Increased Grik4 gene dosage causes imbalanced circuit output and human disease-related behaviors

Article Increased Grik4 Gene Dosage Causes Imbalanced Circuit Output and Human Disease-Related Behaviors Graphical Abstract Highlights dGrik4 is widely expressed in the brain dGluK4 enrichment increases glutamate release probability and AMPAR conductance dGrik4 overexpression leads to persistent unbalanced inhibitory and excitatory activity dGluK4 gain of function alters behaviors common to human disease Authors Vineet Arora, Valeria Pecoraro, M. Isabel Aller, Celia Roma ´n, Ana V. Paternain, Juan Lerma Correspondence [email protected] In Brief Arora et al. show that an increase in Grik4 gene dose enhances the efficiency of synaptic transmission, causing a persistent circuit disequilibrium that alters the main amygdala outputs. This may account for the behavioral abnormalities observed in disorders like autism and schizophrenia. Arora et al., 2018, Cell Reports 23, 3827–3838 June 26, 2018 ª2018 Agencia Estatal Consejo Superior de Investigaciones Cientı ´ficas. https://doi.org/10.1016/j.celrep.2018.05.086 Cell Reports Article Increased Grik4 Gene Dosage Causes Imbalanced Circuit Output and Human Disease-Related Behaviors Vineet Arora, 1,2 Valeria Pecoraro, 1,2,3 M. Isabel Aller, 1 Celia Roma ´n, 1 Ana V. Paternain, 1 and Juan Lerma 1,4, * 1 Instituto de Neurociencias CSIC-UMH, 03550 San Juan de Alicante, Spain 2 These authors contributed equally 3 Present address: UMR5297 Institut Interdisciplinaire de Neurosciences (IINS), University of Bordeaux, Bordeaux, France 4 Lead Contact *Correspondence: [email protected] https://doi.org/10.1016/j.celrep.2018.05.086 SUMMARY Altered glutamatergic neurotransmission is thought to contribute to mental disorders and neurodegenerative diseases. Copy-number variation in genes associated with glutamatergic synapses represents a source of genetic variability, possibly underlying neurological and mental disease susceptibility. The GRIK4 gene encodes a high-affinity kainate receptor subunit of essentially unknown function, although de novo duplication of the 11q23.3-q24.1 locus to which it maps has been detected in autism and other disorders. To determine how changes in the dose of Grik4 affect synaptic activity, we studied mice overexpressing this gene in the forebrain. A mild gain in Grik4 enhances synaptic transmission, causing a persistent imbalance in inhibitory and excitatory activity and disturbing the circuits responsible for the main amygdala outputs. These changes in glutamatergic activity reverse when Grik4 levels are normalized; thus, they may account for the behavioral abnormalities in disorders like autism or schizophrenia. INTRODUCTION Throughout the CNS, excitatory and inhibitory synaptic transmissions are tightly regulated to sustain proper brain function. The correct activity of circuits depends not only on their appropriate wiring but also on the concerted interactions between presynaptically released transmitters and their postsynaptic receptors. Notably, establishing the adequate receptor number and type at synapses is fundamentally important to fine-tune neuronal communication. While changing the number of synaptic contacts and/or the density of synaptic receptors is a physiological mechanism that underlies the brain plasticity associated to learning and memory (e.g., Nicoll, 2017), pervasive brain pathologies can also emerge because of variations in the dosage of certain genes (Kenny et al., 2014; Poot et al., 2010; Sebat et al., 2007). Insertions, deletions, inversions, and duplications may result in loss or gain of gene function, and when this affects genes that are active at synapses, these alterations will modify normal circuits and their performance. This may be the case of GRIK4, a gene encoding a high-affinity kainate receptor (KAR) subunit, GluK4, whose function remains largely unknown. One known GRIK4 insertion or deletion (indel) variant has an altered 30UTR (3-UTR), and this variant is negatively associated with bipolar disorder, reducing the likelihood that a carrier of the deletion will develop this disease (Pickard et al., 2008). Interestingly, GRIK4 mRNA transcripts carrying the deletion generate more stable RNA transcripts in normal subjects and, thus, more GluK4 protein (Pickard et al., 2008; Knight et al., 2012). This situation is associated with enhanced hippocampal activation (Whalley et al., 2009), indicating that GluK4 abundance may significantly affect hippocampal processing. Similarly, a de novo duplication of the chromosome 11q23.3-q24.1 locus in which GRIK4 lies has been identified in a case of autism (Griswold et al., 2012), and, more recently, a genome-wide linkage analysis found this gene to be associated to different endophenotypes in schizophrenia (Greenwood et al., 2016). Understanding brain diseases relies on defining the molecular, synaptic, and cellular alterations that underlie the behavioral features of each disease, although some more general rules may exist. At certain glutamatergic synapses, KARs mediate a small part of the synaptic response, although they are critically important molecules that impart emergent integrative properties on the synapse (Sachidhanandam et al., 2009; Frerking et al., 1999). In addition, KARs participate in complex signaling pathways at both excitatory and inhibitory synaptic contacts in the brain (Lerma, 2003). The five KAR subunits are encoded by five separate genes (GRIK1–GRIK5), and they assemble into heteromeric receptors. KARs not only mediate fast synaptic transmission but also modulate synaptic properties and other events through canonical and non-canonical (metabotropic) signaling pathways (Rozas et al., 2003; Marques et al., 2013; for a review, see Valbuena and Lerma, 2016). Accordingly, these receptors play an instrumental role in the function and activity of neurons in many areas of the brain (Segerstra ˚le et al., 2010; Lerma and Marques, 2013). To further understand how neuronal communication is altered in cases of GRIK4 gene overdose, particularly in relation to the aforementioned human brain diseases, synaptic transmission was evaluated in C57BL/6J-Tg(camk2-grik4)3 mice. This mouse line overexpresses GluK4 in principal cells of the forebrain, and Cell Reports 23, 3827–3838, June 26, 2018 ª2018 Agencia Estatal Consejo Superior de Investigaciones Cientı ´ficas. 3827 This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). these animals display signs of depression, anxiety, and social impairment (Aller et al., 2015), closely reflecting the human endophenotypes associated with autism and schizophrenia (e.g., White et al., 2009; Gillott et al., 2001). We found that modest increases in GluK4 protein are associated with synaptic gain at selected synapses in the amygdala complex, resulting in unbalanced circuit outputs. This effect is attained not only by increasing the synaptic release probability but also by enhancing the postsynaptic a-amino-3-hydroxy-5-methyl-4-isoxazolepropionic-receptor (AMPAR)-mediated component. Indeed, an excess of GluK4 seems to delay the maturation of synapses as they remain enriched in the GluA2-lacking AMPARs typical of young synapses. Interestingly, both the synaptic and behavioral phenotypes recover upon normalization of the Grik4 gene dose. Overall, this remarkable effect on synaptic transmission may account for the behavioral abnormalities evident in diseases like autism spectrum disorders or schizophrenia. RESULTS The brain areas where genes encoding KAR subunits are expressed have been rudimentarily described in radioactive in situ hybridization studies (e.g., Wisden and Seeburg, 1993). While this approach allows the expression of intensely expressed KAR subunits to be defined, the expression of the Grik4 gene remains poorly documented (e.g., the Allen Brain Atlas). Similarly, no specific antibodies are available that recognize KAR subunits—in particular, the high-affinity subunits— further hindering the completion of a comprehensive expression map in the brain, especially in mice. Perhaps, for this reason, KARs made up of GluK5 and GluK2 subunits are considered the most abundant in the brain (Wisden and Seeburg, 1993; Fisher and Fisher, 2014). To explore the expression of Grik4 in the mouse brain, we took advantage of the higher sensitivity of non-radioactive in situ hybridization (Acloque et al., 2008; Paternain et al., 2000). Contrary to what is believed, Grik4 was abundantly expressed in all fields of the hippocampal formation, including the CA1, CA2, and CA3 fields and the dentate gyrus (DG), as well as in the neocortex and the amygdala (Figures 1A and S1). Interestingly, the amygdala has classically been related to anxiety and depression, two of the phenotypes observed in the transgenic mice that overexpress Grik4 (Aller et al., 2015). Synaptic Transmission in the Basolateral Amygdala We examined the influence of GluK4 overexpression in the amygdala. Although Grik4 was abundantly expressed in the amygdala complex, recombinant GluK4 was mostly detected in pyramidal neurons, particularly in the basolateral amygdala (BLA) (Figure 1A). Pyramidal neurons in the BLA were recorded, where they were identified by their low firing frequency and the Figure 1. Synaptic Effects of GluK4 Overexpression in Basolateral Amygdala (A) Top: in situ hybridization showing the expression of Grik4 mRNA in the hippocampus and basolateral amygdala (BLA), centrolateral amygdala (CeLA) and centromedial amygdala (CeMA) of a GluK4 +/+ animal. Bottom: the immunocytochemical detection of transgenic myc-GluK4 in similar coronal slices (full images are the montage of four high-power images) (see also Figure S1). Scale bars: 100 mm. (B) Pyramidal cells were identified in BLA by their morphology, after labeling with biocytin from the patch pipette, and by their firing properties. The panels show the morphology and the response to 50 pA hyperor depolarizing current pulses of a typical BLA pyramidal neuron, reflecting spike adaptation. (C) Synaptic currents evoked by supramaximal stimulation of the external capsule in GluK4 +/+ and GluK4 Over mice. The boxplot below represents the pooled data from n = 49 neurons (8 GluK4 +/+ mice) and n = 54 neurons (10 GluK4 Over mice). ***p<0.001. Mann-Whitney rank-sum test. (D) The amplitude of external-capsule-evoked AMPAR-mediated EPSCs were also referred to the amplitude of NMDA receptor (NMDAR)-mediated EPSCs, which were measured 50 ms after the peak response at +40 mV (gray band) (p < 0.001, Mann-Whitney rank-sum test; n = 30 neurons). (E) Representative AMPARs and/or KARs mediated mEPSCs at a membrane potential of 60 mV recorded from BLA pyramidal cells from each genotype, their quantification using cumulative probability distributions and boxplots showing the higher amplitude and the frequency for GluK4 Over mice relative to GluK4 +/+ mice. The data from n = 37 neurons from 7 slices from 6 GluK4 +/+ and 7 GluK4 Over mice are indicated. ***p < 0.001, Mann-Whitney rank-sum test. See also Figures S1–S3. 3828 Cell Reports 23, 3827–3838, June 26, 2018 large spike frequency adaptation upon membrane depolarization. Their identity was further verified anatomically by filling the recorded neurons with biocytin (Figure 1B). The KainateRmediated excitatory postsynaptic currents (EPSCs KAR ) in BLA pyramidal neurons were larger and presented faster decay kinetics (Figure S2), demonstrating the presence of functional KARs that contain the GluK4 subunit in these neurons. We also noted an increase in the number of neurons in which EPSCs KAR could be resolved (10/41 neurons in the wild-type [WT] as opposed to 13/32 neurons in the transgenic mice: 24.4% versus 40.6%). Moreover, the AMPAR-mediated EPSC (EPSCs AMPAR ) evoked by supramaximal stimulation of the external capsule were 75% larger in the GluK4-overexpressing mice (138.7 ± 12.14 pA versus 241.2 ±18.9 pA; ns = 49 and 54 neurons, respectively; p < 0.001; Figure 1C). To further verify that this increase in EPSC AMPAR amplitude was specific, we measured the ratio of EPSCs AMPAR to NMDAR-mediated EPSCs (EPSCs NMDA ), which certified the enhancement of synaptic AMPAR-mediated responses in transgenic mice (Figure 1D). We then analyzed how GluK4 overexpression affected presynaptic parameters, such as the miniature EPSC (mEPSC) frequency in the presence of tetrodotoxin (TTX). Analysis of these elementary synaptic events revealed a higher frequency in GluK4 over mice (1.96 ±0.15 Hz versus 3.19 ±0.26 Hz for the WT and GluK4 over mice, respectively; p < 0.001), compatible with the presence of KARs at presynaptic boutons, where they could enhance the release probability (e.g., Pinheiro et al., 2007). As mEPSCs are of an unknown source, we looked at the paired-pulse ratio of responses evoked by external capsule stimuli and found that it was significantly decreased (Figure S3), further certifying an increase of the probability of release, at least from these terminals. However, to our surprise, the cumulative plots clearly revealed that the mEPSC AMPAR amplitude was also larger in the transgenic mice (11.26 ±0.95 pA versus 14.14 ±0.47 pA for WT and GluK4 over mice, respectively; p < 0.001; Figure 1E). Hence, GluK4 overexpression in principal cells would appear to have a dual effect. On the one hand, it increases the amplitude of EPSCs mediated by KARs, while on the other, it enhances both the amplitude and frequency of the AMPAR-mediated mEPSCs. Synaptic Effect on Neurons of the Centrolateral Amygdala While BLA is the main entry point of sensory inputs to the amygdala (LeDoux et al., 1990), the central nuclei are the main source of amygdala outputs; particularly, the centromedial amygdala (CeMA) (Duvarci and Pare, 2014). BLA pyramidal cells project to the centrolateral amygdala (CeLA), where they contact two different types of GABA neurons, one projecting outside the amygdaloid nucleus (the regular-spiking cells) while the other contacts CeMA neurons (late-spiking cells: for reviews, see Duvarci and Pare, 2014; Janak and Tye, 2015). These neurons did not overexpress GluK4 (see Figure 1A), because expression of the Grik4 transgene is under the control of the CaMKII promoter in these mice, which is only active in principal cells. To determine how these connections were modulated by an excess of GluK4, the two different types of GABA neurons in the CeLA were identified by their firing pattern and studied separately (Figure 2A). Regular-spiking cells developed larger excitatory responses in the transgenic mice in response to stimuli applied to the BLA (Figure 2B). While the mEPSC AMPAR amplitude was altered slightly (12.0 ±0.45 pA in WT versus 13.8 ±0.45 pA in GluK4 over ; ns = 26 and 36 neurons, respectively; p = 0.027), their frequency was dramatically enhanced (2.1 ±0.16 Hz in WT versus 3.3 ± 0.11 Hz in GluK4 over mice; p < 0.001; Figure 2C). By contrast, the evoked EPSC (eEPSC) AMPAR amplitude remained fairly constant in late-spiking cells, although there was a tendency for it to be smaller than in mice overexpressing GluK4 (Figure 2D). Accordingly, the mEPSC AMPAR frequency diminished by 40% (3.0 ±0.21 Hz versus 1.8 ±0.16 Hz for WT and GluK4 over mice, respectively; p < 0.001), while the amplitude decreased only mildly (14.2 ±0.45 pA versus 12.8 ±0.50 pA for WT and GluK4 over mice, respectively; p = 0.042; Figure 2E). These data indicate that excess of GluK4 has two different effects in the amygdala. While it increases the release probability at BLA pyramidal boutons that contact regular-spiking cells in the CeLA, it also reduces the release probability at contacts established by BLA pyramidal neurons with CeLA late-spiking cells. The overall result of these changes may be a dramatic alteration in the excitatory-inhibitory balance within this structure. Since regularand late-spiking neurons inhibit each other (Lopez de Armentia and Sah, 2004), imbalances in the input to these cells is likely to strongly affect their activity in these transgenic mice (Figure 3A). To further assess this hypothesis, we simultaneously recorded spontaneous EPSCs (sEPSCs) and spontaneous inhibitory postsynaptic currents (sIPSCs) from both types of CeLA neurons, adjusting the intracellular chloride concentration to 5 mM during recordings at a 50-mV holding potential (Zhou et al., 2009). Under these conditions, the sIPSCs were outward currents, while the sEPSCs were inwardly directed deflections (Figure 3B). As expected, the frequency of sEPSCs, but not that of sIPSCs, increased by 39% in regular-spiking cells (2.3 ±0.14 Hz versus 3.2 ±0.21 Hz for WT and GluK4 over mice; ns = 17 and 18 neurons, respectively; p < 0.001; Figure 3B). By contrast, late-firing neurons had a larger frequency of sIPSCs in the transgenic mice (1.5 ±0.14 Hz versus 2.6 ±0.22 Hz for WT and GluK4 over mice; ns = 19 and 20 neurons for WT and GluK4 over mice, respectively; p < 0.001) and a slightly reduced excitatory input (2.8 ±0.28 Hz versus 2.0 ±0.22 Hz for WT and GluK4 over mice, respectively; p = 0.041; Figure 3C). We did not detect any changes in sEPSCs or in the sIPSC amplitudes in these cells (Figures 3C and 3D). Basal NeuronaI Activity Is Altered in the Amygdala Nuclei In Vivo in the Transgenic Animals To further test whether this imbalance occurs during an in vivo situation, we undertook experiments using activity-dependent, immediate-early gene (c-Fos) expression to track ongoing neuronal activity in behaving animals. This method has been reliably used to detect changes in neuronal activation in the amygdala and several other brain structures (e.g., Tye et al., 2011). We quantified the proportion of neurons in the BLA, CeLA, and CeMA in WT (5 mice) and GluK4-overexpressing (7 mice) animals (Figure 4). Consistent with the electrophysiological data, we found not only a significantly 75% higher proportion Cell Reports 23, 3827–3838, June 26, 2018 3829 of c-Fos-positive BLA cells but also a slightly but significantly higher c-Fos immunoreactivity average intensity in transgenic animals (Figure 4C). Although barely significant (p = 0.06, twotailed t test), c-Fos-positive cells were also more abundant in CeLA (53%). In contrast, c-Fos-positive neurons were largely increased in CeMA (73%) (Figures 4D–4F), as could be expected from electrophysiological ex vivo data. Altogether, these data indicate that the excess of GluK4 sets a new level of ongoing activity in the different types of amygdala neurons, which permanently alters the intra-amygdala circuit processing and output. Larger Postsynaptic Responses Are Not Only the Result of a Higher Release Probability The increased frequency of events can be accounted for by an enhanced release probability at presynaptic terminals where high-affinity KARs are overexpressed, producing larger evoked postsynaptic responses. However, the larger amplitude of elementary mEPSCs AMPAR is more difficult to explain. More abundant and/or different types of AMPARs at postsynaptic sites where GluK4 subunits are expressed could account for these effects. However, a significantly higher frequency of elementary events leading to multiquantal responses could contribute to, if not explain, the larger average amplitude of mEPSCs AMPAR observed in transgenic mice. To determine whether the increase in the mEPSC AMPAR amplitude was simply the consequence of an increased release probability, we performed a quantal analysis of these responses in both hippocampal CA3 and BLA neurons. When we classified the events according to their amplitude, and we constructed frequency histograms, a typical distribution was evident, with an initial peak and then a skewed tail over larger amplitudes (Figure S4A), particularly for the GluK4 over mice. To avoid the influence of the skewed distribution when determining the amplitude of the minimum quantal events, we fitted a Gaussian curve to the initial peak by using the initial half of the distribution together with those bars after the peak that showed symmetrical values. In this way, the quantal content in CA3 pyramidal cells was seen to be 10.4 ±0.50 pA (18 cells, 3 mice) in GluK4 WT mice and 12.7 ±0.44 pA in GluK4 over mice (22 cells, 3 mice; p < 0.005), consistent with the average increase observed Figure 2. GluK4 Overexpression in BLA Pyramidal Cells Produces Dissimilar Effects on Excitatory Synaptic Input to the CeLA Principal Neurons (A) CeLA neurons were distinguished according to the firing properties as regular-spiking (top) and late-spiking (bottom) neurons. The intermediate cartoon shows a scheme of position of stimulation and recording electrodes. (B) Top: typical examples of eEPSCs recorded from CeLA regular-spiking cells after stimulating the BLA in the GluK4 +/+ and GluK4 Over mice. Bottom: the stimulus-response curves were obtained by applying stimuli from 6 mA to 10 mA (n = 11 neurons from 9 slices from 4 GluK4 +/+ mice and n = 9 neurons from 8 slices from 5 GluK4 Over mice. ***p < 0.001, Student’s t test with Bonferroni correction for pairwise multiple comparisons). (C) AMPAR-mediated mEPSCs recorded from regular-firing neurons (V m =60 mV). Below are the cumulative probability distributions and boxplots (insets) of the frequency and amplitudes of mEPSCs AMPAR (n = 60 neurons from 60 slices from 13 GluK4 +/+ and 17 GluK4 Over mice; *p < 0.05; ***p < 0.001; Student’s t test). (D) eEPSCs recorded from CeLA late-spiking neurons after stimulating the BLA of the GluK4 +/+ and GluK4 Over mice (top) and the corresponding stimulus response curves (bottom). (E) AMPAR-mediated mEPSCs recorded from late-firing neurons (V m =60 mV). Below are the cumulative probability distributions and boxplots (insets) for the mEPSC AMPAR frequency and amplitudes (n = 63 neurons from 63 slices from 12 GluK4 +/+ and 15 GluK4 Over mice; *p < 0.05; ***p < 0.001, Student’s t test). 3830 Cell Reports 23, 3827–3838, June 26, 2018 after the more conventional mEPSC AMPAR analysis. To further reduce the probability of multiquantal events, we reduced the concentration of extracellular Ca 2+ to 0.1 mM (Figure S4B), and there were dramatically fewer larger events, allowing the initial peak to be more accurately estimated. The minimum peak amplitudes were consistently higher in overexpressing cells (10.4 ± 0.52 pA and 12.5 ±0.54 pA for GluK4 WT and GluK4 over , respectively; p < 0.01), indicating that the larger amplitude of elementary events in GluK4 over mice must be due to some other phenomenon, one probably related to postsynaptic alterations in the number and/or the type of AMPARs. Similar outcomes were obtained when BLA neurons were analyzed, with the quantal size in BLA pyramidal cells estimated to be 8.4 ±0.3 pA in WT neurons (n = 16) and 11.3 ±0.2 pA in transgenic cells (n = 16; p < 0.001), representing a 23% increase in amplitude (Figure S4C). Overexpression of GluK4 Alters the Type of AMPARs at the Synapse Once the summation of more frequent mEPSCs AMPAR had been ruled out as the cause of the larger postsynaptic events, we wanted to determine whether the increased event amplitude observed in GluK4 over mice was due to more Ca 2+ -permeable AMPARs (CP-AMPARs) present at the postsynaptic membrane. CP-AMPARs have larger single-channel conductance (Swanson et al., 1997), so that their enrichment at synapses provokes synaptic responses of larger amplitudes. A hallmark of these CP-AMPARs is their strong inward rectification, which allows them to be detected by simply looking at the rectification index, calculated as the ratio of the responses measured at membrane potentials of +40 and 65 mV. We studied this issue at mossy fiber (MF)-to-CA3 synapses, and since the abundance of these CP-AMPARs varies with neuronal maturation, we looked at two different developmental stages (post-natal day [P]13–P14 and P17–P21), which displayed clear plastic changes in the WT animals (Figure 5A) but not in the transgenic animals overexpressing GluK4 (Figure 5B). Hence, more CP-AMPARs exist at synapses overexpressing GluK4 at both ages. To further substantiate this conclusion, we assessed the effects IEM1460, a CP-AMPAR and KAR open channel blocker (Schlesinger et al., 2005), and while we found this blocker had no effect on WT P17–P21 mice (8.0 ±7.8% blockade), it largely blocked AMPARs in GluK4 Over mice (34.1 ±4.8%; p = 0.009) (Figure 5C). Figure 3. Unbalanced Inhibitory to Excitatory Activity in CeLA Principal Neurons (A) Scheme of the intrinsic amygdaloid circuit showing the mutual inhibition of CeLA principal neurons (regularand late-firing cells; RFC and LFC, respectively) and their output and electrode arrangements. The neurons were identified according to their firing in response to current pulses (insets). (B) Spontaneous IPSCs and EPSCs were recorded simultaneously by arranging the intracellular chloride concentration so that the sIPSCs were seen as outward currents and the sEPSCs were directed inwardly (see Experimental Procedures). (C) The frequency of sEPSCs, but not of sIPSCs, was increased in regular-firing cells from GluK4 Over mice (top). The amplitude remained unaltered (bottom) (***p < 0.001, Student’s t test, ns = 17 and 18 neurons from WT and transgenic mice, respectively). (D) The frequency of sIPSCs was increased in late-firing cells from GluK4 Over mice (***p < 0.001, Student’s t test), while that of the sEPSCs fell slightly (*p = 0.041, Student’s t test; ns = 19 and 20 neurons from each type of mouse). The amplitudes of sEPSC and sIPSCs did not change (plot shown at the bottom). Figure 4. c-Fos Expression in Basolateral Amygdala, Centrolateral Amygdala, and Centromedial Amygdala Neurons in GluK4 Over Mice (A) Coronal section through the amygdala complex showing DAPI-positive cells and regions of interest used for quantification of c-Fos immunoreactivity (M, medial; L, lateral). (B) Representative examples of regions used for quantification arranged in a column per genotype. (C) Density of DAPI-identified neurons positive for c-Fos immunoreactivity in BLA. (D) The average c-Fos intensity in cells measured in (C) was slightly (4%) but significantly (p = 0.039) increased in transgenic mice. (E and F) The density of c-Fos-expressing cells in (E) CeLA and (F) CeMA in both genotypes are shown. Data were obtained from averaging 4–5 section counts per mouse from 5 (GluK4 +/+ ) and 7 (GluK4 Over ) animals. *p < 0.05; ***p < 0.001, two-tailed Student t test. Scale bar: 100 mm. Cell Reports 23, 3827–3838, June 26, 2018 3831 Accordingly, IEM1460 significantly reduced the rectification in neurons from GluK4 over mice (Figure 5D) but did not totally remove it. Together, these data suggest that the presence of more GluK4 subunits increases the ratio of calcium-permeable to -impermeable AMPARs at synapses, which could account for the altered amplitude of EPSCs AMPAR beyond the increase in transmitter release. This led us to hypothesize that GluK4 could regulate synaptic transmission at excitatory synapses beyond its role as an ion channel receptor. The Increased Probability of Release in GluK4 over Mice Is due to a Change in the Affinity of Presynaptic KARs At the presynaptic level, the overexpression of GluK4 increased the release probability, as revealed by a higher mEPSC frequency. This phenomenon might reflect a higher affinity of the presynaptic KARs due to the incorporation of more high-affinity GluK4 subunits into the functional receptors. Such higher affinity KARs would eventually become activated by ambient glutamate, stimulating quantal glutamate release. To address this possibility, we recorded mEPSCs from BLA pyramidal cells in the presence of the KAR antagonist UBP310. This treatment attenuated the mEPSC frequency in WT animals and, as expected, reverted the increase in the mEPSC frequency in GluK4 over mice (Figures 6A and 6B). Experiments carried in CA3 pyramidal cells yielded similar results (Figure S5). Furthermore, the greater release probability in GluK4 Over mice attenuated the short-term plasticity measured as the degree of frequency facilitation in MF-to-CA3 synapses (Figures 6C and 6D), and blocking KARs with UBP310 restored the magnitude of frequency facilitation to values observed in WT mice (Figure 6D). These results led us to conclude that enhanced GluK4 expression enriches presynaptic boutons with higher affinity KARs, making them sensitive to ambient glutamate and facilitating quantal glutamate release. Normalizing the Grik4 Dose Rescues the Synaptic and Behavioral Phenotypes of GluK4 over Mice The data presented earlier strongly suggest that the synaptic phenotypes associated with a higher dose of Grik4 in the transgenic mice are the direct consequence of excess of GluK4 protein. To demonstrate this, we generated GluK4 Over mice with a normal copy number of the Grik4 gene (GluK4 Res ) by crossing GluK4 Over mice with GluK4 / mice, generating animals in which the amount of GluK4 protein was normalized. In western blots Figure 5. Ca 2+ -Permeable AMPARs Are Enriched in GluK4-Overexpressing Synapses (A) Representative AMPAR eEPSCs recorded from CA3 pyramidal neurons at the indicated holding potentials from P13–P14 (juvenile) and P17–P21 (adult)WT mice. The RI was calculated as the peak response at +40 mV relative to the peak response at 65 mV (bottom), and it was significantly different between the two stages of maturation: P13–P14, 0.51 ±0.04 (13 neurons from 13 slices); and P17–P21, 0.87 ±0.12 (22 neurons from 22 slices). (B) Same analysis in GluK4 Over mice revealed no difference at the two developmental stages: 0.23 ±0.04 (18 neurons from 18 slices) and 0.35 ±0.06 (16 neurons from 16 slices). (C) Representative recordings of individual EPSCs evoked by stimulation of mossy fibers at 65 mV in CA3 neurons from GluK4 +/+ and GluK4 Over mice, before and after application of IEM 1460 (upper panel). The bottom panel presents data showing the larger blocking effect of IEM 1460 (50 mM, gray bar) on GluK4 Over mice. The data derive from 15 neurons from 15 slices for WT and 18 neurons from 18 slices for GluK4 Over recorded from 3 adult mice from each genotype. (D) Representative recordings of individual EPSCs evoked by MF stimulation at 65 mV and +40 mV in neurons from GluK4 Over mice, with and without IEM 1460 (upper). The bottom panel presents data showing the lack of effect of IEM 1460 (50 mM, grey bar) on the rectification index of WT neurons yet it decreased the rectification in GluK4 Over neurons (red bar). The data from 22 neurons/22 slices for WT and 15 neurons/15 slices for GluK4 Over recorded from 3 adult mice from each genotype. *p < 0.05; **p < 0.01; Mann-Whitney rank-sum test was used in (A)–(C) and ANOVA on ranks with Dunn’s method was used in (D). 3832 Cell Reports 23, 3827–3838, June 26, 2018 (WBs) of brain tissue, there was a clear lack of endogenous GluK4, with these animals exclusively expressing the protein product of the Grik4 transgene (i.e., myc-GluK4: Figure 7A). The amplitude of the miniature and evoked EPSCs from BLA neurons reverted to the normal WT amplitude (Figure 7B). Interestingly, the release probability—which was increased in GluK4 Over and depressed in GluK4 / mice, as estimated by the paired-pulse ratio—returned to those values observed in WT animals (Figure 7B). The frequency of mEPSCs recorded from both regularand late-spiking cells in the CeLA recovered to normal values (Figure 7C). Similarly, in the hippocampus, the amplitude of eEPSCs AMPAR recorded from CA3 pyramidal neurons in these mice was similar to that of WT, and the sEPSCs KAR had normal values, too. Paired-pulse ratio and frequency facilitation returned to typical values at mossy-fiber-to-CA3 synapses (Figure S6). GluK4 over mice display several behavioral phenotypes, like anxiety, depression, reduced locomotor activity, etc. (see Aller et al., 2015). We confirmed these altered behaviors in new experiments carried out in WT siblings and GluK4 / animals and observed that normalization of GluK4 protein levels also reversed these altered behaviors, as is evident in a series of tests that characterize social interactions (Figure 7D), depression (Figure 7E), and anxiety (Figures 7F and 7G) in mice. Interestingly, the locomotor activity of these animals was also normalized (see Figure 7F). DISCUSSION In this study, we found that the over-dosage of Grik4, which encodes a high-affinity KAR subunit and results in a mild increase of GluK4 protein levels, alters the efficacy of synaptic transmission in the amygdala circuits. This consisted of a remarkably persistent enhancement of synaptic gain at selected synapses, which caused a clear imbalance between the inhibitory and excitatory activity at the amygdala output cells. As de novo duplication of this gene has been described in cases of autism (Griswold et al., 2012) and was associated with different endophenotypes in schizophrenia (Pickard et al., 2006; Greenwood et al., 2016), the synaptic alterations described in this study as a result of enriched GluK4 expression may also occur in these pathologies, possibly accounting for some of the abnormal behaviors observed. Thus, our data reveal that mild, yet reliable, alterations to the properties of particular synapses affect the overall performance of neuronal circuits and, hence, the individual’s behavior. Although great caution needs to be applied when trying to extrapolate from animal models to human diseases, these data highlight that persistent aberrant activity within defined circuits may underlie the abnormal behaviors associated with mental disease. KARs Modify the Circuit Balance in the Amygdala Previous studies circumscribed the expression of Grik4 mostly to the CA3 field in the hippocampus (Wisden and Seeburg, 1993; see also the Allen Mouse Brain Atlas). However, we revisited the expression of Grik4 by non-radioactive in situ hybridization and found Grik4 transcripts to be more widely distributed than anticipated, with strong Grik4 expression in the CA3 and dentate gyrus, as well as in the CA1 field of the hippocampus. Similarly, we also found significant expression in the neocortex and amygdala nuclei. Accordingly, the expression of this gene in this structure and the remarkable effects that its overexpression has on Figure 6. Enhanced Glutamate Release Is Caused by the Tonic Activity of Presynaptic High-Affinity KARs (A) The antagonism of KARs by UBP310 (10 mM) depressed the increased frequency of mEPSC in BLA synapses. (B) Quantification of the effects of UBP310 on the frequency of mEPSC in GluK4 +/+ and GluK4overexpressing mice. While the frequency in GluK4 +/+ mice was reduced from 1.96 ±0.15 Hz to 1.17 ±0.09 Hz in GluK4 Over mice, it was reduced from 3.20 ±0.25 Hz to 1.75 ±0.21 Hz (n = 18 neurons; *p < 0.05, one-way ANOVA on ranks with Dunn’s method). (C) In the CA3 field of the hippocampus, short term plasticity measured as 1-Hz frequency facilitation (FF) was attenuated in GluK4-overexpressing mice: 456 ±39.3% (n = 10 neurons from 10 slices) and 307 ±29.2% (n = 8 neurons from 8 slices) in WT and GluK4 Over mice, respectively (p = 0.01, two-tailed Student’s t test). The insets show the representative averaged (n = 30) synaptic responses in each case. (D) Effect of UBP310 on FF in CA3 pyramidal cells. Note how the degree of facilitation is restored upon KAR antagonism in GluK4 Over mice: 419 ± 41.8% (n = 8 neurons from 8 slices). ***p < 0.001, *p < 0.05, two-tailed Student’s t test). See also Figure S5. Cell Reports 23, 3827–3838, June 26, 2018 3833 the synaptic efficiency at the level of the amygdala, a structure classically related to depression and anxiety, are of particular interest. Cortico-amygdala synaptic transmission was enhanced through two main actions: the increase in the probability of release; and the enhancement of postsynaptic responses through an increase in quantal size, which drives an increase in the activity of BLA neurons in the transgenic animals. This concurs with the observation that most forms of human anxiety disorders are associated with BLA hyperactivity (Felix-Ortiz et al., 2016), and BLA hyperexcitability (or hypertrophy) in rodents is associated with enduring facilitation of anxiety-like behaviors (Roozendaal et al., 2009). In addition, BLA interactions with downstream targets like the central amygdala and ventral hippocampus are sufficient to alter anxiety (Tye et al., 2011; Felix-Ortiz and Tye, 2014). Perhaps the most striking effect of GluK4 overexpression was found in the intra-amygdala circuits, where transmission from the BLA to CeLA was clearly altered. Interestingly, the CeLA contains GABA neurons, and as such, they do not overexpress Grik4. Therefore, the enhanced GluK4 activity could be ascribed exclusively to the enrichment of GluK4-containing presynaptic KARs in BLA pyramidal neuron terminals that contact CeLA neurons. The mEPSC AMPAR frequency increased in synapses contacting CeLA regular-spiking cells, but it was reduced in synapses with the other type of CeLA neuron, the late-spiking cells, which represent the main source of inhibition of CeMA inhibitory neurons (LeDoux et al., 1990; Duvarci and Pare, 2014; Janak and Tye, 2015). Consequently, the evoked responses in the CeLA were enhanced in regular-spiking cells and depressed in latespiking cells. The mEPSC AMPAR amplitudes were also slightly larger or reduced, respectively, probably reflecting the altered frequency of multiquantal release in each case. This intrinsic circuit disequilibrium might dramatically alter the main amygdala ouptputs. This hypothesis was further confirmed 0 50 100 150 200 - GluK4 (Nat) - myc-GluK4 -Tubulin 0 10 15 20 BL Amygdala mEPSC Amplitude (pA) eEPSC Amplitude (pA) 0 100 200 300 400 500 B Density (%) *** *** * * 0.0 1.0 1.5 2.0 2.5 Paired Pulse Ratio * * 0 1 2 3 4 5 6 7 Late Firing *** mEPSC Frequency (Hz) CeL Amygdala C 0 1 2 3 4 5 6 Regular Firing *** ** mEPSC Frequency (Hz) G 0 10 20 30 40 50 60 Time in the open arm (%) *** Elevated Plus Maze 0 60 80 100 120 140 160 180 Open Field Total distance traveled (m) *** *** F GluK4+/+ GluK4over GluK4-/- GluK4Res 0 50 100 150 200 250 300 350 Immobility time (s) Swimming test *** ** ** *** ED Social Interaction Time sniffing (%) ns *** NF NF NF NF F NF NNFF N A *** ** GluK4+/+ GluK4Over GluK4 4 - 4 / - / -GluK4Res 0 5 10 15 20 25 30 Figure 7. Rescue of GluK4 Over Synaptic and Behavioral Phenotypes by Restoring Levels of Grik4 Expression (A) Western blot analysis showing the recovery of the total amount of GluK4 protein after crossing GluK4 Over and GluK4 / mice (i.e., GluK4 Res ). Note that normal GluK4 protein levels are established in these mice at the expense of the native GluK4 protein (Nat). The boxplot shows data obtained from 3 different sets of animals. (B) Amplitudes of AMPAR-mediated mEPSCs (ns = 31, 21, 15, and 27 neurons for each phenotype, respectively from 4–9 mice) and external-capsule-eEPSCs (ns = 26, 17, 15, and 16 neurons from 5–9 mice) in BLA pyramidal neurons in the four different mouse genotypes, including GluK4 deficient. The right panel shows the paired-pulse ratio values from BLA neurons observed in the four genotypes, showing recovery in mice with normalized levels of GluK4 (ns = 29, 17, 19, and 16 neurons, respectively, from 5–9 mice). (C) Frequencies of mEPSCs at late-firing cells (ns = 13, 16, 15, and 19 neurons from 8–9 mice) and regular-firing cells from the CeLA from mice (ns = 11, 20, 11, and 12 neurons from 7–11 mice) of the four genotypes. For (B) and (C): *p < 0.05; **p < 0.01; ***p < 0.001, one-way ANOVA on ranks with Dunn’s method. (D) GluK4 Over mice do not discriminate between novel (N) and familiar (F) mice in social interaction tests, while restoring GluK4 protein levels restores social interaction to normality. Insets are representative heat plots from a 3-chamber social test for each type of mouse, in which mice were confronted with familiar and novel mice. Student’s t test: **p < 0.01; ***p < 0.001, from 14 GluK4 +/+ , 11 GluK4 Over , 9 KO, and 14 GluK4 Res mice). (E) In the forced swimming test, GluK4 Over mice remained immobile for longer periods than WT and KO mice, which is compatible with depressive behaviors. These symptoms were attenuated, but not eliminated, after normalizing the GluK4 protein levels. (F and G) Mice overexpressing GluK4 show indications of severe anxiety in the open field test (F) and elevated plus maze (G) that were eliminated when GluK4 levels were normalized (GluK4 Res ). Insets show the animal’s tracks with different phenotypes. In the open field test, the deficit of total distance walked by GluK4 Over mice was recovered upon normalization of the GluK4 protein levels. For (E)–(G): *p < 0.05, **p < 0.01; ***p < 0.001 one-way ANOVA analysis of 21 GluK4 +/+ , 19 GluK4 Over , 19 KO and 19 GluK4 Res mice. See also Figure S6. 3834 Cell Reports 23, 3827–3838, June 26, 2018