Abstract

   De novo variants in DDX3X account for 1–3% of unexplained intellectual
   disability (ID) cases and are amongst the most common causes of ID
   especially in females. Forty‐seven patients (44 females, 3 males) have
   been described. We identified 31 additional individuals carrying 29
   unique DDX3X variants, including 30 postnatal individuals with complex
   clinical presentations of developmental delay or ID, and one fetus with
   abnormal ultrasound findings. Rare or novel phenotypes observed include
   respiratory problems, congenital heart disease, skeletal muscle
   mitochondrial DNA depletion, and late‐onset neurologic decline. Our
   findings expand the spectrum of DNA variants and phenotypes associated
   with DDX3X disorders.

Introduction

   Intellectual disability (ID) affects 1–3% of the population and is more
   prevalent in males versus females.[113]1 Although over 100 genes on the
   X chromosome were found to be associated with ID in males,[114]2,
   [115]3 relatively less is known about X‐linked ID genes in
   females.[116]4 Whole‐exome sequencing (WES) is finding de novo variants
   in X‐linked ID genes in females of all ages.[117]5, [118]6, [119]7,
   [120]8 However, limited information is available regarding such cases.
   Some of the genes causing ID in females are known to cause disease in
   males, including PHF6, [121]9 NEXMIF, [122]8 and USP9X, with the latter
   causing congenital malformations not observed in affected males.[123]10
   Further evidence for gender‐specific variant pathogenicity comes from
   DDX3X located on Xp11.4, with pathogenic de novo variants causing
   syndromic ID in 39 females; in the same study, three males inherited
   DDX3X variants from apparently unaffected mothers.[124]11 Differences
   in predicted variant severity or X‐chromosome inactivation studies from
   blood DNA did not explain the gender‐specific disease expression. Five
   additional females with DDX3X variants have been described in the
   literature.[125]12, [126]13, [127]14 These reports led us to
   hypothesize that females with de novo variation in DDX3X may show
   additional clinical phenotypes. We report 31 individuals with
   DDX3X‐related disorders, and provide comprehensive clinical
   presentations for 13, including expanding the age range of molecular
   diagnosis with the oldest reported individual and a fetus. These data
   expand the number of DDX3X pathogenic variants and their associated
   phenotypic spectrum.

Methods

   Variants in DDX3X were identified by WES, performed according to
   previously described methods,[128]5, [129]6, [130]13 either on a
   clinical basis at Baylor Genetics (Females 1–24, Males 1–2, Fetus 1) or
   on a research basis by the Baylor Hopkins Center for Mendelian Genomics
   (BHCMG, Females 25–27) or through the Centre de Génétique Humaine,
   Université de Franche‐Comté (Female 28). Deidentified reporting of
   aggregated demographic and molecular data for all clinically referred
   cases was approved by the Institutional Review Board at Baylor College
   of Medicine (BCM). Additional, informed consent for publication of
   clinical details was obtained for a subset of clinically referred cases
   and all research‐based cases according to IRB‐approved protocols: at
   Baylor College of Medicine (Female 8, 14, 17, 23, 24, 26), through the
   Undiagnosed Diseases Network (UDN) protocol (Female 13), and through
   the BHCMG (Females 7, 19, 25, 27); and at Centre de Génétique Humaine,
   Université de Franche‐Comté (Female 28). Females 7 and 19 were
   previously reported in a study of research‐based reanalysis of clinical
   WES data.[131]13 DDX3X variants were annotated using transcript
   [132]NM_001193416. Variant pathogenicity was determined based on the
   ACMG guidelines[133]15 and the internal guidelines developed at Baylor
   Genetics ([134]https://www.baylorgenetics.com/variant-classification/).
   For the interpretation of de novo variants, the PS2 evidence is used if
   rare and/or private variants in the proband were detected in both
   parents by WES (Trio WES) or Sanger sequencing (proband only WES).
   Otherwise the PM6 evidence is used. 0.1–1 μg total RNA from patient
   fibroblast cells was extracted for library preparation with TruSeq
   Stranded mRNA kit and was sequenced by Illumina NextSeq 550. Genes with
   expression at the top/bottom 5% were used for pathway enrichment
   analysis by Ingenuity Pathway Analysis (IPA, QIAGEN Inc.,
   [135]https://www.qiagenbioinformatics.com/products/ingenuity-pathway-an
   alysis).

Results

   Among 4839 (2152 females, 2687 males) patients referred to the Baylor
   Genetics laboratory for clinical WES with developmental delay (DD)
   and/or ID, 26 postnatal individuals (24 females, 2 males) were found to
   carry pathogenic or likely pathogenic variants in DDX3X, and 1 female
   fetus was found to carry a de novo variant of unknown significance in
   DDX3X. Through collaboration with the BHCMG and Centre de Génétique
   Humaine, Université de Franche‐Comté, an additional four unrelated
   female cases (Females 25–28) were identified. The ages at molecular
   diagnosis of the postnatal individuals ranged from 1 to 47 years
   (Table [136]S1). Twenty‐nine unique variants were identified (26 novel
   and 3 reported previously), including 13 missense, 6 frameshift, 3
   splice site, 4 nonsense, and 3 in‐frame deletion/duplication changes
   (Table [137]1 and Fig. [138]1A and B). In 29 individuals with available
   parents (27 female, 1 male, 1 fetus), the DDX3X variants were confirmed
   as de novo, supporting the variant pathogenicity. Two de novo variants,
   c.573_575del (p.I191del) and c.1805G>A (p.R602Q) are mosaic in the
   proband, with allele fractions of 21% and 14%, respectively
   (Table [139]1). The most frequent clinical presentations in the 28
   females include DD and/or ID (28/28), hypotonia (19/28), dysmorphic
   features (19/28), structural brain abnormalities (18/20 who had brain
   MRI, Fig. [140]S1), movement disorders (17/28), visual impairments
   (9/28), and microcephaly (7/28) (Table [141]2). The most commonly
   observed dysmorphic facial features include a high‐arched palate
   (5/19), thin upper lip (5/19), large ears (5/19), and long/smooth/large
   philtrum (4/19). Clinical presentations that are not present in
   published studies include respiratory problems (5/28): obstructive
   sleep apnea, tachypnea, and chronic respiratory failure, as well as
   congenital heart disease (5/7 who had echocardiogram):
   atrial/ventricular septal defect, double orifice mitral valve with
   small patent ductus arteriosus, mild concentric left ventricular
   hypertrophy and bicuspid aortic valve. In comparison to published data,
   autism spectrum disorder and other behavior problems and skin
   abnormalities are underrepresented in our cohort (Table [142]2). For 13
   subjects, we obtained additional informed consent and provide detailed
   clinical descriptions (Table [143]S1), as well as clinical images for
   two subjects (Fig. [144]1D).

Table 1.

   Subjects with causal variants in DDX3X
   Subject Genotype[145]a Inheritance Nucleotide change Amino acid change
   Mutant/total reads by WES WES type Variant interpretation SIFT
   Polyphen2 Mutation taster Scaled CADD score Ref
   Female 1 Het De novo c.14_17delCAGT p.A5 fs 93/203 Trio Pathogenic,
   PVS1, PS2, PM2 NA NA NA 34
   Female 2 Het De novo c.949T>C p.C317R 130/265 Proband Likely
   pathogenic, PS2, PM2, PP3 deleterious probably damaging disease causing
   27.9
   Female 3 Het De novo c.126_129delTTTA p.H42 fs 38/88 Proband
   Pathogenic, PVS1, PM2, PM6 NA NA NA 33
   Female 4 Mosaic 21% De novo c.573_575del p.I191del 69/327 Proband
   Likely pathogenic, PS2, PM2 NA NA NA 22.3
   Female 5 Het De novo c.1244T>A p.I415N 205/435 Proband Likely
   pathogenic, PS6, PM2, PP3 deleterious probably damaging disease causing
   31
   Female 6 Het De novo c.971C>G p.P324R 146/292 Proband Likely
   pathogenic, PS2, PM2, PP3 deleterious probably damaging disease causing
   28.3
   Female 7 Het De novo c.1703C>T p.P568L 239/448 Proband Pathogenic, PS2,
   PS4, PM1, PM2, PP3, PP5 deleterious probably damaging disease causing
   34 [146]13
   Female 8 Het De novo [147]b c.336dupC p.R113 fs 107/233 Proband
   Pathogenic, PVS1, PS2, PM2 NA NA NA 33
   Female 9 Het De novo c.873_874insTATA p.R292 fs 97/257 Proband
   Pathogenic, PVS1, PS2, PM2 NA NA NA 35
   Female 10 Het De novo c.874C>T p.R292* 119/229 Proband Pathogenic,
   PVS1, PS2, PM2 NA NA NA 39
   Female 11 Het De novo c.887G>C p.R296P 183/407 Proband Likely
   pathogenic, PS2, PM2, PP3 deleterious probably damaging disease causing
   34
   Female 12 Het De novo c.1180_1185dupCGTGAT p.R394_D395dup 70/182
   Proband Likely pathogenic, PS2, PM2 NA NA NA 19.5
   Female 13 Het De novo c.1600C>T p.R534C 30/88 Proband Pathogenic, PS2,
   PS4, PM1, PM2, PP3 deleterious probably damaging disease causing 31
   Female 14 Het De novo c.1600C>T p.R534C 50/114 Proband Pathogenic, PS2,
   PS4, PM1, PM2, PP3 deleterious probably damaging disease causing 31
   Female 15 Mosaic 14% De novo c.1805G>A p.R602Q 19/137 Trio Likely
   pathogenic, PS2, PM2, PP3 deleterious probably damaging NA 27.9
   Female 16 Het De novo c.1804C>T p.R602* 67/141 Proband Pathogenic,
   PVS1, PS2, PM2 NA NA NA 51
   Female 17 Het De novo c.453_454del p.S152 fs 173/372 Proband
   Pathogenic, PVS1, PS2, PM2 NA NA NA 34
   Female 18 Het De novo c.173C>A p.S58* 53/100 Proband Pathogenic, PVS1,
   PM2, PM6 NA NA NA 36
   Female 19 Het De novo c.192dupA p.D65 fs 98/210 Proband Pathogenic,
   PVS1, PS2, PM2 NA NA NA 34 [148]13
   Female 20 Het De novo c.1595C>T p.T532M 79/171 Proband Likely
   pathogenic, PS6, PM2, PP3 deleterious probably damaging disease causing
   33
   Female 21 Het De novo c.1033G>C p.V345L 79/155 Proband Likely
   pathogenic, PS2, PM2, PP3 deleterious possibly damaging disease causing
   26.6
   Female 22 Het No parental samples c.1386C>G p.Y462* 159/342 Proband
   Likely pathogenic, PVS1, PM2 NA NA NA 37
   Female 23 Het De novo c.284+1G>A p.? 95/191 Proband Pathogenic, PVS1,
   PS2, PM2 NA NA NA 26.9
   Female 24 Het De novo c.865‐1G>A p.? 58/163 Proband Pathogenic, PVS1,
   PS2, PM2 NA NA NA 24.6
   Female 25 Het De novo c.1021T>C p.C341R 65/126 Trio Likely pathogenic,
   PS2, PM2, PP3 deleterious damaging disease causing 26.4
   Female 26 Het De novo c.1244T>A p.I415N 46/87 Proband Likely
   pathogenic, PS6, PM2, PP3 deleterious damaging disease causing 31
   Female 27 Het De novo c.1206_1208delCTT p.F402del 108/235 Trio Likely
   pathogenic, PS2, PM2, PP3 NA (Provean: deleterious) NA disease causing
   18.88
   Female 28 Het De novo c.1438A>G p.R480G 1821/3806 Trio Likely
   pathogenic, PS2, PM2, PP3 deleterious probably damaging disease causing
   25.6
   Male 1 Hemi From asymptomatic mother c.1052G>A p.R351Q 50/50 Proband
   Likely pathogenic, PS6, PM2, PP5 deleterious benign NA 25.1 [149]11
   Male 2 Hemi De novo c.443+3A>T p.? 87/88 Proband Likely pathogenic,
   PS2, PM2 NA NA NA 13.91
   Fetus 1 Het De novo c.1304T>C p.L435P 260/513 Trio Variant of unknown
   significance, PS2, PM2, phenotypic match uncertain deleterious probably
   damaging NA 28.4
   [150]Open in a new tab

   Variant interpretation column contains the clinical significance of the
   variant and the type of evidences supporting the interpretation based
   on the ACMG guidelines[151]15 and the internal guidelines developed at
   Baylor Genetics
   ([152]https://www.baylorgenetics.com/variant-classification/).

   NA: not applicable.

   WES type: Trio: trio WES; Proband: proband only WES.

   None of the variants above has been seen in ExAC
   ([153]http://exac.broadinstitute.org/) or gnomAD.[154]22
   ^a

   Het: heterozygous. Hemi: hemizygous.
   ^b

   Heterozygous in the similarly affected monozygotic twin sibling,
   negative in two other siblings.

Figure 1.

   Figure 1
   [155]Open in a new tab

   Location of DDX3X variants identified in this study, Female individuals
   (25, 26, 27) ascertained through the BHCMG, and muscle biopsy results
   in Female 17 showing abnormal mitochondrial morphology. (A) Schematic
   view of the DDX3X exon–intron structure based on NM_001193416. Blue
   boxes represent exons and yellow fields represent introns. Exon number
   is listed below each exon. cDNA change is listed for each variant. (B)
   Schematic view of the DDX3X protein structure based on Snijders Blok
   et al.[156]11 Amino acid change is listed for each variant. (C)
   Pedigree and Sanger tracings demonstrate de novo inheritance in three
   unrelated female probands. (D) Female 25 demonstrated synophrys, a
   broad nasal root with upturned nostrils, a long philtrum, and thin
   upper lip. Female 27 demonstrated cupped ears, a long philtrum, and a
   thin upper lip. (E–G) Muscle biopsy results in Female 17 (E) Skeletal
   muscle cross‐section showing mild variation in fiber size (H&E;
   magnification ×400). (F) Skeletal muscle cross‐section showing few
   fibers with mild subsarcolemmal increase in oxidative activity
   [cytochrome oxidase (long arrow) and NADH tetrazolium reductase (inset
   – arrow heads; magnifications ×400)]. (G) Electron microscopic images
   showing mild subsarcolemmal mitochondrial proliferation (long arrow)
   with inset in the upper corner showing pleomorphic abnormally elongated
   and irregularly shaped mitochondria (arrow heads). Variant color in (A)
   and (B): black, first reported in this study; purple, previously
   reported. The c.1304T>C (p.L435P) variant from Fetus 1 was not listed
   in (A) and (B).

Table 2.

   Comparison of clinical presentations in this study and in the published
   cohort
   Clinical features Number of subjects in this study Percentage in this
   study Percentage in the published cohort
   DD and/or ID 28/28 100% 100%
   Hypotonia 19/28 68% 76%
   Dysmorphic features 19/28 68% NA
   Structural brain abnormalities 18/20 90% 81%
   Movement disorders 17/28 61% 45%
   Visual impairments 9/28 32% 34%
   Microcephaly 7/28 25% 32%
   Autism spectrum disorders and other behavior problems[157]a 6/28 21%
   53%
   Respiratory problems 5/28 18% NA
   Congenital heart disease 5/7 71% NA
   Skin abnormalities[158]a 5/28 18% 37%
   [159]Open in a new tab

   NA, not specified or reported in the published study.[160]10
   ^a

   In comparison to published data, autism spectrum disorder and other
   behavior problems and skin abnormalities are underrepresented in our
   cohort: 6/28 versus 20/38 (P = 5.2 × 10^−3) and 5/28 versus 14/38
   (P = 4.6 × 10^−2), respectively. One‐tailed Z score test for two
   population proportions is used.

   In two subjects undergoing muscle biopsy, skeletal muscle mitochondrial
   DNA content was reduced. The first subject (Female 17) is a 6‐year‐old
   nondysmorphic girl with a history of neonatal hypotonia, esophageal
   reflux, and global developmental delay. A quadriceps muscle biopsy
   demonstrated mild fiber type variation and abnormal pleomorphic
   mitochondria on electron microscopy (Fig. [161]1E–G). After correction
   for the reduced citrate synthase activity, respiratory chain enzyme
   activity analysis demonstrated reductions of multiple complexes, with
   relative sparing of complex II activity. Sequencing of mitochondrial
   DNA from the muscle sample did not detect any known or likely
   pathogenic variants. Mitochondrial DNA content in muscle was 39% of
   age‐matched control muscle. Clinical WES demonstrated a de novo
   heterozygous c.453_454del (p.S152 fs) pathogenic variant in DDX3X, with
   no other variants in known disease‐associated genes that explain the
   patient's clinical presentations. The second subject is a 47‐year‐old
   woman (Female 13) with history of global developmental delay,
   intellectual disability, short stature, dysmorphic features,
   microcephaly, and unilateral renal agenesis. She learned to sit at two
   years of age and walk at eight, and she only learned to say simple
   words. In her early 40 sec, she regressed, becoming nonverbal and
   unable to ambulate or to use her arms. She was found to carry a de novo
   heterozygous c.1600C>T (p.R534C) pathogenic variant in DDX3X
   (Table [162]1). The same variant was also observed in Female 14 in our
   cohort, and a variant involving the same codon (p.R534H) has been
   reported in a patient with ID.[163]11 A quadriceps muscle biopsy
   demonstrated severe mitochondrial and lipid depletion, and reduction in
   mitochondrial size similar to Female 17. Mitochondrial DNA content in
   muscle was 26% of the mean value for age‐ and tissue‐matched controls.
   A reduction in all mitochondrial respiratory chain complex activities
   was observed. However, the reduction do not meet diagnostic criteria
   after correction for the low citrate synthase activity.

Discussion

   Normal RNA metabolism requires the function of RNA helicases (RH), and
   yet, the exact function of most human RH remains unknown. There are six
   superfamilies of RH known with more than 50 human members in
   superfamily two that are characterized by a DExH and DExD signature in
   their Walker B motifs, thus termed DHX and DDX proteins. Genetic
   studies have begun to address the role of altered RH function in human
   disease (e.g., DHX37 and DHX30).[164]16, [165]17 DDX3X encodes a
   DEAD‐box RNA helicase important in transcription, splicing, RNA
   transport, and translation.[166]18, [167]19 In a diagnostic laboratory
   referral cohort of 4839 subjects with DD and/or ID, we have identified
   26 postnatal individuals (24 females, 2 males) with syndromic ID or DD
   carrying pathogenic or likely pathogenic variants in DDX3X, and one
   fetus with abnormal ultrasound findings carrying a de novo variant of
   unknown significance in DDX3X; an additional four females were
   identified through research WES at BHCMG. The overall frequency of
   pathogenic or likely pathogenic DDX3X variations in our diagnostic
   laboratory referral cohort is 0.54% of the total (26/4839) and 1.12% of
   females (24/2152), similar to a previous report (0.6% and 1.5%
   respectively),[168]10 confirming mutations in DDX3X are one of the most
   common genetic causes of unexplained ID in females. In our diagnostic
   laboratory, DDX3X ranks third among approximately 450 genes for the
   occurrence of de novo variants, with ARID1B first (43 individuals) and
   ANKRD11 sec (29 individuals).

   In addition to confirming and extending published mutational data, our
   phenotypic analyses expand the phenotypic spectrum associated with
   DDX3X variants in females. For instance, we found respiratory problems
   and congenital heart disease in 5/28 and 5/7 of our subjects,
   phenotypes not previously described in the original description of
   DDX3X related disorders,[169]11 although observed in a subsequent
   report of two females.[170]12 We found no evidence for
   genotype–phenotype correlations between the mutations we identified and
   age at onset or phenotypic severity. Previously reported individuals
   ranged in ages from 1 to 33 years. We report the phenotype of a
   47‐year‐old woman (Female 13) who had manifestations consistent with
   DDX3X disorder and a clinical picture of previously unreported
   late‐onset neurologic decline. The decline is unrelated to intercurrent
   illness, and her motor function is at least in part responsive to
   physical therapy. Of note, other X‐linked DD/ID loci, exemplified by
   female FMR1 premutation [MIM: 300623] and MECP2 duplication
   carriers,[171]20 are notable for late adult onset neurological or
   neurocognitive phenotypes.

   Two variants reported in this study, c.1600C>T (p.R534C) and c.1703C>T
   (p.P568L), and three previously reported, c.641T>C (p.I214T), c.931C>T
   (p.R311*), and c.1084C>T (p.R362C),[172]11 have also been observed to
   occur somatically in association with medulloblastoma, malignant
   melanoma, and esophageal squamous cell carcinoma
   ([173]http://cancer.sanger.ac.uk/cosmic). Malignancy has not been
   reported in the 31 patients included in this study. However, pathway
   analysis for the highest 5% and lowest 5% genes expressed in RNAseq
   data from dermal fibroblasts obtained in one subject (Female 13) showed
   enrichment in cell cycle control of chromosomal replication and
   double‐strand break repair pathways (Table [174]S2). Future studies
   will elucidate whether individuals carrying DDX3X variants are at risk
   for the development of malignancies.[175]21 In summary, we identified
   31 unrelated patients with causal variants in DDX3X and expanded the
   genotypic and phenotypic spectrum of DDX3X‐related disorders. The
   collective data suggest that DDX3X defects are a frequent cause of
   syndromic ID in females, and the causal variants are likely to be
   loss‐of‐function (ExAC database showed pLI = 1.00 for DDX3X).[176]22

Author Contributions

   Conception and design of the study: X. W., B. L., J. R. L., B. H. G.,
   P. M. Acquisition and analysis of data: X. W., J. E. P., J. A. R., B.
   L., J. R. L., B. H. G., P. M. C. A. B., F. S., L. I., J. M. H., S. E.
   H., T. M. M., A. S., J. Z., J. B., M. S. L., W. H., F. V., M. A. W., W.
   B., R. X., P. F. L., Y. S., A. G., E. Y. G., Y. J., S. A. D., A. W. H.,
   M. M. K., D. P., J. P., D. M. M., N. H., J. W. B., L. V. M., R. A. G.,
   M. K. E., Z. C. A., T. H., A. M. A., S. C., C. M. E., F. X., Y. Y.
   Drafting a significant portion of the manuscript or figures: X. W., J.
   E. P., J. A. R., J. R. L., B. H. G., P. M.

Conflicts of Interest

   The Department of Molecular and Human Genetics at Baylor College of
   Medicine receives revenue from clinical genetic testing done at Baylor
   Genetics. J. R. L. has stock ownership in 23 and Me, is a paid
   consultant for Regeneron Pharmaceuticals, has stock options in
   Lasergen, Inc and is a co‐inventor on multiple United States and
   European patents related to molecular diagnostics for inherited
   neuropathies, eye diseases, and bacterial genomic fingerprinting.

Supporting information

   Figure S1. Brain MRI images of subjects with DDX3X variants.
   [177]Click here for additional data file.^ (4.7MB, tif)

   Table S1. Clinical features and DDX3X variants in subjects enrolled in
   this study. Detailed clinical features are only reported for the
   subjects in whom we were able to obtain additional informed consent.
   [178]Click here for additional data file.^ (16.9KB, xlsx)

   Table S2. RNAseq pathway analysis for Female 13.
   [179]Click here for additional data file.^ (8.8KB, xlsx)

   Table S3. Members of the Undiagnosed Diseases Network.
   [180]Click here for additional data file.^ (24.1KB, xlsx)

Acknowledgments