Deletion of CASK in mice is lethal and impairs synaptic function

Abstract

CASK is an evolutionarily conserved multidomain protein composed of an N-terminal Ca2+/calmodulin-kinase domain, central PDZ and SH3 domains, and a C-terminal guanylate kinase domain. Many potential activities for CASK have been suggested, including functions in scaffolding the synapse, in organizing ion channels, and in regulating neuronal gene transcription. To better define the physiological importance of CASK, we have now analyzed CASK "knockdown" mice in which CASK expression was suppressed by approximately 70%, and CASK knockout (KO) mice, in which CASK expression was abolished. CASK knockdown mice are viable but smaller than WT mice, whereas CASK KO mice die at first day after birth. CASK KO mice exhibit no major developmental abnormalities apart from a partially penetrant cleft palate syndrome. In CASK-deficient neurons, the levels of the CASK-interacting proteins Mints, Veli/Mals, and neurexins are decreased, whereas the level of neuroligin 1 (which binds to neurexins that in turn bind to CASK) is increased. Neurons lacking CASK display overall normal electrical properties and form ultrastructurally normal synapses. However, glutamatergic spontaneous synaptic release events are increased, and GABAergic synaptic release events are decreased in CASK-deficient neurons. In contrast to spontaneous neurotransmitter release, evoked release exhibited no major changes. Our data suggest that CASK, the only member of the membrane-associated guanylate kinase protein family that contains a Ca2+/calmodulin-dependent kinase domain, is required for mouse survival and performs a selectively essential function without being in itself required for core activities of neurons, such as membrane excitability, Ca2+-triggered presynaptic release, or postsynaptic receptor functions.

Fig. 1.

Gene targeting strategy for CASK: protein levels in knockin and KO mice. (A) Targeting strategy. Genomic clones containing the 5′ end of the murine CASK gene (top diagram) were used to construct a targeting vector in which the first coding exon is flanked by loxP sites (triangles) and a neomycin resistance gene cassette (Neo) is inserted into the intron following the exon. In addition, a diphtheria toxin gene (DT) is attached to the long arm of the vector. Homologous recombination of the targeting vector with the endogenous CASK gene replaces the endogenous with the modified genomic sequence and eliminates the diphtheria toxin gene. The resulting “floxed allele” retains the exon in the knockin mice but includes a neomycin resistance gene cassette in the intron, both of which are excised by cre recombinase in the KO. The scale bar on the lower right applies to all panels. (B) Immunoblotting analysis of CASK, Velis, and GDI (GDP dissociation inhibitor) in WT (+/+), heterozygous floxed mutant (+/flox), and homozygous floxed mutant (flox/flox) mice. (C) Immunoblotting analysis of CASK and selected proteins in WT and homozygous KO mice. (D) Quantitation of CASK levels in knockin and KO mice and of various indicated proteins in KO mice. Protein levels were measured by using quantitative immunoblotting with ¹²⁵I-labeled secondary antibodies and phosphoimager detection. Samples were derived from littermate offspring of heterozygous matings.

Fig. 2.

Cleft palate and increased cell death in CASK KO mice. (A) Sagittal section of WT and CASK KO mice at first day of birth (P1) stained with hematoxylin and eosin. Shown are transverse sections through the heads of one WT and two different CASK KO mice. Note the cleft palate in the CASK KO animals (asterisk). (B) Thalamic slices of littermate WT and CASK KO mice at day P1 stained with TUNEL. The scale bar applies to both panels. (C) Quantitation of TUNEL-positive cells WT and CASK KO thalamic slices. Note the 3-fold increase in TUNEL-reacting cells in CASK KO slices.

Fig. 3.

Electrical properties of CASK KO neurons. (A) Comparative analysis of membrane conductance in neurons at 14 days in vitro from littermate WT and CASK-deficient mice. Neurons were examined in current-clamp mode in the presence of 1 μM tetrodotoxin (mean input resistance: 357.08 ± 19.4 MΩ). The neuronal membrane potential was measured in response to 200-ms current injections, with an 800-ms interval between current injections. The graph plots the membrane potential as a function of injected current; in coincident values for WT and KO neurons, the symbol for the KO neuron is superposed on the symbol for the WT neuron (n = 9 mice used for cultures). (B and C) Analysis of action potential generation in WT and CASK-deficient neurons. Neurons held in current-clamp mode were injected with 200-ms pulses of current in the absence of tetrodotoxin, and the amplitude and firing threshold of the resulting action potentials were analyzed. B shows representative recordings from a KO neuron, and C shows summary graphs from WT and KO neurons (n = 8 and 5 independent cultures, respectively; data are not corrected for junction potential). (D) Representative traces of HVA Ca²⁺ currents evoked by a step depolarization from −70 mV to 0 mV in brainstem pre-Bötzinger complex neurons of WT and CASK KO mice. (E) Summary graph depicting average HVA Ca²⁺ current densities.

Fig. 4.

Analysis of spontaneous release events in CASK KO mice. (A) Representative traces from recordings of the spontaneous miniature synaptic events (minis). Cultured cortical neurons at 14 days in vitro were analyzed in voltage-clamp mode in the presence of 1 μM tetrodotoxin. Glutamatergic and GABAergic responses were examined separately upon addition of 50 μM picrotoxin or 10 μM CNQX and 50 μM AP-5, respectively. The box at the bottom displays a representative single event at high resolution. (B and C) Comparison of the minifrequencies (B) and miniamplitudes (C) for glutamatergic and GABAergic minievents in WT and CASK-deficient neurons (means ± SEMs; n = 3 independent cultures for glutamatergic, n = 15 neurons for WT, and n = 19 neurons for KO; n = 4 independent cultures GABAergic, n = 33 neurons for WT, and n = 35 neurons for KO). Statistical significance was assessed in pairwise comparisons by using Student's t test.

Fig. 5.

Evoked synaptic responses in CASK KO mice. Whole-cell recordings in voltage-clamp mode were obtained from cultured cortical neurons; responses were triggered by field stimulation. (A–F) Amplitudes of evoked synaptic responses in WT and CASK-deficient neurons. Synaptic responses to isolated action potentials were measured in cultured cortical neurons (14 days in vitro) in voltage-clamp mode by whole-cell recordings. Responses were monitored at a holding potential of −70 mV in the absence of receptor blockers (A and B show total responses; n = 83 CASK KO and n = 72 WT neurons), in the presence of 10 μM CNQX and 50 μM AP-5 (C and D show glutamatergic responses; n = 12 CASK KO and n = 11 WT neurons), or in the presence of 50 μM picrotoxin (E and F show GABAergic responses; n = 11 CASK KO and WT neurons). (G and H) NMDA receptor-dependent responses were recorded from a holding potential of +40 mV in 50 μM picrotoxin (H; n = 11 CASK KO and WT neurons). (I) Paired-pulse facilitation. Shown is a summary graph [size of the second response divided by the size of the first response to two closely spaced stimuli (paired-pulse ratio); n = 24 KO and n = 23 WT neurons]. All data shown are means ± SEMs.