Cannabinoid overrides triggers of GABAergic plasticity in vestibular circuits and distorts the development of navigation

Abstract

Early life exposure to cannabis can result in long-lasting deficits in spatial navigation. We ask if the development of this behavior is subject to early life activity of type I cannabinoid receptor (CB1R) in the vestibular nucleus. In rodents, we found that local exposure to CB1R agonist within the first postnatal week, but not thereafter, led to a decline in the induction efficacy of long-term depression at GABAergic synapses (LTD_GABA), a key step in the hard-wiring of vestibular circuits. Within this critical period, endocannabinoid-mediated LTD_GABA at inhibitory neurons was selectively triggered by cholecystokinin, whereas that at excitatory neurons was by serotonin. Neonatal exposure to cannabinoids extended the phase of high GABAergic synaptic plasticity and overrode the synapse-specific, modulatory mechanism for plasticity. Such treatment delayed the postnatal emergence of vestibular-dependent reflexes and deranged adult navigational behavior. Deficits in higher functions are thus attributable to the maldevelopment of sensory processing circuits resulting from early cannabis exposure.

Keywords: Biological sciences; Natural sciences; Neuroscience.

Figure 1

eCB shifts the maturation program of VN circuits through tuning synaptic plasticity at GABAergic synapses (A) The emergence of negative geotaxis (A₁) and air-righting reflex (A₂) were delayed with the pretreatment of the VN at P1 with CB1R agonist WIN55 or GABA_AR antagonist bicuculline, but advanced with CB1R antagonist AM251 or CaN-BP. (B) Gramicidin-perforated whole-cell patch-clamp recording of GABA reversal potential. Postnatal shift of I-V curve indicated that GABAergic response of MVN neurons changed from excitatory to inhibitory between P9 and P12. Vertical axis intersects the horizontal axis at −70 mV (average resting membrane potential). (C₁) Average response of MVN neurons to TBS in brain slices obtained from P5-8, P9-11 and P5-8 treated with CaN-BP. (C₂) Following TBS, mPSC_GABA frequency was increased (n = 19 cells, p < 0.001) but amplitude was unchanged (n = 19 cells, p > 0.05) and PPR was decreased (n = 17 cells, p < 0.01), indicating presynaptic plasticity. Lines join PPR values from the same cell. Red line indicates change in mean PPR value before and after TBS. Representative tracings of paired-pulse responses before (black) and after TBS (green) are shown above. (C₃) Top: Normalized PSC amplitudes of each recorded cell after TBS under various experimental conditions. Pale gray area denotes no change in PSC_GABA amplitude, the area below for LTD in PSC amplitude, and that above for LTP response. Bottom: Bar chart summarizing the percentage of MVN neurons showing LTD (light gray), LTP (white) or no change in PSC amplitude (dark gray). In P5-8 brain slices, 80% of sampled MVN neurons exhibited TBS-induced LTD_GABA, while the remaining 20% showed no change. LTP was not observed. With the bath addition of CaN-BP, 65% of sampled MVN neurons showed no response to TBS (red circles in C₁, p < 0.001, vs. untreated P5–8 cells). In P9–11 slices, the occurrence of LTD decreased to 29% while another 32% of sampled MVN neurons exhibited LTP (p < 0.001 vs. untreated P5–8 cells). (D) LTD was abolished in the averaged PSC response of all recorded MVN cells in P5–8 brainstem slices after blocking lipase activity with the bath addition of Orlistat. Closer examination of the response of each cell further revealed that the proportion of LTD_GABA-expressing MVN neurons was significantly decreased to 38% (p = 0.008, vs. untreated P5–8 cells, see C₃). LTP was induced in another 22% of sampled MVN neurons. (E) With the bath addition of JZL184 to P9–11 brainstem slices to inhibit 2-AG degradation, the percentage of MVN neurons that showed LTD_GABA increased to 78% (p < 0.001 vs. untreated P9–11 cells). Mean ± SEM are shown. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.01. two-way ANOVA for behavioral tests in (A), pair t-test for mPSC_GABA recordings in (C₂), Fisher’s exact test with Bonferroni’s correction for multiple measurements was used to evaluate change in responses to TBS in (C₃).

Figure 2

eCB system governs the induction efficacy of LTD_GABA and sets the developmental trajectory of LTD_GABA in developing VN neurons (A) In P5–8 brain slices, MVN neurons that responded to an initial TBS with LTD_GABA could no longer respond to a second TBS with further LTDGABA after the bath addition of AM251 (purple triangles). In a parallel preparation of P5–8 slices (red circles), a second TBS could induce further LTD_GABA in LTD_GABA-expressing MVN neurons. (B) Bar charts summarizing the individual PSC_GABA (top) and the percentages (bottom) of P5–8 MVN neurons to TBS after sham or P1 a.m.251 application, as well as the percentage of cells showing each type of response, viz. LTD (light gray), LTP (white) or no change in PSC amplitude (dark gray). (C) In P5–8 rats pre-treated with AM251 at P1, 74% of sampled MVN neurons in brain slices did not exhibit LTD_GABA following an initial TBS (purple triangles), contrasting sham control rats in which 79% of sampled neurons showed LTD_GABA (green circles, p < 0.001). With the bath addition of JZL184, a second TBS could induce LTD_GABA in 80% of sampled MVN neurons that were non- LTD_GABA-expressing with the first TBS (purple triangles). (D) In P9–11 slices, all sampled MVN neurons that did not respond to an initial TBS could then be induced to express LTD_GABA with a second TBS after the bath addition of WIN55. (E) Bar charts summarizing the individual PSC_GABA (top) and the percentages (bottom) of P9–11 MVN neurons to TBS after sham or P1 WIN55. (F) In P9–11 rats pre-treated with WIN55 (blue squares) at P1, 86% of sampled MV neurons in brain slices exhibited LTD_GABA after the first TBS, contrasting sham control rats in which only 23% of sampled MVN neurons showed LTD_GABA (p < 0.001). These LTD_GABA-expressing cells could not respond to a second TBS with further LTD_GABA after the addition of Orlistat to the bath. Mean ± SEM are shown. Fisher’s exact test Bonferroni’s correction for multiple measurements was used to evaluate change in responses to TBS in (B) and (E).

Figure 3

eCB release is triggered by 5-HT and CCK receptors but not metabotropic glutamate receptors in developing VN neurons (A₁) With the bath addition of MDL11939 (a 5-HT receptor antagonist) to P5-8 brain slices, the percentage of P5–8 MVN neurons showing LTD_GABA decreased to 30% (green triangles, p = 0.003), contrasting 79% in controls. (A₂) With the bath addition of CI988 (a CCKBR antagonist) the percentage of P5–8 MVN neurons showing LTD_GABA decrease slightly to 50% (CI988, blue triangles), but was not statistically significant (p = 0.096). (A₃) With the bath addition of LY367385 (a mGLuR1 receptor antagonist, black circles) or MPEP (a mGluR5 antagonist, pink triangles), the percentage of P5–8 MVN neurons showing LTD_GABA was similar to the control (LY367385: 73%, p = 1; MPEP: 62%, p = 0.347). (A₄) Bar charts summarizing the individual PSC_GABA (top) and the percentages (bottom) of P5-8 MVN neurons showing each type of response (LTD, light gray; LTP, white; no change, dark gray) in each treatment group after TBS. (B₁) In P9–11 brain slices, no LTD or LTP was observed in the averaged response of recorded MVN cells (green circles). This was in agreement with the majority (61%) of cells showing no change after TBS (see B₄). With the bath addition of α-m-5-HT (a 5-HT receptor agonist), the percentage of MVN neurons showing LTD_GABA after an initial TBS increased to 78% (yellow triangles, p < 0.001). These LTD_GABA-expressing cells could not respond to a second TBS with further LTD_GABA with the addition of Orlistat to the bath. (B₂) With the bath addition of CCK (a CCKBR agonist), 57% of the P9–11 MVN neurons showed LTD_GABA (purple triangles, p = 0.011), contrasting control rats in which only 11% of sampled MVN neurons showed LTD_GABA. (B₃) With the bath addition of DHPG (a mGluR1 agonist), the percentage of MVN neurons showing LTD_GABA after TBS did not change significantly (27%, blue circles, p = 0.484). (B₄) Bar charts summarizing the individual PSC_GABA (top) and the percentages (bottom) of P9–11 MVN neurons in each treatment group after TBS. Mean ± SEM are shown. Fisher’s exact test with Bonferroni’s correction for multiple measurements was used to evaluate change in responses to TBS in (A₄) and (B₄).

Figure 4

Distinct triggers for eCB-mediated LTD_GABA at inhibitory and excitatory neurons in VN (A) In P5–8 VGAT mice (filled circles), the majority of sampled VGAT (A₁, 40%) and non-VGAT neurons (A₂, 80%) in the MVN showed LTD_GABA after TBS. While at P9–17 (open circles), the majority of neurons showing no change in response amplitude after TBS (A₂). (B) Cnr1^−/− mice had higher percentage of LTP-expressing MVN neurons at both P5–8 (filled triangles, p = 0.001), but the difference was no longer significant at P9–17 (open circles, p = 0.067) compared with VGAT-Venus mice. (C) In P5–8 mice, LTD_GABA-expressing neurons in both VGAT (C₁) and non-VGAT (C₂) populations could not be induced to express LTD_GABA with a second TBS after the bath addition of Orlistat. (D) Bath addition of MDL11939 (a 5HT_2AR antagonist) to P5–8 brain slices did not affect LTD_GABA in VGAT neurons in the MVN (p = 0.560) (D₁). However, the percentage of LTD-expressing non-VGAT cells was significantly decreased to 17% (p = 0.008) after the bath addition of MDL11939 (D₂). (E₁) With the bath addition of CI988 (a CCK_BR antagonist) to P5–8 slices, the percentage of LTD_GABA-expressing VGAT neurons in the MVN decreased to 50% (p = 0.009 vs. P5–8 control). (E₂) Such treatment did not significantly reduce the percentage of LTD_GABA-expressing non-VGAT neurons in the MVN (70%, p = 0.293). (F₁) With the bath addition of CCK to P9–17 slices, the percentage of LTD_GABA-expressing VGAT neurons in the MVN was increased to 73% (p = 0.004 vs. P9–17 control). (F₂) On the other hand, the percentage of non-VGAT neurons responding to TBS with LTD_GABA remained similar to controls at 31% (p = 0.857). (G_1,2) Bar charts summarizing PSC_GABA amplitudes of each recorded cell and percentages of MVN neurons showing the response described above for P5–8 (F₁) and P9–17 (F₂) VGAT-Venus and Cnr1^−/− mice. (H) Schematic diagram showing differential control of eCB-mediated LTD_GABA by CCK_BR and 5HT_2AR at VGAT and non-VGAT neurons in the P5–8 and P9–17 MVN. Mean ± SEM are shown. Fischer’s exact test with Bonferroni’s correction for multiple measurements was used to compared response profiles to TBS between various treatment groups.

Figure 5

Use of Cnr1^−/− to show that GABA release from VN neurons is limited by CB1R. mPSC_GABA of MVN cells from control, VGAT and Cnr1^−/− transgenic mice of two age groups, viz. P5–18 and P9–17 (A₁ and B₁) Upper panels: Representative tracings of mPSC_GABA from P7 (A₁) and P14 (B₁) neurons voltage clamped at −70 mV. Lower panels: Cumulative distribution of amplitude and inter-event interval of mPSC_GABA from control (black line) and Cnr1^−/− mice (red line). (A₂ and B₂) The mPSC_GABA amplitude of MVN neurons was similar between all groups at both ages, but the mPSC_GABA frequency and PPR of MVN neurons was significantly increased in Cnr1^−/−. This suggested a presynaptic site of action for Cnr1^−/−. Decay time of mPSC_GABA was significantly shorter in Cnr1^−/− mice of the P5–8 group but was not different at P9–17 group. Mean ± SEM are shown. ∗p < 0.05, ∗∗∗∗p < 0.0001. K-S test to compare cumulative distributions in (A₁) and (B₁). two-way ANOVA for comparison of mPSC_GABA properties in (A₂) and (B₂).

Figure 6

Perturbation of eCB signaling in the neonatal VN leads to long-lasting deficits in vestibular-dependent navigation (A) Excursion paths of adult rats pretreated at P1 with sham implantation, AM251, WIN55, or CaN-BP in the light, dark and new location test for spatial reckoning. Red lines represent the trajectories of homeward paths (light/dark test: 8 trails/rat are superimposed; new location: 4 trails/rat are superimposed). Filled black circle on the edge of each round table surface represents the location of the home base; green circles in bottom panels represent the location of the old home base. Histograms showing the average heading angle (B), time spent in the quadrant containing food (C), errors made in locating the homebase (D), and training time required for rats pretreated with AM251 (n = 6 rats), WIN55 (n = 6 rats), CaN-BP (n = 5 rats), or sham operated (n = 5 rats) at P1 or P12 (E). Means ± SEM are shown. ∗p < 0.05, ∗∗p < 0.01. two-way ANOVA for comparison of behavioral test performance in (B–E).