Rubinstein–Taybi syndrome (RSTS) is a multisystemic pathology characterized by intellectual disability and delayed postnatal physical development, accompanied by a complex of phenotypic signs, such as broad thumbs and toes, microcephaly, and facial dysmorphisms such as downslanted palpebral fissures, high-arched eyebrows, a broad and “beaked” nasal bridge, a columella below the alae nasi, a narrow and high-arched palate, dental abnormalities, and a “grimacing” smile. This is the classic portrait of RSTS. However, after discovering a second gene linked to the syndrome–EP300, it became clear that the phenotypic manifestations vary widely. Currently, there are two genetic variants of RSTS: type 1, associated with the CREBBP gene, was first identified as the cause of RSTS in 1995 and was cloned 2 years later; type 2 is associated with EP300, with the first cases with variants described in 2005 (Petrif et al., 1995; Roelfsema et al., 2005). Characterized by a high degree of homology, both genes have the same domain structure and encode proteins with more than 70% similarity in their amino acid sequences. The protein products of these genes, CBP (CREB-binding protein, for short - CBP) and p300 (E1A-associated protein), are involved in the regulation of transcription and the function of many regulatory proteins in the cell. For a long time, sequencing these genes was difficult due to their large size, but since their association with RSTS was discovered, the development of new methods, such as next-generation sequencing (NGS) has facilitated ongoing efforts to unravel the molecular mechanisms of pathogenesis.

The purpose of this study was to identify the spectrum of mutations in genes responsible for RSTS in Russian patients. This article presents the results from a cohort of 158 probands referred to the Research Center for Medical Genetics (RCMG). The methods included searching for large deletions and duplications in CREBBP and EP300 by multiplex ligation-dependent probe amplification (MLPA). In cases where large deletions or duplications were not found, patients underwent a sequence analysis of the coding regions of CREBBP and EP300 using NGS.

Using the consecutive search of large intragenic rearrangements and NGS of three selected genes, pathogenic and likely pathogenic variants were identified in 67 out of 158 patients examined (42.4%). In 91 patients (57.6%), the molecular genetic diagnosis could not be confirmed after two stages of the study. However, of these 91 cases, variants of uncertain clinical significance were identified in five patients (3.2%). Additional segregation analysis in their families could enable the reclassification of the clinical significance of these variants.

The distribution of variants in genes shows that the majority of the identified changes—62 variants—were found in CREBBP; 4 variants were found in EP300, and 1 previously described variant was found in SRCAP (Figure 1). Clinical data for this patient were not available, but this finding supports the expediency of differential diagnosis among RSTS-like phenotypes.

Large deletions and duplications leading to a complete loss of one copy of the gene or a critical change in the gene structure were identified only in CREBBP (Table 1). For this gene, their share was 31% (19 cases) relative to the entire examined cohort—12%. Some samples showed a repeating pattern in the results of MLPA imaging (1 exon, 4–31 exons, and the entire CREBBP gene); however, in this work, the exact boundaries of the rearrangements were not established and may be different in each case. Neither deletions nor duplications in EP300 were detected, but the incomplete coverage of this gene by the probes available in the MLPA kit used does not allow us to completely exclude the presence of such changes in uncovered regions. In addition, in two patients, two different likely pathogenic variants were identified in CREBBP, both leading to the premature termination of transcription due to a decrease in signal intensity for only one exon, which was subsequently examined by Sanger sequencing. The diagnostic search for these patients was completed at this stage.

Pathogenic and likely pathogenic point variants in this study were identified using NGS in a total of 48 patients (30.37%). In 43 cases (27.2%), CREBBP was involved; in four cases, EP300 was found; and one variant was found in SRCAP. Parental DNA samples were available for only five cases, in which the de novo emergence of the variants was confirmed. The vast majority of cases of RSTS represent the first occurrence in a family, but parental testing is important, at least for the reclassification of variants of uncertain clinical significance, such as missense variants, deletions, or insertions without frameshift and any changes outside the canonical splice sites. Table 2 presents all pathogenic and likely pathogenic variants identified in this work. Figures 2, 3 show the distribution of variants in CREBBP and EP300 by exons, according to the domain structure of the encoded protein.

Six variants in CREBBP and one variant in EP300 identified in this cohort were previously described in the literature in patients with RSTS. Two of these variants—the nonsense variant c.1447C > T p (Arg483*) and the missense substitution c.4439A > G, p. (Asp1480Gly) in CREBBP—were each found in two patients from two unrelated families (Cross et al., 2020; Coupry et al., 2002).

The sample in this study is highly heterogeneous, making it impossible to establish genotype–phenotype correlations. Clinical information, including the phenotypic characteristics of the patients, was available for only 24 probands with identified pathogenic and likely pathogenic variants but was insufficiently structured for proper analysis. All patients were sent directly for confirmation or exclusion of the diagnosis, and we assume that each patient demonstrated some of the main clinical features of RSTS.

Figure 4 shows the range of pathogenic and likely pathogenic variants in CREBBP identified in this study.

In addition, in the studied cohort, four VoUSs in CREBBP were identified (Table 3)—two missense variants in exons 19 and 31, one variant that presumably affects splicing due to a single nucleotide substitution at the +5 position of intron 9, and an in-frame deletion of 15 bp. In silico predictors of variant impact suggested that each of these variants was pathogenic. Moreover, three variants that change the amino acid sequence itself (missense and in-frame deletion) are located in functionally important domains; therefore, they could not be confidently excluded from the list of variants that may affect the formation of the RSTS phenotype. One missense variant in EP300, located in the PHD domain, was classified as a VoUS because it met only the PM2 and PP3 criteria (Ryzhkova et al., 2019).

In the cohort of patients with RSTS studied in this work, the proportion of identified pathogenic and likely pathogenic variants was 43.7%. Currently, according to worldwide studies of various cohorts of patients with RSTS, the detection of the cause of the disease ranges from 37% to 75% (Coupry et al., 2002—47.6%; Bartsch et al., 2005—56% in patients with classical RSTS and 25% in those with a putative but “incomplete” phenotype; Bentivegna et al., 2006—61.3%; Schorry et al., 2008—56%, Wincent et al., 2016—70.6%; and Cross et al., 2020—37%) (Cross et al., 2020; Coupry et al., 2002; Bartsch et al., 2005; Bentivegna et al., 2006; Schorry et al., 2008; Wincent et al., 2016). These highly variable results are likely due to differences in cohort size, selection criteria, and the detected genetic and phenotypic heterogeneity of the syndrome.

The lack of structured clinical data in this study likely contributed to the relatively low detection rate compared to that in previously reported global data based on studies of clinically well-described cohorts of patients with classical RSTS. In patients without an identified cause, variants in large non-coding regions of genes, along with large rearrangements (for example, translocations) or deletions in regions not covered by the MLPA set (for example, in the gene EP300), cannot be excluded. Thorough clinical analysis and description of the phenotype could increase the proportion of cases with a confirmed diagnosis. Identifying groups of key and less specific features, while counting their occurrences could probably allow for a more confident distinction between classical RSTS with a clear phenotype and less typical cases that still exhibit certain manifestations. Recently, an international consensus statement on RSTS was published, in which experts proposed cardinal and supportive diagnostic criteria and a key to assessing their combination in a patient (Lacombe et al., 2024). Despite the possible variability of the phenotype and the fact that some cases of RSTS remain without molecular–genetic confirmation (including patients with identified VoUSs), a clinical diagnosis is necessary to build an optimal diagnostic algorithm that provides genetic counseling to patients’ families and manage patients throughout their lives. VoUSs also require a reassessment of the clinical hypothesis or additional studies to establish their causality for the specific patient. The expansion of diagnostic capabilities, along with the search for potential therapeutic methods, requires further study of RSTS in global practice.

Attempts have been made to identify genotype–phenotype correlations for RSTS, but contradictory data exist regarding the impact on phenotype of various mutations located in different regions of the gene and deletions of various ranges. For example, variants in the proximal part of CREBBP lead to the formation of a premature stop codon that may not be associated with severe intellectual disability, which is usually one of the key features of the syndrome. On the contrary, variants in exon 31 of EP300 lead to a severe form of classic RSTS developmental defects (Pérez‐Grijalba et al., 2019; Cohen et al., 2020). In the same study, the authors described splice site disruption in the HAT domain in a patient with intact intelligence and an unremarkable phenotype with severe growth retardation as the main clinical manifestation. Despite such contradictions, in our cohort, the predominant localization of missense variants in conserved regions of functional domains is consistent with that in the literature, which suggests that the disruption of the main functions of the protein is a key mechanism in the pathogenesis of RSTS. Missense variants in exons 30 and 31 of CREBBP and EP300 were recently linked to Menke–Hennekam syndrome (MHS) in patients with an uncertain diagnosis before whole exome sequencing (Menke et al., 2016). All variants described in MHS affect the amino acids in the TAZ2 and ZZ domains. However, in the same region, a large number of missense variants have been reported in patients with classical RSTS, although phenotypically classical forms of these syndromes show visible differences. The reason for this heterogeneity has not yet been established; it is assumed that a disruption of protein–protein interactions occurs, which is probably different from the effects of the variants that lead to the development of RSTS (Banka et al., 2019). The majority of the described pathogenic variants in RSTS are still associated with the premature termination of translation, resulting in the synthesis of a shortened protein or with haploinsufficiency due to early termination or deletion of the genetic material.

The variability in the clinical manifestations of RSTS, along with its overlap with other chromatinopathy syndromes, highlights the challenges of differential diagnosis and underscores the need for larger-scale studies involving patients. Potentially, it can also expand our understanding of the molecular mechanisms underlying the development of pathology. Currently, there is limited information in the literature regarding the results of whole exome and whole genome sequencing in patients with suspected RSTS without identified changes in CREBBP and EP300. Several studies have described cross-sectional changes between Rubinstein–Taybi, Wiedemann–Steiner, Coffin–Siris, Kabuki, Bohring–Opitz, and Cornelia de Lange syndromes (Enomoto et al., 2022; Negri et al., 2019; Di Fede et al., 2020; Woods et al., 2014; Cucco et al., 2020). The aforementioned nosologies and RSTS are clinically characterized by common features such as delayed physical development, intellectual disability, limb anomalies, and, sometimes, partially similar dysmorphic facial features. Each of these syndromes is associated with genes involved in the epigenetic regulation of cellular processes at different levels. The shared pathogenic effect, characterized by a shift in the equilibrium toward a more closed state of chromatin and changes in the expression levels of regulated genes, can manifest in overlapping clinical signs (Bjornsson, 2015). For example, the product of the KMT2A gene is a transcriptional activator with methyltransferase activity targeting the fourth lysine of the H3 histone subunit of the nucleosome (H3K4). Pathogenic variants in KMT2A are associated with the clinical manifestations of Wiedemann–Steiner syndrome (OMIM:605130). KMT2D activates transcription through trimethylation of the same amino acid, H3K4, and the disruption of its function is associated with Kabuki syndrome type 1 (OMIM:147920). ASXL1 encodes a Polycomb-group enhancer protein, and it is presumed to play a role in the differential enhancement of transcription of some genes while weakening that of others involved in embryonic development; mutations in this gene have been identified in Bohring–Opitz syndrome (OMIM:605039). The SRCAP protein is involved in chromatin remodeling and is one of the components that initiates CREB-mediated transcription by activating CBP (OMIM:136140). In each of the described cases, follow-up clinical examination reveals the signs of these syndromes but not of RSTS: contractures and exophthalmos in the case of ASXL1 mutation and eversion of the lateral third of the lower eyelid with mutations in KMT2D, despite the predominance of features initially suggesting RSTS. The only pathogenic variant we identified in SRCAP, which is typically associated with Floating–Harbor syndrome, highlights the possibility that similar phenotypes can develop for genes that are part of the same regulatory network. The described cases of this type remain sporadic and usually include patients in whom the initially presumed diagnosis was not exactly confirmed, and therefore, the scope of the search was expanded. At the same time, the strategy of initial whole exome sequencing of patients with the RSTS phenotype is more likely to detect variants in two main genes: CREBBP and EP300 (Yu et al., 2019; Lee et al., 2022).

The evaluation of patients should include different methods that complement each other’s limitations. Clinical evaluation also plays an important role in formulating the fastest and most effective diagnostic algorithm when a diagnosis can be phenotypically suggested. Despite the available research methods, there are still a large number of cases of RSTS in which the molecular defect cannot be identified. These patients undoubtedly require further study, which may help not only in making an accurate diagnosis but also in the timely recognition and prevention of possible syndrome-specific complications.