We analysed samples of 31 unrelated patients sent to our laboratory with parental consent with the request of genetic analysis for Rubinstein-Taybi syndrome. Studies and procedures were in accordance with the ethical standards of the host institutions. The group consisted of 16 female and 15 male patients, aged 2 to 42 years. Clinical diagnoses were established by different Centers and eventually confirmed by the coordinating clinical Center (Dept of Pediatrics, University of Milan) based on clinical records and photographs, when no pathogenetic mutation could be found. Careful examination of the patients' phenotypes was done, and the presence of malformations and medical complications were confirmed or excluded.

Thirty subjects had a normal karyotype by Giemsa-trypsin banding at 450-band resolution; patient 38 showed a 46, XY, der(14)t(9;14)(p11.2;p11.2) karyotype.

Phytohemagglutinin-stimulated peripheral blood lymphocytes were set up in culture from samples using Chromosome Kit "Synchro" (Celbio) and modified RPMI (Irvine Scientific) plus 5% fetal calf serum (Gibco). The cultures were blocked with colchicine after 72 h. Chromosome preparations were obtained using a standard technique.

BAC clones RP11 95J11, RP11 461A8 and RP11 566K11 were supplied by Mariano Rocchi 13; BAC RP11 1072J2 was purchased from the CHORI BAC PAC Resource Center (Oakland, USA). DNA was obtained starting from a single colony grown in 4 ml of LB medium supplemented with 20 μg/ml chloramphenicol (Sigma) following standard procedures.

BAC clones were labelled with digoxigenin-dUTP (Roche Diagnostic) using a nick translation kit (Roche Diagnostic). The FISH experiments were performed according to standard procedures 14. The chromosomes were counterstained with DAPI in antifade (Vectashild), and then visualised using a Leitz DM-RB microscope equipped for DAPI and FITC/TRITC epifluorescence optics. The images were captured by means of a CCD camera (Hamamatsu 3CCD Camera, C5810) and visualised using Highfish software (Casti Imaging).

Almost 50 metaphases and 100 nuclei were scored for each sample in order to establish the mosaic condition. Only intact and undamaged nuclei free of cytoplasm were analyzed. Nuclei with low signal intensities, diffuse signals, or absence of signals on both homolog chromosomes were considered to be hybridization failures and were not scored. Two small focal (or paired) signals of the same color and the same intensity, separated by a distance of less than the area of one signal, were considered to be a split signal from one chromosome. Interphase nuclei with one large signal of the same color and increased intensity of fluorescence with the absence of a second hybridization signal in an interphase nucleus were considered to be overlapping (or over-position) of two signals and were not scored. Informative mosaic samples were defined as those in which more than 5% of the nuclei had a reproducible abnormal pattern of signals different from normal chromosomal signals.

Fluorescent genotyping was performed as previously described 11. The following fluorescently dye-labeled chromosome 16p microsatellite markers: MS4, MS2 and D16S3065 (from centromere to telomere) were used for segregation analysis from parents to probands. Fluorescence detection was performed on an ABI 3100 sequencer. ABI PRISM software (Genescan) was used for gel analysis.

RT-PCR analysis was performed as previously described 11 for patients 14, 16, 20, 25, 28 and 29 for whom mRNA was available.

The DNA sequencing of all coding exons of the CREBBP gene was performed as previously described 11. Twentysix CREBBP PCR genomic fragments were amplified (see Additional file 1). The DNA sequencing protocol was updated for use with the BigDye Chemistry, Versions 2.0 and 3.1, and the ABI PRISM 310 and 377 Genetic analyzer systems (Applied Biosystems, Darmstadt, Germany). If a mutation or polymorphism was found, a second sequencing run was performed for confirmation. The origin of the mutation or polymorphism, whether de novo or familial, was established if parental DNA was available. Paternity testing was not performed due to legal restrictions. The Human Gene Mutation Database (HGMD) 15, recent publications 8101112, and the dbSNP database 16 were consulted to check whether sequence variations were previously reported or not. Missense mutations of patients 15 and 46 identified in this study were absent in 100 controls, while that carried by patient 23 was absent in 50 controls.

Thirty-one Italian RSTS patients, whose main signs are reported in Table 1, were carefully clinically analysed. Sex ratio was F:M = 16:15. All patients had psychomotor/mental retardation, which ranged from mild to severe in 17 cases. Microcephaly was evident in 20 (64.5%); a subset (12 cases) accounting for 61.3% of RSTS patients manifested poor growth. The main facial features were specifically described and/or evident from photos (see Figure 1) and the overall facial gestalt was absolutely typical in all but one patient. As reported in Table 1, the typical large thumb was evident in 93% (27/29 patients), and in 11 cases it was also abducted; the hallux was large in 89% (26/29), and only in 4 cases it was valgus/varus. Only 11 individuals had major malformations of internal organs or limbs: congenital heart disease (patent ductus arteriosus, ventricular/atrial septum defects, bicuspid aorta) was reported in 6 of them, cryptorchid testes in 3, coloboma of the optic nerve, renal hypoplasia, duodenal malrotation and hexadactyly in one. In 5 patients supernumerary nipples were evident. Only in two cases (pt 16 and 17) lacrimal ducts stenosis and sensorineural deafness (pt 14 and 38) were signaled. In case 46 the presence of multiple epitheliomas was reported. In another single case (pt 15) Perthe's disease was found.

We screened the 31 recruited patients for point mutations, small deletions or insertions, and large deletions and duplications of the CREBBP gene. First, by genotyping the polymorphic intragenic repeats MS4, MS2 and D16S3065 11 plus one SNP (c.5454G>A, 12) we could identify two deletions spanning the whole gene in patients 30 and 41. Both deletions affected the maternal chromosome. Figure 2 shows for patient 30: (A) the haplotype reconstruction attesting the lack of three maternally contributed intragenic polymorphic loci; (B) FISH of the CREBBP spanning BAC 1072J2 evidencing a signal of reduced intensity on one chromosome 16. By means of FISH analysis we detected three additional deletions, all in a mosaic condition. Two (pt 40, pt 66) affect the 5' of the gene, while one (pt 38), spans mainly the central region of the gene and its flanking 3' sequences. The last one is present in a structurally rearranged karyotype (see subjects in the methods section). The mosaic condition has been confirmed in two patients from which buccal smears were available. Scoring of CREBBP specific signals on both lymphocytes (> 150 cells- metaphases and nuclei) and buccal smears (>100 nuclei) gave comparable results in showing an average 20% of deleted cells in both cases. A fraction of 30% deleted cells was estimated on the third case for whom a sample of 300 cells from peripheral blood was scored. Current studies are in progress aimed to evaluating the extent and boundaries of all the deletions. These data will be reported elsewhere. A schematic view of the identified deletions mapped on the encompassed genomic CREBBP region is shown in Figure 3.

Then we screened the remaining 26 patients for point mutations by direct sequencing of 26 PCR genomic fragments encompassing the complete coding sequence and splice sites of CREBBP. We identified 14 sequence alterations of pathogenic significance: all are listed in Table 2 and shown in Figure 4 in their location within exons encoding different CREBBP domains. Although the identified mutations are distributed across the whole gene without apparent hot-spots, six, including two missense (p.Tyr1482Cys; p.Asp1543Tyr), and four truncating mutations (p.Lys1239ValfsX14; p.Gly1479X; p.Asp1543GlnfsX39; p.Pro1655CysfsX89) are clustered within the HAT domain. Out of the remaining mutations, which map outside the HAT domain, five of eight are predicted to lead to premature translation stops upstream of the HAT region, thus indirectly spoiling its function. They comprise p.Arg370X and p.Arg424X in the N-Terminal Transactivation Domain (TAD), p.Gln662X in the CREB Binding Domain and p.Gln1118ProfsX13 and p.Arg1173X in the Bromodomain. The remaining three mutations, p.Arg14Gly in the 5' Nuclear Receptor Binding Domain, p.Ser2015AlafsX25 and p.Gln2022ArgfsX16 in the C-Terminal TAD, do not affect the HAT domain at all.

Twelve of 14 mutations are novel, while p.Arg370X has been reported by several groups 8101117 and p.Arg424X has been also recently described 12. As to the nine different single nucleotide substitutions they consist of five nonsense (pts 24, 35, 11, 32 and 34), three missense (pts 23, 15 and 46) and one splice site mutation (patient 20). Out of the five sequence alterations producing a frameshift in the reading frame, one is a 2 bp-duplication (pt 2) leading to the potential incorporation of 12 anomalous aminoacids, three are deletions sized 1, 2 and 7 nucleotides respectively (pts 22, 17 and 21) predicting 88, 13 and 15 anomalous aminoacids and one is a splice site mutation (c.4728+1G>A) affecting IVS 28 donor site. Analysis of this mutation with two different softwares, the Splice Site Prediction by Neural Network (BDGP) 18 and the SpliceView 19 indicated that the splice donor site was suppressed (predicted scores = 0%) and a novel candidate donor site 260 bp upstream the mutation was activated. The SpliceView software, also detected a near donor site, 104 bp upstream the splice mutation. We were able to identify CREBBP transcripts from patient 20 also including a smaller transcript, in addition to the normal one (Figure 5). Direct sequencing of the anomalous transcript confirmed the activation of the nearest upstream cryptic donor site predicted by the SpliceView software. The frameshift produced by missplicing should lead to the incorporation of 38 anomalous aminoacids.

All three missense changes, p.Arg14Gly (pt 23), p.Asp1543Tyr (pt 46) and p.Tyr1482Cys (pt 15) were not previously reported. They were established to have occurred de novo and were found absent in 50 (p.Arg14Gly) and in 100 (p.Asp1543Tyr and p.Tyr1482Cys) healthy individuals, respectively. Furthermore all of them affect aminoacids that are highly conserved both through evolution and in the related p300 (Figure 6) 20. p.Arg14Gly in the 5' Nuclear Receptor Binding Domain is predicted by PSORTII 21 to cause the destruction of a monopartite Nuclear Localization Sequence (NLS) (Figure 7). This NLS is a typical monopartite cluster of basic residues, with a conserved pattern starting with Pro, followed by Lys or Arg in 3 out of 5 residues (PKRAKLS) 2223. In addition the same software predicted another central bipartite Nuclear Localization Signal at 1591 (KKTNKNKSSISRANKKK). Both the NLSs are highly conserved in mouse CREBBP and in p300 protein. p.Tyr1482Cys and the p.Asp1543Tyr are both in the HAT domain. Finally we detected 18 CREBBP polymorphisms, including 12 single nucleotide polymorphisms (SNPs) and a 3'UTR 1 bp-duplication (see Additional file 2). Three of the sequence variations have not been previously reported, including two intronic SNPs (c.3983-22C>T, c.4560+14A>G) and the 3'UTR 7329+25dupC. Conversely the sequence variation (c.5933A>G), was reported in other patients with RSTS 1112. This substitution, causing the p.Asn1978Ser change in a conserved glutamine-rich region, downstream the HAT domain, has been inherited by our patient 22 by his apparently normal mother. The same patient was found to bear a de novo truncating mutation in the exon upstream the p.Asn1978Ser change (c.4963delC). We were able to establish that both mutations occurred on the maternal allele in a cis configuration (data not shown). Moreover we detected the same nucleotide substitution in one of 50 controls analysed.