Nerve growth factor affects Ca²⁺ currents via the p75 receptor to enhance prolactin mRNA levels in GH₃ rat pituitary cells

Abstract

In clonal pituitary GH₃ cells, spontaneous action potentials drive the opening of Ca_v1 (L-type) channels, leading to Ca²⁺ transients that are coupled to prolactin gene transcription. Nerve growth factor (NGF) has been shown to stimulate prolactin synthesis by GH₃ cells, but the underlying mechanisms are unknown. Here we studied whether NGF influences prolactin gene expression and Ca²⁺ currents. By using RT-PCR, NGF (50 ng ml⁻¹) was found to augment prolactin mRNA levels by ∼80% when applied to GH₃ cells for 3 days. A parallel change in the prolactin content was detected by Western blotting. Both NGF-induced responses were mimicked by an agonist (Bay K 8644) and prevented by a blocker (nimodipine) of L-type channels. In whole-cell patch-clamp experiments, NGF enhanced the L-type Ca²⁺ current by ∼2-fold within 60 min. This effect reversed quickly upon growth factor withdrawal, but was maintained for days in the continued presence of NGF. In addition, chronic treatment (≥ 24 h) with NGF amplified the T-type current, which flows through Ca_v3 channels and is thought to support pacemaking activity. Thus, NGF probably increases the amount of Ca²⁺ that enters per action potential and may also induce a late increase in spike frequency. MC192, a specific antibody for the p75 neurotrophin receptor, but not tyrosine kinase inhibitors (K252a and lavendustin A), blocked the effects of NGF on Ca²⁺ currents. Overall, the results indicate that NGF activates the p75 receptor to cause a prolonged increase in Ca²⁺ influx through L-type channels, which in turn up-regulates the prolactin mRNA.

The development of the anterior pituitary gland leads to the appearance of six major types of endocrine cells, each of them defined by the expression of one or more hormone genes (Dasen & Rosenfeld, 2001). The last two cell types to emerge from the nascent gland are somatotropes and lactotropes, which produce growth hormone (GH) and prolactin, respectively. These cells arise from a common lineage (Burrows et al. 1999) as first suggested by the discovery of the somatolactotrope, a cell group that secretes both GH and prolactin (Frawley et al. 1985). There is now ample evidence that one of the functions of somatolactotrope cells is to serve as bipotential progenitors, which can give rise to either somatotropes or lactotropes in response to extrinsic signals (Frawley & Boockfor, 1991; Missale & Spano, 1998; Burrows et al. 1999). It is of interest to note that although mature lactotropes do not express the GH gene, they are able to secrete prolactin at higher basal rates than their bihormonal precursors (Felix et al. 1993; Missale et al. 1995). A third key feature of lactotropes is that they are equipped with D₂ receptors for dopamine (Missale et al. 1995), the major physiological inhibitor of prolactin secretion (Missale & Spano, 1998).

The GH₃ cell line is a clonal strain of rat somatolactotrope cells and has been used as a model to study the role of growth factors and other signals in somatotrope and lactotrope cell differentiation (Missale & Spano, 1998). In this context, the role of nerve growth factor (NGF), the first reported member of the neurotrophin family (Huang & Reichardt, 2001), has attracted attention for three reasons. First, NGF is present at high levels in the anterior pituitary (Lathinen et al. 1989; Patterson & Childs, 1994b; Missale et al. 1996a). Second, the two types of NGF receptors described in neurons, TrkA and p75 (Chao, 2003), are expressed in cells of the somatolactotrope lineage (Patterson & Childs, 1994a; Missale et al. 1994, 1995, 1996a,b). Finally, these cells themselves are sources of NGF and hence this growth factor could act in an autocrine or paracrine fashion to influence the cell phenotype (Patterson & Childs, 1994a; Missale et al. 1995, 1996a,b). Chronic exposure of GH₃ cells to exogenous NGF has been shown to increase the rate of prolactin synthesis while decreasing GH production (Missale et al. 1994). Treatment of GH₃ cells with NGF also results in the expression of D₂ receptors, which are absent in the control condition (Missale et al. 1994). Thus, GH₃ cells differentiate into lactotrope-like cells when grown in the presence of NGF. The use of both rat pituitary cells in primary culture (Missale et al. 1995) and human prolactinoma cell lines (Missale et al. 1996b) has confirmed that NGF is a differentiation factor for lactotropes. Furthermore, a recent study in prolactinoma cells has shed light on the signalling pathway involved in the induction of D₂ receptors by NGF (Fiorentini et al. 2002). However, studies on the mechanism of action of NGF as a regulator of prolactin and GH production are lacking.

With respect to the control of prolactin synthesis, it is known that GH₃ cells spontaneously fire Ca²⁺-dependent action potentials from a resting potential close to −50 mV (Corrette et al. 1995). The analysis of Ca²⁺ currents under voltage clamp has shown that GH₃ cells are endowed with Ca_v3 (T-type) and high voltage-activated (HVA) Ca²⁺ channels (Matteson & Armstrong, 1986; Ritchie, 1993; Meza et al. 1994; Perez-Reyes, 2003). The T-type channels mediate low voltage-activated Ca²⁺ currents that may help regulate the spike frequency (Matteson & Armstrong, 1986; Ritchie, 1993), whereas HVA channels are important contributors to the upstroke of the action potential (Glassmeier et al. 2001). Ca_v1 (L-type) channels, which are highly sensitive to dihydropyridine (DHP) drugs (Perez-Reyes, 2003), carry a large fraction (50% or more) of the HVA Ca²⁺ current (Piros et al. 1995; Glassmeier et al. 2001). Each action potential is thus followed by a transient rise in cytosolic Ca²⁺ concentration that is mostly due to Ca²⁺ entry through L-type channels (Schlegel et al. 1987; Charles et al. 1999; Giráldez et al. 2002). This Ca²⁺ influx is a major determinant of the basal rate of prolactin production (Enyeart et al. 1987; Laverriere et al. 1988), because it is positively coupled to prolactin gene expression (Laverriere et al. 1989; Day & Maurer, 1990; Enyeart et al. 1990; Wang & Maurer, 1999). Based on these data, we have examined whether NGF regulates prolactin mRNA levels, prolactin content and Ca²⁺ currents in GH₃ cells. The results indicate that NGF enhances the expression level of prolactin by causing a prolonged increase in Ca²⁺ influx through L-type channels. In line with this, we show that NGF rapidly (within tens of minutes) increases the L-type Ca²⁺ current amplitude and that it also potentiates the T-type Ca²⁺ current over a longer time period. We also present evidence that NGF acts through the p75 neurotrophin receptor to affect both types of Ca²⁺ currents.

Methods

Cell culture

GH₃ cells from the American Type Culture Collection (Rockville, MD, USA) were grown as a monolayer culture as previously described (Monjaraz et al. 2000). The standard culture medium consisted of Ham's F10 medium supplemented with 15% horse serum, 2.5% fetal bovine serum, 2 mml-glutamine, 100 i.u. ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin (Life Technologies, Grand Island, NY, USA). The maintenance culture was grown in 25-cm² flasks (Corning Costar, Cambridge, MA, USA). Once a week, the cells were removed from the flask and diluted 5-fold for propagation. For RNA isolation and Western blot analysis, cells were seeded into 60-mm culture dishes (∼3 × 10⁶ cells per dish). On day 2 after plating, control cells were fed with fresh standard medium. Sister cultures of GH₃ cells received standard medium supplemented with either 50 ng ml⁻¹ NGF, 0.5 μm (−)-Bay K 8644, 1 μm nimodipine or 50 ng ml⁻¹ NGF plus 1 μm nimodipine. The medium, with or without additions, was replenished on days 3, 4 and 5. NGF consisted of the 2.5S form of the purified peptide from mouse submaxillary glands and was obtained from Alomone Laboratories (Jerusalem, Israel). In some experiments, epidermal growth factor (EGF) was used instead of NGF. The DHP drugs (Bay K 8644 and nimodipine) and EGF were purchased from Sigma Chemical Co. (St Louis, MO, USA). At the concentrations used, these compounds have been shown to elicit nearly maximal effects on prolactin production (Enyeart et al. 1987, 1990; Laverriere et al. 1989; Aanestad et al. 1993; Missale et al. 1994). For current recording, cells were seeded into 35-mm culture dishes (∼2 × 10⁵ cells per dish) containing poly-l-lysine-coated glass coverslips. NGF was applied to these cells for variable periods (from 15 min to up to 5 days) before the recordings, which were carried out on day 5 or 6 after plating. In each experiment, data obtained from NGF-treated cells were matched with measurements performed on the same day on control cells. Other compounds used for cell treatments were actinomycin D (Sigma), the anti-p75 monoclonal antibody MC192 (Chemicon International, Temecula, CA, USA) and the tyrosine kinase inhibitors K252a (Alomone), lavendustin A (Sigma) and genistein (Sigma).

RNA isolation and RT-PCR analysis

The cells were recovered from the culture dishes on day 5 after plating. Harvested cells were collected by centrifugation and total RNA was then extracted from each cell group using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). The amount of total RNA obtained was always close to 100 μg per dish. The integrity of the RNA was assessed by electrophoresis in 1% agarose gels. Total RNA samples were stored at −70°C until required for RT-PCR analysis. RT-PCR was performed as previously described (Vega et al. 2003) using the SuperScript one-step kit (Invitrogen) and primers specific for the prolactin and cyclophilin mRNA sequences (GenBank accession no. NM012629 and M19533, respectively). The primers for prolactin were as follows: 5′-CTTCTGTTCTGCCAAAATGTGC-3′ (upper primer) and 5′-CGTTAGCAGTTGTTTTTATGGAC-3′ (lower primer). The primer set for cyclophilin was the same as used by Vega et al. (2003). The reaction mixtures were processed in a GeneAmp 2400 thermocycler (Perkin-Elmer, Shelton, CT, USA). The reverse transcriptase reaction was carried out at 50°C for 30 min. For amplification, an initial denaturation step at 94°C for 2 min was followed by the number of PCR cycles indicated below. Each cycle consisted of three temperature steps: 30 s at 94°C, 30 s at 57°C and 90 s at 72°C. The reaction ended with a final incubation at 72°C for 7 min. To check primer performance, reaction mixtures containing 1 μg total RNA in a final volume of 50 μl were subjected to 40 cycles of amplification. PCR products were resolved by electrophoresis on 1% agarose gels stained with ethidium bromide. The use of the primer set for prolactin in these reactions led to the amplification of a cDNA of 623 bp, as expected. Likewise, the cyclophilin cDNA ran at the predicted size of 257 bp. The identity of both cDNAs was verified by sequencing. For semiquantitative analysis, the prolactin cDNA was amplified for 23 cycles from reaction mixtures containing 3 ng total RNA in a final volume of 25 μl. The cyclophilin cDNA served as an internal control and was amplified from each RNA sample using 20 ng RNA per reaction and 22 PCR cycles. Given that the number of cycles used was within the exponential phase of the PCR reaction, these conditions allowed detection of both cDNAs in the linear range of the assay. In all cases, aliquots of PCR products (5 μl) were run in triplicate on 1% agarose gels containing ethidium bromide. For a given RNA sample, both PCR products were run on the same gel to avoid intergel variability. Digitized images of the gels, which are presented in the negative form, were obtained with a Kodak EDAS 120 system. The intensity of PCR signals was measured by densitometry using Kodak Digital Science software.

Western blot

To evaluate the prolactin content, cells were lysed in ice-cold buffer containing (mm): NaCl 150, Tris-HCl 10 (pH 7.5), NaF 20 and Na₃VO₄ 0.5, with 1% Nonidet P-40 and a cocktail of protease inhibitors (Roche, Mannheim, Germany). The lysates were cleared from insoluble material by centrifugation at 12 000 g for 15 min at 4°C, and total protein concentration was determined using the BCA assay kit (Pierce, Rockford, IL, USA). Aliquots of cell proteins (50 μg) were boiled in Laemmli buffer, resolved by SDS-PAGE and electro-transferred onto polyvinylidene difluoride membranes. Non-specific binding was blocked by 60-min incubation at room temperature (22–24°C) in the presence of Tris-buffered saline with 0.1% Tween 20 (TBST) containing 5% non-fat dry milk, 3% bovine serum albumin and 2% normal goat serum. Membranes were then exposed overnight at 4°C to a rabbit anti-prolactin polyclonal antibody (NIDDK antirPRL IC-5) diluted 1 : 5000 in blocking buffer. To control for variation in the amount of protein loaded, membranes were also probed with a mouse antiβ-actin monoclonal antibody (1 : 300; a gift from Dr M. Hernández, Cinvestav-IPN, Mexico). After being washed with TBST, membranes were incubated with the corresponding secondary antibody conjugated with peroxidase. Enhanced chemiluminescence reagents were then applied for 2 min to reveal bound peroxidase activity. Finally, membranes were exposed to photographic films and digitized images of the obtained blots were analysed as described above to measure band intensity.

Current recording

Ca²⁺ currents were recorded from GH₃ cells using the whole-cell mode of the patch-clamp technique. The cells were placed in a chamber mounted on the stage of an inverted microscope and superfused with an external solution containing (mm): NaCl 134, TEA Cl 10, CaCl₂ 10, Hepes-NaOH 10 and glucose 5, with 1 μm TTX; pH 7.3. Recordings from NGF-treated cells were made in the absence or continued presence of the growth factor. In the latter case, the external solution was supplemented with NGF (50 ng ml⁻¹). Patch pipettes were pulled from borosilicate glass capillaries. They had a resistance of ∼2 MΩ when filled with the internal recording solution containing (mm): CsCl 135, EGTA-CsOH 10, CaCl₂ 1, MgCl₂ 2, Na₂ATP 2, GTP 0.05, Hepes-CsOH 10 and glucose 5; pH 7.3. Currents were measured with an Axopatch 200A amplifier (Axon Instruments, Union City, CA, USA) at room temperature. Data acquisition and voltage pulse generation were performed with a TL-1 interface controlled by pCLAMP software (Axon Instruments). Test pulses were applied from a holding potential of −80 mV. Current responses were filtered at 5 or 10 kHz and digitized at intervals of 14, 20 or 100 μs, depending on pulse duration. Capacitive transients were reduced to a minimum with the amplifier's circuit. The remaining linear components in the current signals were subtracted with scaled pulse routines. Values for cell capacitance (C_m) and series resistance (R_s) were obtained by using the transient cancellation feature on the amplifier. The C_m value was verified by integrating capacitive transients as described by Meza et al. (1994). The average C_m in control and treated cells was always in the range 10–12 pF. Thus, the product R_s × C_m was usually less than 35 μs; when it exceeded this value, R_s was compensated by 40–60% with the amplifier's circuit. Recording was started ∼1 min after rupturing the membrane patch and was generally concluded within the next 2 min. Each current measurement was made in triplicate and averaged.

Data analysis

Data are given as means ± s.e.m. for the number (n) of cells or separate experiments indicated. ANOVA and Student's unpaired t test were used to assess the significance of differences between distinct groups. P < 0.05 was considered significant. Analysis of current traces and curve fittings was performed using pCLAMP (Axon Instruments) and SigmaPlot software (SPSS, Chicago, IL, USA). Ca²⁺ tail currents were fitted by the sum of two exponentials following the procedure of Matteson & Armstrong (1986). The initial amplitudes of the fast and slow components in the tail currents were taken as the amplitudes of the fitted exponentials measured 120 μs after the onset of repolarization. The voltage dependence of tail current amplitude was described by a Boltzmann function:

where I is the current, I_max is the maximal tail current at positive voltages, V_m is the activating potential, V_1/2 is the mid-point potential and k is the slope factor.

Results

NGF regulates the prolactin mRNA through a mechanism involving L-type Ca²⁺ channels

We first examined whether the long-term effects of NGF on GH₃ cells include regulation of prolactin mRNA levels. Total RNA was extracted from control cells and cells that were exposed to NGF (50 ng ml⁻¹) for 72 h in culture. Optimal conditions for semiquantitative RT-PCR were then used to amplify the cDNAs for prolactin and cyclophilin from the RNA samples. Whereas the cyclophilin mRNA signal appeared to be unaffected by NGF treatment, the intensity of the prolactin mRNA signal was clearly enhanced (Fig. 1A). To quantify this effect, the ratio of prolactin to cyclophilin signal intensity was determined for each cell group and then expressed as a percentage of its control value. In three separate RT-PCR reactions using RNA from different batches of cells, NGF was found to augment the standardized levels of the prolactin mRNA by ∼80% (Fig. 1B). Our second objective in this first series of experiments was to test the role of L-type Ca²⁺ channels in the action of NGF. To this end, the RT-PCR analysis was extended to GH₃ cells treated with Bay K 8644 (0.5 μm) or nimodipine (1 μm). These DHP drugs, which are two well-known modulators (an agonist and a blocker, respectively) of L-type Ca²⁺ channels (Enyeart et al. 1987, 1990; Piros et al. 1995; Perez-Reyes, 2003), were also applied for 72 h. The effect of adding NGF in combination with nimodipine was studied in a similar way. Consistent with previous reports (Laverriere et al. 1989; Enyeart et al. 1990), we found that prolactin mRNA levels were increased ∼1.9-fold by the L-type channel agonist, and were reduced more than 3-fold by the L-type channel blocker (Fig. 1B). These results confirmed that the basal expression of the prolactin gene in GH₃ cells is largely sustained by L-type Ca²⁺ channel activity. Moreover, they showed that long-term exposure to Bay K 8644 up-regulates prolactin mRNA to about the same level as NGF. The close similarity of both stimulatory effects suggested that NGF may act via L-type channels. Indeed, when NGF was added in the presence of nimodipine, the resultant prolactin mRNA levels were as low as those observed after treatment with the L-type channel blocker alone (Fig. 1B). Thus, blocking Ca²⁺ influx through L-type channels completely abolished the effect of NGF on the prolactin mRNA.

Figure 1. The effect of NGF on prolactin mRNA and protein levels is mimicked by Bay K 8644 and prevented by nimodipine.

A, representative gel showing the effects of long-term exposure to NGF or DHP drugs on PCR signals for the prolactin and cyclophilin messages. Total RNA was extracted from sibling cultures of GH₃ cells that were grown in the absence (control) or presence of 50 ng ml⁻¹ NGF, 0.5 μm (−)-Bay K 8644 (BayK), 1 μm nimodipine (NIM), or NGF plus nimodipine (NGF + NIM) for 72 h. The cDNAs for cyclophilin and prolactin were then amplified in parallel from the RNA samples using optimal conditions for semiquantitative analysis. B, average intensity of the prolactin mRNA signal, normalized to the cyclophilin signal and expressed as a percentage of control values, in the different cell groups. The results (mean ± s.e.m) of three independent experiments are shown. Distinct letters denote significantly different (P < 0.05) means. C, example of Western blot designed to measure changes in the intracellular content of prolactin. Before protein extraction, cells were treated with NGF and DHP drugs as described in A. D, relative prolactin levels in the indicated cell groups. Data were derived from the densitometric analysis of three separate blots, with prolactin signals normalized to the corresponding β-actin levels.

The regulation of prolactin mRNA levels by NGF and DHP drugs would be expected to be followed by parallel changes in prolactin production. It has been reported, in fact, that in the GH₄C₁ cell line, a subclone of GH₃ cells, both the amount of secreted prolactin and the intracellular content of this hormone can be markedly altered by chronic treatment with L-type Ca²⁺ channel modulators (Enyeart et al. 1987). GH₃ cells also secrete much more prolactin after NGF treatment (Missale et al. 1994), but a corresponding increase in prolactin content has not been observed. To address this issue, cell proteins were extracted before and after long-term exposure (72 h) to NGF and DHP drugs. The prolactin content was then evaluated by Western blotting using β-actin signals for normalization, as shown in Fig. 1C. The results of the analysis of three separate blots are shown in Fig. 1D. Prolactin was found to be present at relatively high levels in cells treated with NGF or Bay K 8644, and at very low levels in cells treated with either nimodipine alone or NGF plus nimodipine. Overall, there was a strong correlation between the prolactin content (Fig. 1D) and the expression level of the prolactin mRNA (Fig. 1B).

For comparative purposes, RT-PCR was also performed on total RNA from GH₃ cells treated with EGF, which has been shown to share the ability of NGF to promote the lactotrope phenotype (Schonbrunn et al. 1980; Felix et al. 1995; Missale & Spano, 1998). In agreement with previous work (White & Bancroft, 1983; Gilchrist & Shull, 1993), a 72-h exposure to EGF (0.15 nm) enhanced prolactin mRNA levels by ∼90%. Although this effect of EGF was quantitatively similar to that of NGF, the use of nimodipine revealed that the mechanisms involved are different. Figure 2 shows that the EGF-induced elevation in prolactin mRNA was prevented only in part by the L-type channel blocker. As compared to cells exposed to nimodipine alone, prolactin mRNA levels were increased ∼5-fold by the combined treatment with EGF and nimodipine. From the results in Figs 1 and 2, we conclude that L-type Ca²⁺ channels are essential for the stimulation of prolactin expression by NGF, but not by EGF.

Figure 2. EGF elevates the prolactin mRNA even in the presence of nimodipine.

A, PCR signals for the prolactin and cyclophilin messages in a representative gel. Semi-quantitative RT-PCR was performed on total RNA from control GH₃ cells and cells that were treated with 0.15 nm EGF, 1 μm nimodipine (NIM), or EGF plus nimodipine (EGF + NIM) for 72 h. B, relative levels of the prolactin mRNA in these cells as determined from four separate experiments.

NGF rapidly enhances the L-type Ca²⁺ current

The results in Fig. 1 prompted us to carry out voltage-clamp experiments to study the influence of NGF on whole-cell Ca²⁺ currents. As a first approach, GH₃ cells were treated with NGF for 60 min at 37°C and current recordings were then made at room temperature in the continued presence of the growth factor. Figure 3A shows examples of Ca²⁺ current traces obtained from control and NGF-treated cells. In each cell, Ca²⁺ currents were activated by 10-ms step depolarizations from a holding potential of −80 mV. At pulse end, the membrane was repolarized to −80 mV to monitor tail current behaviour. In Fig. 3B, average peak current values during the activating pulses are plotted as a function of the test potential for seven cells of each group. The 60-min exposure to NGF had no effect on Ca²⁺ current amplitude at low depolarizations, but resulted in larger Ca²⁺ currents at −10 mV or more positive potentials. At +20 mV, for example, peak Ca²⁺ current was increased from −166 ± 23 pA in control cells to −242 ± 18 pA in NGF-treated cells. The average initial C_m was 10.6 ± 0.5 and 10.4 ± 0.5 F, respectively. In the experiments to be described below, mean C_m values also did not differ significantly between control and treated cells. Thus, in no case were the reported changes in Ca²⁺ current amplitude due to significant variations in cell surface area.

Figure 3. A 60-min exposure to NGF increases the activity of HVA Ca²⁺ channels.

A, representative Ca²⁺ current recordings in a control cell and a cell that was exposed to NGF for 60 min at 37°C before being examined with NGF present in the bathing solution (1 h NGF). Step depolarizations of 10 ms were applied from a holding potential of −80 mV. Currents are shown for potentials between −50 and +30 mV in 20 mV increments, as indicated by the diagram in the left panel. Tail currents recorded on repolarization are also shown. B, dependence of peak Ca²⁺ current on the membrane potential (V_m) during the depolarizing steps in control and NGF-treated cells (n = 7 in each case). From −10 to +50 mV, the magnitude of the current was significantly larger in NGF-treated cells. C, plots of the initial amplitude of the tail current for HVA (fast tail) and T-type (slow tail) channels against V_m. Same cells and symbols as in B. Each data set was fitted with a Boltzmann function (continuous lines) to determine the maximal tail current (I_max), V_1/2 and k. Values for I_max are given in the text. The values for V_1/2 were 7 ± 1 and 9 ± 1 mV for HVA channels, and −33 ± 1 and −34 ± 2 mV for T-type channels in control and NGF-treated cells, respectively. The values for k were 12.8 ± 0.9 and 13.1 ± 0.7 mV, and 8.3 ± 1.2 and 7.6 ± 0.7 mV, respectively.

T-type and HVA Ca²⁺ currents have been previously distinguished in GH₃ cells (Matteson & Armstrong, 1986; Ritchie, 1993; Meza et al. 1994). In our experiments, the external recording solution contained 10 mm Ca²⁺, and Cs⁺ was the main internal cation. Under similar conditions, the T-type current has been found to be dominant in the voltage range from −50 to −20 mV (Ritchie, 1993; Meza et al. 1994). Thus, the current–voltage curves in Fig. 3B suggest that NGF selectively increased the activity of HVA Ca²⁺ channels. This view was supported by the analysis of tail currents. In control and NGF-treated cells, Ca²⁺ tail currents declined either slowly or in two phases, fast and slow, depending on the voltage level during the preceding activating pulse (Fig. 3A). It is known that the slow phase of these current signals marks the closing or deactivation of T-type channels, whereas the fast phase reflects the rapid closing of HVA channels (Matteson & Armstrong, 1986; Ritchie, 1993; Meza et al. 1994). The initial amplitude of each tail phase, which provides an index of the respective Ca²⁺ conductance, is plotted against the activating potential in Fig. 3C for the same cells as in Fig. 3B. Sigmoidal relations were obtained in all cases due to the voltage dependence of Ca²⁺ channel activation. Boltzmann fits to data points showed that the maximal tail current (I_max) for HVA channels was −455 ± 59 pA in control cells and −703 ± 64 pA in NGF-treated cells. The I_max for T-type channels was −132 ± 34 and −135 ± 40 pA, respectively. For each channel set, the other two parameters of the fits (V_1/2 and k) were similar in both cell groups (see the legend to Fig. 3C). Thus, NGF increased the HVA current amplitude ∼1.5-fold at all test potentials without significantly affecting the T-type current. NGF also did not alter the rate of Ca²⁺ channel deactivation. After the steps to +20 mV, for example, the HVA channels deactivated at −80 mV with a time constant of 0.13 ± 0.01 and 0.12 ± 0.01 ms in control and NGF-treated cells, respectively; the deactivation time constant for T-type channels was 3.0 ± 0.1 and 2.8 ± 0.1 ms, respectively. In summary, treatment of GH₃ cells with NGF for 60 min had no impact on gating properties of T-type and HVA Ca²⁺ channels, but enhanced the HVA current by ∼50%.

It has been recently shown that in GH₃B₆ cells the HVA Ca²⁺ current is carried not only through L-type channels but also by other Ca²⁺ channel subtypes, which are insensitive to DHP drugs at concentrations in the low micromolar range (Glassmeier et al. 2001). To determine whether NGF specifically targeted L-type channels when applied to GH₃ cells for 60 min, Ca²⁺ currents evoked by 10-ms pulses to +20 mV were recorded from control and NGF-treated cells in the absence and presence of 1 μm nimodipine. The results are illustrated in Fig. 4A and summarized in Fig. 4B. As can be seen, nimodipine inhibited peak Ca²⁺ current at +20 mV by ∼47% in control cells and by ∼67% in NGF-treated cells. The associated fast tail at −80 mV was reduced by ∼52% and ∼68%, respectively, with no significant change in the slow tail. Therefore, in the presence of nimodipine, the Ca²⁺ currents were identical, on average, in control and NGF-treated cells. The main conclusion from these results is that the 50% increase in HVA Ca²⁺ current produced by NGF was due to a nearly 2-fold enlargement of the L-type current component (Fig. 4C and D). It is worth noting that nimodipine did not block a significant fraction of T-type Ca²⁺ channels, as deduced from the lack of change in the slow tail current. Furthermore, the T-type current measured at −30 mV was also unaffected by nimodipine (data not shown).

Figure 4. NGF targets L-type Ca²⁺ channels.

A, effects of including 1 μm nimodipine (NIM) in the bathing solution on Ca²⁺ currents recorded from control cells (left traces) and cells that were treated with NGF for 60 min before being examined in the continued presence of the growth factor (right traces). The pulse protocol is shown in the left panel. Scale bars apply to all traces, including those in C. B, average Ca²⁺ current values obtained from the indicated number of control and NGF-treated cells before and after application of nimodipine (NIM). ^*Significant difference from control values determined in the absence of the Ca²⁺ channel blocker. C, currents blocked by nimodipine in the examples shown in A. D, average amplitude of the L-type Ca²⁺ current in control and NGF-treated cells. These data were derived from the results in B. In each cell group, the mean value of current observed in the presence of nimodipine was subtracted from the current traces recorded in the absence of the Ca²⁺ channel blocker to yield the L-type current amplitude.

To test the reversibility of the action of NGF on HVA Ca²⁺ channels, GH₃ cells that had been exposed to the growth factor for 60 min were subjected to voltage clamp with or without NGF present externally. Currents began to be recorded ∼10 min after immersion of the cells into the bathing solution. The presence of NGF throughout the recording session allowed us to detect again a significant increase in peak Ca²⁺ current at +20 mV, as compared to control cells (Fig. 5). Likewise, the fast tail current at the end of the pulse was increased ∼1.5-fold. These effects, however, were not seen when the external recording solution was not supplemented with NGF. Thus, the positive influence of NGF on HVA Ca²⁺ channel activity fully reversed within some minutes after withdrawal of the growth factor.

Figure 5. The effect of NGF on HVA Ca²⁺ current is rapidly reversible.

A, relative Ca²⁺ current values in control cells and two cell groups that were exposed to NGF for 60 min at 37°C before the recordings. Test depolarizations of 10 ms to +20 mV were used. A treated cell group was voltage clamped in the continued presence of the growth factor (1 h NGF), whereas the other one was examined within 10–30 min after NGF removal (Wash). Each current measurement was divided by C_m and then converted to a percentage of the corresponding mean value of current density in control cells. This was done to facilitate comparison of the present data with those shown in Fig. 6, which were obtained from a different batch of cells. ^*Significant difference from controls. B, examples of Ca²⁺ currents recorded from one cell of each group.

NGF induces a slow increase in the T-type Ca²⁺ current

We next assessed the consequences of a shorter (15 min) or much longer (72 h) exposure to NGF on Ca²⁺ current amplitude. The application of NGF for 15 min at 37°C, followed by current recording in the continued presence of the growth factor, failed to produce a significant increase in the Ca²⁺ currents elicited by test pulses to +20 mV (data not shown). When, on the other hand, the treatment with NGF was prolonged to 72 h, a pronounced increase in peak Ca²⁺ current at +20 mV was obtained (to ∼190% of control). The analysis of tail currents revealed an enhanced activity of both HVA and T-type channels in this case (Fig. 6). The increase in the fast tail (∼1.55-fold) was similar in magnitude to that induced by a 60-min exposure to NGF, and once more reversed rapidly upon growth factor removal. By contrast, the increase in the slow tail (∼2.2-fold) was a persistent event, as it did not show any sign of reversion when recordings were made in the absence of NGF.

Figure 6. Chronic NGF treatment stimulates the activity of both HVA and T-type Ca²⁺ channels.

A, Ca²⁺ current density values obtained in response to 10-ms depolarizations to +20 mV in control cells and two cell groups treated with NGF for 72 h in culture. Recordings from treated cells were made in the presence (3 d NGF) or absence (Wash) of NGF in the bathing solution. Data are presented as in Fig. 5. B, representative Ca²⁺ current recordings from control and NGF-treated cells.

In the following experiments we characterized the regulation of T-type channels in GH₃ cells by chronic NGF treatment. To minimize the effect of NGF on HVA Ca²⁺ current, the growth factor was not present in the external recording solution. Figure 7 compares Ca²⁺ current–voltage curves in control cells and cells that were exposed to NGF for 4 days. This treatment persistently increased peak Ca²⁺ current at all test potentials (Fig. 7A and B). The increase was larger at −40 or −30 mV (∼2.5-fold) than at −10 mV (∼2-fold) or +20 mV (∼1.4-fold) as a result of the distinct contribution of T-type channels to the total Ca²⁺ current at different voltages. This contribution reduces with increasing depolarization as more and more HVA channels open. In fact, as shown by tail current analysis, the T-type current was enhanced ∼2.5-fold over the range of test potentials investigated (Fig. 7C). The 10–15% increase in the fast tail observed in NGF-treated cells at positive voltages was not significant. As in Fig. 3C, the tail current data in this experiment were well fitted by Boltzmann equations. The I_max for T-type channels was −166 ± 29 pA (n = 7) in control cells and −452 ± 43 pA (n = 7) in NGF-treated cells, whereas the I_max for HVA channels was −508 ± 70 and −578 ± 45 pA, respectively. The fitted values for V_1/2 and k, which are given in the legend to Fig. 7C, indicated no persistent changes in the voltage dependence of T-type or HVA Ca²⁺ channel activation.

Figure 7. NGF increases the T-type Ca²⁺ current without affecting its voltage dependence.

A, families of Ca²⁺ currents recorded from control and NGF-treated cells using the pulse protocol shown in the left panel. In this experiment, treated cells were exposed to NGF for 4 days and then examined in the absence of the growth factor (Wash after 4 d NGF). B, voltage dependence of the average Ca²⁺ current measured during 10-ms step depolarizations in control and NGF-treated cells (n = 7 for both). C, voltage dependence of tail current amplitude for HVA (fast tail) and T-type (slow tail) Ca²⁺ channels. Same cells and symbols as in B. Continuous lines correspond to Boltzmann fits to data points. The I_max values are given in the text. The values for V_1/2 were 10 ± 3 and 12 ± 1 mV for HVA channels, and −30 ± 2 and −32 ± 1 mV for T-type channels in control and NGF-treated cells, respectively. The values for k were 13.5 ± 0.7 and 12.9 ± 0.4 mV, and 7.7 ± 0.6 and 6.9 ± 0.5 mV, respectively.

The time course of the effect of NGF on T-type channel activity was explored by measuring Ca²⁺ currents in response to 200-ms depolarizing pulses to −30 mV (Fig. 8A). A 24-h exposure to NGF was found to produce a significant increase in T-type Ca²⁺ current density (to ∼150% of control). The magnitude of the current continued to increase during the subsequent days and was maximal on day 5 (Fig. 8B). Following a 5-day treatment, little recovery was observed 6 h after NGF withdrawal. When the recovery period was extended to 25 h, the T-type current density was still well above the control level (Fig. 8C). Concerning the mechanism involved, actinomycin D (1 μm), an inhibitor of gene transcription, was able to prevent the NGF-induced increase in T-type current when measured over a 24-h period (Fig. 8D), suggesting that RNA synthesis is required for this increase. In addition, chronic NGF treatment did not seem to modify the kinetic properties of T-type channels. For example, in the experiment of Fig. 6, the slow tail current declined with a time constant of 3.3 ± 0.2 ms (n = 15) or 3.2 ± 0.1 ms (n = 32) in control and NGF-treated cells, respectively. Thus, T-type channels deactivated at similar rates in both cell groups, despite the 2.3-fold difference in current density. There were also no major changes in activation and inactivation kinetics, as suggested by the Ca²⁺ current records in Fig. 8A. To compare the inactivation rate, the decaying phase of the T-type current at −30 mV was fitted with a single exponential. The corresponding inactivation time constant was similar in control cells (26 ± 2 ms; n = 15) and cells that were treated with NGF for 4 days (25 ± 1 ms; n = 15).

Figure 8. Time course and block by actinomycin D of the NGF-induced increase in T-type current.

A, typical T-type Ca²⁺ currents evoked by 200-ms test pulses to −30 mV in a control cell and a cell that was treated with NGF for 4 days. NGF was not included in the bathing solution. B, T-type current density as a function of the duration of NGF treatment. Data were derived from three different batches of cells, which are indicated by distinct symbols. The number of cells investigated is given next to each data point. As in panels C and D, the peak amplitude of the T-type current at −30 mV was divided by C_m and then normalized to the corresponding control value. C, slow decline in T-type current density after NGF withdrawal. The first point in this graph corresponds to the last point in B. The two additional data points were also obtained from the batch of cells that was exposed to NGF for 5 days. These cells were allowed to recover at 37°C for the indicated times before the recording session. D, T-type current density after 24-h exposures to NGF, 1 μm actinomycin D (Act D) or NGF in combination with actinomycin D (NGF + Act D). Nine or 10 cells were examined per group.

To further compare the properties of T-type currents in control and NGF-treated cells, nimodipine (1 μm) and Cd²⁺ (30 μm) were added to the external recording solution. Preliminary tests revealed that the combined use of both Ca²⁺ channel blockers could largely suppress HVA Ca²⁺ currents, while blocking only a minor fraction (∼15%) of T-type channels. This allowed us to record well-resolved, isolated T-type currents over a wide range of membrane voltages (Fig. 9A). Under these conditions, peak T-type current amplitude was ∼2.5-fold larger, on average, in cells that had been exposed to NGF for 4 days than in control cells, regardless of the activating pulse used (Fig. 9A and B). This finding reinforced the conclusion derived from tail current measurements in the absence of nimodipine and Cd²⁺ (Fig. 7C, lower graph). The data in Fig. 9B were used to construct peak Ca²⁺ conductance–voltage relationships, and they are plotted in Fig. 9C. As shown by Boltzmann fits, NGF treatment increased the maximal conductance for T-type channels to 2.8 ± 0.2 nS (n = 7), from a control value of 1.1 ± 0.1 nS (n = 7), without altering the activation parameters (Fig. 9C). Analysis of current traces like those in Fig. 9A also confirmed that NGF had no significant effect on the inactivation rate of T-type channels (Fig. 9D). Finally, the voltage dependence of T-type channel inactivation was measured by applying a constant test pulse to −20 mV after 500-ms prepulses of varying amplitude. The resulting inactivation curves were practically identical in control and NGF-treated cells, with V_1/2 close to −45 mV in both cases (Fig. 9E).

Figure 9. Comparison of isolated T-type Ca²⁺ currents in control and NGF-treated cells.

A, representative traces showing increased T-type currents after a 4-day treatment with NGF compared to controls. Currents were evoked by 100-ms pulses to the indicated voltages. HVA Ca²⁺ channels were blocked by adding 1 μm nimodipine and 30 μm Cd²⁺ to the external recording solution. B, average peak amplitude of the T-type current measured in control and NGF-treated cells (n = 7 in both cases) as a function of V_m. C and D, voltage dependence of the conductance and inactivation time constant of T-type channels in same cells as in B. Ca²⁺ conductance was calculated by dividing peak current by V_m−V_rev, with V_rev (the reversal potential) equal to +50 mV. Continuous lines in C are Boltzmann functions with V_1/2 and k values as indicated; values for the maximal conductance are given in the text. E, inactivation curves for T-type channels in control and NGF-treated cells (n = 4 for both). The normalized peak amplitude of the T-type current at −20 mV is plotted against V_m during 500-ms prepulses. V_1/2 and k values derived from Boltzmann fits (continuous lines) were −45 ± 2 and 7.6 ± 0.3 mV, respectively, in control cells, and −46 ± 2 and 7.5 ± 0.6 mV, respectively, in NGF-treated cells.

NGF regulates Ca²⁺ channels through the p75 receptor

NGF can signal through two types of cell surface receptors, the TrkA tyrosine kinase receptor and the p75 neurotrophin receptor (Chao, 2003). These receptors are often present on the same cell and both have been described in pituitary cells of the somatolactotrope lineage, including GH₃ cells (Missale et al. 1994). It was thus important to explore the role of TrkA and p75 receptors in the two distinct effects of NGF on Ca²⁺ currents described above. Based on previous work (Berg et al. 1992; Lei et al. 1998; Fiorentini et al. 2002), the involvement of TrkA was examined by exposing GH₃ cells to K252a (50 nm), lavendustin A (3 μm) or genistein (5 μm), which are inhibitors of tyrosine kinases. To inhibit p75-mediated responses, cells were treated with MC192 (50 ng ml⁻¹), a mouse monoclonal antibody that is specific for rat p75. MC192 has been used with this same purpose in several previous studies (e.g. Bui et al. 2002; Numakawa et al. 2003; McCollum & Estus, 2004). Figure 10A and B shows that MC192 fully blocked the enhancement of HVA Ca²⁺ current produced by a 60-min exposure to NGF. By contrast, this effect of NGF persisted in the presence of K252a or lavendustin A (Fig. 10A), and similar results were obtained when genistein was used (data not shown). The increase in T-type Ca²⁺ current induced by long-term exposure (48 h) to NGF was also effectively prevented by MC192 (Fig. 10C and D), but not by K252a or lavendustin A (Fig. 10C). No significant changes in basal Ca²⁺ current densities were noted in response to short-term or long-term treatment with K252a, lavendustin A or the anti-p75 antibody (Fig. 10A and C). These results indicate that activation of p75, but not of TrkA, is required for the NGF-induced increase of L-type and T-type Ca²⁺ currents in GH₃ cells.

Figure 10. The effects of NGF on Ca²⁺ currents are prevented by MC192, but not by tyrosine kinase inhibitors.

A, summary of Ca²⁺ current measurements from control cells and seven groups of treated cells using 10-ms test pulses to +20 mV. Before the recordings, some cells were treated with 50 ng ml⁻¹ MC192, 50 nm K252a or 3 μm lavendustin A (LA) for 90–120 min at 37°C, or with 50 ng ml⁻¹ NGF for 60 min. Other cell groups were first incubated for 60 min with MC192, K252a or lavendustin A, and then exposed to NGF for a second 60-min period in the continued presence of those compounds. The external recording solution was supplemented accordingly. The number of cells examined is given above left-hand bars. B, examples of Ca²⁺ current recorded from the indicated cell groups. Same experiment as in A. C, peak T-type Ca²⁺ current density measured from the number of cells given above each bar using 200-ms test pulses to −30 mV. Cells were treated with the indicated compounds for 48 h, then examined in the presence of external recording solution without additions. D, examples of current traces obtained in experiment shown in C.

Discussion

The major aim of this study was to gain insight into the mechanisms by which NGF increases the capacity of GH₃ cells to produce prolactin. The results show that chronic exposure to NGF up-regulates the prolactin mRNA, leading to a marked rise in the prolactin content of GH₃ cells. The action of NGF is mimicked by Bay K 8644 and can be completely prevented by nimodipine. This suggests that NGF enhances the influx of Ca²⁺ through L-type channels caused by spontaneous action potentials. Indeed, the L-type Ca²⁺ current recorded under voltage clamp is enhanced by NGF within tens of minutes. In addition, NGF is able to augment the T-type Ca²⁺ current over days and thereby may induce an increase of cell excitability. Finally, our data point to the p75 neurotrophin receptor as a key signalling element that mediates the effects of NGF on Ca²⁺ currents. Taken together, these findings indicate that NGF stimulates expression of the prolactin gene through activation of the p75 receptor and subsequent regulation of two main sets of Ca²⁺ channels.

NGF regulates L-type and T-type channels over different time scales

A 15-min exposure of GH₃ cells to NGF failed to enhance the HVA Ca²⁺ current. This is in contrast to previous studies in other excitable cells showing that NGF can acutely (within 3–5 min) amplify L-type currents (Wildering et al. 1995; Jia et al. 1999; Rosenbaum et al. 2001). Nevertheless, we found that NGF does increase the L-type component of HVA Ca²⁺ current by ∼2-fold when applied to GH₃ cells for 60 min. The L-type component was taken as the Ca²⁺ current blocked by 1 μm nimodipine and comprised about half of the total HVA current in the absence of NGF. Similar results have been obtained before on GH₃ cells (Piros et al. 1995) and GH₃B₆ cells (Glassmeier et al. 2001). NGF did not affect the HVA current that is insensitive to nimodipine, which at least in GH₃B₆ cells is carried in part by P/Q-type Ca²⁺ channels (Glassmeier et al. 2001), and so the total HVA current was increased by ∼50%. This effect was readily reversible, indicating a short-term control of L-type channels by NGF. The effect, however, was maintained for days in the continued presence of the growth factor. Thus, NGF is likely to produce a sustained increase in the amount of Ca²⁺ that enters through L-type channels with each action potential.

It is well established that the expression levels of HVA Ca²⁺ channels can be subjected to long-term regulation by NGF in PC12 cells (Usowicz et al. 1990; Colston et al. 1998; Black et al. 2003) as well as in neurons (Levine et al. 1995; Lei et al. 1998; Baldelli et al. 2000) and pancreatic β cells (Rosenbaum et al. 2002). Apparently this was not the case in GH₃ cells, as a significant persistent change in HVA Ca²⁺ current amplitude was not observed after a 4-day exposure to NGF. Instead, chronic NGF treatment promoted the functional expression of T-type Ca²⁺ channels, leading to a 2- to 3-fold increase in the corresponding current. The slow onset and reversion of this effect suggests the involvement of gene expression regulation. Indeed, NGF was unable to augment the T-type current when actinomycin D was used to block gene transcription. Recently, mRNAs for two distinct isoforms of the T-type channel (Ca_v3.1 and Ca_v3.3) were detected in GH₃ cells (Mudado et al. 2004). Thus, it would be interesting in the future to test whether these mRNAs are responsive to NGF. As in previous studies (Ritchie, 1993; Meza et al. 1994), we observed that the activation of T-type channels became significant at −50 mV, which is near the resting potential of GH₃ cells (Corrette et al. 1995). At this voltage, the steady-state inactivation of T-type channels is considerable, but not complete. Therefore, the small fraction of T-type channels that are still available at the resting potential could lower the threshold and change the pattern of action potential firing, as has been shown in other cell types (Huguenard, 1996; Novara et al. 2004). Thus, by promoting the functional expression of T-type channels, NGF may induce a delayed increase in the firing frequency. In this way, an additional, indirect enhancement of Ca²⁺ influx through L-type channels would be obtained.

A positive effect of NGF on T-type Ca²⁺ current density was previously described in PC12 cells by Garber et al. (1989). To our knowledge, there are no other precedents for the long-term regulation of T-type channels by NGF shown here. It is interesting that GH₃ cells also respond with a persistent increase in T-type current to the chronic application of 17β-oestradiol (Ritchie, 1993) or insulin (Meza et al. 1994). Like NGF, these other agents stimulate prolactin synthesis by GH₃ cells (Stanley, 1988; Arroba et al. 2005). Conversely, when GH₃ cells are grown in the presence of glucocorticoids for some days, T-type current density (Meza et al. 1994) and the rate of prolactin synthesis (White et al. 1981) are both greatly reduced. Thus, in this system, T-type Ca²⁺ channels appear to be involved in the control of prolactin production by a number of external signals, including NGF.

The p75 receptor is critical for induction of the lactotrope phenotype by NGF

NGF initiates its biological actions by interacting with two types of cell surface receptors, TrkA and p75 (Chao, 2003), both of which are present in GH₃ cells (Missale et al. 1994) and prolactinoma cells (Fiorentini et al. 2002). The TrkA receptor has been implicated in the regulation of HVA Ca²⁺ channels by NGF in several cell types (Lei et al. 1998; Jia et al. 1999; Baldelli et al. 2000; Black et al. 2003). At variance with this, our data indicate that the effects of NGF on Ca²⁺ currents in GH₃ cells are mediated by p75 in a TrkA-independent way. Neither the early increase in the L-type current nor the late increase in the T-type current could be observed when NGF was applied in combination with MC192, an anti-p75 antibody that has proved to be a valuable tool to identify p75-mediated responses in rat neurons (Blochl & Sirrenberg, 1996; Numakawa et al. 2003; McCollum & Estus, 2004) and PC12 cells (Bui et al. 2002). By contrast, both increases persisted in the presence of K252a, a potent inhibitor of the tyrosine kinase activity of Trk receptors (Berg et al. 1992). Two other compounds that are widely used to block tyrosine kinases (lavendustin A and genistein) were also unable to prevent the NGF-induced increase in L-type current. As the effect of NGF on prolactin mRNA levels is linked to Ca²⁺ channel regulation, we conclude that p75 plays a crucial role in mediating the ability of this growth factor to stimulate prolactin synthesis by GH₃ cells. A p75-dependent pathway has been shown to be responsible for the rapid modulation of delayed rectifier K⁺ channels by NGF in rat sensory neurons (Zhang & Nicol, 2004). There are also some indications that the p75 receptor may support, in part, the acute effect of NGF on L-type Ca²⁺ current in PC12 cells (Jia et al. 1999). Our data, however, represent the first evidence that NGF binding to p75 can lead to both rapid and slow enhancements of Ca²⁺ currents in the same target cell without the involvement of TrkA.

The NGF-induced increase in the rate of prolactin production is known to reflect the conversion of GH₃ cells into lactotrope-like cells. The associated events include the suppression of GH synthesis and the activation of D₂ receptor expression (Missale et al. 1994; Missale & Spano, 1998). Since the GH gene has been found to be unresponsive to chronic changes in the external Ca²⁺ concentration (White et al. 1981), the inhibition of GH production by NGF is probably independent of Ca²⁺ channel regulation. On the other hand, NGF may induce expression of D₂ receptors in GH₃ cells by activating the nuclear factor-κB (NF-κB), as has been demonstrated by Fiorentini et al. (2002) in prolactinoma cells. Studies in neurons indicate that this transcription factor is controlled by [Ca²⁺]_i (West et al. 2002; Lilienbaum & Israel, 2003), and that Ca²⁺ influx through L-type channels is responsible for the basal NF-κB activity (Lilienbaum & Israel, 2003). Therefore, it seems likely that L-type channels are also involved in the induction of D₂ receptors by NGF. In line with this idea, preliminary data show that GH₃ cells express D₂ receptors in response to Bay K 8644 (Espinosa et al. 2004). Furthermore, as in other cell types (Chao, 2003), the activation of NF-κB in prolactinoma cells by NGF occurs via the p75 receptor (Fiorentini et al. 2002). These data, together with our present findings, argue in favour of a pivotal role for p75 in the induction of the lactotrope phenotype by NGF. More precisely, it can be suggested that NGF acts through the p75 receptor to enhance L-type and T-type Ca²⁺ currents, and that the consequent increase in Ca²⁺ entry not only up-regulates the prolactin mRNA, but also helps to inducing the appearance of D₂ receptors. The influence of Ca²⁺ influx on D₂ receptor expression, however, has yet to be firmly established. In addition, further work will be needed in order to define the steps linking p75 to Ca²⁺ channels.

Comparison with the effects of EGF and thyrotropin-releasing hormone

GH₃ cells also develop towards a lactotrope-like phenotype when grown in the presence of EGF or thyrotropin-releasing hormone (TRH) (Schonbrunn et al. 1980; Felix et al. 1995; Missale & Spano, 1998). Both EGF and TRH have been shown to stimulate prolactin synthesis at the transcriptional level in a Ca²⁺-dependent manner (White & Bancroft, 1983; Ramsdell & Tashjian, 1985; Laverriere et al. 1988). In this study, we confirmed that long-term exposure to EGF enhances prolactin gene expression in GH₃ cells. In addition, we found that nimodipine was unable to prevent the up-regulation of prolactin mRNA observed during EGF treatment. Clearly then, the action of EGF differed in L-type Ca²⁺ channel dependency from that of NGF. The long-term regulation of prolactin gene expression by TRH also persists in the presence of DHP blockers (Laverriere et al. 1989; Wang & Maurer, 1999). These findings raise the possibility that some DHP-insensitive Ca²⁺ channel type contributes to the effects of EGF and TRH on the prolactin gene. There are reasons to believe that the N-type channel, which is a member of the Ca_v2 channel family and can be selectively blocked by the peptide toxin ω-conotoxin GVIA (Perez-Reyes, 2003), could be a good candidate for this role. N-type Ca²⁺ channels are not usually expressed in GH₃ cells (Liévano et al. 1994) or GH₃B₆ cells (Glassmeier et al. 2001). It is most likely that they were also absent in NGF-treated GH₃ cells because NGF did not persistently increase the HVA Ca²⁺ current. In contrast, we have previously shown that chronic treatment of GH₃ cells with EGF causes a long-lasting increase in HVA Ca²⁺ current density (Meza et al. 1994; Cota et al. 1997). Indirect evidence suggests that this is due to the recruitment of a new set of Ca²⁺ channels that are sensitive to ω-conotoxin GVIA (Monjaraz et al. 1996). Similarly, it has been reported that the TRH-induced accumulation of prolactin mRNA can be partially decreased by ω-conotoxin GVIA with no change in the basal prolactin mRNA levels (Laverriere et al. 1989; Wang & Maurer, 1999). Thus, unlike NGF, EGF and TRH may activate the expression of N-type Ca²⁺ channels in GH₃ cells. More generally, our results and these previous observations suggest that the pathways for Ca²⁺ influx involved in the induction of the lactotrope phenotype by EGF and TRH are not identical to those used by NGF.

Conclusion

We have found that NGF can act via the p75 receptor to positively affect L-type and T-type Ca²⁺ currents in GH₃ cells. This novel ability of NGF is likely to result in a prolonged increase in Ca²⁺ entry through L-type channels during the spontaneous spiking activity of the cells. In turn, the increased Ca²⁺ influx enhances prolactin gene expression, as shown by the measures of prolactin mRNA levels. These findings greatly improve our knowledge about the mechanism of action of NGF as a promoter of lactotrope cell differentiation.