Non-cell-autonomous mechanism of activity-dependent neurotransmitter switching

Abstract

Activity-dependent neurotransmitter switching engages genetic programs regulating transmitter synthesis, but the mechanism by which activity is transduced is unknown. We suppressed activity in single neurons in the embryonic spinal cord to determine whether glutamate-gamma-aminobutyric acid (GABA) switching is cell autonomous. Transmitter respecification did not occur, suggesting that it is homeostatically regulated by the level of activity in surrounding neurons. Graded increase in the number of silenced neurons in cultures led to graded decrease in the number of neurons expressing GABA, supporting non-cell-autonomous transmitter switching. We found that brain-derived neurotrophic factor (BDNF) is expressed in the spinal cord during the period of transmitter respecification and that spike activity causes release of BDNF. Activation of TrkB receptors triggers a signaling cascade involving JNK-mediated activation of cJun that regulates tlx3, a glutamate/GABA selector gene, accounting for calcium-spike BDNF-dependent transmitter switching. Our findings identify a molecular mechanism for activity-dependent respecification of neurotransmitter phenotype in developing spinal neurons.

Figure 1. Misexpression of hKir2.1 suppresses Ca²⁺ spike generation in single neurons in situ without changing the identity of their neurotransmitter

(A) Confocal image of a stage 25 ventral spinal cord. Dashed oval outlines a hKir2.1-mCherry/Fluo-4 AM-expressing neuron (red and yellow). Neurons loaded with Fluo-4 AM (green) are internal controls. (B) Spike activity is absent in these hKir2.1-mCherry-expressing neurons and present in internal controls. Ca²⁺ spikes were identified as fluorescence transients greater than 20% of ΔF/F₀ (dashed lines), more than twice the standard deviation of the baseline. (C) Spike incidence during one hr imaging periods. n≥7 stage 23–25 embryos per group; 37–63 neurons were analyzed per group. Kir-mCh cells, hKir2.1-mCherry-expressing neurons; mCh cells, mCherry alone-expressing neurons. (D,E) VGluT1 and GABA staining and quantitative analysis of stage 41 larvae from Kir-mCherry DNA- and mCherry DNA-injected embryos. Arrowheads identify a VGluT1-immunoreactive and a GABA-immunoreactive neuron. n≥7 embryos per group; 36–58 neurons were analyzed for each group. VGluT1/mCherry and GABA/mCherry represent the fraction of mCherry-labeled neurons that are VGluT1-immunoreactive and GABA-immunoreactive. ***, p<0.001; ns, not significant. Mann-Whitney U test. See also Figure S1.

Figure 2. Increasing the number of silenced neurons in vitro decreases the number of both silenced and active GABA-IR neurons equally

(A–B) A representative view of active (wild-type) and silenced (hKir2.1-mCherry labeled) neurons in a dissociated 50% silenced neuron-enriched culture. (A) High magnification views of individual active or silenced neurons. (B) Low magnification view showing 2 active and 2 silenced neurons (arrowheads), one of which is GABA-IR+. hKir2.1-mCherry: red; GABA-IR/hKir2.1-mCherry: yellow. (C) The percent of neurons expressing GABA-IR depends on the percent of silenced neurons expressing hKir2.1-mCherry. Black: total number of GABA-IR neurons/total number of neurons. Red: number of GABA-IR Kir+ neurons/number of Kir+ neurons. Grey: number of GABA-IR Kir− neurons/number of Kir− neurons. Blue: Total GABA-IR% of 100% silenced cultures with BDNF addition. Simple linear regression analysis was performed for the first 3 groups. Total: slope= −0.19 ± 0.02, significantly non-zero (p<0.0001); Kir+: slope= −0.15 ± 0.03, significantly non-zero (p<0.0001); Kir-: slope= −0.14 ± 0.04, significantly non-zero (p=0.002). ANOVA. The slopes for active (Kir-) neurons, silenced (Kir+) neurons, and the total population are not significantly different. ANOVA. (D) Bars represent the percent of neurons expressing GABA-IR from 6 groups (0, 25, 50, 75, 100% silenced and 100% silenced plus BDNF). Cultures with more than 35 neurons post-staining were selected, and n=4–8 cultures per group were analyzed. Data are mean±SEM. *, p<0.05; ***, p<0.001; ****, p<0.0001. One-way ANOVA Tukey’s multiple comparisons test.

Figure 3. BDNF is expressed throughout the critical period for neurotransmitter specification and release depends on spontaneous Ca²⁺ spike activity

(A) BDNF transcript expression in embryos at stages 18, 24 and 28. (B) BDNF release by Ca²⁺ spike activity. Left: superimposed fluorescence and bright field images of a neuron and muscle and undifferentiated cells loaded with BDNF-pHluorin before and after depolarization with 2 mM Ca²⁺ + 100 mM KCl (2CaKCl). Right: representative trace showing time course of cell body fluorescence changes of a neuron and an undifferentiated cell. The culture was first depolarized with KCl in the absence of Ca²⁺ as control (0CaKCl), and then depolarized 3 times in the presence of Ca²⁺ (2CaKCl). Each 30s depolarization, mimicking a Ca²⁺ spike, causes a decrease in BDNF-pHluorin fluorescence selectively in the neuron in the presence of Ca²⁺. Arrows indicate the start of perfusion: 0Ca, 0 mM Ca²⁺ + 0.67 mM KCl; 0CaKCl, 0 mM Ca²⁺ + 100 mM KCl; 2Ca, 2 mM Ca²⁺ + 0.67 mM KCl; 2CaKCl, 2 mM Ca²⁺ + 100 mM KCl. (C) Quantification of the decrease of BDNF-pHluorin fluorescence in response to 2CaKCl. n=5 neurons from 5 independent experiments. Data are mean±SEM. **, p<0.01. Mann-Whitney U Test. (D) BDNF protein in culture medium in the absence (Kir) or in the presence (control) of spontaneous activity. Cultures were prepared from embryos injected either with hKir2.1 mRNA and Cascade Blue dextran or Cascade Blue dextran alone (control). Medium was collected at the indicated times and released BDNF levels (pg/ml) were measured with a conventional two-site ELISA system. n ≥4 cultures per condition; ≥40 neurons/culture. Kir, hKir2.1 misexpression. Data are mean±SEM. *, p<0.05. Mann-Whitney U test.

Figure 4. BDNF regulates neurotransmitter phenotype downstream of Ca²⁺ spike activity

(A) Embryos were electroporated with lissamine-tagged BDNF MO or CMO. Lissamine distribution (red) in a live stage 41 larva demonstrates targeting of the spinal cord. (B) Single agarose beads (blue) loaded with BDNF, K252a or K252b were implanted adjacent to the nascent neural tube (white dashed circle) at stage 19 and stage 41 larvae were sectioned for immunocytochemistry. (C–E) GABA (C,D) and glutamate (C,E) staining following implantation of beads containing BDNF (100 ng/ml), K252a or K252b (50 μM), a membrane impermeable analog of K252a, or BDNF MO and CMO electroporation. Beads were located adjacent to the most rostral 100 μm of the spinal cord. 101 to 195 neurons were analyzed per condition per 200 μm. (F,G) GABA and glutamate expression in embryos in the presence of BDNF, in embryos misexpressing hKir2.1, and in embryos misxpressing hKir2.1 in the presence of BDNF. C–G, n≥5 embryos per condition. 128–255 neurons were analyzed per condition per 200 μm. Data are mean±SEM. *, p<0.05; **, p<0.01. ns, not significant. ANOVA test with Bonferroni post-hoc analysis. See also Figure S2.

Figure 5. BDNF regulates cJun phosphorylation

(A) P-cJun expression in the presence or absence of hKir2.1 expression, BDNF (100 ng/ml) or Trk receptor blocker (50 μM). Protein levels in neuronal cultures were determined by Western blot. Graph presents the ratio between the phosphorylated protein and total protein expressed in optical density units (O.D.U). n≥5 cultures per condition. Kir, hKir2.1 misexpression. Full-length Western blot is shown in Figure S5. (B) P-cJun and TrkB staining of embryos (stage 28) showing colocalization. n ≥5 embryos per condition. 132 to 151 neurons were analyzed per condition per 100 μm. (C) P-cJun and (D) cJun staining of embryos (stage 28) in the presence of control or BDNF beads or electroporated with BDNF MO or CMO. n≥5 embryos per condition. 137 to 158 neurons were analyzed per condition per 100 μm. Data are mean±SEM. *, p<0.05; **, p<0.01. Mann-Whitney U test.

Figure 6. Signaling pathway regulating BDNF-dependent cJun phosphorylation

(A) P-cJun expression in the presence or absence of BDNF (100 ng/ml) and inhibitors of ERK1/2 (MEK, U0126), JNK (SP600125) or Trk receptors (K252a) (all 10 μM). Graph as in Figure 4. n≥5 cultures per condition; ≥40 neurons/culture. Data are mean±SEM. *, p<0.05. ANOVA test with Bonferroni post-hoc analysis. Full-length Western blot is shown in Figure S5. (B,C) P-JNK and JNK staining of embryos in the presence of control or BDNF beads or electroporated with BDNF MO or CMO. n≥5 stage 28 embryos per condition; 137 to 151 neurons were analyzed per condition per 100 μm. Data are mean±SEM. *, p<0.05. Mann-Whitney U test.

Figure 7. JNK regulates cJun phosphorylation and neurotransmitter phenotype

(A,B) P-cJun and cJun staining of embryos (stage 28) electroporated with JNK MO or CMO. 128–152 neurons were analyzed per condition per 100 μm. (C,D) GABA and glutamate staining of embryos (stage 41) following JNK MO and CMO electroporation. n ≥5 embryos per condition. 152–215 neurons were analyzed per condition per 200 μm. Data are mean±SEM. *, p<0.05. Mann-Whitney U test.

Figure 8. Model of non-cell-autonomous regulation of transmitter expression

Transmitter identity is initially specified genetically and then modulated by environmental influences through changes in spontaneous Ca²⁺-dependent activity. Ca²⁺ spikes regulate the release of BDNF that initiates the TrkB/MAPK signaling cascade. JNK-mediated phosphorylation of cJun regulates transcription to select neurotransmitter fate.