De novo, deleterious sequence variants that alter the transcriptional activity of the homeoprotein PBX1 are associated with intellectual disability and pleiotropic developmental defects

Abstract

We present eight patients with de novo, deleterious sequence variants in the PBX1 gene. PBX1 encodes a three amino acid loop extension (TALE) homeodomain transcription factor that forms multimeric complexes with TALE and HOX proteins to regulate target gene transcription during development. As previously reported, Pbx1 homozygous mutant mice (Pbx1^−/−) develop malformations and hypoplasia or aplasia of multiple organs, including the craniofacial skeleton, ear, branchial arches, heart, lungs, diaphragm, gut, kidneys, and gonads. Clinical findings similar to those in Pbx mutant mice were observed in all patients with varying expressivity and severity, including external ear anomalies, abnormal branchial arch derivatives, heart malformations, diaphragmatic hernia, renal hypoplasia and ambiguous genitalia. All patients but one had developmental delays. Previously reported patients with congenital anomalies affecting the kidney and urinary tract exhibited deletions and loss of function variants in PBX1. The sequence variants in our cases included missense substitutions adjacent to the PBX1 homeodomain (p.Arg184Pro, p.Met224Lys, and p.Arg227Pro) or within the homeodomain (p.Arg234Pro, and p.Arg235Gln), whereas p.Ser262Glnfs*2, and p.Arg288* yielded truncated PBX1 proteins. Functional studies on five PBX1 sequence variants revealed perturbation of intrinsic, PBX-dependent transactivation ability and altered nuclear translocation, suggesting abnormal interactions between mutant PBX1 proteins and wild-type TALE or HOX cofactors. It is likely that the mutations directly affect the transcription of PBX1 target genes to impact embryonic development. We conclude that deleterious sequence variants in PBX1 cause intellectual disability and pleiotropic malformations resembling those in Pbx1 mutant mice, arguing for strong conservation of gene function between these two species.

Introduction

PBX1, the gene encoding the Pre-B-cell leukemia homeobox (PBX) transcription factor (1), is the mammalian homologue of the extradenticle (exd) gene in Drosophila melanogaster (2). PBX1 shares 71% identity to exd and encodes a three amino acid loop extension (TALE) homeodomain transcription factor (2,3). PBX1 and its related family members, PBX2–4 (3), dimerize with other TALE class homeodomain proteins from the Myeloid Ecotropic Integration Site (MEIS) and PBX regulating protein (PREP) families through a PBC domain to form nuclear complexes that enhance the binding specificity of HOX proteins to DNA, thus regulating target genes that control segment identity and organ patterning during embryogenesis (3–6). We previously reported that Pbx1 is widely transcribed throughout the developing murine embryo and Pbx1-deficient mice that are homozygous for a null allele (Pbx1^−/−) develop pleiotropic developmental defects. These comprise malformations, severe hypoplasia, or aplasia of multiple organs, including craniofacial skeleton, ear pinnae, branchial arch-derived structures, heart, lungs, diaphragm, liver, stomach, gut, pancreas, kidneys and gonads (4,7–9). Heterozygous mice are viable and fertile, but are smaller in size than wildtype (WT) mice, with a 30% reduction in weight prior to eight weeks of age (4). Pbx1^−/− embryos also exhibit abnormal derivatives of the second branchial arch, with elongated and deformed hyoid bones, dysplastic styloid processes and lack of stapes and oval windows (4).

Recently, eight patients were reported with congenital anomalies affecting the kidney and urinary tract (CAKUT), and interstitial deletions of varying sizes that all encompassed PBX1 at chromosome 1q23.3q24.1 (10). All patients manifested renal hypoplasia, with or without renal dysplasia that presented as a loss of corticomedullary differentiation and renal cysts (10) As expression of the PBX1 gene was detected in the fetal and adult kidneys in humans, it was concluded that PBX1 was a new causative gene for CAKUT (10). Extrarenal manifestations in those patients were described as developmental delays and autism spectrum disorder, outer ear anomalies, hearing impairment, and cardiac defects. However, the report could not unequivocally associate the observed abnormalities with haploinsufficiency of PBX1 due to the inclusion of additional genes in the large interstitial deletions (10). A recent study also examined 204 patients with CAKUT syndrome for variants in 330 genes (11) The authors identified three, de novo, heterozygous PBX1 variants and two deletions for a detection frequency of 2.5%. Out of those cases, a female with hearing loss and renal hypoplasia had a frameshift variant in PBX1 (p.Asn143Thrfs*37), an 11-year-old female with renal cystic dysplasia, developmental and growth delays and a long and narrow face, had a nonsense variant (p.Arg184*) and a fetus with oligohydramnios and renal hypoplasia had a splice site variant (c.511–2 A > G). In addition, two patients with PBX1 deletions manifested renal dysplasia with hearing impairment and renal agenesis/dysplasia, developmental differences, microcephaly together with facial anomalies, respectively (11).

We now present eight patients who are heterozygous for de novo, deleterious sequence variants in PBX1 and establish pathogenic mutations of this gene as causative in the etiology of a pleiotropic intellectual disability syndrome with ear, branchial arch, cardiac, pulmonary, diaphragmatic, renal, and gonadal malformations. Importantly, functional assays testing five of the sequence variants demonstrated overall abnormal transactivation capabilities for the mutant PBX1 proteins.

Results

Clinical features of the eight patients with de novo, PBX1 sequence variants

The clinical findings for the eight patients, ordered according to the position of their PBX1 sequence variants (patient 1–7; while patient 8 carries an identical mutation to patient 5), have been summarized in Table 1. Further clinical descriptions have been provided as Supplementary Files.

Table 1.

Clinical findings in eight patients with de novo sequence variants in PBX1

	Patient 1	Patient 2	Patient 3	Patient 4	Patient 5	Patient 6	Patient 7	Patient 8
Aural findings
Dysplastic ears	−	−	−	−	+	+	+	−
EAC stenosisa	−	−	−	−	−	+	+	−
Microtia	−	−	−	−	+	+	−	−
Hearing loss	−		−	−	−	+; unilat.b	−	−
Branchial findings
Cartilaginous rests	−	−	+	−	−	−	−	−
Cardiac/Pulmonary
Ebstein anomaly	−	−				−	+	−
PDAc	−	+d		+	+	−
Tetralogy of Fallote	+				+; unilat.	−		−
Eventration diaphragm	−	+; unilat.				−		−
Lung hypoplasia	−	+/−f			+; unilat.	−		−
Renal
Renal hypoplasia	−g	−	−	−	−	+	+; UTIsh	−
Pyelocaliectasis	−g	−	−	+; unilat.	−	−	−	−
Increased echogenicity	−g	−	−	+	−	−	−	−
Dilated fetal ureters	−g	−	−	−	+	−	−	−
Craniofacial
Dysmorphism	−	+		−	+i	+	+j	−
Micrognathia	–	+			+	+		−
Wide neck/nuchal fold	−	−		+	+			−
Other
Brachydactylyk	−	−			+	−	+	−
Cryptorchidism	−	+l	+	NA	+	NA	NA	+m
PBX1variant	p.Arg184Pro	p.Met224Lys	p.Arg227Pro	p.Arg234Pro	p.Arg235Gln	p.Ser262Glnfs*2	p.Arg288*	p.Arg235Gln

EAC stenosis^a = external auditory canal stenosis; unilat.^b = unilateral; PDA^c = patent ductus arteriosus; +^d = PDA and secundum ASD; Tetralogy of Fallot^e = Tetralogy of Fallot with absent pulmonary valve, severe pulmonary stenosis, ventricular septal defect; +/−^f = probable; UTI^h = urinary tract infections; −^g = no renal ultrasound scan was performed in Patient 1; +ⁱ = blue sclera, ear microtia with prominent lobules, prominent nuchal folds, and micrognathia; +^j = Low anterior/posterior hairlines; coarse hair, deep-set eyes, short philtrum, an exaggerated Cupid‘s bow, bifid uvula, prominent mandible, brachydactyly type D, pes planovalgus and genu valgum; absent axillary hair; Brachydactyly^k = brachydactyly and/or nail hypoplasia; +^l = microphallus, rugated tissue superior and lateral to phallus, intra-abdominal testicular tissue and persistence of the Müllerian ducts; +^m = undervirilized external genitalia with a micropenis (2.2 x 0.8 cm), labial folds without palpable gonads and two perineal openings comprising a vaginal introitus and urethra meatus. Pelvic ultrasound showed uterus didelphys.

The patients ranged from 3 months to 27 years of age. During the pregnancies, oligohydramnios, increased nuchal folds, and unilateral diaphragmatic eventration were each present in two patients. Developmental delays were present in all but two children (patient 1 and patient 8) who were 3 and 12 months of age respectively. In those with delays, the age of independent walking ranged from 18–36 months and single words were first uttered at around 2 years of age. Only one patient was severely delayed and carried the additional diagnoses of autism and obsessive/compulsive disorder. Hypotonia was present in four patients. All patients except one had unremarkable growth parameters at birth, but postnatal short stature and failure to thrive developed in two, with weight and height measurements less than the fifth centile. Microtia and/or dysplastic ear helices were noted in three patients (Fig. 1A), with absent tragus and prominent lobes and attached earlobes. One patient had unilateral, mild to moderate conductive hearing loss with bilateral stenosis of the external auditory canals. Craniofacial dysmorphic findings, apart from ear anomalies and micrognathia, did not delineate a recognizable facial phenotype. One male had cartilaginous neck rests, a rare finding caused by aberrant development of the branchial arches (Fig. 1B) (12). Cardiac findings comprised Ebstein anomaly, Tetralogy of Fallot and patent ductus arteriosus that required surgery in two patients. One male had right lung hypoplasia with right diaphragm eventration/paralysis and another patient with a male karyotype and atypical sexual development also had a unilateral diaphragmatic eventration. Renal anomalies comprised hypoplasia, left pyelocaliectasis and dilated fetal ureters in seven patients; patient 1 did not undergo a renal ultrasound scan. The oldest patient also presented with recurrent urinary tract infections. Four males with karyotype 46, XY had cryptorchidism and two males had atypical sexual development: one with intra-abdominal testes and retained Müllerian structures and the other with micropenis and undervirilized male external genitalia, a vaginal introitus and uterus didelphys.

Figure 1.

Abnormalities in ear and branchial arch-derived structures in patients with heterozygous, de novo, PBX1 sequence variants. (A) Right ear from Patient 6, showing a dysplastic helix with absence of the tragus and attached earlobe. (B) Cartilaginous neck rests seen in Patient 3. (C) Mapping of mutations onto the homeodomain structure. (D) Protein sequence analysis showing sites of altered residues in PBX1 in relation to an interactive DNA strand. (wildtype; top row)

Identification of PBX1 sequence variants using whole exome sequencing

For Patient 1, whole exome sequencing (WES) identified a de novo, heterozygous variant in PBX1: c.551 G > C, predicting (p.Arg184Pro) that was the only potential disease-causing candidate using a combination of in silico approaches. This sequence variant, which was confirmed by Sanger sequencing, affects the PBC-B domain that is required for interaction with MEINOX proteins (3). The variant was identified in a single trio from a cohort comprising a total of 120 individuals (65 females and 55 males) from 30 families affected by different types of sporadic or inherited congenital heart disease (CHD) for a total of 57 affected individuals and 63 unaffected controls.

For Patient 2, WES revealed a de novo, heterozygous variant in PBX1: c.671 T > A, predicting (p.Met224Lys). The mean depth of coverage was 150x with a quality threshold of 97.8%. This variant is a non-conservative amino-acid substitution, occurring at a position that is conserved across species. In view of the anomalies in the ear, renal, and branchial arch derivatives in these patients, we examined the genomic regions corresponding to the genes involved in branchio-oto-renal (BOR) syndrome (13) and to HNF1B that is mutated in CAKUT. No mutations were detected in these genes or any other CAKUT gene, despite high sequence coverage (EYA1 100%, SIX5 99.1%, SIX1 100%, and HNF1B 100%).

Patient 3 had a de novo, heterozygous sequence variant, c.680 G > C, predicting (p.Arg227Pro) in PBX1. The variant GATK QUAL score was 673 (>500 considered high quality) and coverage was >80x. The Arg227 residue is proximal to the predicted DNA-binding homeodomain of PBX1 (amino acids 233–295; Fig. 1C). Arg227 is in a region of high sequence identity among PBX family members, at a residue with complete conservation across species (Fig. 1D). No mutations were detected in the genes involved in BOR syndrome and HNF1B, despite high sequence coverage (EYA1 100%, SIX5 99.1%, SIX1 100%, and HNF1B 100%).

In Patient 4, WES identified a de novo, heterozygous variant, c.701 G > C, predicting (p.Arg234Pro), in the homeodomain of PBX1 that exhibits complete conservation across species (Fig. 1C and D). No mutations were detected in three genes associated with BOR syndrome and HNF1B, all of which had coverage of 100% (EYA1, SIX5, SIX1, and HNF1B).

For Patient 5, WES demonstrated a de novo, heterozygous variant, c.704 G > A, predicting (p.Arg235Gln), in PBX1. This variant is in a genomic region of high conservation (Fig. 1C and D). Despite good coverage, no mutations were identified in the BOR genes (EYA1 100%, SIX5 91.5%, SIX1 100% and HNF1B 100%).

In Patient 6, mean depth of coverage was 109x and quality threshold (percentage of captured region covered by at least 10 sequence reads) was 96.1%. The results showed heterozygosity for a de novo, sequence variant in PBX1, c.783dupC, predicting (p.Ser262Glnfs*2). This frameshift mutation is expected to lead to nonsense-mediated decay and to result in premature protein truncation. Coverage of the genes involved in BOR syndrome was high (EYA1 100%, SIX5 99.8%, SIX1 100% and HNF1B 100%) and no mutations were detected.

In Patient 7, clinical WES was performed with a mean depth of coverage of 95x and a heterozygous, de novo, PBX1 (p.Arg288*) sequence variant with a GATK score of 814 and a read depth of 84x was detected. The variant was confirmed by Sanger sequencing and results in a premature protein truncation.

In Patient 8, whole exome sequencing showed a de novo, heterozygous variant in PBX1, c.704 G > A, predicting p.Arg235Gln, that was identical to the variant identified in Patient 5. Patient 8 was also heterozygous for a variant in the steroidogenic factor-1 (NR5A1) gene, c.1379 A > G, predicting (p.Gln460Arg). However, six individuals were reported in the Exome Aggregation Consortium (ExAC) database to have heterozygous missense variants at the Gln460 residue in NR5A1: five with (p.Gln460Pro) and one with (p.Gln460Leu). In addition, the above substitutions occur at a position that is not conserved across species. Lastly, PolyPhen-2 predicted that the NR5A1 variant identified in Patient 8 is likely to be benign (PolyPhen-2 score of 0). Given all of the above, the NR5A1 variant was considered of unknown significance and the PBX1 variant deemed as the causative mutation in Patient 8.

All of the PBX1 sequence variants were heterozygous and de novo, and none of them were found in 1000 genomes or the ExAC browser databases. Furthermore, all variants were predicted to be pathogenic and the majority had high Genomic Evolutionary Rate Profiling (GERP) scores (Table 2). All patients carried only sequence variants in PBX1 except for patient 8, as discussed above, and no additional variants in genes associated with CAKUT were identified.

Table 2.

Nucleotide variants in PBX1 (NM_002585.3)

Chromosome	Nucleotide	Protein	Exon	ExACa /dbSNPb	SIFTc	Poly-Phen-2d	Mutation Taster	CADDe score	GERPf score
chr1(GRCh37): g.164768976G>C	c.551G>C	(p.Arg184Pro)	4	−/−g	Damaging	Prob damaging	DCh	34	5.76i
chr1(GRCh37): g.164769096T>A	c.671T>A	(p.Met224Lys)	4	−/−	Damaging	Poss damaging	DC	31	5.64
chr1(GRCh37): g.164769105G>C	c.680G>C	(p.Arg227Pro)	4	−/−	Damaging	Prob damaging	DC	34	5.64
chr1(GRCh37): g.164769126G>C	c.701G>C	(p.Arg234Pro)	4–5j	−/−	Damaging	Prob damaging	DC	34	5.64
chr1(GRCh37): g.164776781G>A	c.704G>A	(p.Arg235Gln)	5	−/−	Damaging	Prob damaging	DC	26.4	4.16k
chr1(GRCh37): g.164776860_ 164776861insC	c.783dupC	(p.Ser262Gln fs*2)	5	−/−	ndl	nd	DC	nd	5.6
chr1(GRCh37): g.164781251C>T	c.862C>T	(p.Arg288*)	6	−/−	nd	nd	DC	nd	3.7

ExAC^a = Exome aggregation consortium browser; dbSNP^b = Single nucleotide polymorphism database; SIFT^c = Sorting tolerant from intolerant; Poly-Phen-2^d = Polymorphism Phenotyping v2; CADD^e = Combined Annotation Dependent Depletion; GERP^f = Genomic evolutionary rate profiling; −/−^g = absent in both ExAC browser and dbSNP; DC^h = predicted to be disease-causing; 5.76ⁱ—Exon 4 average GERP score was 4.19; 4–5^j = codon crosses a splice site; 4.16^k—Exon 5 average GERP score 4.71; nd^l = no data.

Prediction of protein structure perturbations

Structural mapping of the sites of the four, most downstream (3’), human sequence variants in PBX1 showed that the homeodomain region was disordered and charged, two characteristic properties of DNA-binding proteins that can contribute to specific, DNA sequence selectivity for DNA interactions (14) The mutations affecting Arg234, Arg235, and probably Arg227 are hypothesized to affect the ability of PBX1 to bind DNA, whereas Arg288 comes into direct contact with the DNA and may directly affect the mode of this interaction (Fig. 1C). The amino acids that were substituted in the PBX1 homeodomain of these patients were conserved in many species (Fig. 1D).

Functional assessment of the PBX1 sequence variants identified in six patients

The PBX1 sequence variants identified in six out of the eight patients described in this study (Patients 3–8; Table 1) were cloned in a pSG5 expression vector and confirmed by Sanger sequencing. After transient transfection in HEK293 cells in culture, western blot analysis showed that all of the constructs ectopically expressed PBX1 proteins of various sizes, according to the mutation introduced (Fig. 2). Mutations present in Patients 3, 4 and 5/8 resulted in the production of PBX1 protein of the same size as the WT (42 kDa). In contrast, mutations found in Patients 6 and 7 yielded truncated proteins of shorter size than WT (approximately 35 and 37 kDa, respectively). These results demonstrated that all of the PBX1 mutations identified in these six patients do not affect the production of the relative proteins.

Figure 2.

Analysis of ectopic expression of wildtype and mutant PBX1 cDNAs in HEK293 cells. Human HEK293 cells were transfected in duplicate with either PBX1 wildtype (WT) or PBX1-Mutant cDNAs. Two separate western blots (divided by vertical line) detect the five different PBX1 mutant proteins identified in the six patients (Patient 3 and 4, left blot; Patient 5/8, 6 and 7, right blot). The blots were revealed with the PBX1 Ab. PBX1 WT and untransfected HEK293 cells were used as controls in both experiments. Exposure of the western blots was reduced to a few seconds in order to visualize the ectopic proteins without saturation of the signal; this did not allow concomitant detection of the endogenous PBX1 protein, which is present in HEK293 cells as we previously reported (16). Anti-ACTIN was used to normalize for protein loading. Protein sizes are indicated on the left of the blots (kiloDaltons; kDa).

To assess the transactivation capabilities of the PBX1 proteins carrying the sequence variants described above, we performed luciferase reporter assays following our established protocols (15,16) We employed a regulatory element present on human chromosome 1 within the PBX1 promoter (chr1: 164, 527, 984–164, 528, 440; hg19), which is highly conserved within vertebrates (thereafter named Pbx1 Prom) (Fig. 3A). Using chromatin immunoprecipitation coupled with quantitative polymerase chain reaction (ChIP-qPCR) on mouse embryonic hindlimbs and midfaces at embryonic day (E)10.5, we unequivocally demonstrated that the orthologous element within the mouse Pbx1 promoter (chr1: 170, 362, 511–170, 362, 668; mm9) is strongly bound by Pbx1 itself, suggesting the presence of an autoregulatory loop (Fig. 3B). PBX1 Prom, which contains two putative PBX1 binding sites, was cloned upstream of the luciferase reporter gene (Fig. 3C) within a pGL3 vector (pGL3-PBX1 Prom). We then used the human HEK293 cell line, which expresses endogenous PBX1 protein (16), for luciferase reporter assays. These revealed that when WT PBX1 and its cofactor PREP1 (5) were co-transfected in these cells, luciferase activity driven by the pGL3-PBX1 Prom was increased by almost two fold versus the pGL3 empty vector (Fig. 4A). These results are consistent with the moderate, but reproducible and statistically significant, transactivation activity of PBX1 with its cofactor PREP1 versus empty vector in similar reporter assays in cell culture, as we previously reported (15). In contrast, when we co-transfected the PBX1 mutant cDNAs together with the cDNA for PREP1 in HEK293 cells, luciferase activity driven by the pGL3-PBX1 Prom was significantly reduced for all five PBX1 mutations analysed (Fig. 4A). Furthermore, in order to evaluate the intrinsic transactivation capability of the PBX1 mutations described above without the interference of endogenous PBX1 protein, we generated a murine cell line with Pbx1 loss-of-function (LOF). We used an available primary mesenchymal cell line derived from E15.5 mouse embryos (15) and targeted exon 3 of Pbx1 (mc4-Pbx1 LOF) via clustered regularly interspaced short palindromic repeats (CRISPR)/Cas-mediated genome editing (17). After targeting of exon 3, western blot demonstrated a striking reduction of Pbx1 protein [both Pbx1a and Pbx1b (16)] in the clone with insertions/deletions (indels) affecting both alleles (Pbx1 LOF), as compared with WT and clones with heterozygous indels (Fig. 4B). Interestingly, reporter assays performed as described above using Pbx1 LOF mc4 cells revealed that luciferase activity driven by the pGL3-PBX1 Prom was significantly reduced only for the mutations present in Patients 4 and 6 (Fig. 4C). These results establish that, in a cellular system with markedly reduced levels of endogenous PBX1, only the PBX1 proteins carrying the mutations identified in Patients 4 and 6 intrinsically alter the transactivation capability of the PBX1 transcription factor. These findings also strongly suggest that the endogenous WT PBX1 protein present in HEK293 cells, which resembles human haploinsufficiency, interacts with the ectopic mutant proteins. This abnormal interaction of WT and mutant PBX1 could adversely affect target gene transactivation.

Figure 3.

Identification of a PBX1 regulatory element within PBX1 own promoter. (A) Adapted, UCSC genome browser view of the human PBX1 locus (hg19). Regulatory element present on chromosome 1 within the human PBX1 promoter (chr1: 164, 527, 984–164, 528, 440) is highlighted in grey. This region displays high conservation within vertebrates. The light-colored bar reflects uncertainty in the relationship between the DNA of human and chimp, due to lack of sequence in relevant portions of the alignments. (B) Murine Pbx1 binding to Pbx1 Prom by ChIP-qPCR. Histograms represent the enrichment of each region following immunoprecipitation with the Pbx1 Ab and the negative control Ab (IgG) calculated as percentage of input. Pbx1 Prom is strongly bound by Pbx1 itself in both E10.5 midfaces and hindlimbs. An unbound region was employed as a negative control (Neg. control; chr11: 103, 623, 380–103, 623, 491, mm9). (C) A 457 bp fragment within the human PBX1 promoter was cloned into the pGL3-TK vector using the KpnI restriction site. Left, vector map; Right, PBX1 promoter sequence containing two putative PBX binding sites: TGATTG and the canonical decameric Pbx-Prep motif CTGTCAATCA (27).

Figure 4.

Functional assessment of the transactivation capability of PBX1 mutant proteins carrying the mutations identified in six patients. (A) Luciferase activity driven by PBX1 Prom in human HEK293 cells transfected with PBX1 and PREP1 expression constructs. Values represent fold activation of pGL3-hPBX1 Prom over basal promoter activity (pGL3 empty vector, set to 1) normalized by total amount of protein. Each value is obtained from the average of at least five independent experiments, each performed in triplicate. Error bars represent the standard error of the mean (SEM). Asterisks represent P-values: (***P-value < 0.001), (****P-value < 0.0001). (B) Western blot analysis of murine mc4 cells subjected to CRISPR/Cas-mediated gene targeting of Pbx1. The blot was revealed with the PBX1 Ab. Anti-Actin was used to normalize for protein loading. Loss-of-function (LOF); presence of indels in both Pbx1 alleles. WT; wildtype. Het; presence of indel in one Pbx1 allele. (C) Luciferase assays were conducted as described in (A) using the murine cell line with Pbx1 LOF obtained via CRISPR/Cas-mediated genome editing (mc4-Pbx1 LOF). Error bars represent the SEM. Asterisks represent P-values: (*P-value < 0.05). ns; not significant.

It is also worthy of note that to execute their developmental programs on specific target genes, PBX homeodomain proteins must enter the nucleus, a process that is concomitantly regulated by the presence of two nuclear localization signals (NLS) within the homeodomain (18) and by PBX dimerization with TALE and HOX cofactors (5). To assess the ability of the PBX1 mutant proteins to translocate into the cell nucleus, we further analysed the cytoplasmic and nuclear fractions of HEK293 cells transfected with cDNAs carrying the PBX1 mutations identified in the six patients together with the PREP1 cofactor. Notably, levels of the PBX1 proteins carrying the mutations present in Patients 6 and 7 (with truncations that delete the homeodomain 3’ region comprising the second NLS) were remarkably reduced in the nuclear fraction as compared with WT control (Fig. 5). However, in these mutants, the PBX1 cytoplasmic protein fraction was not reduced, ruling out confounding effects due to degradation or efficiency of transfection. These results support a scenario whereby the altered PBX1-dependent transactivation capabilities observed in Patients 6 and 7 can result, at least in part, from defective nuclear translocation.

Figure 5.

Mutations identified in Patients 6 and 7 result in lower levels of PBX1 proteins in the nucleus. Western blot analysis of PBX1 protein levels in cytoplasmic (left) and nuclear (right) extracts from human HEK293 cells transfected for 24 h with either PBX1 wildtype (WT) or PBX1-Mutant cDNAs together with the cofactor PREP1. Cytoplasmic and nuclear fractions separated by empty lane. Patients 6 and 7 show remarkably reduced PBX1 levels in the nuclear fraction (black empty asterisks). Equal loading and nucleo-cytoplasmic fractionation were monitored with anti-Nucleolin and anti-Tubulin antibodies. Minor nuclear contamination of the cytoplasmic fraction is detectable. Neg. Ctrl, Negative control: untransfected HEK293 cells.

Discussion

We have described eight individuals with intellectual disability, developmental delays, and variable abnormalities (including craniofacial, ear, branchial arch, cardiac, pulmonary, diaphragmatic, renal, and genital defects) carrying de novo, heterozygous, deleterious sequence variants in the PBX1 gene. The clinical findings in individual patients overlapped sufficiently to delineate a recognizable syndrome. The pleiotropic developmental phenotype is consistent with the broad expression of Pbx1 during murine embryogenesis, with predominant localization to neural tissues [brain, spinal cord, and ganglia (6,19)] as well as head and olfactory epithelium, heart, lung, gut, spleen, pancreas, kidney, and gonads (4,8,15). Interestingly, in view of the dysplastic ear helices observed in these patients, murine Pbx1 craniofacial expression at E13.5 is highest in the auricular pinna (4). Pbx1 is also detectable in the ectoderm of the second branchial arch and around the first pharyngeal groove at E11.5 (4). The human phenotype closely recapitulates the abnormalities previously described in Pbx1^−/− mouse embryos that included cardiac outflow tract defects with absence of the ductus arteriosus at E14.5; diaphragmatic hernia at E15.5; bilateral or unilateral renal hypoplasia; axial malpositioning of the kidneys; unilateral renal agenesis and impaired renal function (4,7–9,16,20). Pbx1^−/− mouse embryos also failed to develop typical branchial arches and, at E10.5, the pharyngeal groove separating arches 3 and 4 was absent, and the fourth branchial arch was small compared with WT embryos (8), resulting in absence or hypoplasia of organs derived from the caudal pharyngeal region (21). Conditional mutant mice with loss of Pbx1 function in the ventricular zone and in postmitotic neurons demonstrated abnormal laminar patterning of the neocortex, with hypoplasia and dyslamination of the frontal cortex (19) Accordingly, the developmental delays observed in our patients are likely to result from impaired PBX1-dependent transcription of critical target genes in the developing brain. The ear and branchial arch anomalies noted in our patients also bear some resemblance to those present in BOR syndrome (13). However, despite high sequence coverage, no mutations were found in any of the genes implicated in BOR syndrome (EYA1, SIX5 and SIX1) in our patients. Interestingly, the ear anomalies, including stenosis of the external auditory canals, and the renal hypoplasia observed in the two patients with truncating mutations of PBX1 closely resemble the phenotypes of previously reported cases with large interstitial chromosome deletions encompassing PBX1 (10). Recently, mutations in PBX1 were also identified in five cases out of a cohort of 204 patients with CAKUT syndrome (11). Clinical findings in the five patients included renal hypoplasia, renal dysplasia with a horseshoe kidney, deafness, developmental delays, growth delays with microcephaly and facial dysmorphic features (11). Three of the reported cases exhibited PBX1 variants, while two had deletions encompassing PBX1. However, no functional studies were provided for any of the patients.

One striking observation concerning the patients described in this study is the high variability of their phenotypes, with disparate clinical findings including branchial arch rests, diaphragmatic eventration, and ambiguous genitalia (Table 1). Possible explanations for the phenotypic variability observed in patients with PBX1 mutations may include allelic heterogeneity. Indeed, it would appear that loss-of-function alleles (deletions, frameshift and stop variants) may result in CAKUT, hearing impairment, and developmental differences, whereas missense variants may result in more complex phenotypes that include cardiac defects, diaphragmatic eventration, and ambiguous genitalia. However, the regulatory function of PBX1 as a HOX gene cofactor is also likely to result in substantial phenotypic heterogeneity. It is worthy of note that two patients with phenotypes of different complexity and severity carry the same PBX1 variant (Patients 5 and 8), suggesting that interactions of PBX1 with TALE or HOX cofactors can be perturbed in a different fashion and with varied outcomes. Lastly, the presence of different genetic modifiers interacting with PBX1 or of mutations in non-coding regulatory elements of the genome is also likely to play a role in driving genetic heterogeneity.

In humans, PBX1 encodes a 430 amino acid protein with a homeobox domain at residues 233–305 that serves as a eukaryotic DNA-binding motif. Four of the sequence variants described in our patients consisted of substitutions involving closely spaced residues within the homeodomain; in addition one variant involving p.Arg227 was immediately adjacent to the homeodomain and within the PBC domain, which is conserved among Pbx, Exd and Ceh-20, and interacts with Meis/Prep cofactors (3). This clustering of mutations is striking and argues for a critical role of the homeodomain residues for PBX1 function in humans. In support of a phenotype associated with haploinsufficiency for PBX1 is the dearth of loss-of-function variants listed in the ExAC browser and the pLI score (0.91) of the gene (22). Interestingly, population missense variants are not only rare in PBX1, but we noted that PBX1 has a ∼120 amino-acid missense-depleted region (MDR) encompassing the homeodomain (23). Notably, the PBX1 missense mutations identified in our patients are located within or just proximal to this MDR, further supporting their pathogenicity. It will be of interest to further ascertain why this region of PBX1 is sensitive to missense variation. It is also unclear if certain missense substitutions may act as dominant negative mutations, given the propensity of PBX1 complexes to form multimers with both TALE and non-TALE cofactors (3). Indeed, as PBX1 acts in a transcription complex with MEIS1 (3,24) and murine Meis1 is expressed in the second branchial arch (25), aberrant function of MEIS1 may also prove to be an important contributor to this aspect of the PBX1 phenotype. It is worthy of note that two of the five PBX1 mutations we analysed resulted in premature protein truncation due to the generation of early in-frame stop codons, but none of the identified mutations significantly perturbed protein production, as shown by our western blot analyses.

The identification of a strong auto-regulatory element within the conserved PBX1 promoter, which we characterized in vivo by ChIP analysis in different embryonic tissues, provided a powerful reagent for functional assessment of the PBX1 mutations under analysis. Parallel evaluation of the transactivation capability of PBX1 mutant proteins carrying the five different mutations in a system containing WT PBX1 (human HEK293 cells) versus a system with marked reduction of endogenous Pbx1 protein (murine mc4-Pbx1 LOF), afforded an assessment of two distinct scenarios. The former cellular assay reflects the transactivation capability of the different PBX1 mutations in the presence of WT protein, which resembles PBX1 haploinsufficiency in the patients. The latter assay, in contrast, allows the evaluation of the intrinsic transactivation capability of the different PBX1 mutations in a system with low levels of endogenous WT PBX1 protein. Interestingly, in the presence of WT PBX1, all five mutant proteins exhibited a significant decrease in transactivation capability, despite the different locations of the mutations within the protein domains (homeodomain or PBC) (3,24). In this system, the mutant proteins might either be directly responsible for the decrease of transactivation activity observed or might affect the capability of the endogenous WT protein to physiologically transactivate target genes, a scenario that closely resembles the human disease with PBX1 haploinsufficiency. Parallel evaluation of the PBX1 mutant proteins in the system with marked reduction of WT Pbx1 demonstrated that only two of the five mutant proteins (Patients 4 and 6) exhibited significantly diminished transactivation activity. These findings strongly suggest that these two mutations can directly affect the intrinsic capability of PBX1 to transactivate downstream transcriptional targets. Interestingly, while the mutation in Patient 7 shows a similar trend, but without reaching statistical significance (P-value = 0.08), mutations in Patients 3 and 5/8 do not exhibit this behavior. It is noteworthy that, while Patient 3 carries a mutation outside of the homeodomain, Patient 5/8 bears a mutation inside it, like Patients 4, 6 and 7, making it difficult to directly link the transactivation abnormalities of the mutated proteins solely to the presence of an abnormal homeodomain. Furthermore, the mutations identified in Patients 6 and 7 were associated with lower levels of PBX1 proteins in the cell nucleus, as compared with WT. These results highlight that in these cases altered transactivation capabilities can result, at least in part, from abnormal nuclear translocation of PBX1 and the consequent impaired formation of complexes comprising PBX1 and its cofactors, which is similar to the reported behavior of EYA1 (26). Taken together, our experiments suggest that the phenotypes observed in PBX1 mutant patients likely result from a combination of perturbed biological processes including: 1) intrinsic alterations of transactivation ability of the PBX1 mutant proteins, at least in Patients 4 and 6; 2) broadly abnormal interactions between mutant PBX1 and WT proteins (including PBX1 itself and other TALE or HOX cofactors) (24,27); 3) defective translocation of mutant PBX1 proteins to the nucleus in Patients 6 and 7. Overall, these scenarios can affect physiological transcription of PBX1 target genes, thus causing aberrations of normal embryonic development.

We demonstrate for the first time that de novo, deleterious PBX1 sequence variants can cause a syndromic form of intellectual disability with pleiotropic developmental defects that involve multiple organ systems. The novel phenotype we report here is highly variable in expressivity and severity between patients, consistent with findings we previously reported in mouse embryos with constitutive or conditional loss of Pbx1, suggesting high functional conservation of this essential developmental gene between these two species. It is likely that the precise definition of the human syndrome with PBX1 mutations will be refined as an increasing numbers of cases are reported and the mechanisms of action for the sequence variants are further understood.

Materials and Methods

Case reports

Clinicians were contacted after matching in GeneMatcher (28) and informed consent for sequencing was obtained from all patients. Medical history and phenotypic features were extracted from clinical records as per standard practice.

Whole exome sequencing

Exome sequencing was performed with a trio approach on probands (patients 1–8) and biological parents for all families. All sequence variants are described with reference to PBX1 transcript NM_ 002585.3.

For Patient 1, whole exome sequencing (WES) was performed as a research test. Exomes libraries were prepared using the Nextera Rapid Capture Exome kit (Illumina, CA). Exome libraries were sequenced on Illumina HiSeq2000 and HiSeq2500 instruments typically multiplexing 12 barcoded libraries per sequencing lane. Sequencing coverage analysis was performed using Bedtools (29) in conjunction with custom scripts.

Following exome sequencing, DNA sequence reads were aligned to the human reference genome build hg19 using the Burrows-Wheeler Aligner (30) to create Binary Alignment Map (BAM) files. Single Nucleotide Variant (SNV) calling was performed on the BAM files using Genome Analysis Toolkit (GATK) software (31) with the option of HaplotypeCaller1 (32). This generated a genomic Variant Call Format (VCF) file for each sample. VCF files from multiple samples were then processed together via a Joint Genotyping call with GATK to produce one multi-sample VCF file for all DNA variants. Variants in the VCF were annotated with various metrics using ANNOVAR (33). The VarSifter tool (34) was used to filter variants based on genomic location, allele frequency, inheritance pattern and predicted pathogenicity. Given the absent family history for congenital heart disease (CHD), only rare variants at a Minor Allele Frequency (MAF) < 0.01 with respect to the ExAC database, fitting de novo, recessive, or compound heterozygous inheritance models were considered. Variants passing the population frequency thresholds were checked with the Integrative Genomics Viewer (IGV) tool (35) to confirm read quality. Variants were subsequently considered pathogenic if they exceeded both in silico pathogenicity prediction thresholds: Combined Annotation-Dependent Depletion (CADD) > 15 (36), and PolyPhen2-HVAR ≥ 0.446 (37).

For Patients 2–6 and 8, who underwent WES as a clinical test, sequencing technology and the variant interpretation protocol have been previously reported (38).

For Patient 7, WES and variant calling were performed as previously described (39). Briefly, the exome was captured using the Agilent SureSelect v4 kit (Agilent, Santa Clara, CA, USA). Exome libraries were sequenced on an Illumina HiSeq2000 instrument (Illumina, San Diego, CA, USA) with 101 bp paired-end reads at a median coverage of 75x. Sequence reads were aligned to the hg19 reference genome using BWA version 0.5.9-r16. Variants were subsequently called by the GATK unified genotyper, version 3.2–2 and annotated using a custom diagnostic annotation pipeline.

Protein modeling studies

We mapped the deleterious sequence variants in PBX1 onto the PBX1 homeodomain structure (PDB code 1puf). This structure starts with residue 233, and thus does not include the Arg184, Met224 or Arg227 residues of PBX1. The mapped region is disordered and charged (14).

Site-directed mutagenesis

We generated expression vectors containing five PBX1 cDNAs carrying the sequence variants predicting p.Arg227Pro, p.Arg234Pro, p.Arg235Gln, p.Ser262Glnfs*2 and p.Arg288* that were identified in six of the eight patients described in this study. We utilized a polymerase chain reaction (PCR)-based approach using specific primers for each single mutation (Supplementary Material, Table S1). The pSG5-PBX1b (40) expression vector was used as a template for the PCR reaction. High-fidelity Taq polymerase enzyme Q5 (New England BioLabs, MA) was employed. After purifying the PCR product from agarose gel, T4 Polynucleotide Kinase (New England BioLabs, MA) was used in order to add a 5’-phosphate to the DNA, following manufacturer‘s instructions. 50 ng of DNA was used in a ligation reaction with T4 ligase (New England BioLabs, MA), followed by bacterial transformation. Positive colonies carrying the desired sequence variants were identified with Sanger sequencing.

Western blot analysis

Total protein lysates, as well as nuclear and cytoplasmatic fractions, were resolved by 10% SDS-PAGE and blotted to polyvinylidene difluoride (PVDF) membranes (Millipore, CA) as previously described (41). Membranes were first incubated with the primary antibody (Ab) overnight and then with a peroxidase-conjugated secondary Ab. Peroxidase activity was measured by using an enhanced chemiluminescence (ECL) kit (Pierce/Thermo Fisher Scientific), following the manufacturer’s instructions. The Abs used in this study were: Pbx1 (Cell Signaling, 4342), Actin (Proteintech, IL), Tubulin (Proteintech, IL), and Nucleolin (Santa Cruz Biotechnology, CA).

Chromatin immunoprecipitation (ChIP)-quantitative (q)PCR

Embryonic tissues (midfaces and hindlimbs) were isolated from embryonic day (E)10.5 mouse embryos after timed-matings of Swiss-Webster mice. The embryonic tissues were immediately crosslinked for 10 min in 1% formaldehyde (Electron Microscopy Sciences, #15710). ChIP assays were performed as previously described (42). The crosslinked material was sonicated to 200–500 basepair (bp) fragments (Diagenode Bioruptor). Immunoprecipitation (IP) was performed from 12–14 midfaces or 24–26 hindlimbs from E10.5 embryos, by overnight incubation at 4 °C with 1 µg of anti-Pbx1 Ab or control IgG (Cell Signalling, 2729), followed by Dynabeads protein A (Invitrogen). After washing, IP and input DNA was de-crosslinked and purified using QIAquick PCR purification kit (Qiagen, 28106). qPCR was performed to compare enrichment of specific genomic locations after immunoprecipitation on a QuantStudio 6 Flex Real-Time PCR System (Life Technologies). Enrichment was calculated as percentage of the input. Primers used for qPCR amplification (40 cycles) were:

Pbx1 Prom forward 5’- GCCCCCTTCTATGATTGGCTAGT-3’;

reverse 5’- TGGGTCATAGCTCCGCCTCC-3’;

Negative control forward 5’- GCCAGCCTGAGCTATATGGA-3’;

reverse 5’-TGGACGCTGGGAACTAAATC-3’.

Cell culture and transfections

The human HEK293 cell line and a primary murine mesenchymal cell line derived from E15.5 embryonic spleens (mc4) were cultured in DMEM (Dulbecco‘s modified Eagle‘s medium; Corning) supplemented with 10% fetal bovine serum (FBS; Gibco) and penicillin (100 U/ml)/streptomycin (100 U/ml; Lonza) at 37 °C, under 5% CO₂. Cell transfections were carried out using Lipofectamine 2000 (Thermo Fisher Scientific), following the manufacturer‘s instructions.

Clustered regularly interspaced short palindromic repeats (CRISPR)/Cas-mediated genome editing of Pbx1 in mc4

The following guide RNAs (gRNAs; CRs) were designed to target exon 3 of murine Pbx1 by employing the CRISPR design software (http://tools.genome-engineering.org): CR1: AGACCCCCAGCTCATGCGAC, CR2: TATTCGGGGAGCCCAAGAAG. These 2 CRs were independently cloned in the piCRg vector, which contains Cas9 and the tracrRNA (17). The targeting vectors were then delivered to the primary mesenchymal cells by lipofection. Single clones were picked and Sanger sequencing was used to identify the presence of the mutations introduced by CRISPR-Cas.

Luciferase assays

A 457 bp fragment encompassing two of the PBX1 binding sites within the PBX1 promoter was amplified from human genomic DNA using the following primers:

Forward 5’-GGGGTACCCCTCAGCCTCCTGCTTTTGTTT-3’

Reverse 5’- GGGGTACCCCCCTCAAGGACGCAACCTAAG-3’.

KpnI restriction sites (underlined) were used to clone the fragment into the pGL3-TK vector (Promega). Transfections were conducted in 24-well plates, as previously reported (15,16). A total amount of 350 ng DNA per well was transfected into the cells and comprised 50 ng of the reporter vector (either pGL3-TK-PBX1 promoter or empty pGL3-TK), 150 ng of the pSG5-PREP1 construct expressing the cDNA of the PBX1 cofactor PREP1, and 150 ng of the pSG5-PBX1 construct carrying either WT PBX1 cDNA, or one of the five PBX1 cDNAs containing the human mutations. For all of these experiments, the short isoform of WT PBX1 (PBX1b) (40) was used. 48 h after transfection, cells were harvested and lysed in Passive Lysis Buffer (Promega). Lysates were analysed by using the Luciferase Assay System (Promega) in a Glomax Explorer System (Promega). Normalization was performed after measuring the protein content for each sample using a Bradford protein assay (Bio-Rad Laboratory). The results represent an average of five replicate experiments, with each experiment performed in triplicate. An unpaired t-test was conducted to determine statistical significance.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

We are grateful to the patients and families for their participation and to the UCLA Clinical Genomics Center for their help with sequencing Patient 3. L.S. wants to personally thank Mr. Robert Aho for assistance with preparation of the figures.

Conflict of Interest statement. Megan Cho and Kristin Monaghan are employees of GeneDx.

Funding

The National Institutes of Health [R01 grant DE024745 to L.S.], the Dr. Caroline H. Damsky Award to M.R., and the Australian National Health and Medical Research Council Fellowship [ID1042002 to S.L.D].

Author notes

†

The authors wish it to be known that, in their opinion, Anne Slavotinek and Maurizio Risolino authors should be regarded as joint First Authors.

‡

The authors wish it to be known that, in their opinion, Licia Selleri and Joseph Shieh authors should be regarded as joint Senior Authors.