Bi-allelic Variants in IQSEC1 Cause Intellectual Disability, Developmental Delay, and Short Stature

Abstract

We report two consanguineous families with probands that exhibit intellectual disability, developmental delay, short stature, aphasia, and hypotonia in which homozygous non-synonymous variants were identified in IQSEC1 (GenBank: NM_001134382.3). In a Pakistani family, the IQSEC1 segregating variant is c.1028C>T (p.Thr343Met), while in a Saudi Arabian family the variant is c.962G>A (p.Arg321Gln). IQSEC1-3 encode guanine nucleotide exchange factors for the small GTPase ARF6 and their loss affects a variety of actin-dependent cellular processes, including AMPA receptor trafficking at synapses. The ortholog of IQSECs in the fly is schizo and its loss affects growth cone guidance at the midline in the CNS, also an actin-dependent process. Overexpression of the reference IQSEC1 cDNA in wild-type flies is lethal, but overexpression of the two variant IQSEC1 cDNAs did not affect viability. Loss of schizo caused embryonic lethality that could be rescued to 2^nd instar larvae by moderate expression of the human reference cDNA. However, the p.Arg321Gln and p.Thr343Met variants failed to rescue embryonic lethality. These data indicate that the variants behave as loss-of-function mutations. We also show that schizo in photoreceptors is required for phototransduction. Finally, mice with a conditional Iqsec1 deletion in cortical neurons exhibited an increased density of dendritic spines with an immature morphology. The phenotypic similarity of the affecteds and the functional experiments in flies and mice indicate that IQSEC1 variants are the cause of a recessive disease with intellectual disability, developmental delay, and short stature, and that axonal guidance and dendritic projection defects as well as dendritic spine dysgenesis may underlie disease pathogenesis.

Introduction

Much progress has been made in the past few years in the identification of genes responsible for intellectual disabilities (ID), yet more than half of all cases remain undiagnosed.^{1, 2} It has been estimated that the total number of genes involved in autosomal-recessive ID could be in the thousands.^{1, 3, 4} Autosomal-recessive disorders are common in consanguineous populations^{5, 6, 7} and we have therefore focused on studying consanguineous families with more than a single affected sibling to identify recessive genes that cause ID and improve genetic diagnosis.⁷

Here, we present compelling data that IQSEC1 (MIM: 610166) is the cause of a syndrome with ID. IQSEC1 is part of a family of three genes. The ortholog in flies,⁸ schizo, was shown to affect axon guidance⁹ and myoblast fusion.^{10, 11} IQSEC1, 2, 3 and Schizo all have an IQ-like motif and a Sec7-PH tandem domain. They are members of a family of guanine nucleotide exchange factors (GEFs) for ARF family GTPases and have been shown to regulate endosomal trafficking and actin cytoskeletal remodeling. IQSEC1/BRAG2 has been found in multiple tissues and is implicated in a variety of functions including myoblast fusion, integrin trafficking, angiogenesis, and cancer metastasis.¹²

In the mammalian brain, IQSEC1/BRAG2 and IQSEC2/BRAG1 are concentrated at the postsynaptic density of glutamatergic synapses.^{13, 14, 15, 16} Both proteins have been shown to control activity-dependent removal of AMPA receptors from glutamatergic synapses, a process that is required for long-term synaptic depression^{15, 17, 18} and involved in forgetting.¹⁹ Moreover, both proteins are sequentially recruited by NMDA receptors to activate Arf6 during neuronal development and this is crucial for the maturation of glutamatergic synapses.²⁰ Pathogenic variants in human IQSEC2 (MIM: 300522) are the cause of nonsyndromic and syndromic forms of X chromosome-linked intellectual disabilities (XLIDs),^{21, 22, 23, 24} whereas pathogenic variants in the autosomal IQSEC1 have not been described yet.

Given that there is a single fly gene, schizo, and three human IQSEC homologs with high DIOPT scores,⁸ the phenotypes of loss of schizo in flies are anticipated to be more severe than in human. Indeed, paralogs often functionally compensate for each other, providing genetic robustness and revealing tissue-specific phenotypes.²⁵

The main known phenotype associated with loss of schizo is a growth cone guidance defect. Growth cones of neurons often follow precise paths to find and connect with their targets by sampling cues produced by numerous cell types.²⁶ The Slit-Robo (Roundabout) pathway is one of main axon guidance pathways discovered in Drosophila.^{27, 28, 29, 30} Slit is secreted by the midline CNS glial cells in embryos where it binds to Robo receptors expressed on growth cones of neurons to prevent them from crossing the midline in tracks named commissures.³¹ The levels of Robo are then downregulated by expression of commissureless, an E3 ligase, expressed in neurons, permitting them to cross the midline as they do not interact with the Slit repellant and are attracted by Netrins.³² In the midline glia, schizo was shown to downregulate Slit secretion.⁹ Hence, in the absence of schizo, the increase in Slit prevents many growth cones from crossing the midline. Slits secreted by glia in mice similarly regulate the crossing of axons in the brain to form the corpus callosum, a process mediated via Robo proteins in neurons.³³ Some affected individuals with IQSEC2 variants exhibit a thinning of the corpus callosum,²³ suggesting that the function of schizo may be conserved. In summary, the above data suggest that IQSECs and schizo may play similar roles in flies and vertebrates.

In this study, we report five affected individuals from two unrelated families with similar phenotypes of intellectual disability, developmental delay, short stature, aphasia, and hypotonia. By combining exome sequencing and genotyping of family members, we identified two different recessive likely pathogenic variants in IQSEC1, segregating in each family with the disease phenotype. Studies of the Drosophila ortholog and expression of cDNAs encoding human reference and variant IQSEC1 in flies suggest that the variants are loss- or partial loss-of-function mutations. Finally, loss of IQSEC1 in neurons of the mouse cortex leads to defects in the maturation of dendritic spines. The model organism data underscore the importance of Schizo in flies and IQSEC1 in mice, orthologs of human IQSEC1, in neural development and function.

Material and Methods

Families Studied

Family 1 (F208) was enrolled, recruited, and sampled by the Institute of Basic Medical Sciences (IBMS), Khyber Medical University, Peshawar, Pakistan, and was studied at the Department of Genetic Medicine and Development, University of Geneva, Switzerland. The current study was approved by the ethical committee of the Khyber Medical University, Peshawar, Pakistan and by the Bioethics Committee of the University Hospitals of Geneva (Protocol number: CER 11-036). Family 2 was studied at the National Neuroscience Institute, King Fahad Medical City, Riyadh, Saudi Arabia, after the ethical approval of the institution. Informed consent forms were signed by guardians of both families. Peripheral blood samples were obtained from individuals of both families and the genomic DNA was extracted according to standard protocols.

Genetic Analysis

In family 1 (F208), whole-exome sequencing of the proband (IV-2, Figure 1A) was performed by using SureSelect Human All Exon v6 reagents (Agilent Technologies) and sequenced on an Illumina HiSeq4000 platform. The exome-sequencing data were analyzed by using a customized pipeline that includes the published algorithms, Burrows-Wheeler aligner tool (BWA),³⁴ SAMtools,³⁴ PICARD, and the Genome Analysis Toolkit (GATK).³⁵ Sequenced reads were aligned to the GRCh37/hg19³⁶ reference human genome and the filtering of variants was performed as described in previous studies.^{37, 38, 39} The Illumina 720K SNP array (HumanOmniExpress Bead Chip by Illumina) was performed to genotype the members of the family (III-1, III-2, IV-2, IV-3, and IV-6 in Figure 1A) that were sampled. The genotyping data were analyzed to calculate the run of homozygosity (ROH) by using PLINK.⁴⁰ A homozygous region of 50 consecutive homozygous SNPs, allowing a maximum of one mismatch and demarcated by the first heterozygous SNP at the edge, was defined as the ROH. Variants from the exome-sequencing data, present in the ROH segregating with the disease phenotype in the family, were isolated and filtered by using CATCH.⁴¹ Features like zygosity, quality score, minor allele frequency, conservation scores, and pathogenicity prediction were used to filter and prioritize the likely pathogenic variants as described previously.^{37, 38, 42} In two affected individuals (V-6 and V-8) in family 2, enrichment, sequencing, mapping, and variant calling were done by Beijing Genome Institute Europe (Copenhagen, Denmark). Capture of exons was done using an Agilent SureSelect Human All Exon 50 Mb Kit (V4). Sequencing was performed using an Illumina Hiseq 2000. Read mapping and variant calling were done using BWA and GATK, respectively. Annotation and variant prioritization were done as described.⁴³ In particular, due to parental consanguinity, homozygous variants were selected and filtered for low frequencies (less than 1%) in multiple control populations (in-house database with > 5,000 European exomes and 450 exomes from individuals from the Middle East, and the ExAC and gnomAD database) and pathogenicity of the variant (LoF variants or evolutionary conserved missense variants). This led to the identification of homozygous variants in IQSEC1, DNAJC25-GNG10 (a readthrough transcript with uncertain biological relevance), and WNT5A (MIM: 164975) (involved in autosomal-dominant Robinow syndrome [MIM: 180700]) in family 2. Heterozygosity of the IQSEC1 variant (the only likely candidate variant) in the parents and homozygosity in the affected sibling was done by Sanger sequencing. All candidate variants were validated by Sanger sequencing (Figure S1).

Figure 1.

Segregation of IQSEC1 Variants in Two Families with Overlapping Neurodevelopmental Phenotypes

Pedigree trees showing segregation of rare homozygous pathogenic IQSEC1 variants c.1028C>T (p.Thr343Met) in family 1 (A) and c.962G>A (p.Arg321Gln) in family 2 (B). Dashed line indicates the relationship by history; exact relationship is not clear (B). Unfilled shapes denote healthy while shaded shapes denote affected individuals; squares are males; circles are females; small triangle represent stillborn and double horizontal lines denote consanguineous marriages.

Drosophila Genetics

The following stocks were obtained from the Bloomington Drosophila Stock Center (BDSC) at Indiana University: y¹ w^∗; Mi{MIC}siz^MI02011 (RRID: BDSC_38550⁴⁴), y¹ w^∗; Mi{MIC}siz^MI13078/TM3, Sb¹ Ser¹ (RRID: BDSC_58652⁴⁴), y¹ w^∗; Mi{MIC}siz^MI01727/TM3, Sb¹ Ser¹ (RRID: BDSC_35071⁴⁴), y¹ w^∗; Mi {PT-GFSTF.2} is siz^{MI01727-GFSTF.2} (RRID: BDSC_60287⁴⁵), y¹ w^∗; P {Act5C-GAL4} 25FO1/CyO, y+ (RRID: BDSC_4414⁴⁶), P {w [+mW.hs ] = GawB} elav[C155] (RRID: BDSC_458⁴⁷), y¹ v¹; P{TRiP.HMS01980} attP40/CyO (RRID: BDSC_39060⁴⁸), and P{KK103616} VIE-260B (RRID: VDRC⁴⁹).

To test whether the RNAis are specific for schizo, we tested two independent RNAi stocks (TRiP.HMS01980 and KK103616) that should target all transcripts. Both cause embryonic lethality when ubiquitously expressed with Act-Gal4 or Repo-Gal4 but are viable when crossed to elav-Gal4, suggesting that they target schizo.

Generation of IQSEC1 Drosophila Transgenes

Transgenic stocks were generated as previously described.⁵⁰ Briefly, the IQSEC1 cDNA entry clone (GenBank: NM_001330619.1) was shuttled to the pGW-attB-HA⁵¹ using Gateway cloning (Thermo Fisher Scientific). Q5 site-directed mutagenesis (NEB) followed by Sanger verification was used to create IQSEC1 variants with the following primers: IQSEC1-T235M, FW 5′-GACACGGACATGAGCTGCCGG-3′, RV 5′-AGCCTTGTCCTCTTTGTGGG-3′; IQSEC1-R213Q, FW 5′-CTGCGGCTACAGGCTGGGGGC-3′. RV 5′-GTCCGACTCGGTGCTGGAC-3′ Constructs were inserted into the VK37 (PBac {y [+] – attP } VK00037) docking site by Phi-C31-mediated transgenesis.⁵²

Hatching Rate Experiments and Immunohistochemistry of Embryos

To quantify the hatching rate of embryos, we collected freshly laid eggs for 4 h on grape juice plates.⁵³ The hatched larvae were counted after 2 days. Larval stages were assessed based on the size of the mouth hooks.

For immunohistochemistry, embryos were collected and processed using the formaldehyde-based fixation protocol.⁵³ Embryos were incubated in CY3 conjugated goat anti-HRP antibody (Jackson ImmunoResearch #123-165-021) diluted 1:250 in PBS-Tween (0.1 %) with 10% NGS solution on a rotating platform overnight at 4°C in the dark. Embryos were washed in PBS-Tween (0.1%) and mounted in ProLong Gold Antifade Mountant (ThermoFisher). Confocal images were taken using a SP8 confocal microscope (Leica) with a 20× or 63× oil immersion lens. Image analysis and processing was performed using Imaris 9.3.0 (Bitplane).

ERG Recording of Fly Eye

ERG (electroretinogram) recordings were performed as described in Verstreken et al.⁵⁴ In brief, flies were glued to a slide with Elmer’s Glue. A recording electrode filled with 100 mM NaCl was placed on the eye, and a reference electrode was placed on the fly head. During the recording, a 1 s pulse of light stimulation was given, and the ERG traces of ten flies for each genotype were recorded and analyzed with WinWCP v.5.3.3 software.

Dendritic Spines in Brain Sections of Mice with a Conditional Iqsec1 Deletion in Cortical Neurons

All animal procedures were in accordance with the European Union’s Directive 86/609/EEC and the Regional Boards in Hamburg and Berlin (T-0269/11). To deplete IQSEC1 in cortical projection neurons and to label a subset of neurons, we crossed Iqsec1^fl/fl mice¹⁵ with NEX-Cre mice⁵⁵ and thy1-GFP line M mice.⁵⁶ IQSEC1 was detected in cortex protein samples (7 μg total protein/lane) from wild-type, Iqsec1^fl/fl:NEX-Cre-negative (ctrl) and Iqsec1^fl/fl:NEX-Cre-positive (ΔIQSEC1) mice on an immunoblot as described.¹⁵

Sagittal slices (100 μm) of 4 pairs of male littermate mice (2–4 months) were prepared using a vibratome (VT 1000S, Leica), imaged, and analyzed blind to genotype. GFP-positive layer V pyramidal neurons of the somatosensory cortex were identified on an SP5 inverted confocal laser-scanning microscope system (Leica). Z stack images (step size: 0.5 μm) of 488 nm-argon-laser-scanned image fields (pixel size x,y: 0.05 μm) with segments of secondary dendrites branching into cortical layer II/III were acquired using a 63-times magnifying oil-immersion objective with a 1.25 numerical aperture and a 5-times optical zoom. Blind 3D deconvolution was performed on all z stacks using Autoquant X3 (Media Cybernetics). Spines were detected and their body lengths (largest distance to the surface of the dendrite model) and head diameters evaluated by NeuronStudio (CNIC, Mount Sinai School of Medicine) until an equal number of spines (1,917) per genotype was reached. Only automatically recognized spines were analyzed.

Spine densities, body lengths, and head diameters determined in 47 (ctrl) and 39 (ΔIQSEC1) dendritic segments of 4 mice per genotype were plotted as boxplots with Min-to-Max-whiskers. Graph Pad Prism version 5 was used for graphs and statistical analysis. Statistical significances (^∗p < 0.05) for differences between genotypes were assessed via t test.

Results

Clinical Evaluation of Affected Individuals

As shown in Figure 1A, family 1 (F208) is a consanguineous family. It originates from the region near Kohat in Khyber Pakhtunkhwa, Pakistan. In this family there were four affected individuals and two of them died at an early age with unknown etiology except that they shared many features, including ID, short stature, microcephaly, aphasia, and aggressive behavior, with the two surviving individuals. As indicated in the pedigree, the unaffected parents (III-1 and III- 2) are first cousins (Figure 1A). Of their seven children, four are affected (IV-2, IV-4, IV-6, and IV-8) and three are healthy (IV-1, IV-3, and IV-7); one pregnancy was interrupted naturally for unknown reason (IV-5). Among the affected siblings, two (IV-2 and IV-6) were alive at the time of evaluation. Both individuals exhibit similar phenotypes of severe intellectual disability, short stature, aphasia, hypotonia, facial dysmorphisms, and aggressive behavior (Table 1). Three affected siblings (two brothers and one sister) of the father (III-1) had died for an unknown genetic disorder.

Table 1.

Clinical Features of Affected Individuals with Recessive IQSEC1 Variants

Family ID	Family 1 (F208)		Family 2
Variation in IQSEC1 (NM_001134382.3)	c.1028C>T (p.Thr343Met) Homozygous		c.962G>A (p.Arg321Gln) Homozygous
Origin	Pakistani		Saudi Arabia
Individual ID	IV-2	IV-6	V-1 (not genotyped)	V-6	V-8
Sex	female	female	male	male	male
Age at last evaluation (years)	36	23	6	11	6
Height (cm)	146	150	120	148	119
Stature	short	short	short	short	short
Head circumference (cm)	51 (<1 percentile)	51 (<1 percentile)	48 (25th percentile)	53 (50th percentile)	49 (25th percentile)
Intellectual disability	severe	severe	severe	severe	severe
Developmental delay	yes	yes	yes	yes	yes
Speech	aphasia	aphasia	few words	few words	few words
Motor milestones	delayed	delayed	delayed	delayed	delayed
Epilepsy	no	no	yes (early on-set)	yes (early on-set)	yes (early on-set)
Hypotonia	yes (mild)	yes (mild)	yes	yes	yes
Behavioral problems	aggressive	aggressive	inattention and hyperactivity	aggressive, ADHD and hyperactivity	inattention and hyperactivity
MRI	not done	not done	normal	normal (figure s2)	normal
Other symptoms	low vision	lower limbs weakness	multiple hyperpigmented café au lait spots in his lower back and upper thigh, truncal ataxia and ataxic gait with frequent falls	unsteady gait	unsteady gait

Abbreviations: ADHD, attention deficit hyperactivity disorder; MRI, magnetic resonance imaging.

In family 2 from Saudi Arabia, three male affecteds (V-1, V-6, and V-8), aged 6, 11, and 6 years, respectively, presented with a history of severe epileptic encephalopathy, delayed psychomotor development and ID (Figure 1B). The phenotypically unaffected parents (III-4 and IV-3) are consanguineous (first cousins once removed). The affected individual (V-7) died at age 6 without genetic diagnosis, but his presentation of disease and symptomatic course were similar to his affected brothers. There is a paternal family history of seizure disorder and brain atrophy. The unaffected siblings (V-2, V-3, V-4, and V-5) showed neither evidence of epilepsy nor major cognitive or motor deficits. The seizures in affected individuals (V-1, V-6, V-7, and V-8) started at 2 months of age with brief jerky-like movements. They developed febrile seizures, which were soon followed by other seizure types including predominant generalized tonic-clonic seizures and absence seizures. They also exhibited repeated episodes of status epilepticus triggered by illnesses or fever. Despite multiple medication trials, the affected individuals continued to experience multiple seizures of different semiology many times per day. They were able to walk independently at age 3 and started to use short phrases before the age of 4. With time they showed behavioral changes in the form of hyperactivity and/or lack of attention. Brain MRI scans and EEG (electroencephalography) performed in the two affected individuals of family 2 did not reveal any abnormality (no thinning of the corpus callosum) (Figure S2). Follow-up EEG studies showed a slowing of the background activity, diffuse delta, and some theta activity. Low-amplitude sharp discharges were seen at the front central regions of brain.

In summary, affected individuals from both families shared similar clinical features, including intellectual disability, developmental delay, short stature, hypotonia, and aphasia.

Exome Sequencing and Genetic Analyses Reveal Missense Variants in IQSEC1

Exome sequencing was first performed in proband IV-2 of family 1 (F208) with a coverage of >95% at 20× and a total mean coverage of ∼100×. The data were analyzed by using our in-house pipeline and prioritization algorithm.^{37, 38} This yielded no pathogenic variants in genes known to be implicated in intellectual disability, developmental delay, or short stature. We then intersected the exome-sequencing data with the genotyping data of five members of family 1 (F208) to select variants in the regions of homozygosity (ROH) segregating recessively with the disease phenotype. All segregating variants were filtered as described previously.^{37, 38} We included homozygous exonic and splicing variants (±6 bp), excluded all synonymous variants, except in the splicing regions, selected variants with a minor allele frequency of <0.02 in gnomAD database⁵⁷ and our local control subjects, filtered out if found within duplications of the genome, conservation score (by GERP++),⁵⁸ and predicted pathogenicity scores (by SIFT,⁵⁹ PolyPhen2,⁶⁰ and Mutation Taster⁶¹) were also taken into account. Using the above mentioned filters, only one homozygous missense variant (GenBank: NM_001134382.3; c.1028C>T [p.Thr343Met]) in IQSEC1 was identified in IV-2 and IV-6 of family 1 (F208) (Figures 1A and S1A). The number of heterozygous IQSEC1 loss-of-function variants in gnomAD is low. Hence, the pLI score of IQSEC1 calculated in gnomAD is 1, indicating that the gene is intolerant to loss-of-function variants. The absence of homozygous loss-of-function variants in gnomAD also suggests that the presence of bi-allelic loss-of-function variants is incompatible with normal life. As such, the homozygous missense variants described here may represent partial loss-of-function variants causing recessive disease. The IQSEC1 variant p.Thr343Met is present twice in 230,690 alleles in gnomAD and missing in the Bravo database (125,568 alleles). However, no homozygous variant was found in these databases. Moreover, the variant was not present in our local cohort of 300 control subjects from the same Pakistani ethnicity. Segregation of the variant was confirmed by Sanger sequencing; the variant is heterozygous in parents (III-1 and III-2) and homozygous in both affected individuals (IV-2 and IV-6) and the unaffected sibling (IV-3) is normal and not carrying the variant (Figures 1 and S1). By sharing our findings through GeneMatcher,⁶² we found the second family (family 2) with three affected individuals from Saudi Arabia with a different homozygous missense variant (c.962G>A [p.Arg321Gln]) and similar clinical findings. Again, both parents (III-4 and IV-3) were carriers and the affected children (V-6 and V-8) were homozygous (Figure S1B). This variant is present 2-fold in Bravo and three times in gnomAD databases. The phenotype of the affected individual (V-1) from different parents is very similar to individuals V-6 and V-8, but a DNA sample was unavailable to document the segregation of IQSEC1 variant p.Arg321Gln.

The Human Variants of IQSEC1 Behave as Loss-of-Function Mutations in Drosophila

The fly homolog of IQSEC1, IQSEC2, and IQSEC3 (MIM: 612118) is schizo; however, Schizo shows a slightly higher sequence similarity to IQSEC1 than to IQSEC2 and IQSEC3 (DIOPT score of 12, 11, and 9 out of 15,^{8, 63, 64} respectively). Schizo and IQSEC1 share 45% similarity and 32% identity over their entire proteins and contain a well-conserved Sec7 and PH domain (Figure 2A). To examine whether the IQSEC1 variants impair the function of the protein, we generated UAS-transgenes with the IQSEC1 variants (p.Thr343Met, p.Arg321Gln) as well as the reference IQSEC1 cDNA. Ubiquitous overexpression of the reference IQSEC1 cDNA in wild-type flies is lethal, showing that overexpression is toxic. However, wild-type animals overexpressing either of the two IQSEC1 variant cDNAs were viable, showing that the variants are likely to be loss-of-function variants (Figure 2B).

Figure 2.

Schizo Is a Functional Homolog of IQSEC1 in Fly

(A) Schizo and IQSEC1 are evolutionarily conserved and share IQ-like motif, Sec7, and PH domains.

(B) Overexpression of IQSEC1 at 25°C is toxic, but the human IQSEC1 variants are not toxic when expressed under the same conditions.

(C) Structure of schizo and its transcripts. MiMIC transposable element insertions in schizo are embryonic lethal. The MiMIC siz^MI01727 was converted using RMCE and an artificial exon that encodes GFP was introduced (siz^{MI01727-GFSTF.2}) to determine protein localization.

(D) Embryonic lethality of siz^M13078 and siz^MI01727 are partially rescued by human IQSEC1 WT but not by IQSEC1 p.Thr343Met or IQSEC1 p.Arg321Gln. 0% of eggs of homozygous mutants (siz^MI13078 and siz^MI01727, n > 500) hatched. At room temperature, 70% (n = 120) of the eggs hatched when the UAS-IQSEC1 reference was ubiquitously overexpressed (siz^MI01727/siz^MI01727;act-Gal4/UAS-IQSEC1 ref) but 80% (n = 84) of hatched larvae died as 2^nd instar larvae (∼20% of hatched larvae die in 1^st instar larvae stage). None of embryos (n = 150) hatched when the UAS-IQSEC1 variants were ubiquitously overexpressed (siz^MI01727/siz^MI01727;act Gal4/UAS-IQSEC1 [p.Thr343Met or p.Arg321Gln]).

To study the function of schizo, we used four different Minos Mediated Integration Cassette (MiMIC) insertions.^{44, 65} As shown in Figure 2C, MiMICs contain the yellow+ dominant body-color marker to identify the transgenic animals, a mutagenic gene-trap cassette consisting of a SA followed by stop codons, the coding sequence of the fluorescent protein EGFP, and an SV40 polyadenylation signal sequence. The sequences between the attP sites can be replaced through Recombination Mediated Cassette Exchange (RMCE) in vivo with any DNA cassette flanked by two inverted ΦC31 attB sites. siz^MI13078 is inserted in a coding exon that targets the four transcripts of schizo, and siz^MI01727 is inserted in an intron truncating all schizo transcripts. Both insertions cause embryonic lethality, but siz^MI02011, which only targets one transcript (RB), does not cause lethality indicating that RB is not critical for survival (Figure 2C). As expected, siz^MI13078 fails to complement siz^MI01727. To assess the endogenous subcellular localization of the Schizo protein, we integrated the artificial exon SA-EGFP-FlAsH-StrepII-TEV-3xFlag tag-SD flanked by two inverted attB sites (siz^{MI01727-GFSTF.2}; siz-GFP) (Figure 2C). These flies are homozygous, viable, and healthy, indicating that the internal GFP tag does not obviously affect protein function and that the siz^MI01727 chromosome carries no other lethal or visible mutations.

Given that overexpression of the reference IQSEC1 cDNA is toxic, we lowered the overexpression of human reference and variant cDNAs by raising the mutant animals at 18°C and found that the reference human cDNA can partially rescue the lethality from embryos of siz^MI01727 and siz^MI13078 to 2^nd instar larvae. However, both IQSEC1 p.Arg321Gln and IQSEC1 p.Thr343Met failed to rescue embryonic lethality, again suggesting that these variants are loss-of-function mutations (Figure 2D).

To determine where the Schizo protein levels in embryos, we costained them with anti-Repo (a glial marker) and anti-Elav (a neuronal marker) in GFP siz^{MI01727-GFSTF.2} (siz-GFP)-expressing embryos. As shown in Figure 3A, Schizo is broadly localized in the nervous system of stage 15 embryos, including CNS and PNS neurons and glial cells. In the adult brain (Figure 3B, half of a Drosophila brain), the protein is present throughout the neuropil and cell bodies and is also localized in glia (not shown). Analysis of lysates from siz-GFP flies probed with a polyclonal antibody against GFP revealed three different proteins (145 kDa, 160 kDa, and 162 kDa) in agreement with the molecular mass calculated for RB, RD, and RA when fused with GFP. These proteins are not detected in siz^MI0127/TM3 animals (Figure 3C). In summary, schizo is localized in glia and neurons at all stages tested, including third instar larvae (not shown).

Figure 3.

Schizo Is Localized in Glia and Neurons in Fly

(A) siz^{MI01727-GFSTF.2} (siz-GFP) embryos were stained with anti-Repo (glia) and anti-Elav (neurons) antibodies. Siz-GFP is broadly localized in the nervous system in fly embryos, including neurons and glia. The scale bar is 50 μm.

(B) In adult heads, Siz-GFP (green) is localized in the neuropil and cell bodies of neurons and glia. Elav staining labels the nuclei of the neurons (magenta). The scale bar represents 50 μm.

(C) Analysis of lysates from siz-GFP flies revealed three different kinds of transcripts.

Schizo regulates Slit expression by promoting its endocytosis in CNS midline glia of embryos and hence affects commissure formation.⁹ To assess the phenotype of the siz^M01727 mutants, we collected embryos and stained them with an antibody to HRP, a marker for neuronal membranes. We observe numerous commissural defects, consistent with previous findings⁹ (Figure 4A), suggesting that siz^M01727 is a strong loss-of-function allele. However, we also noted very severe axonal guidance and dendritic projection defects in the PNS of embryos that were not previously reported (Figure 4B).

Figure 4.

Schizo Is Required for Proper Axon Guidance in CNS and PNS

(A) Homozygous siz^MI01727 embryos show commissural defects (white arrows).

(B) Homozygous siz^MI01727 embryos exhibit severe axonal guidance defects in the peripheral neurons of all segments (yellow arrows). The scale bar represents 50 μm.

IQSEC1, IQSEC2, and IQSEC3 are localized in the mouse retina.^{66, 67} Given that Schizo is also localized in photoreceptors (Figure 3B), we performed electroretinography (ERG) to determine whether the protein is required in the fly visual system. While ubiquitous or glia-specific downregulation of schizo is lethal, neuron-specific downregulation of schizo produces viable flies. We lowered expression of schizo using two independent RNAis (elav > siz RNAi-1 (TRIP.HMS01980) and elav > siz RNAi-2 (P{KK103616})) and performed ERG recordings in 5-day-old flies. As shown in Figure 5, reduced expression of schizo dramatically affect ERG amplitude and abolished the on- and off-transients when compared to controls (elav > Luciferase RNAi). Hence, vision of these flies is either very severely impaired or lost.

Figure 5.

Schizo Is Required in the Fly Visual System

Lowering the level of Schizo in neurons (elav > siz RNAi-1 and elav > siz RNAi-2) caused dramatically reduced amplitudes and loss of on- and off-transients of ERGs when compared to control (elav > Luciferase RNAi). Statistical analyses are one-way ANOVA followed by a Tukey post hoc test. Mean ± SEM; ^∗∗∗∗p < 0.001.

Targeted Depletion of IQSEC1/BRAG2 in Mouse Cortex Neurons Affects the Maturation of Dendritic Spines

In the mammalian CNS, IQSEC1 is concentrated in postsynaptic densities and plays a critical role in the development and plasticity of glutamatergic synapses.^{15, 16, 20} Dysgenesis of dendritic spines,^{68, 69, 70} which house the postsynaptic specializations of glutamatergic synapses, represents a morphological hallmark of disorders associated with intellectual disability and autism.^{71, 72} To explore a role of IQSEC1 in the development of dendritic spines in the neocortex, we generated Iqsec1-floxed mice¹⁵ carrying both a NEX-Cre allele that drives Cre expression in projection neurons of the cortex⁵⁵ and a thy1-GFP allele (thy1-GFP line M) that labels a subset of cortical cells.⁷³ In cortices of NEX-Cre-positive Iqsec1-floxed mice, we detected only traces of IQSEC1/BRAG2 protein (Figure 6A), confirming that NEX-Cre-mediated deletion of exon 2 in Iqsec1 early in development suppressed IQSEC1 expression in cortical neurons to a large extent. We chose to analyze spines on secondary apical dendrites of GFP-positive somatosensory cortex layer V neurons in brain sections of adult Cre-positive mice (ΔIQSEC1) and Cre-negative littermates (ctrl) (Figures 6B–6H). Neurons lacking IQSEC1 had an increased density of dendritic spines (Figure 6C). Morphological measurements of 1,917 spines from each ΔIQSEC1 and ctrl mice revealed that the average body length of the spines was not altered by IQSEC1 depletion (Figure 6D). In contrast, the spines of IQSEC1-lacking neurons had on average smaller heads than the spines of ctrl neurons (Figure 6E). Depletion of IQSEC1 increased the density of spines over the whole range of the spine body lengths (Figure 6F), but the increased density was limited to spines with small heads (Figure 6G). Together, IQSEC1-deficient layer V neurons contained excess spines with small heads in adult mice. We divided the spine head diameter by the spine body length to reveal a maturity index for each spine and evaluated the effect of the genotype (Figure 6H). The leftward shift of the cumulative frequency curve upon IQSEC1 depletion indicates that lack of IQSEC1 in cortical neurons leads to an increase in immature-appearing dendritic spines.

Figure 6.

IQSEC1 Depletion in Mouse Cortex Results in Dendritic Spines of Immature Morphology

(A) Western blot revealing IQSEC1 in the cerebral cortex of wild-type (WT), NEX-Cre-negative (ctrl), and NEX-Cre-positive (ΔIQSEC1) Iqsec1^fl/fl mice.¹⁵

(B) Representative images of dendritic segments of cortical layer V neurons of thy1-GFP-positive ctrl and ΔIQSEC1 mice. Scale bar: 5 μm.

(C) Spine density was increased upon IQSEC1 depletion. n = 39–47 dendrites of four mice per genotype, ^∗p < 0.0001.

(D) Spine body length was not changed upon IQSEC1 depletion. n = 39–47 dendrites, p = 0.868.

(E) Spine head diameter was decreased upon IQSEC1 depletion. n = 39–47 dendrites, ^∗p < 0.0001.

Data in (C)–(E) are represented as total range (whiskers), interquartile range (box), and median (line inside box).

(F and G) Density of spines with small heads (spine head diameter < 0.6 μm) was selectively increased upon IQSEC1 depletion. Shown are the densities of spines grouped into body length bins of 0.2 μm (F) or grouped into head diameter bins of 0.1 μm (G). Values on x-axes indicate the maximum of each bin.

(H) Cumulative frequency distribution of spine maturity index (0.0–1.5), calculated as the ratio of head diameter (d_h) to body length (l_b) of each individual spine of the analyzed population (ctrl, n = 1,812; ΔIQSEC1, n = 1,821), shifted to the left upon IQSEC1 depletion.

Discussion

We report five affected individuals from two unrelated families having different pathogenic bi-allelic recessive variants in IQSEC1; all case subjects present similar clinical features: intellectual disability, short stature, and speech problems. Additionally, the three affected individuals from family 2 harboring IQSEC1 p.Arg321Gln present with seizures. The inclusion of the IQSEC1 locus in clinical diagnosis may identify more affected individuals with pathogenic variants in IQSEC1 that would enable the description of this gene-phenotype link and potentially reveal genotype-phenotype correlations.

Mutations in IQSEC2 have been linked to XLIDs. Initially, inherited mutations in IQSEC2 were identified in four families with non-syndromic ID.²⁴ These missense mutations resulted in amino acid changes in the Sec7 domain or IQ-like motif and in a compromised GEF activity of IQSEC2 on ARF6.^{21, 74} Subsequently, IQSEC2 mutations have also been found in affected individuals with syndromic forms of ID with seizures, and many of the pathogenic variants in IQSEC2 were shown to occur de novo.^{22, 23, 75, 76, 77, 78, 79} Taken together, familial and de novo mutations in IQSEC2 are associated with moderate to profound ID and with seizures. Mutations in IQSEC3 have recently been found in families with ID as well.⁸⁰

The three IQSECs in human (1, 2, and 3) are highly similar in structure and sequence. They also show similar sequence identities to the fly Schizo protein even though IQSEC1 has the highest DIOPT score when aligned with Schizo.^{8, 63} Loss of schizo in flies causes embryonic lethality whereas homozygous loss-of-function variants in IQSEC1 (this work), IQSEC2,^{22, 23} and IQSEC3⁸⁰ are recessive and do not appear to cause lethality. This is anticipated as essential genes in flies with neural phenotypes with more than one human homolog were three times more likely to be associated with human diseases reported in OMIM.^{81, 82, 83} In contrast, most of the essential fly genes with a single human homolog were not reported in OMIM⁸² as they may be essential in human as well. Hence, one simple interpretation would be that the three IQSECs are partially redundant in vital functions. It is also worth noting that the phenotypes associated with the mutations in the three human IQSEC genes are quite similar: intellectual disability, short stature, and speech problems.^{22, 23, 80}

The claim that the pathogenic variants in IQSEC1 are causative is supported by the overexpression assays, rescue experiments, protein localization, and phenotypic data associated with the loss of the homolog of IQSEC1, schizo, in fly embryos. The human IQSEC1 variants result in amino acid changes in the region between the IQ-like motif and the Sec7 domain. While Schizo and IQSEC1 show high sequence similarities between the functional domains, this region is not well conserved between Drosophila and human. Nevertheless, both variants present a clear functional difference when overexpressed compared to overexpression of reference cDNA. The mutations are unlikely to affect the basal GEF activity of IQSEC1, since they are located outside of the Sec7 domain. However, the regulation of the GEF activity by Ca²⁺/Calmodulin and by surface receptors^{18, 20} may be impaired in the IQSEC1 variants. Alternatively, they may affect protein folding, alter subcellular localization, or reduce protein stability.

Loss of schizo causes severe axonal guidance defects in the CNS⁹ but we also observe severe axon and dendritic defects in the PNS. In the PNS the axons are often short and highly aberrant in structure and aberrant axon outgrowth was also described for in vitro cultured cortical neurons derived from Iqsec2 knockout mice.⁸⁴ Note that the Slit and Robo proteins as well as Arf6 are known to play a critical role in the formation of the corpus callosum in mice⁸⁵ and these proteins control crossing of growth cones across the midline in mice,³³ similar to the formation of commissures in flies. Finally, a conserved function in myoblast fusion has been described for Schizo and IQSEC1.^{11, 86}

Given that the mouse homologs of IQSECs are localized in the eye^{66, 67} and schizo is localized in the fly photoreceptors, we also recorded ERGs upon knockdown of schizo in photoreceptor neurons. These flies exhibit almost complete loss of ERG amplitudes and lack of on-off transients, showing that they have no or severely impaired sight. Interestingly, vision defects have also been documented in affected individual IV-2 of family 1 (this study) and in IQSEC2 case subjects.^{22, 23}

IQSEC1 and IQSEC2 are core components of the postsynaptic density at excitatory synapses in the brain.^{13, 14, 15, 16} Genes for postsynaptic density proteins are frequently linked to cognitive disorders.⁸⁷ Both IQSEC1 and IQSEC2 are controlled by NMDA receptors and are critical for the activity-dependent reduction in the efficacy of AMPA receptor-mediated synaptic transmission during long-term depression.^{15, 17, 18, 20, 88, 89} IQSECs function as GEFs for the small GTPase Arf6 that controls endosomal protein traffic and actin cytoskeleton reorganization at the plasma membrane. Arf6 regulates the development of dendritic spines by promoting the conversion of filopodia into spines.^{90, 91} We previously observed that a targeted reduction of IQSEC1 or IQSEC2 levels interferes with the maturation of glutamatergic synapses.²⁰ Here, we show that a depletion of IQSEC1 in neurons of the mouse cortex affects the density and morphology of dendritic spines. In hippocampal neuron cultures, a knockdown of Iqsec2 expression resulted in a transiently increased density of dendritic spines at DIV15, whereas overexpression of IQSEC2 resulted in an increased proportion of mature spines at DIV21.⁸⁴ The increased density and immature-appearing morphology of the dendritic spines of IQSEC1-depleted neurons of the cortex support a direct role of IQSEC1 in the development of dendritic spines. Spine dysgenesis has been connected with ID previously.⁷¹ The link between lack of IQSEC1 and aberrant spine development therefore supports that loss-of-function variants of IQSEC1 can cause ID.

In summary, we describe a syndrome associating intellectual disability, developmental delay, short stature, aphasia/speech problems, hypotonia, and epilepsy in some case subjects by reporting five individuals from two unrelated families harboring bi-allelic variants in IQSEC1. Based on experiments in Drosophila and mice, IQSEC1 plays a critical role in neuronal development, supporting that the bi-allelic pathogenic variants in the IQSEC1 gene cause the above-mentioned syndrome in humans.

Declaration of Interests

The authors declare no competing interests.

Published: October 10, 2019

Web Resources

• Bravo database, https://bravo.sph.umich.edu/freeze5/hg38/

• DIOPT, https://www.flyrnai.org/cgi-bin/DRSC_orthologs.pl

• ExAC Browser, http://exac.broadinstitute.org/

• FlyBase, http://flybase.org/

• GenBank, https://www.ncbi.nlm.nih.gov/genbank/

• GeneMatcher, https://genematcher.org/

• gnomAD Browser, https://gnomad.broadinstitute.org/

• GTEx Portal, https://www.gtexportal.org/home/

• MARRVEL, http://marrvel.org/

• OMIM, https://www.omim.org/

• Picard, http://broadinstitute.github.io/picard/

• RRID, https://scicrunch.org/resources