Study subjects

A total of 137 individuals from 125 families were enrolled (Additional file: Table S1). Age ranged from 3 weeks old (BAB3053, who died at 3 weeks and carried a triplication including MECP2) to 53 years old (BAB14550, who carried other complex duplication). We were able to obtain updated clinical information from 18 out of 38 published individuals. Eighty-two subjects were clinically evaluated at TCH-BBC Rett Center. The remaining 55 subjects’ clinical information were gathered through a mixture of telemedicine and local provider information. One hundred thirty-six of the subjects were male and only one female was included since she was an unbalanced translocation carrier between chromosome 13 and X, thus clinical presentation was classical MRXSL phenotype.

Genomic rearrangement studies

Samples from all probands were submitted to a customized aCGH to confirm the clinical and molecular diagnosis of MECP2 duplication. After confirmation, short-read sequencing was performed in all unpublished, newly enrolled probands and for previously published samples that did not have their breakpoint junction resolved by Sanger sequencing (N= 74) [15, 18, 20]. Long-read genome sequencing (N = 42) and Optical Genome Mapping (OGM, N = 53) were performed on all samples displaying aCGH profile suggesting a complex rearrangement involving the MECP2 locus (i.e., interspersed duplications or triplications flanked by duplications) as well as those with an apparently simple duplication for whom short-reads did not resolve breakpoint junctions. The workflow of the genomic approaches is shown in Additional file: Fig S1. In all, combined analysis of customized aCGH, short-read and long-read genome sequencing and OGM allowed us to resolve the genomic structure of 118 (94%) out of 125 unique MECP2 duplications. Of the seven unsolved structures, two are terminal duplications for whom none of the methodologies applied could resolve, four only had biological material for aCGH, one had aCGH and short-read sequencing without resolving the genomic structure. Additional file: Table S2 details the genomic platform utilized for each sample in this cohort and Additional file: Table S7 provides the detailed coordinates of identified CNVs.

The resolved genomic structure of the Xq28 duplication spanning MECP2 encompasses the following five categories or rearrangement types: head-to-tail duplication (tandem duplication, 48%), inverted triplication flanked by duplications (DUP-TRP/INV-DUP, 20%) [15], terminal duplications (i.e., duplications including the pseudoautosomal region 2- PAR2, 22%), interspersed duplications (DUP-NML-DUP, 5%) and other types of complex duplication rearrangements (5%) (Fig. 1, Additional file: Fig S2 and Additional file: Table S7). Most of the terminal duplication structures consist of translocations (65%), followed by recombinant chromosomes (23%), large tandem duplications (8%) and inverted duplication (4%). Of the 17 subjects with translocation, 10 of them involve chromosome Y (59%). Triplication encompassing MECP2 were observed in two groups, four individuals carry DUP-TRP/INV-DUP while one carries a terminal duplication with translocation to Yq.

Fig. 1

Distribution of genomic structures observed in the MRXSL cohort. a Pie chart distribution of the genomic structure of 118 unrelated individuals carrying MECP2 duplication. b Violin plot representing the size distribution of the CNV encompassing MECP2 in each genomic subgroup. Length in bp, log10 scale. c Pie chart distribution of the genomic structure of 23 de novo MECP2 duplication events. d Violin plot representing the size distribution of inherited and de novo structural variants in MRXSL. CGR: Complex Genomic Rearrangement, DUP: Duplication, INV: Inverted, NML: Normal, rec: Recombinant, SV: Structural Variant, TRP: Triplication

Duplication size varied from 64.6 kb (BAB12190, partial MECP2 duplication) to 16.5 Mb (BAB3212) with hg19 genomic coordinates spanning from chrX:138,733,695 to the telomere. Median size of the MECP2 duplications varies from 484 to 541 kb in all groups with at least 75% < 700 kb except for Terminal duplications in which median is 6.2 Mb with 75% of the duplications < 10.5 Mb (Fig. 1b). CNV inheritance was investigated by ddPCR using probes targeting MECP2 as well as by checking the presence of breakpoint junctions of the SV. The latter was used to rule out balanced events in the maternal X chromosome either by breakpoint junction Sanger PCR or by short-read sequencing when available. Maternal samples were available for 110 families indicating that 87 out of 110 (79%) CNVs were inherited from an apparently unaffected mother, whereas 23 events occurred de novo in the proband (Additional files: Table S1/S2 and Fig. 1d). Notably, 15 out of 23 (65%) de novo events occurred in the Terminal duplication group, 13 out of 15 (87%) consisting of translocations which reveals a strong bias towards that specific genomic structure to the de novo events in MRXSL. Only 17% of probands with Tandem duplication carry a de novo event, whereas complex rearrangements such as DUP-NML-DUP (8%) and DUP-TRP/INV-DUP (4%) are rarely de novo (Fig. 1c). Importantly, de novo events of MRXSL duplications tend to be larger than inherited CNVs with mean size of 2.9 Mb compared to 550 kb (Fig. 1d).

Genomic content in MRXSL varies largely and partially depends on whether the structure is a terminal duplication or any other SV. In all, the largest duplication events will include 40 genes with disease association in OMIM, a few of them known to affect the nervous system when disrupted or dysregulated including FMR1 (*309550), L1CAM (*308840), FLNA (*300017), GDI (*300104), RAB39B (*300774) (Additional file: Fig S2a). The SRO in our cohort of probands with a clinical diagnosis consistent with MRXSL (N = 136 probands; BAB12190 was excluded since he was healthy due to partial MECP2 Duplication) is 127 kb [chrX:153,259,853 (BAB3039)−153,386,785 (BAB15790)]. Importantly, this SRO includes both MECP2 and IRAK1 but also the majority of the MECP2 noncoding cis-regulatory elements (CREs) identified by interspecies sequence comparison and reporter plasmid transfection assays [46] or more recently Mecp2 mouse brain CRE that are conserved in humans [47] (Additional file: Fig S2b).

Clinical characteristics of MRXSL

We used the genomic structure information as a general guide to perform the genotype–phenotype analysis: Tandem duplication, Other complex duplications (this group includes genomic rearrangements that do not fit into the remaining subgroups such as DUP-TRP/INV-DUP, DUP-NML-DUP and other CGRs), Terminal duplications, Translocations and Triplications including MECP2. Of note, the Translocation subgroup includes translocations to other chromosomes but not insertional translocations (N = 1). All translocations in this subgroup also had terminal duplication, thus Translocations are a subgroup of Terminal duplication. Previous studies have shown that individuals carrying triplication encompassing the entire MECP2 coding region present with a more severe phenotype [15, 16], therefore, we analyzed MECP2 triplications all together regardless of the genomic structure. It has also been suggested that individuals carrying duplication of RAB39B present a more severe phenotype as well as translocations [12, 13]. To independently investigate the potential contribution of RAB39B and translocation to phenotypic severity we separated patients carrying terminal duplication from those with terminal duplication with translocations. With the purpose of analyzing the clinical data, we have included seven individuals with unsolved genomic structures to the groups as following: apparently simple duplications were included in the Tandem duplication group and terminal duplication/unsolved structure were included in the Terminal duplication group.

Pre-/peri-/post-natal

Birth weight was available in 100 subjects, 83 had normal birth weight, 11 subjects were small for gestational age (SGA) and six subjects were large for gestational age (LGA). The rate of normal birth weight gradually decreased in the following order Tandem duplication (92.1%), Other complex duplication (88.2%), Terminal duplication (66.6%), Translocation (71.4%) and Triplication (40.0%). Statistically significant correlation was observed between groups (p-value = 0.014, Table 1). The main difference was between Tandem duplication and Triplication, and Other complex duplication and Triplication (Additional file: Table S8).

Table 1 Clinical features with frequency (%) for each genetic subcategory and statistical difference between groups (last column)

Birth height measurements were available in 58 subjects. Fifty-two of them had normal length, four subjects were shorter, and two subjects measured longer for gestational age. Head circumference was available in only 29 subjects, 27 subjects were normocephalic, and there was one microcephalic and one macrocephalic subject.

Postnatal complications requiring extended hospital stay including NICU admission, were observed in 66 out of 123 (53.6%) subjects and this ratio increased in the order Tandem duplication (21/57 = 36.8%), Other complex duplication (19/35 = 54.3%), Terminal duplication (9/11 = 81.8%), Translocation (13/16 = 81.3%) and Triplication (4/4 = 100%). There was a strong statistical correlation between groups (p-value = 0.001, Table 1). The statistical differences were observed between Tandem duplication and Terminal duplication, Translocation and Triplication (Additional file: Table S8).

One of the defining features of MRXSL is congenital hypotonia (considered cut off as 4 months or younger) and it is observed in 74 out of 121 (61.1%) of subjects, and the rate increased in the order of Tandem duplication (30/59 = 50.8%), Other complex duplication (16/31 = 51.6%), Terminal duplication (9/11 = 81.8%), Translocation (14/15 = 93.3%) and Triplication (5/5 = 100%). There was a strong statistical correlation between groups (p-value = 0.004, Table 1) and this significance was mainly due to difference between Tandem duplication/Other complex duplication and Translocation (Additional file: Table S8).

Developmental parameters

All patients have severe to profound DD/ID with highest developmental skills mostly not exceeding 24 months in all three domains; calculated gross and fine motor skills are provided in Additional file: Table S1. Highest achieved developmental skills, and fine and gross motor DQs gradually declined from Tandem duplication to Triplication (Fig. 2). As a reflection of this, while two of the Terminal duplication individuals achieved independent walking, none of the Translocation and Triplication subjects achieved this important milestone.

Fig. 2

Developmental Comparisons of Different MRXSL structural variants. Highest achieved gross motor skills (a), gross motor DQ (b), highest achieved fine motor skills (c), fine motor DQ (d). Overall, the developmental delay severity worsens in the following order: Tandem duplication < Other complex duplication < Terminal duplication < Translocation < Triplication. Medians are provided in gray boxes/black lines, means are given in blue lines. MRXSL: MECP2 Duplication Syndrome, DQ: Developmental Quotient

Subject BAB12190 carries a ~ 64 kb partial duplication of MECP2 including the first two exons who presented with normal development and unremarkable physical examination.

Recurrent infections

One of the core clinical features of MRXSL is the frequent infections [6, 22] which are often reported as recurrent respiratory infections. We evaluated recurrent infections in four categories including pneumonia (PNA), upper respiratory infections (URI), urinary tract infection and others. Pneumonia and upper respiratory infections were frequent and ranged between 67–82% of groups with no statistical difference between groups (Table 1). Urinary tract infections (UTIs) were identified in 34 out of 99 (34.3%) of individuals. The frequency of UTIs gradually increased in the following order Tandem duplication (11/48 = 22.9%), Other complex duplication (9/29 = 31.0%), Terminal duplication (5/8 = 62.5%) and Translocation (9/13 = 69.2%). There is not enough information for individuals carrying Triplication except for one who did not have UTI. Difference between groups was statistically meaningful (p-value = 0.01, Table 1); it mainly stemmed from the difference between Tandem duplication and Translocation (Additional file: Table S8). Also, additional infections including acute otitis media (N = 42), recurrent skin infections (N = 8), sepsis (N = 7), and meningitis (N = 3) along with low immunoglobulins (N = 11) were reported. Only 17 individuals had undergone some immune workup and 15 of them had some abnormalities. Details of these workups are included in the Column S of Additional file: Table S1. The most common abnormality was low IgA level and it is found in eight individuals. Two individuals had poor antibody formation response to vaccines. The remaining immune abnormalities differed and were unique to individual patients.

Anthropometric measurements

Weight, height and head circumference (OFC) were gathered from subjects’ clinical notes. Percentiles scattered throughout the growth chart, but majority had normal weight (88 out of 106 subjects = 83.0%), height (78 out of 102 subjects = 76.4%) and OFC (62 out of 101 subjects = 61.3%) measurements. Underweight and overweight were observed in 11 (10.3%) and seven (6.6%) subjects, respectively. Short stature was observed in 23 (22.5%) subjects whereas, tall stature was seen in only two (1.9%) subjects. Microcephaly and macrocephaly were observed in 21 (20.7%) and 18 (17.8%) of subjects, respectively. The rate of microcephaly progressively increased from Tandem duplication to Triplication group: Tandem duplication (2/53 = 3.7%), Other complex duplication (2/23 = 8.6%), Terminal duplication (5/9 = 55.5%), Translocation (10/13 = 76.9%) and Triplication (2/3 = 66.6%). There was a strong statistical correlation between groups (p-value = 0.001, Table 1). These differences can be attributed to many differences between groups including Tandem duplication-Terminal duplication, Tandem duplication-Translocation, Tandem duplication-Triplication, Other complex duplication-Terminal duplication, Other complex duplication-Translocation and Other complex duplication-Triplication.

Dysmorphia

Dysmorphic features were reported in all groups and highly common (> 80%) without significant difference between groups (Table 1). There was no distinctive facial gestalt, recognizable dysmorphic features; however, the most common dysmorphic features include facial hypotonia/open mouth, brachycephaly, plagiocephaly, high forehead/frontal bossing, midface hypoplasia/flat nasal bridge, dysplastic ear including large ear, hypo- or hyper-telorism and tapering fingers.

Gastrointestinal system

We split gastrointestinal (GI) system problems into four domains including: chewing/swallowing difficulties, gastroesophageal reflux, constipation/diarrhea, and other GI issues. For chewing/swallowing difficulties, we attempted to gather information on the severity of chewing/swallowing difficulty by obtaining history on whether patients can eat regular/soft/chopped diet and G-tube dependence status. Feeding and chewing difficulties were highly prevalent in MRXSL (over 80% in all groups without statistical difference, Table 1); however, requirement for a tube feeding increased from Tandem duplication to Triplication group: Tandem duplication (21/60 = 35.0%), Other complex duplication (12/32 = 37.5%), Terminal duplication (6/9 = 66.6%), Translocation (8/14 = 57.1%) and triplication (5/5, 100%). The correlation between groups was statistically significant (p-value = 0.019, Table 1). This difference was mainly due to Tandem duplication/Other complex duplication and Triplication (Additional file: Table S8).

Constipation was highly prevalent in all groups with > 80% frequency. Similarly, gastroesophageal reflux disorder (GERD) was reported in > 70% of individuals with MRXSL in all groups and there was no statistically significant difference between groups (Table 1). Of note, intestinal pseudo-obstruction was reported in two out of three Triplication individuals.

Genitourinary system

Genitourinary (GU) system abnormalities were investigated in two categories including structural [i.e., cryptorchidism, Congenital Anomalies of the Kidney and Urinary Tract (CAKUT)] and functional GU anomalies (e.g., urinary retention and kidney stone). One-hundred and twenty-six subjects had information on GU abnormalities. Anatomical GU defects were reported in 45.2% of individuals (57/126). Importantly, the frequency and complexity of GU defects increased with genomic complexity: Tandem duplication (17/59 = 28.8%), Other complex duplication (15/36 = 41.7%), Terminal duplication (8/11 = 72.7%), Translocation (12/15 = 80.0%) and Triplication (5/5 = 100%). There was a strong statistical correlation between groups (p-value = 0.001, Table 1). This significance was due to multiple difference between groups including Tandem duplication and Terminal duplication/Translocation/Triplication, and Other complex duplication and Translocation/Triplication (Additional file: Table S8). Majority of Tandem duplications had relatively minor GU anomalies including cryptorchidism (N:8/17 = 47.0%), while hypospadias, hydrocele and hypogenitalia/microphallus were observed in three individuals (17.6%). On the other hand, Terminal duplication, and Translocation and Triplication patients presented with hypogenitalia/microphallus more often (3/8 = 37.5%, 3/12 = 25% and 2/5 = 40%, respectively). Additionally, urinary retention and kidney stone were observed in 16 and 7 subjects, respectively. Similarly, urinary retention and kidney stone were more common in the Tandem duplication group (N = 7/59 and 4/59, respectively), while 3/5 of the Triplication subjects had hydronephrosis/vesicoureteral reflux.

Neurological system

Neurological evaluation was performed in eight categories including tone, epilepsy, movement disorders, behavioral problems, dysautonomia, bruxism, sensory abnormalities/high pain tolerance and others.

a)

Tone: Abnormal tone such as hypotonia, hypertonia and central hypotonia-appendicular hypertonia was widely prevalent across all groups ranging from 80 to 100%. There was no statistical difference between groups (Table 1).

b)

Epilepsy: The frequencies of epilepsy in Tandem duplication, Other complex duplication, Terminal duplication, Translocation duplication and Triplication were 59.0%, 50.0%, 45.4%, 58.8% and 40.0%, respectively. Difference between groups was not statistically significant (Table 1). We then investigated the age of seizure onset in these groups and identified that seizures start earlier as the complexity increases. Age of onset for seizure in Tandem duplication, Other complex duplication, Terminal duplication, Translocation duplication and Triplication are 8.4 years, 8.2 years, 5.6 years, 4 years 10 months, and < 2 years, respectively.

We also investigated the presence of neuromotor regression and whether there was a link between regression and epilepsy. Regression was attributed to seizure onset in 12 individuals, seizures becoming refractory in 17 individuals, infections in 6 individuals, and antiseizure medication side effect in 4 individuals.

iii)

Movement disorders: There are no MRXSL-specific recognizable movement anomalies. We observed a broad range of abnormal movements including dystonia, ataxia, choreiform movements and spasticity. The frequency varied significantly. One out of eight (12.5%) individuals of Terminal duplication had ataxia, while three out of three (100%) individuals with Triplication had spasticity. There was borderline statistical significance between groups with p-value of 0.04 without a specific difference between subgroups (Table 1 and Additional file: Table S8).

iv)

Neurobehavioral disorders: Neurobehavioral traits particularly autism spectrum disorder (ASD) was shown to be prevalent in MRXSL [24, 26]. We investigated the neurobehavioral traits in our large cohort. We did not deeply investigate whether patients fulfill ASD criteria however we documented whether there were repetitive movements, poor eye contact and sensitivity to stimulation (all patients have lack of or poor speech thus we queried other features of ASD). One hundred and nine individuals out of 127 (85.8%) had additional at least one of the three core features on top of poor speech. There was no difference between genetic subgroups (Table 1). Of note, individuals with Triplication were mostly too young to assess the neurobehavioral trait. In addition to ASD phenotype, there was only one individual with attention deficit hyperactivity disorder in the entire cohort.

v)

Dysautonomia/Bruxism/High Pain Tolerance/Self-mutilation: Dysautonomia features including drooling, blood-flow dysregulation to extremities and abnormal breathing (breath-holding or hyperventilation) are commonly reported in the allelic Rett syndrome and MRXSL [13, 48]. We identified the frequency of dysautonomia in 105/121 (86.7%) of individuals. Bruxism was reported in 81/112 (72.3%) of subjects. High pain tolerance was present in 85/109 (77.9%) of MRXSL individuals. There was no statistical difference between groups for dysautonomia, bruxism and high pain tolerance (Table 1). While 18 individuals reported self-mutilation, 16 subjects stated no self-mutilation.

Musculoskeletal system

Musculoskeletal anomalies are common in neurodevelopmental disorders due to deconditioning, immobility and nutritional deficiency. We obtained data on 93 subjects for their musculoskeletal problems and 49 (52.6%) reported musculoskeletal abnormalities, with the most common ones including bone fractures (26 subjects), osteopenia/osteoporosis (13 subjects including 3 requiring alendronate infusion), scoliosis (13 subjects), joint contractures (nine subjects). Also, one individual had osteosarcoma. Difference between groups were not statistically significant (Table 1).

Sleep disorders

Sleep problems in MRXSL were reported in two studies [9, 13]. We evaluated sleep disorders in two categories: insomnia and sleep apnea. We had information on insomnia from 118 subjects and 62 of them (52.5%) were found to have insomnia. Data on sleep apnea status were available in 114 subjects and 63 subjects (55.2%) were reported to have sleep apnea. Only two subjects were reported to have central sleep apnea and the remaining had obstructive sleep apnea. The frequency of sleep apnea increased in the following order: Tandem duplication 27/56 = 48.2%, Other complex duplication 16/32 = 50.0%, Terminal duplication 6/9 = 66.6%, Translocation duplication 12/15 = 80.0% and Triplication 2/2 = 100%. Despite the gradual increase in percentage with worsening complexity, there was no statistical correlation (Table 1).

Vision and hearing

Ascertainment of vision and hearing abnormalities have been limited in MRXSL. Miguet et al. studied presence of hypermetropia and hearing loss without further details. [7] Information on visual abnormalities were present in 117 subjects and 71 of them (60.6%) reported various, relatively minor visual abnormalities including refraction errors and strabismus. Differences between groups were not statistically significant (Table 1). Individuals who carry Triplication had more serious visual problems including one individual with nystagmus and large cornea, a second individual with nystagmus and hazy cornea, and a third individual with optic nerve hypoplasia. Hearing abnormalities are found to be less common.

Hearing evaluation data were gathered on 113 subjects and only 14 subjects (12.3%) were found to have hearing deficit including two of them with sensorineural hearing deficit. Similar to eye findings, all three Triplication subjects who had information on hearing had hearing loss (100%) and one of them was documented as sensorineural hearing deficit. Thus, a statistical difference was found between groups (p = 0.001, Table 1). The difference between groups were due to Triplication and all other subgroups (Additional file: Table S8).

Neuroimaging findings

Neuroimaging abnormalities were reported in MRXSL previously. Brain MRI or CT results from 61 subjects were available. In 50 of them, imaging studies revealed various non-specific abnormalities, with the most common including corpus callosum hypoplasia, delayed myelination, cerebral atrophy and ventriculomegaly. Of note, all Translocation and Triplication subjects had abnormalities. There was no statistical correlation between groups (Table 1).

Other clinical findings

We also investigated other problems not commonly reported in MRXSL. Eczema, asthma, anemia (mostly iron deficiency anemia), milk protein allergy were the most common additional findings. Of note, four subjects had hyponatremia, three subjects had bleeding diathesis or thrombocytopenia, four subjects had various short stature disorders including Leri-Weill Dyschondrosteosis (BAB2684), rhizomelic shortening (BAB3037) and dwarfism (BAB3212 and BAB14298). BAB2684, BAB3037 and BAB14298 probands carry Terminal duplications resulting from a recombinant X chromosome therefore short statures are likely related to the deletion of SHOX. The fourth individual with short stature (BAB3212) does not have Xp (SHOX) deletion but a 3q29 translocation. Patient was clinically diagnosed as atypical dwarfism due to growth hormone deficiency. Thus, the short stature in this individual has a different etiology. Lastly, three subjects had various autoimmune disorders including relapsing polychondritis (BAB2806), recurrent pleural/pericardial effusions (BAB14598) and ichthyosis (BAB3037).

Survival analysis reveals evidence for influence of genomotype

Since all Triplication subjects either died or were gravely ill within the first few years of life, we investigated whether different genomic subgroups play a role in the survival of individuals with MRXSL. Kaplan Meier/Cox Regression survival analysis showed that survival duration gradually decreases from Tandem duplication to Triplication (Fig. 3). The difference between groups were due to Triplication and all other subgroups, and Tandem duplication and Translocation (Additional file: Table S8). Cox regression for survival analysis indicated that different structural variations are statistically significant predictor of survival duration (p-value < 0.001). Strikingly, the likelihood of death compared to Tandem duplication had increased 147 times (95% Confidence Interval, 22.30–978.37) in the Triplication group and 4.26 times (95% Confidence Interval, 1.13–16.12) in the Translocation group. Of note, none of the Terminal duplication (N = 11) individuals died. However, six out of 11 of the Terminal duplication individuals were under age four. Overall, 16 out of 136 individuals (11.7%) died in the entire cohort and eight of them were either Tandem or Other complex duplication. All of the Tandem and/or Other complex duplication deaths occurred between 12 to 23 years, while Translocation and Triplication deaths were 10 or under. We also investigated the cause of death (Column J of Additional file: Table S1) in these individuals. Cause of death was related to the combination of poorly controlled seizures and frequent respiratory infections in seven out of 12 individuals. One individual (BAB11979) died of recurrent lung infections without seizure and one individual (BAB15677) died due to dehydration/septic shock. In the Triplication group, the etiology of death was identified in three individuals and all three individuals died due to discontinuation of supportive care (i.e., ventilator support) given the gravity of their disease which is different than the remaining groups.

Fig. 3

Survival curve analysis for different genomic subgroups. Cox regression for survival analysis indicate that SVs are good predictor of survival probability: Tandem (5/59) > Other complex (3/41) > Translocations (4/16) > Triplications (4/5) (Additional file: Table S1). Overall p-value < 0.0001. SV: Structural Variant

Gene expression and dosage analyses

To assess the impact of the heterogeneity of the genomic rearrangements on the molecular profile of a given patient’s cell, we performed RNA-sequencing on LCLs and fibroblast cells derived from a subset of our cohort (N = 64 LCL [24 Tandem, 19 Other complex, 8 Terminal, 10 Translocation, and 3 Triplication] and 18 fibroblast lines [9 Tandem, 6 Other complex, 2 Terminal, and 1 Translocation]). First, we evaluated expression of the SRO genes, MECP2 and IRAK1, between the clinical groups. We observed that MECP2 expression is indeed increased in MRXSL LCLs regardless of rearrangement type compared to unaffected control LCLs. Importantly, MECP2 expression is significantly (FDR < 0.1) increased in Triplication LCLs compared to Tandem, Other complex, Terminal, and Translocation rearrangements (Fig. 4a and Additional file: Table S9). IRAK1 expression is likewise increased in MRXSL LCL regardless of rearrangement type compared to unaffected control LCLs. IRAK1 is included in the triplicated region, and we remarkably observed an increased expression of IRAK1 (FDR < 1 × 10–10) in triplication lines compared to other MRXSL rearrangements (Fig. 4b). These data suggest that gene expression correlates tightly with copy number supporting the gene dosage hypothesis.

Fig. 4

Transcriptomic heterogeneity amongst MRXSL patient-derived lymphoblastoid cell lines. RNA from lymphoblasts from 64 MRXSL individuals from the clinical cohort was collected and processed for RNA-sequencing transcriptomic analyses. Patient lines were collected as triplicate RNA preparations. a Normalized MECP2 expression from RNA-sequencing data. b Normalized IRAK1 expression from RNA-sequencing data. c Gene expression along Xq28. Expressed genes are ordered from more centromeric (left) to telomeric (right) within the maximal genomic region spanning the cohort of samples collected. Samples are scaled by column, and rows are clustered using hierarchical clustering using Euclidean distance. d Principal component representation of global gene expression changes between MRXSL and unaffected control lines

We then asked if correlation with copy number persisted for additional genes on Xq28 outside the SRO, i.e., MECP2 and IRAK1. We clustered gene expression of the detected genes within the genomic coordinates encompassing the entire cohort of measured samples. We found that gene expression in Terminal and Translocation rearrangements clustered separately from the remainder of the cohort (Fig. 4c). Additionally, the Triplication individuals clustered separately as well, while the Tandem and Other complex rearrangement samples were intermixed. We identified a few notable exceptions, for example BAB3161, harboring a complex DUP-NML-DUP/INV rearrangement clustered by itself. The genes distal to the MECP2 gene were all expressed at a higher level (F8 through TMLHE). We observed a similar phenomenon in BAB12480, which is a complex DUP-NML-DUP/INV rearrangement with a smaller second duplicated segment than BAB3161. Indeed, the genes with the second duplicated segment of BAB12480 were overexpressed (GAB3 through FUNDC2), but not genes mapping to a phased control set just outside of the second duplicated segment like MTCP1 (Fig. 4c).

Next, we asked if these patterns of gene expression were consistent in an additional cell type by culturing patient-derived fibroblasts. We observed an increased expression of both MECP2 and IRAK1 in the MRXSL patient lines, though we were unable to obtain fibroblasts from an individual with a MECP2 triplication. Like in LCLs, MECP2 and IRAK1 expression is increased in MRXSL individuals compared to unaffected controls (Additional file: Fig S3a, b and Table S10). On a patient-by-patient basis, we also found gene expression along Xq28 ordered by the genes in the rearrangement. For example, two siblings (BAB3274 and BAB3275) share the same genomic rearrangement, and these two gene expression measurements clustered together. Furthermore, these patients harbor a triplication and both cell lines have higher expression of the genes within the triplicated region (e.g., FLNA) (Additional file: Fig S3c). Taken together, these data demonstrate that genes contained within a region altered by copy number gain increase in expression in multiple cell types.

Given that MECP2 globally regulates gene expression in the brain, we next sought to determine if the global pattern of gene expression was altered in MRXSL individuals compared to unaffected controls. Like the gene dysregulation observed in the brain, we identified thousands of low-magnitude gene expression alterations between MRXSL and unaffected control cell lines. This is reflected by a global separation between genotypes in PCA space in both cell types (Fig. 4d and Additional file: Fig S3d). Lastly, as subtle changes to MECP2 protein levels causes neurological dysfunction [47, 49], we measured MECP2 protein levels matched to the RNA-sequenced samples using quantitative capillary electrophoresis. We first observed that MECP2 protein levels were elevated in MRXSL patient cells. Next, we observed a significant (p < 0.05) correlation (Pearson R = 0.6, Spearman p = 0.63) between the log-transformed MECP2 RNA levels and MECP2 protein levels (Fig. 5). All together, these data demonstrate that genomic aberrations spanning MECP2 lead to altered MECP2 RNA and MECP2 protein levels, leading to global transcriptional dysregulation in MRXSL patient cells.

Fig. 5

Correlation between MECP2 RNA and MECP2 protein measurements in patient LCLs. a MECP2 protein measurements were made using capillary electrophoresis from matched lysates as collected for RNA-sequencing. MECP2 signal intensity was measured using a Jess Western instrument and normalized to total protein. b Correlation between MECP2 RNA and MECP2 protein; Pearson and Spearman correlation coefficients are displayed in upper left

Quantitative clinical phenotyping

Similar to what was identified in the clinical severity, developmental skills and survival, there were patterned differences in the heatmap of phenotype clustering as captured by HPO analyses. Out of all five groups, subjects with MECP2 triplication had a distinct pattern compared to other groups. On the other hand, there were clear distinction in phenotypic patterns between Tandem duplication-Other complex duplication vs. Terminal duplication-Translocation group. In the Triplication group, some clinical features such as high pain tolerance, bruxism, drooling, stereotypies, self-mutilation, musculoskeletal anomalies were not observed as several of these features evolve over time. However, some features such as genitourinary anomalies, eye findings and brain imaging abnormalities are more common/severe in the Triplication group (Fig. 6a). Overall, these findings support our clinical observations.

Fig. 6

Phenotypic analysis of five different structural variant groups of MRXSL and quantitative similarity analysis of MRXSL features with known OMIM genes on the p and q terminals of X-chromosome. a Prevalence of certain features ranged from 0 (light yellow) to 1.0 (black). Individuals carrying Tandem Duplication and Other complex duplication, Terminal duplication and Tr