The Rubinstein–Taybi syndrome: modeling mental impairment in the mouse


  • A. Barco

    Corresponding author
    1. Instituto de Neurociencias de Alicante (UMH-CSIC), Campus de Sant Joan, Sant Joan d’Alacant, Spain
      *A. Barco, Instituto de Neurociencias de Alicante (UMH-CSIC), Campus de Sant Joan, Apt. 18, Sant Joan d’Alacant, 03550 Spain. E-mail:
    Search for more papers by this author

*A. Barco, Instituto de Neurociencias de Alicante (UMH-CSIC), Campus de Sant Joan, Apt. 18, Sant Joan d’Alacant, 03550 Spain. E-mail:


Mental impairment syndromes are diagnosed based on below-average general intellectual function originated during developmental periods. Intellectual abilities rely on the capability of our brain to obtain, process, store and retrieve information. Advances in the past decade on the molecular basis of memory have led to a better understanding of how a normal brain works but also have shed new light on our understanding of many pathologies of the nervous system, including diverse syndromes involving mental impairment. The recent multidisciplinary analysis of various mouse models for Rubinstein–Taybi syndrome has shown the power of animal models to produce an important leap forward in our understanding of a complex mental disease while simultaneously opening new avenues for its treatment. These studies also suggest that some of the cognitive and physiological deficits observed in mental impairment syndromes may not simply be caused by defects originated during development but may result from the continued requirement of specific enzymatic activities throughout life.

Important advances in human genetics and clinical research in the past decade have led to the identification of genes responsible for various mental impairment and learning disorders, whereas, in parallel, progress in mouse genetics and molecular biology has showed the enzymatic activities and signaling cascades that underlie the formation of memories in the brain. In particular, the generation and use of genetically modified mice to model neurological and neurodegenerative diseases is opening up new therapeutic avenues for the treatment of these conditions and clarifying their molecular etiology (Watase & Zoghbi 2003). A remarkable and encouraging example of this progress can be found in the recent characterization by different research groups of various mouse models for Rubinstein–Taybi syndrome (RTS), a poorly understood hereditary disease that affects only 1 out of 125 000 newborns (Wiley et al. 2003). Taken together, these studies not only have provided important insight into the etiology of this condition but have also shown some new aspects of the molecular basis of cognition. Although a number of important questions remain unanswered, these advances clearly highlight the usefulness of mouse models to investigate and tackle the treatment of neurological and psychiatric disorders.


Rubinstein–Taybi syndrome (OMIM #180849) is a complex autosomal dominant disease characterized by mental impairment, diverse skeletal abnormalities and a high incidence of neoplasia (Rubinstein & Taybi 1963). Children with RTS have distinctive anatomical features that can include apparent hypertelorism, prominent nose, malpositioned ears and especially broad thumbs and toes, which are frequently used for diagnosis. These early manifestations will later lead to postnatal growth retardation and in some cases microcephalia. There are also a number of medical conditions frequently associated to this disease (see exhaustive review by Wiley et al. 2003), such as constipation and other gastric dysfunctions, congenital heart disease, renal abnormalities and ophthalmologic problems. Behavioral studies of individuals suffering from RTS have shown that they experience difficulty in planning and executing motor acts and have a short attention span. Their intelligence level is low with an average IQ of 51, although mental range actually goes from 25 to 80 (Gotts & Liemohn 1977; Hennekam et al. 1992).

Petrij et al. (1995) discovered that about 40% of the individuals diagnosed with RTS carried heterozygous mutation in the gene encoding the CREB-binding protein (CBP). The wide variation in the severity of mental impairment and other clinical features observed in patients with RTS may be because of the heterogeneity of chromosomal rearrangements that include microdeletions, inversions and translocations. Rubinstein–Taybi syndrome can also be caused by substitutions or point missense mutations into the cbp gene (Kalkhoven et al. 2003; Petrij et al. 1995). Two different mechanisms for transcriptional dysfunction have been associated to these mutations: (1) haplo-insufficiency and (2) dominant-negative action of truncated CBP protein (Coupry et al. 2002; Petrij et al. 1995, 2000). Mutations leading to the expression of a truncated CBP protein have been associated to severity of symptoms and are found in less than 10% of the patients.

Recently, a new study on RTS has shown that a small percentage (∼3%) of mental impairment cases diagnosed as RTS are caused by mutations into the gene encoding for p300 highly homologous to cbp (Roelfsema et al. 2005). Although the primary sequence of p300 and CBP are more than 70% similar and they have many common interaction partners, these two proteins have distinct cellular functions and cannot always replace one another (Kalkhoven 2004).

The CBP, a multifunctional transcriptional activator

CREB-binding protein obtained its name because it was originally identified as an interaction partner for CREB (Chrivia et al. 1993). It is now known that CBP also interacts with many other transcription factors. Indeed, more than 100 proteins have been reported to interact either with CBP or with the closely related co-activator p300 (Janknecht 2002).

Several structural and functional domains have been identified in these two co-activators, these include a histone acetyltransferase (HAT) domain, a bromodomain that binds acetylated lysine residues in histones and other proteins, two transactivation domains located in both ends of the protein, three cysteine/histidine-rich regions (CH1 to CH3) that bind zinc and are involved in protein–protein interactions, and multiple specific interaction domains for different transcription factors (Fig. 1a).

Figure 1.

Structure and function of CBP. (a) Linear representation of wild-type CBP and mutant variants expressed in mouse models of RTS: The location of some important domains and sites in CBP structure are labeled, these include an HAT domain, a bromodomain (BD), two transactivation domains, three cysteine/histidine-rich regions (CH1 to CH3) and multiple specific interaction domains for different transcription factors such as the CREB-interacting domain KID and the nuclear receptor (NR) interacting domain at the N-terminus of the protein. (b) CBP regulates transcription at different levels: The capability of CBP to co-activate transcription depends on four different activities: (1) enabling the interaction between transcription factors and the RNApol II complex, (2) serving as a molecular scaffold that brings a variety of enzymatic activities together to the promoter, and catalyzing the transfer of acetyl groups to lysine residues in histones (3) and other proteins (4). Ac, acetyl group; PK, protein kinase; TF, transcription factor.

At least four different activities related to the activation and regulation of gene expression have been associated to CBP and p300 functions (Fig. 1b) (Kalkhoven 2004):

  • 1To bridge DNA-binding transcription factors to components of the basal transcription machinery, such as the TATA-box-binding protein (TBP).
  • 2To serve as a molecular scaffold that recruits a variety of enzymatic activities together to the promoter. For example, the interaction of CBP with mitogen-activated protein kinases (MAPKs) and the E–Cdk2 complex promotes the phosphorylation of CBP and enables the phosphorylation of several CBP-interacting transcription factors (Liu et al. 1999; Perkins et al. 1997).
  • 3To catalyze the transfer of acetyl groups to lysine residues in the N-terminus of histones. This modification decreases the strength of the interaction between the positively charged histones and the negatively charged DNA backbone resulting in the relaxation of the nucleosome and facilitating transcription.
  • 4To catalyze the transfer of acetyl groups to lysine residues in proteins other than histones, such as the tumor suppressors p53 and pRb, diverse transcription factors, and components of the RNApol II complex (Boyes et al. 1998; Chan et al. 2001; Gu & Roeder 1997; Hung et al. 2001).

These activities are regulated at least at two different levels. First, CBP is the target of several posttranscriptional modifications that modulate its activity. The transactivation potential of CBP is increased by the kinases PKA, PKC, CaMKIV and p42/44 MAPKs (Impey et al. 2002; Zanger et al. 2001). Also, CBP is methylated at Arg residues by CARM1, a modification that interferes with its binding to CREB and prevents CREB-dependent gene expression (Xu et al. 2001). Second, CBP is in limited supply within the cell and several transcription factors compete for available CBP-binding sites.

CBP and memory formation

The interaction between the phosphorylated form of the transcription factor CREB and CBP is required to trigger the transcriptional activity of CREB and links several kinase pathways, driven by neuronal activity, to activation of specific promoters (Barco et al. 2003; Kandel 2001). Genetic, pharmacological and molecular evidence indicates that the CREB pathway is a core component of the molecular switch that converts short- to long-term memory (LTM). Furthermore, other transcription factors known to be activated during activity-dependent neuronal gene expression, such as c-Fos (Janknecht & Nordheim 1996), c-Jun (Bannister et al. 1995) or NF-κβ (Gerritsen et al. 1997; Perkins et al. 1997), also interact with CBP and compete with CREB for specific binding sites. The convergence of multiple intracellular cascades on CBP together with its multiple roles regulating transcription ideally positions this protein to integrate different signals and regulate neuronal responses during the formation and maintenance of memory.

Mouse models for RTS

Several potential mouse models for RTS have been generated since the discovery of cbp as the major gene responsible for RTS (Table 1). The first one, a conventional knockout of the cbp gene, was generated by Tanaka et al. (1997) a decade ago. Few months later, Yao and colleagues reported the generation of a second mouse strain carrying a targeted mutation on the cbp gene, although in this case the deletion affected different exons (Kung et al. 2000; Yao et al. 1998). Oike et al. (1999a) produced a third mouse model for RTS during a gene trapping screening. In this case, the disrupted cbp allele encoded a truncated CBP protein (residues 1–1084, Fig. 1a). In all three mutant strains, homozygous embryos died during embryonic development at E8-E10, whereas heterozygous mice survived and exhibited some anatomical features also found in patients with RTS, such as growth retardation and specific skeletal abnormalities. In the case of the mutant expressing a truncated CBP protein, the phenotypes were more pronounced than in the null allele mutants (see comparison at Tanaka et al. 2000). This effect is likely because of the interference of the truncated CBP protein with the activity of full-length CBP (Parker et al. 1996).

Table 1.  Comparison of mouse models for RTS
  • CNS, central nervous system; dox, doxycycline; E-LTP, early phase of long-term potentiation; FC, fear conditioning; LTD, long-term depression; ND, not determined; OR, object recognition; PDE4, phosphodiesteresa 4; SAHA, suberoylanilide hydroxamic acid; TSA, trichostatin A;=, no significant effect; (−), significant defect; (+), significant recovery.

  • *

    In vitro assay showed that the mutant protein inhibits this activity. See also comparison by Josselyn (2005).

MutationNull allele heterozygous: 50% reduction of CBP levelTruncated allele heterozygous: expression of dominant-negative truncated proteinTetracycline-inducible transgene encoding a CBP protein lacking HAT activityTransgenic overexpression of dominant negative-truncated proteinTriple point mutation in the KIX domain in homozygosis
Promotercbp (gene-targeting)cbp (gene-targeting)tetO combined with pCaMKIIα-tTAFragment of the CaMKIIα promotercbp (gene-targeting)
Time–course of expression of mutationAs CBPAs CBPInduction by dox removal shortly before experimentsSupposedly postnatalAs CBP
Pattern of expression of the mutationAs CBP (ubiquitous)As CBP (ubiquitous)CA1, dentate gyrus, caudate putamen and neocortexHippocampus, amygdala, striatum and cortexAs CBP (ubiquitous)
FeaturesSkeletal abnormalities, growth retardation, genetic background effect on viabilitySkeletal abnormalities, growth retardation, reduced viabilityNo apparent differenceNo apparent differenceSmaller body size and reduced thymus volume
 Exploratory activity=(−)==ND
 Sensorimotor gating=NDNDNDND
 Motor coordination(−) (rotarod)NDNDNDND
 Working memory= (radial maze)= (Y-maze)NDNDND
 Passive avoidanceND(−) LTM (24 hr)NDNDND
 OR= STM (3 hr);= STM (3 hr);= STM (0.5 hr);ND= STM (1 hr);
 (−) LTM (24 hr)(−) LTM (24 hr)(−) LTM (24 hr)ND(−) LTM (24 hr)
 FC(−) context LTM (24 hr);= context LTM;=(−) context LTM (24 hr)(−) context LTM (24 hr)
 = cued LTM(−) cued LTM (24 hr) = cued LTM= cued LTM
 Spatial memory at the MWM= visible and hidden platform tasks; (−) swimming speed= visible and hidden platform tasks= visible platform task; (−) hidden platform task (= if intense training)= visible platform task; (−) hidden platform taskND
 Basal synaptic transmission=NDND=ND
 L-LTP(−) 4 × 100 Hz L-LTPNDND(−) dopamine-induced L-LTP, but normal 4 × 100 Hz L-LTPND
 CRE-driven transcription= in primary neuronal cultureND(−) c-Fos expression in CA1 neuronsND*(−) gene expression of CREB target genes in the hippocampus
 HAT activity(−) AcH2B in the brain + effects at specific lociNDND*NDND
Drug assays(+) FC LTM by SAHA; (+) L-LTP by SAHA; (+) L-LTP by PDE4 inhibitor(+) OR LTM by PDE4 inhibitor(+) OR LTM by TSANDND
 Studies on CNSAlarcon et al. (2004), Levine et al. (2005)Oike et al. (1999a), Bourtchouladze et al. (2003)Korzus et al. (2004)Wood et al. (2005)Wood et al. (2006)
 Non related to the CNSTanaka et al. (1997, 2000)Oike et al. (1999b), Yamauchi et al. (2002)NoneNoneKasper et al. (2002)

More recently, Kasper et al. (2002) used gene-targeting techniques to introduce a triple point mutation in the KIX domain of CBP that disrupts the binding surface for CREB and other transcription factors (Fig. 1a). In this case, homozygous mice were viable, although tended to be smaller than their littermate controls (Kasper et al. 2002). Also, Zhang et al. (2004) and Kang-Decker et al. (2004) independently generated mutant strains that contain a floxed cbp allele and therefore can be used to produce tissue-restricted CBP knockout mice eluding the early lethality of conventional homozygous mutant. These mice will likely allow investigating the consequences of complete lack of CBP function in specific tissues in adult animals.

Using a different approach based in transgenesis, and more specifically in the tTA/tetO system of double transgenics, Korzus et al. (2004) generated a line of regulatable RTS mutants expressing a CBP variant lacking HAT activity. The expressed CBP variant, CBP{HAT−} (Fig. 1a), has been shown to act as a dominant-negative mutant in vitro (Korzus et al. 1998). Crossing tetO-CBP{HAT−} mutants with CaMKII-tTA mice (Mayford et al. 1996) produced mice with restricted expression of this dominant-negative variant in the dentate gyrus and CA1 areas of the hippocampus in a temporally inducible manner. A similar approach was taken by Wood et al. (2005) who expressed the CBP truncated variant produced by Oike and colleagues, referred as CBPΔ1, under the direct control of the CaMKII promoter. The expression of this dominant-negative variant was restricted to neurons in the hippocampus, striatum, amygdala and cortex, although in this case its expression was not regulatable (Wood et al. 2005).

Cognitive deficits in mouse models for RTS

Although some of the behavioral features and neurological deficits observed in patients with RTS cannot be modeled in mice, others, such as cognitive deficits and reduced motor skills, can be easily assessed in mouse models for this disease. Cognition and more specifically memory formation and stability have been investigated in five different mouse models for this mental impairment syndrome [see also comparisons by Josselyn (2005) and Hallam & Bourtchouladze (2006)].

  • 1CBP+/Δ mice: Oike et al. (1999a) first showed that mice expressing a truncated CBP were deficient in LTM but not in short-term memory using step-through passive avoidance and fear-conditioning tasks. More recently, Bourtchouladze et al. (2003) confirmed this impairment using an object recognition task. The severe phenotype of truncated CBP mutants includes cardiac anomalies and hypolocomotion (Oike et al. 1999a, 1999b) that may confound the interpretation of these behavioral experiments.
  • 2Conventional CBP+/− knockout mice: Alarcon et al. (2004) carried out a comprehensive behavioral analysis of cbp+/− heterozygous mice generated by Tanaka et al. (1997) that included emotional behavior, motor abilities, and learning and memory. They found that mutant mice had normal exploratory behavior, levels of anxiety and sensorimotor gating but showed a significant deficit in motor coordination and learning in the rotarod task. Interestingly, humans with RTS also experience difficulty executing motor tasks (Gotts & Liemohn 1977). As observed in the truncated CBP mutants, these mice were impaired in LTM for both object recognition and fear-conditioning tasks. Conversely, cbp+/− mice did not shown any learning impairment in spatial navigation in the Morris water maze (MWM) or in a working memory task in the radial maze.
  • 3CBP{HAT−} mice: The behavioral analysis of these mice also showed deficits in long-term, but not short-term, recognition memory (Korzus et al. 2004). Furthermore, CBP{HAT−} mice showed impaired learning in the MWM but intact memory for contextual fear conditioning.
  • 4CaMKIIα-CBPΔ1 mice: Wood et al. (2005) found that CaMKIIα-CBPΔ1 transgenic mice were impaired in spatial learning in the MWM and showed a mild deficit in LTM for contextual fear conditioning.
  • 5CBPKIX/KIX mice: Wood et al. (2006) have recently investigated the memory phenotype of the strain of knockin mice generated by Kasper et al. (2002) bearing an inactive KIX domain. They found that these mice exhibited impaired LTM for object recognition and contextual fear (Wood et al. 2006).

All these studies were carried out in adult mice. Given that the symptoms of mental impairment in RTS patients emerge during the developmental phase, it would be also interesting to test the performance in cognitive tasks of pups and juvenile mutant mice (Branchi & Ricceri 2002).

Synaptic plasticity in mouse model for RTS

The late phase of long-term potentiation (L-LTP) in the Schaffer collateral hippocampus, an enduring change in the strength of synaptic connections between CA3 and CA1 neurons, is thought to underlie spatial learning and memory storage. This process depends on de novo gene expression and specifically involves the activation of CRE-driven transcription (Martin et al. 2000). Despite the consistent memory deficits found in mouse models for RTS, only two of these models have been analyzed for LTP defects:

  • 1CBP+/− mice: Alarcon et al. (2004) found that these mice had a specific defect in L-LTP induced by four trains at 100 Hz, which might account for the observed deficits in LTM. In contrast, they did not observe alterations in basal synaptic transmission, early phase of long-term potentiation (E-LTP) or long-term depression.
  • 2CaMKIIα-CBPΔ1 mice: Wood et al. (2005) also found that their transgenic mice had normal basal synaptic transmission and E-LTP but could not detect any alteration in L-LTP induced by four trains at 100 Hz. However, these mice had impaired L-LTP in response to one train at 100 Hz in the presence of a D1 dopaminergic agonist (Wood et al. 2005).

Comparison of RTS mouse models: Similarities stronger than differences

Notably, the behavioral characterization of these five mutant strains led to the same conclusion: a reduction of CBP function causes specific impairments in LTM but does not affect STM (Josselyn 2005; Table 1). Similarly, this reduction also caused specific deficits in the late phase of some forms of LTP in the two mouse models in which this phenomenon was investigated. However, a number of disparities emerge from the comparison of the results obtained with the different models. This is not surprising if we consider two essential differences among these mouse models:

  • 1Different temporal and regional patterns of expression of the mutation: Developmental defects are more likely to contribute to the phenotype of cbp+/−, cbp+/Δ and cbpKIX/KIX mutants than that in the case of the two transgenic lines in which expression is driven by the CaMKIIα promoter. Furthermore, the expression of the mutation is restricted to specific cell types in transgenic models but ubiquitous in the models generated by gene-targeting techniques. These features make the conventional knockout mice closer models to the human pathology but also limit the type of questions that can be addressed using these mutants.
  • 2Level of inhibition of CBP functions: cbp+/− mice would likely have a reduction of 50% on CBP activity both as a scaffold protein and as an HAT. However, the multifunctionality of CBP and the poorly characterized dominant inhibitory effect of the proteins expressed in cbp+/Δ, CBP{HAT−} and CaMKIIα-CBPΔ1 mice make difficult the precise estimation of the impact of these mutations on CBP function. In the case of cbpKIX/KIX, the activity of CBP as CREB co-activator is likely to be impaired but HAT activity can be intact.

Differences in the genetic background and in the behavioral and stimulation paradigms used in the five studies may also contribute to these disparities.

Several conclusions can be drawn from the comparison of these models:

  • • The comparison of the phenotypes of cbp+/− and cbp+/Δ heterozygous mice suggests that haploinsufficiency can, by itself, causes deficits on LTM, whereas the dominant-negative effect of truncated CBP may lead to more severe phenotypes not directly related to cognition.
  • • At least two functional domains of CBP, KIX and HAT contribute to the role of this transcriptional co-activator in memory storage.
  • • The threshold for detecting spatial memory impairments may depend on the level of HAT activity inhibition. In contrast to the results obtained in transgenic lines, both knockout strains showed robust memory deficits in tasks dependent on a one-time experience, such as object recognition or fear conditioning, but intact learning at the MWM. It is possible that the repetitive nature of the training in this task made it especially resistant to CBP dysfunction. In fact, intensive training overcame the mild MWM deficits of CBP{HAT−} mutants.
  • • All mutants express the mutation in amygdala except CBP{HAT−} mice, which are the only ones that did not exhibit a defect in fear conditioning. This result suggests that the contribution of this brain structure to the formation of fear memories may be especially sensitive to the reduction of CBP function.
  • • The different results obtained in the case of L-LTP elicited by 4 × 100 Hz stimulation can be because of several reasons, including different levels of CBP function blockade, developmental defects and the contribution of cell types not affected in CaMKIIα-CBPΔ1 mice. Additional experiments will be necessary to fully explore the contribution of CBP to L-LTP formation.

Each of these RTS mouse models has different strengths and weaknesses. The two strains of heterozygous mutant mice (cbp+/− and cbp+/Δ) appear to be more appropriate to model the different symptoms of this complex syndrome and to assess therapeutic approaches. The null allele heterozygous mutant seems to be a more realistic model for the major form of RTS, whereas its comparison with cbp+/Δ may distinguish the contribution of haploinsufficiency and dominant-negative effects to cognitive deficits. On the other hand, restricted knockouts and transgenic mice represent precise tools for understanding the role of CBP in learning and memory and for identifying the anatomical substrates of specific neurological defects. The inducible transgenic line generated by Korzus and colleagues, and the still uncharacterized restricted knockout mutants generated by Kang-Decker et al. (2004) and Zhang et al. (2004), will be especially useful in gaining new insight on the brain functions of CBP both during development and adulthood.

New insights into the molecular etiology of RTS

The characterization of mouse models for RTS has shown a direct role of both the HAT activity of CBP and the activation of the CREB pathway in this pathology. The analysis of the bulk acetylation stage of the different histones in the null allele model of RTS showed that the reduction on CBP function causes a specific deficit in H2B acetylation (Alarcon et al. 2004). The presence of AcH2B seems to be a feature of only the most active genes and has been associated with the maintenance of the overall transcriptional competence of specific loci (Myers et al. 2003). This altered chromatin configuration may affect the transcriptional activity of one or several genes important for LTP and memory storage in response to neuronal stimulation; therefore, it might underlie the LTM and LTP deficits observed in these mice. Additional differences may affect specific residues in H2A, H2B, H3 and H4 at specific loci. In fact, the recent study by Levine et al. (2005) has showed that the chronic reduction of CBP activity causes a decrease in the cocaine-triggered acetylation of H4 at the fosB promoter detectable only by ChIP assays. On the other hand, pharmacological and genetic experiments on the null allele and the truncated allele models of RTS indicated that the reduction in the level of CBP available to regulate changes in CRE-driven gene expression may also contribute to the molecular etiology of this pathology (Alarcon et al. 2004; Bourtchouladze et al. 2003). A view supported by the memory deficits observed in cbpKIX/KIX mice (Wood et al. 2006). The analysis of cell lines derived from patients with RTS showed that it is possible to dissociate the activities of CBP as HAT and CREB co-activator. Mutations located in the HAT domain of CBP completely blocked HAT activity but only reduced its activity as a CREB co-activator by about 50% (Kalkhoven et al. 2003; Murata et al. 2001), suggesting that the loss of both activities might also contribute to this pathology in humans.

These studies did not only clarify some critical aspects of the etiology of RTS but also provided new insights into normal brain function. Although it has been known for more than a decade that the acetylation of histones is involved in cellular memory and the adaptation of cellular responses to transient environmental signals, the discovery of its relevance in adult brain function is very recent and has provoked great interest in the memory field. Histone acetylation and other epigenetic mechanisms for marking chromatin might well underlie the long-term transcriptional effects in specific loci required for long-term modification of synaptic function and, in consequence, have lasting effects on diverse aspect of behavior.

Possible therapeutic approaches for the treatment of RTS

The gained knowledge on the molecular basis of this disease and the availability of these mouse models for RTS has allowed the exploration of the efficacy of new therapeutic approaches for the treatment of this condition.

Experiments in mice with a truncated cbp allele have shown the efficacy of the inhibitors of phosphodiesteresa 4 rolipram and HT0712, which enhance CREB-dependent gene expression (Barad et al. 1998; Bourtchouladze et al. 2003), in ameliorating the memory deficits for object recognition in a dose-dependent manner (Bourtchouladze et al. 2003). Moreover, preincubation of hippocampal slices with rolipram, which is known to facilitate the establishment of LTP (Barad et al. 1998), also reduced the L-LTP deficit observed in cbp+/− mice (Alarcon et al. 2004). These results support a role for CRE-driven gene expression in the etiology of RTS and identified this pathway as a suitable target for therapeutic drugs (Hallam & Bourtchouladze 2006).

On the other hand, studies in the null allele mouse model for RTS showed that the inhibitor of histone deacetylases (HDAC) suberoylanilide hydroxamic acid reversed the L-LTP and LTM deficits observed in these mice (Alarcon et al. 2004), whereas parallel experiments in CBP{HAT−} transgenic mice showed an amelioration of memory deficits by trichostatin A, another HDAC inhibitor (Korzus et al. 2004). Moreover, two recent studies carried out in wild-type mice showed that this family of drugs also enhances LTP and memory in the normal brain (Levenson et al. 2004; Yeh et al. 2004). Interestingly, valproic acid, an antiepileptic drug of frequent use in psychiatry, was recently rediscovered as inhibitor of HDACs. Its potential use in RTS, however, conflicts with its independent inhibitory effect in LTP mediated by the enhancement of GABAergic inhibitory neurotrasmission (Zhang et al. 2003). Various HDAC inhibitors are currently being tested for the treatment of different forms of cancer and neurodegeneration in Huntington disease (Ferrante et al. 2003; Hockly et al. 2003; Kouraklis & Theocharis 2002; McLaughlin & La Thangue 2004). If the results of these clinical trials are positive, it would be worthwhile to test the effectiveness of these new drugs in the treatment of RTS patients. The experiments in mouse models are certainly encouraging.


The author thanks Sally Till and Petra Gromová for critical reading of the manuscript and helpful comments. Research by A.B. is supported by the Marie Curie Excellence grant MEXT-CT-2003-509550 and the Spanish MEC grants BFU2005-00286 and SAF 2005-24584-E.