Neurogenic role of the depolarizing chloride gradient revealed by global overexpression of KCC2 from the onset of development

Abstract

GABA- and glycine-induced depolarization is thought to provide important developmental signals, but the role of the underlying chloride gradient has not been examined from the onset of development. We therefore overexpressed globally the potassium-chloride cotransporter 2 (KCC2) in newly fertilized zebrafish embryos to reverse the chloride gradient. This rendered glycine hyperpolarizing in all neurons, tested at the time that motor behaviors (but not native KCC2) first appear. KCC2 overexpression resulted in fewer mature spontaneously active spinal neurons, more immature silent neurons, and disrupted motor activity. We observed fewer motoneurons and interneurons, a reduction in the elaboration of axonal tracts, and smaller brains and spinal cords. However, we observed no increased apoptosis and a normal complement of sensory neurons, glia, and progenitors. These results suggest that chloride-mediated excitation plays a crucial role in promoting neurogenesis from the earliest stages of embryonic development.

Figure 1.

Zebrafish zfkcc2 expression is absent in embryonic neurons. A, RT-PCR using zebrafish total RNA from animals aged 1–5 dpf, 10 dpf, and 1 and 6 months postfertilization, and 1 year old brain tissue. The top band represents zfkcc2 mRNA (778 bp), and the bottom band corresponds to the zfnkcc1 mRNA (413 bp). zfkcc2 is initially absent (at 1 dpf) and appears at 2 dpf, whereas zfnkcc1 mRNA is detectable at all ages tested. B, Activity of the wild-type KCC2, the EGFP-tagged KCC2, and the C568A mutation. Xenopus oocytes were transfected with cRNA from the three isoforms or with water. Activity of the cotransporters was assessed in isotonic conditions, in both the presence (Control) and absence (−Cl⁻) of chloride, by measuring oocyte uptake of ⁸⁶rubidium⁺.

Figure 2.

KCC2 overexpression reverses the depolarizing chloride gradient. A, EGFP fluorescence of spinal cord neurons from EGFP–KCC2-overexpressing embryos. Note that the expression is localized to the surface and that all neurons express the cotransporter. B, Pie charts illustrating the proportions of silent neurons (white) associated with immaturity and spontaneously active neurons (black) in control, KCC2-, and KCC2–C568A-overexpressing embryos. The diagrams illustrate the experimental paradigm. C, Representative traces of spontaneous activity recorded from neurons in control, KCC2-, and KCC2–C568A-overexpressing embryos. In each case, examples of PDs, which are gap junction mediated, are highlighted by gray and include a large depolarization followed by afterhyperpolarization. Examples of SBs, which are chloride mediated, are highlighted in blue. Note that, in neurons from control and KCC2–C568A embryos, SBs are depolarizing from rest (−40 mV), although they are hyperpolarizing from rest in neurons from KCC2 embryos, as illustrated by extended superimposed SBs traces (Ci, taken from shaded examples in C, −40 mV traces). Cii, The reversal potential of SBs (taken from the examples in C) in control (black circles) and KCC2–C568A neurons (blue squares) was approximately −30 mV [above resting potential (Rp)], whereas that of the KCC2-overexpressing neuron (red squares) was more negative (−47 mV; below resting potential). Ciii, The average driving force calculated for each group is positive for control and KCC2–C568A-overexpressing embryos, whereas that of KCC2 embryos is negative, suggesting a reversal of the depolarizing chloride gradient in KCC2 neurons. D, To verify whether KCC2 cotransporter was functionally expressed in silent neurons, we compared the change in resting potential (Δ resting potential) in response to application of bumetanide, a more specific blocker of NKCC1 cotransporter, with that of furosemide and DIOA, which are more specific blockers of KCC2. In control neurons, as expected in the absence of KCC2, furosemide and DIOA did not affect the resting potential, whereas bumetanide shifted it to more positive values (bumetanide control vs bumetanide KCC2, Student's t test, p = 0.2). In contrast, in KCC2-overexpressing embryos, furosemide and DIOA each caused a negative shift in the resting potential (furosemide or DIOA in control vs furosemide or DIOA, in KCC2-overexpressing embryos, p < 0.05, Student's t test). In KCC2–C568A embryos, such a negative shift was not observed.

Figure 3.

KCC2 overexpression alters the gross morphology and the early swimming behavior of the embryo. A, Morphological phenotype of control, KCC2-, and KCC2–C568A-injected embryos. B, Response to touch of control, KCC2-, and KCC2–C568A-overexpressing embryos, whose head was immobilized in agarose. To elicit an escape response, the freely moving tip of the tail was gently touched using a pair of forceps. Note the different timescales in control and KCC2–C568A embryos, which responded with rapid alternating contractions, versus KCC2 embryos, which produced slow coils in response to the stimulus.

Figure 4.

KCC2 overexpression impairs the growth and maturation of brain and hindbrain structures. A, In control embryos, the acetylated tubulin staining highlights the postoptic, anterior, and posterior commissures (open arrowheads), the ventral and dorsal longitudinal tracts (filled arrowheads), the trigeminal ganglion (open arrow), and the pigmented optic cup forming the retina (filled arrow). In contrast, only the ventral longitudinal tract (filled arrowhead) and the optic vesicle (filled arrow) can be seen in KCC2-overexpressing embryos. B, Top-down view of the large Mauthner cells, which are easily discernable in control embryos. When KCC2 is overexpressed, the Mauthner cells appear less well defined but their axons cross the midline and project caudally as in control embryos. C, The hindbrain reticulospinal neurons (side view) are present in KCC2 embryos. However, their axon projections to the spinal cord (top-down view) are much less elaborated in KCC2-overexpressing embryos.

Figure 5.

KCC2 overexpression perturbs axonal maturation and neurogenesis in the zebrafish spinal cord. A, Acetylated tubulin staining highlights the impairment in maturation and elaboration of axonal networks in the spinal cord during KCC2 overexpression. The arrowheads point to motoneuron axons that failed to extend completely out of the spinal cord. B, Anti-Hu staining shows that KCC2-overexpressing embryos possess less newly born spinal neurons. When looking at specific neuronal types using post-differentiation markers, we observed decreased populations of interneurons expressing Pax 2 (C) and motoneurons expressing HB9 (D). However, sensory neurons expressing NGN1 (E) were not affected by the overexpression of KCC2.

Figure 6.

Effects of KCC2 overexpression on apoptosis proliferation and differentiation in the zebrafish spinal cord. A, There is no obvious change in the levels of neurodegeneration during KCC2 overexpression because acridine orange-labeled apoptotic cells are not more abundant in the spinal cord of KCC2 embryos. Proliferation was not drastically modified by treatment because the number of BrdU-labeled cells (B) and the number of phospho-histone-3-labeled cells (C) is not significantly different between KCC2 and control embryos. In contrast, (D) neural differentiation was compromised in KCC2 embryos because a subset of committed progenitors marked with Dbx1 was significantly fewer in the spinal cord of these embryos. E, Zrf-1 staining of radial glial cells in the spinal cord of control and KCC2 embryos (n = 26 and n = 14, respectively). F, Fluorescent oligodendrocyte precursors in the spinal cord of PLP–GFP transgenics in non-injected zebrafish and zebrafish overexpressing KCC2 (n = 9 and n = 9, respectively). The arrowheads point to such precursors, which were hard to distinguish at this stage of development in the spinal cord.