be decreased by a hypertonic shift, not increased as we observed. We do not know why filopodia respond in this way, but suggest that it could shed light on the normal regulation of actin polymerization in the growth cone.

It is interesting that the behavior of filopodia described above is directly opposite to that of the acrosomal process of Thy one sperm, which grows faster in dilute sea water and is retarded by the addition of osmolytes (Tflney and Inoué, 1985). Our results therefore throw some doubt on the utility of the acrosomal reaction of Thyone as a model for actin-based surface extensions of vertebrate cells.

The other category of extension seen at the growth cone - the lamellipodium - showed a slight reduction in area following a rise in osmolality of 50 mosM or more (Table 1). A similar effect was earlier seen at the leading edge of corneal epithelial cells following the addition of 0.5 M sorbitol to the culture medium (an increase of about 500mosM; DiPasquale, 1975). However, the reduction in area of growth cone lamellipodia is not mirrored by a corresponding increase in hypo-osmotic medium (Table 1), so that they do not show the reversible sensitivity to small osmotic shifte that would be expected if they were in fact driven by an osmotically driven internal pressure.

In contrast to the behavior of the actin-rich extensions of the growth cone, the elongation of the axon as a whole was consistent with an osmotically driven mechanism. Addition of osmotically active substances to the culture medium caused a decrease in axonal growth rates, while reduction of medium osmolality through addition of water increased the rates of growth above those seen in normal culture medium. The latter phenomenon was seen in its most striking form when an equal volume of water was added to the culture medium. This treatment induced a proliferation of fine axonal branches within 20 min, each branch being tipped with a growth cone and growing at rates that were fivefold greater than that of controls (Table 2).

The response to reduced osmolality appeared to reach a maximum 40-60 min after water addition (Fig. 6), but we did not follow it for a sufficiently long period to know if growth rates eventually return to the previous values. One day after plating in 50 % normal medium, sympathetic ganglia displayed apparently healthy growth with many long axons. Five days after plating in 50 % normal medium, sympathetic ganglia were still apparently healthy and had many long axons tipped with growth cones (Fig. 9).

It is possible that under normal conditions the elongation of the axonal cylinder may be derived in part from a small positive internal pressure. The elongation of eucaryotic flagella - another type of microtubule-rich cell extension - is also sensitive to changes in osmolality and increases to above normal rates following a shift to hypo-osmotic conditions (Telser, 1977; Solter and Gibor, 1978). However, experiments with longer-term cultures are complicated by many factors. It is possible, to begin with, that osmolytes such as sucrose, mannitol and polyethylene glycol might interfere with axonal growth for reasons other than changes in osmotic pressure. Second, it seems likely that neurons will, like other cells, have the ability to adapt to an imposed change in osmolality, changing their transport of ions and small molecules so as to restore the original difference in osmolality across the plasma membrane (and hence any associated turgor pressure). Finally, the fact that in our experiments a decrease in osmolality was generated by addition of water means that there was also a concomitant reduction in ionic strength, which therefore could have contributed to or even caused the observed stimulation of growth.

This work was supported in part by a sabbatical leave from the Medical Research Council, UK (D.B.), NTH grants NS28338, GM36126 and RR04131, and NSF grant INT8814146 (J.R.B.), and NSF grant DCB 86-18694 (F.M.H.). The authors gratefully acknowledge the assistance of Barbara Bernstein in the use of the image analysis programs and thank Mike Mosier for statistical analyses.