Introduction
Williams-Beuren syndrome (WBS, OMIM 194050) is a rare neurodevelopmental disorder caused by a microdeletion of 26-28 genes on chromosome 7q11.23 with an estimated prevalence of 1 in 7500 [
1]. Individuals with WBS present mild to moderate intellectual disability with an average intelligence quotient (IQ) of 55 (ranging from 40 to 100) [
2‐
4]. The syndrome is characterized by an unusual cognitive profile that includes relatively preserved expressive language and facial processing abilities but dramatic deficits in spatial cognition [
4‐
6]. Processing of spatial navigational information and verbal long-term memory, domains highly dependent on hippocampal function, are also severely affected in WBS [
7‐
9]. In addition to this clinical evidence of hippocampal dysfunction, structural and functional abnormalities have also been reported in WBS [
10]. Although the global volume of the hippocampus is preserved, WBS individuals present a shape abnormality in the midsection of the hippocampus and functional studies revealed an overall depression of hippocampal energy metabolism and synaptic activity [
10].
The complete deletion (CD) mouse model was generated in order to mimic the most common and recurrent deletion found in WBS patients, encompassing all single copy genes of the syntenic interval from
Gtf2i to
Fkbp6 [
11]. CD mice presented with many features reminiscent of WBS such as growth deficiency, craniofacial and cardiovascular abnormalities, and several behavioral alterations including hypersociability. In addition, CD mice presented an increase proportion of immature neurons in the dentate gyrus together with shorter dendrites and decreased spine density in CA1 pyramidal neurons of the hippocampus [
11].
Among the genes deleted in WBS,
GTF2I, CLIP2 LIMK1 and STX1A are good candidates for some of the neurobehavioral features of WBS [
12‐
17]. Knockout mouse models of
Clip2,
Limk1 and
Stx1a presented impairment in hippocampal long-term potentiation (LTP) that correlated with behavioral alterations, specifically hippocampal-dependent memory deficits [
14,
16,
18,
19]. A
Gtf2i mouse model presented several behavioral alterations such as hypersociability and anxiety-related behavior together with a reduction in spine density in hippocampal pyramidal neurons [
20,
21].
Although single-gene knockout mouse models have helped to elucidate the contribution of individual genes to the complex phenotype associated with WBS, the CD mouse model might be more closely related to the human phenotype. Therefore, we focused our investigations on hippocampal-dependent synaptic plasticity and memory in the CD model. We have studied cognitive function with a test sensitive to hippocampal function and analyzed the synaptic function on hippocampal slices. CD mice showed deficits in the spontaneous alternation test, indicating impairment in spatial working memory. In addition, electrophysiological experiments showed a significant reduction in the LTP in CA1 hippocampus synapses of CD mice, which could be associated with the reduced levels of BDNF observed in these mice. Taken together, these findings further reinforce the notion that functions controlled by the hippocampus are impaired in CD mice and underlie some of the behavioral deficits present in this animal model of WBS.
Discussion
LTP is considered the main cellular mechanism underlying learning and memory [
39]. As such, this form of synaptic plasticity has been shown to be impaired in a wide variety of genetic models of neurological disease with cognitive deficits [
40‐
43]. LTP deficits in the hippocampal CA1 region have been described in single gene knockout mice for genes included in the WBS region such as
Limk1,
Clip2 and
Stx1a [
14,
16,
18,
19]. Here we also evaluated LTP at synapses between Schaffer collaterals and commissural neurons in CA1 of the hippocampus in the CD model of WBS. Our data show that CD mutant mice also present a significant reduction in LTP that correlates with memory impairment, consistent with the results obtained in the single genes knockout mice. Thus, the additive effects of haploinsufficiency at these genes might be in part responsible of the synaptic dysfunction observed in the CD model. In addition, these data highly suggest that individuals with WBS may also have deficits in hippocampal LTP as part of their complex phenotype.
In addition to synaptic plasticity alterations, all these single gene knockout models (
Limk1,
Clip2 and
Stx1a) exhibited similar memory deficits in the contextual fear conditioning test [
14,
16,
18]. Similarly, CD mice showed impaired fear memory performance in the fear conditioning test, since they exhibited a slight reduction in the freezing time after the conditioning stimulus [
11]. In this work, we showed that the LTP deficits observed in CD mice were also accompanied by cognitive dysfunction revealed by impaired spatial working memory. CD mice performed poorly in the spontaneous alternation test, a simple but demonstrated sensitive method in detecting hippocampal dysfunction [
24], specifically deficits in spatial working memory [
23,
25]. We chose a continuous trial procedure, in which the animal is left in the maze until it completes 15 trials/choices. In the discrete trial method, in which the mouse is placed back into the starting box after each choice, the constant handling between trials is stressful for the mouse and it can affect the results [
44]. We found that CD mice displayed significantly reduced cognitive behavioral performance on the T-maze task compared to WT mice, without differences in locomotion levels. Therefore, the worse performance of CD mice in this test indicates spatial working memory impairment, which might be correlated with the synaptic plasticity defects observed in these mice.
Several LTP-related phenomena could be the responsible for the deficits observed in CD mice. A plausible explanation would be the presence of changes in the probability of transmitter release from the presynaptic terminal. Nevertheless, consistent with
Limk1 and
Stx1a mutant mice [
18,
45], the magnitude of PTP and the degree of PPF in CD mice were indistinguishable from WT. Changes in LTP have been related to alterations in mEPSCs frequency in
Limk1 knockout neurons [
18]. However, in agreement with electrophysiological studies performed in human induced pluripotent stem cell derived neurons of a WBS individual [
46], we could not appreciate any change in the frequency or amplitude of mEPSCs, suggesting that LTP deficits in CD mice cannot be attributed to presynaptic defects or changes in the biophysical properties of postsynaptic AMPA receptors. The study of theta bursts responses showed no differences between genotypes and our analysis of the AMPA/NMDA ratio suggests intact NMDA receptor function in CD mice. Collectively, these results point to a process other than initial induction as the element in LTP production that is affected in CD mice.
Given the above conclusion, it seems reasonable to hypothesize that CD mice exhibit defects in LTP maintenance and expression mechanisms that result in a reduced ability of CA1 synapses to sustain the increase in synaptic strength after the theta burst stimulation. There is overwhelming body of evidence demonstrating that BDNF plays an important role in synaptic plasticity [
47‐
49]. Specifically, BDNF expression seems to have a crucial role in generating sustained structural and functional changes at hippocampal synapses by activating multiple pathways [
50]. Mutant mice deficient for BDNF exhibited a significantly reduced LTP in the CA1 region of the hippocampus [
27,
51]. In addition, several cognitive diseases have associated deficits in
Bdnf expression or BDNF signaling with LTP and memory disturbances [
40,
43,
52]. Significantly decreased mRNA levels of
Bdnf have been documented in the hippocampus of CD mice [
21]. We showed that protein levels of this neurotrophin are also reduced in the hippocampus of CD mice. Patterson and colleagues demonstrated that BDNF is required for the late LTP (L-LTP) produced by theta burst stimulation. BDNF appeared to modulate the translocation of the activated MAPK from the dendrites to the soma and then to the nucleus, where it has access to several nuclear substrates that contribute to the persistence of potentiation, such as the transcription factor cAMP response element binding (CREB) [
53]. Curiously, the absence of LIMK1 resulted in reduced plasticity-dependent CREB activation, and by increasing the activity of CREB, LTP and memory deficits in
Limk1 knockout mice could be rescued [
19]. All together these data indicate that deficiencies of BDNF and LIMK1 present in the CD model could account for the LTP deficits via plasticity-dependent CREB activation, and suggests that LTP deficits in WBS might be treatable by enhancing CREB activation and/or BDNF signaling in the brain. On the other hand, the PI3K pathway, which has been demonstrated to be deregulated in several cognitive diseases [
54‐
56] is also activated by BDNF. Interestingly, WBS has also been related to the PI3K pathway since the regulatory subunit of PI3K (
Pik3r1) is a direct target of GTF2I [
57], previously described as a good candidate for the neurobehavioral features of WBS [
12,
13].
In addition to molecular mechanisms, LTP is accompanied by changes in cytoskeletal organization and in the morphology of dendritic spines [
58]. Dendritic spines contain the majority of excitatory synapses on CA1 pyramidal neurons and changes in spine density or morphology have been associated with aberrations in synaptic plasticity [
18,
59,
60]. CD mice present a reduction in spine density in apical proximal dendrites of CA1 pyramidal neurons, fact that share with single knockout mice of
Gtf2i [
11,
21].
Clip2 and
Limk1 are two synaptic genes that regulate cytoskeletal dynamics, either via the actin filament system (LIMK1) or through the microtubule network (CLIP2) [
17]. Alterations in gene dosage of these molecules may lead to defects in neuronal structure and hence, in synaptic plasticity. In fact,
Limk1 knockout mice presented abnormal morphology of dendritic spines of pyramidal neurons, which correlated with alterations in LTP [
18]. It is thus possible that spine abnormalities found in CD mice contribute to disrupt the production of LTP. Therefore, cytoskeletal defects might be involved in the neurological symptoms of WBS patients.
To sum up, this study highlights the utility of the CD model to study the mechanisms underlying the complex WBS neurocognitive profile. We report that LTP elicited by TBS was significantly impaired in hippocampal field CA1 of CD animals, which may contribute to the cognitive and behavioral phenotype of these mice. LTP deficit was not associated with changes in presynaptic function, LTP induction or AMPA and NMDA receptor function. Important issues for future investigation raised by this study will include determining the potential mechanistic underpinnings of synaptic plasticity deficits in CD mice, which might be related to BDNF-related–(MAPK and PI3K) pathways. It will be extremely important to further study the functioning of these pathways since they open new potential therapeutic approaches, currently unavailable for WBS.
Abbreviations
aCSF, artificial cerebral spinal fluid; BDNF, brain-derived neurotrophic factor; CD, complete deletion; CREB, cAMP response element binding; fEPSP, field excitatory postsynaptic potential; IPSP, inhibitory postsynaptic potential; IQ, intelligence quotient; LTP, long-term potentiation; mEPSCs, miniature excitatory postsynaptic currents; MLPA, multiplex ligation-dependent probe amplification; PPF, paired-pulse facilitation; PTP, post-tetanic potentiation; TBS, theta burst stimulation; WBS, Wiliams-Beuren syndrome; WT, wild-type
Acknowledgements
The authors gratefully acknowledge Mario Carta and Pei Zhang for their generous help and suggestions in electrophysiological experiments.