Lysophosphatidic Acid Stimulates Neurotransmitter-Like Conductance Changes that Precede GABA and l-Glutamate in Early, Presumptive Cortical Neuroblasts

MATERIALS AND METHODS

Primary cultures. All animal protocols were approved by the University of California, San Diego Animal Subjects Committee and are in accordance with National Institutes of Health guidelines and public law. Timed-pregnant Balb/c mice [day of plug, embryonic day 0 (E0)] were euthanized by cervical dislocation, and embryonic cortical regions were dissected as described previously (Chun and Jaenisch, 1996) in cold serum-free OptiMEM-I (Life Technologies, Gaithersburg, MD) supplemented with 20 mmd-glucose, 1% penicillin–streptomycin, either with or without 55 μm β-mercaptoethanol. Cortices were triturated using a sterile long-nose Pasteur pipette into small clusters (<50 cells) and plated at a density of ∼10,000–50,000 cells/12 mm coverslip previously coated with Cell-Tak (Collaborative Research, Bedford, MA). Cultures were grown at 37°C and 5% CO₂. Cells in clusters appeared healthy and remained proliferative (87% incorporated BrdU at over the first 24 hr within clusters), consistent with previous reports of rodent cortical cultures (Ghosh and Greenberg, 1995). The basic morphology of cells in clusters did not change over the duration of the experiment, whereas single cells more often appeared to differentiate into neurons with large sodium currents, although some single cells appeared to remain undifferentiated. The most responsive cells appeared to be those maintained in low-density cultures that resided within the clusters, and electrophysiology was performed on cells within small clusters.

Electrophysiology. The whole-cell patch-clamp technique (Hamill et al., 1981) was used to record ligand-induced currents from embryonic VZ cells maintained for 1–20 hr in serum-free media. Cells were visualized using a Nikon Diaphot 300 with DIC Nomarski optics. Cells were continuously perfused in a physiological saline (∼0.5 ml/min) unless otherwise indicated. The standard physiological saline (Tyrode’s solution) contained (in mm): 130 NaCl, 4 KCl, 5 CaCl₂, 1.2 MgCl₂, and 10 hemi-Na-HEPES (pH 7.37, 295–300 mOsm as measured using a Wescor 5500 vapor-pressure osmometer; Wescor, Logan, UT). In some experiments, the Ca²⁺ concentration of the physiological saline was decreased to 1 mm, and similar results were obtained. Recording electrodes were fabricated from borosilicate capillary tubing (BF100; Sutter Instrument Co., Novato, CA), the tips were coated with dental periphery wax (Miles Laboratories, South Bend, IN), and had resistances of 8–15 MΩ when containing intracellular saline [in mm: 100 K-gluconate, 25 KCl, 0.483 CaCl₂, 3 MgCl₂, 10 hemi-Na-HEPES, and 1 K₄-BAPTA (100 nm free Ca²⁺), pH 7.4, with dextrose added to achieve 290 mOsm]. Series resistances (R_s) were 20–40 MΩ. To mark cells that had been physiologically recorded, lysine-fixable rhodamine dextran (10 kDa; Molecular Probes, Eugene, OR) was included at 500 μm in the pipette solution. To determine the TTX sensitivity of voltage-activated inward currents, a Cs⁺-based intracellular saline was used to block outward K⁺ currents (in mm): 25 CsCl, 102.5 methansulfonate, 0.483 CaCl₂, 3 MgCl₂, 10 hemi-Na HEPES, and 1 BAPTA free acid, pH 7.4 with CsOH. In Cl⁻ ion substitution experiments, the cells were continuously perfused with low Cl⁻extracellular saline composed of (in mm): 140 Na-isethionate, 3 K-gluconate, 5 Ca²⁺-acetate, 1 MgCl₂, and 10 hemi-Na-HEPES (305 mOsm, pH 7.37, with acetic acid). In experiments designed to shift the reversal potential (V_rev) of nonselective cation (NSC) currents, an intracellular saline containing impermeant cations (ZIS) was used (Delay et al., 1997) [in mm: 100 NMDG, 100 methanesulfonate, 5 NaCl, 1 BAPTA free acid, 0.483 CaCl₂ (100 nm free Ca²⁺), 2 MgCl₂, 10 hemi-Na-HEPES; pH 7.4, with dextrose added to achieve 290 mOsm]. Liquid junction potentials were −18 mV using standard pipette and bath solutions as determined both empirically and using the computer program JPCalc (Barry, 1994); in the low Cl⁻ extracellular saline, voltages were corrected by +3 mV, and in experiments using ZIS and Tyrode’s, voltages were corrected by −15 mV. All voltages shown are corrected for liquid junction potential. Current and voltage signals were detected and filtered at 5 kHz with an Axopatch 200A patch-clamp amplifier (Axon Instruments, Foster City, CA), digitally recorded with a TL-1 DMA laboratory interface (Axon Instruments) and personal computer-compatible system, and stored on magnetic disk for off-line analysis. Data acquisition and analysis were performed with pClamp software. Slow changes in holding current were detected and filtered at 5 kHz and recorded with a LPF200A DC amplifier (Warner, Hamden, CT) and VR-10B digital data recorder (Instrutech, Great Neck, NY) onto video tape. The signal was later analyzed at 10 Hz using Axotape software.

The total membrane capacitance (C_m) was determined as the difference between the maximum current after a 30 mV hyperpolarizing voltage ramp from −68 mV generated at a rate of 10 mV/msec and the steady-state current at the final potential (−98 mV) (values refer to V_m when using standard saline solutions). The principles underlying this method derive from the equation describing the current I_C flowing onto (and off of) the membrane, I_C =C_m [dV_m/dt] and are discussed in Armstrong and Gilly (1992). I_Cwas measured as the difference between the current at the end of the voltage ramp and the steady-state current achieved after completion of the ramp. The voltage at the end of the ramp protocol was maintained for 47 msec to ensure that the decay of the maximum current had reached steady-state. Because dV/dt is 10 V/sec in the experiments here,C_m = 0.1[I_C]. TheC_m calculated using this method was similar to the C_m determined using the classical method of calculating the time constant of the current response elicited by a 10 mV voltage step (data not shown).

Cell capacitance measurements provided an assessment of cell coupling (de Roos et al., 1996). The mean single cell C_mdepended on the cell morphology. Single neuroblast-like cells hadC_m values of ∼2 pF similar to the lowest values obtained for cells in clusters (23 of 107 hadC_m values of <3 pF). To assess the effects of cell coupling, 1-Octanol (OCT, 1 mm) was used; however, it was found to decrease the C_m from 10.3 to 4.1 pF in only one of 10 examined clusters. The C_m of these OCT-insensitive cells was 1.7, 3.3, 4.5, 4.9, 4.9, 5.7, 7.8, 8.7, and 14.4 pF. Based on the C_m values for a single cell, only one cell was not electrically coupled (1.7 pF), whereas all of the other observed cells in clusters were coupled; this offered the best biological approximation of the normal state of neuroblasts in the VZ, based on previous analyses demonstrating that VZ cells are coupled in groups of 15–90 cells (LoTurco and Kriegstein, 1991; Bittman et al., 1997).

Current densities were calculated as the peakI_in or maximal I_out at a voltage step to +32 mV normalized to the cell capacitance. Voltage-activated currents were obtained by voltage steps between −88 and +112 mV from a holding potential of −98 mV. Membrane resistance (R_m) values were determined by voltage-ramp (the inverse of the slope of the ramp-induced current at negative potentials) or current-clamp protocols using small current injections and were similar (n = 9). EstimatedR_s errors were no more than 1.5 mV, on average <1 mV.

Apparent reversal potentials (V_rev) of ligand-induced conductance changes were determined using a voltage-ramp protocol (Dubin and Dionne, 1994). Voltage ramps were applied every 6 sec, and the resulting whole-cell ramp-induced currents were recorded. Usually the voltage was ramped from positive to negative values to avoid the nonlinearity introduced by the voltage-activatedI_in when present at high density; however, in cells with large outward currents, the protocol was also reversed to enable the detection of small LPA-induced currents. The current required to clamp the cells at −68 mV was continuously monitored. Ligand-induced conductances were determined from whole-cell currents elicited by a voltage-ramp protocol in the presence and absence of ligand. Control ramp currents were subtracted from those obtained in the presence of ligand to produce difference currents, revealing the component of the whole-cell current–voltage relationship contributed by ligand-induced currents. The voltage at which there was no net ligand-induced current was determined (V_rev). A criterion used to avoid spurious V_rev determinations was that at least three individual ramps during the response revealed a similarV_rev. This was possible because the responses were quite long-lasting. The figures show the difference ramp comparing the initial ramp after onset of the response and the average of control ramps. The concentrations of the major ions in the intracellular and extracellular salines were set such that each ion had a unique equilibrium potential. The equilibrium potentials for Na⁺, K⁺, and Cl⁻ under the normal recording conditions were +85, −85, and −40 mV, respectively. The recorded current trace was filtered off-line (at 1 Hz), and in some traces only part of the positive voltage-ramp deflection is shown (see Figs.4a,c,e, 5a,6a–c).

The LPA-induced conductance was calculated from voltage-ramp data. The conductance was calculated from the linear portion of ramp-induced current–voltage curve at negative potentials (I = G * V, where G is the slope). The control conductance (usually the average of two or three traces before the LPA-induced response) was subtracted from the largest conductance measured in the presence of LPA. The LPA-induced conductance determined by this method was underestimated because the peak response often was not captured during the ramp protocol.

Most values are presented as the arithmetic mean ± SEM. BecauseI_Na values had a skewed distribution, geometric means were determined, and the reported values are the arithmetic means with the upper and lower error limits in parentheses (Hancock et al., 1988). For the latter calculation, a value of 0 pA/pF was given to cells with no detectable inward I_Na and averaged with the natural log of the other values.

Ligand reagents. LPA was thoroughly dissolved in Milli-Q water and used fresh or stored frozen as a 5 mm stock solution and used within 5 d. It was diluted to the indicated concentrations in extracellular saline and applied at concentrations ranging from 10 nm to 1 μm from nearby puffer pipettes. As a control for possible nonspecific lipid effects, a structurally related compound, lysophosphatidyl glycerol (LPG), was applied at concentrations between 10 nm and 10 μm. Stocks of GABA (100 mm) andl-glutamate (300 mm) were stored frozen and thawed only once and diluted in extracellular saline just before use. Ligand was applied (3–10 psi) together with 0.05% fast green dye to mark the plume using a Picospritzer (General Valve, Fairfield, NJ) for durations indicated in the figures (typically 15–25 sec).

BrdU and nestin immunofluorescence. To determine the proliferative capacity of primary cells, cultures were pulsed at the time of plating with bromodeoxyuridine (BrdU), and BrdU immunofluorescence was detected 24 hr later (cell proliferation kit; Boehringer Mannheim, Indianapolis, IN). For nestin immunofluorescence, coverslips containing cells that had been filled with lysine-fixable rhodamine-dextran during recording (see above) were fixed for 15 min with 4% paraformaldehyde in PBS. After PBS rinses, coverslips were incubated with blocking solution (2.5% BSA, 0.3% Triton X-100 in PBS) for 1 hr, followed by monoclonal anti-nestin antibody (PharMingen, La Jolla, CA) at 1:500 in blocking solution overnight. Coverslips were rinsed with PBS and incubated with FITC-conjugated anti-mouse IgG secondary antibody (Pierce, Rockford, IL) at 1:100 for 1 hr. After PBS rinses, all coverslips were counterstained with the fluorescent nuclear stain 4′6-diamidino-2-phenylindole (DAPI; Sigma, St. Louis, MO), air-dried, and mounted on glass microscope slides in GelMount (Biomeda, Foster City, CA). Coverslips were examined and photographed using a Zeiss Axioskop fluorescence microscope with rhodamine, fluorescein, or DAPI filters.

RESULTS

Cells comprising small clusters of 10–50 cells were examined for features consistent with a neuroblast identity in primary cultures derived from E11 and E12 dissociated murine cortices. Single cells that were part of clusters were studied because they offered the closest biological approximation of the actual electrical state of neuroblasts in the VZ, based on extensive previous studies demonstrating that VZ cells are normally coupled in groups of 15–90 cells (LoTurco and Kriegstein, 1991; Bittman et al., 1997). In addition, the resting membrane potentials of the examined cells in clusters were very similar to those reported previously, as were responses to GABA andl-glutamate (see below), demonstrating that this culture system supported physiological parameters of previous reports. Moreover, single cells from a cortical neuroblast cell line that were not electrically coupled showed similar results to cells in clusters. Changes in membrane conductance elicited after rapid, local exposure of these primary cells to LPA or both GABA and l-glutamate were assayed between 1 and 20 hr in vitro under serum-free conditions using the whole-cell variation of the patch-clamp technique. In addition to being tested for electrophysiological properties, recorded cells were labeled using fluorescent dextran, fixed, then double-labeled for nestin immunofluorescence. In some cases, BrdU pulses were also used to determine the percentage of cells undergoing DNA synthesis. Two distinct LPA-induced conductance changes were observed in putative neuroblasts: a Cl⁻ conductance increase and a NSC conductance increase. LPA responsiveness always preceded that of GABA or l-glutamate during development.

DISCUSSION

An expectation extending from LPA receptor gene expression in the cortical VZ (Hecht et al., 1996) was the existence of receptor-mediated actions of LPA influencing the physiology of cortical neuroblasts. In this study, LPA was found to stimulate two distinct conductances in separate populations of nestin-immunoreactive, putative neuroblasts present in cell clusters. These effects commenced at the earliest stages of cortical development and preceded both GABA andl-glutamate sensitivity that appeared at subsequent ages. A majority of cells examined were LPA-responsive, consistent with a prominent role for this lysophospholipid in cortical development.

Examined cells were identified as early cortical neuroblasts based on multiple criteria. At the ages of cell isolation, E11–E12, very few neurons have become postmitotic, and few glia appear to be generated (Caviness, 1982; Blaschke et al., 1996, 1998; Weiner and Chun, 1997). The examined cells could also be easily distinguished by their spherical shape and association in clusters, contrasting with rare, flat, fibroblast-like cells that were also present in the culture. The presence of extensive electrical coupling within the examined clusters is the normal, physiological state of cells in the VZ (LoTurco and Kriegstein, 1991), and the resting membrane potential of these cells was consistent with previous measurements of cortical neuroblasts within living tissue slices (LoTurco et al., 1995). At the older ages examined, 12–20 hr after isolation at E12 (E12–E13), prominent GABA responses were also observed in the spherical cells, further consistent with previous observations on cortical neuroblasts (LoTurco and Kriegstein, 1991; LoTurco et al., 1995). Cells responding to LPA were also nestin-immunoreactive which, combined with the aforementioned criteria, is again consistent with a neuroblast identity (Cataneo and McKay, 1990). In addition, the high percentage of cells responding to LPA (>60% at E11; see Table 2) was consistent with the extent oflp_A1/vzg-1 expression in the VZ observed by in situ hybridization at comparable ages, where most cells appeared to express this LPA receptor gene (Hecht et al., 1996). Moreover, the electrical properties and LPA-dependentG_Cl changes of the putative neuroblasts were similar to those of neuroblast cell lines derived from the embryonic cortex (Chun and Jaenisch, 1996) in which single (noncoupled) cells were examined, indicating that electrical coupling per se was not necessary for the observed physiological responses. Taken together, these data show that the examined primary cells possess developmental and phenotypic characteristics of cortical neuroblasts while responding to LPA stimulation.

LPA exposure induced responses that cannot be explained by nonspecific or known secondary effects. Nonspecific effects could hypothetically be produced by detergent properties of LPA through micelle formation, but such effects are not seen until the critical micelle concentration is reached (>1 mm) (Durieux, 1995), whereas all of our studies were performed in the nanomolar concentration range. In addition, a structurally similar ligand having potential detergent properties, LPG, did not elicit these responses (Fig. 4a). We examined the membrane for possible damage by LPA by monitoring the reversal potential after LPA exposure; these studies demonstrated an intact membrane (Fig. 5c,d). Moreover, if the conductance changes were a nonspecific perturbation, all cells should respond, and the response should be stereotyped. In fact, ∼ $\frac{1}{3}$of the cells did not respond at all (Table 2). Similarly, as seen in Figure 6c, application of LPA after stimulation of a cell by GABA/l-glutamate produced no conductance changes, again inconsistent with a lack of specificity. Combined, these data argue against nonspecific activation of conductance changes by LPA.

Another possibility is that GABA or l-glutamate could produce the conductance changes as a secondarily released factor. This explanation is inconsistent with the previous literature and our data. Based on studies by Kriegstein (LoTurco et al., 1995) in which GABA responses among embryonic VZ cells were extensively characterized, only GABA_A responses were observed, i.e., the fast, ligand-gated ion channel responses, whereas GABA_B responses were not observed in any of these VZ cells, and our data confirm this observation. Similarly, l-Glu responses were also mediated by ligand-gated ion channels. We observed both GABA_A and ionotropic l-Glu receptor responses, both of which have a relatively brief latency (Fig. 6). This brief latency fundamentally contrasts with the prolonged latency of LPA responses, 30–45 sec, that is nearly identical to LPA latencies observed in two different cell systems (Holtsberg et al., 1997; Thoreson et al., 1997). Migrating, postmitotic cortical cells have been reported to respond through GABA_B receptors (Behar et al., 1996), however GABA_B-type responses were not reported in VZ cells of previous studies (LoTurco et al., 1995) and in the present study, never seen even in the presence of clearly detectable GABA_Aresponses (demonstrating that GABA was active); these data are compatible with GABA_B receptors being expressed on at least some newly postmitotic neurons, but not being expressed on mitotic neuroblasts that are electrically coupled in the cell clusters that were recorded here. In addition, no GABA responses (fast or slow) were ever observed at E11. Still further, the increase in chloride conductance with a latency of 30 sec is not typical for a GABA_B receptor (which generally reduces calcium, and increases potassium conductances) (Rang et al., 1995). The direct experimental paradigm of exposing a cell to GABA/l-Glu followed by LPA, or vice-versa (Fig. 6) again illustrates that secondary phenomena cannot account for the LPA responses. Our combined experimental data along with the previous literature show that the responses we observe cannot be explained by secondary release of the previously identified factors, GABA or l-glutamate.

LPA activated two distinct conductances in separate populations of cells. Half of the responsive cells showed increased Cl⁻ conductance based on reversal potential and ion substitution experiments. LPA has been reported to increase Cl⁻ conductances in non-neuronal cells, including defolliculated Xenopus oocytes (Durieux et al., 1992;Fernhout et al., 1992), wounded corneal keratocytes (Watsky, 1995), and quiescent Rat-1 fibroblasts (Fischer et al., 1995). There are differences in observed responses that include homologous desensitization, apparent ligand affinity, pertussis toxin sensitivity, and Ca²⁺ sensitivity. These likely reflect differences in cell type, lysophospholipid (including LPA) receptors (Hecht et al., 1996; Fukushima et al., 1998; Chun et al., 1999; Zhang et al., 1999), and/or downstream mechanisms. The mechanisms underlying activation of the LPA-induced Cl⁻ conductance in cortical neuroblasts are under investigation. One possibility involves increases in intracellular Ca²⁺ and subsequent activation of Ca²⁺-activated Cl⁻channels. LPA increases intracellular Ca²⁺ in some neuroblasts in these primary cultures (Fluo-3 and Fura-2 imaging; A. Dubin, T. Bahnson, and J. Chun, unpublished observation). Alternatively, Cl⁻ currents may be activated by cytoskeletal remodeling. Alterations in the actin cytoskeleton have been shown to gate Cl⁻ channels in neocortical astrocytes (Lascola et al., 1998), embryonic human skeletal muscle myoballs (Häussler et al., 1994), and human vascular endothelial cells (Oike et al., 1994). Actin depolymerization opens CFTR Cl⁻ channels with a time course similar to that observed after LPA stimulation (Fischer et al., 1995). Rearrangements of the actin cytoskeleton could provide such a mechanism and are consistent with LPA-dependent cell-rounding observed in neuroblast cell lines (Jalink et al., 1994; Hecht et al., 1996; Fukushima et al., 1998;Chun et al., 1999).

An NSC conductance rather than a Cl⁻ conductance was activated by LPA in a second population of cells. Ca²⁺ may permeate the LPA-activated channels underlying this NSC response, and its contribution to the inward current may be considerable (Frings et al., 1995). NSC conductances that allow Ca²⁺ entry could participate in the extensive programmed cell death among cortical neuroblasts within the VZ (Blaschke et al., 1996, 1998; Staley et al., 1997) and could also lead to some of the myriad effects associated with Ca²⁺ signaling.

The conductance changes produced by LPA in neuroblasts are likely to be depolarizing. The Cl⁻ equilibrium potential has been shown to be approximately −30 mV in rat neuroblasts using gramicidin perforated-patch techniques (Owens et al., 1996), and NSC conductances have reversal potentials near 0 mV. Thus, LPA-dependent increases in NSC as well as Cl⁻ conductances would produce a depolarization. There were no detectable differences in resting physiological properties between cells responding to LPA with an increased Cl⁻ or NSC conductance. The resting membrane potentials, magnitudes of I_Na and outward current densities, C_m, andR_m were not significantly different. The biological significance of these distinct ionic groups awaits further investigation. Their existence, however, provides a potential link between LPA signaling and several outcomes known to occur within the VZ: cell proliferation, differentiation, and death. In addition to the relatively rapid effects, seconds to minutes, documented here, LPA also has longer term growth factor-like properties in other systems that are manifest over a period of hours to days (Moolenaar et al., 1992). These comparatively long-term effects complement the LPA-dependent responses observed. We note that this scenario is undoubtedly incomplete, and it will be crucial to determine which of these LPA effects take placein vivo through genetic and cell biological strategies. Regardless of the actual roles of the ionic conductance responses observed here, it is notable that they resemble neurotransmitter stimulation among neuroblasts at an early developmental stage that precedes GABA and/or l-glutamate responsiveness, thus providing an additional mechanism for producing ionic conductance changes at the earliest stages of cortical neurogenesis.

The existence of G-protein-coupled receptor-mediated LPA signaling in the embryonic cortex adds to receptor tyrosine kinase and neurotransmitter-gated ion channels as a new and potentially important ligand-receptor system influencing cerebral cortical neurogenesis. A difficult but important issue to address is the in vivocellular source of extracellular LPA. Because LPA is also produced intracellularly as an intermediate in phospholipid biosynthesis (Durieux, 1995), this source must be differentiated from the pool that acts as an extracellular signal. The presented data indicate that the stimulatory activity of LPA affects the majority of cortical neuroblasts, consistent with the spatial and temporal expression of the LPA receptor gene lp_A1/vzg-1 (Jalink et al., 1994; Hecht et al., 1996; Fukushima et al., 1998; Chun et al., 1999). As the first electrophysiological evidence of extracellular lysophospholipid signaling during CNS development, these data indicate that LPA, along with other related lipid molecules and their cognate receptors, could have important influences on the development and function of the cerebral cortex as well as other parts of the nervous system.