Variant-specific pathophysiological mechanisms of AFF3 differently influence transcriptome profiles

Abstract

Background: We previously described the KINSSHIP syndrome, an autosomal dominant disorder associated with intellectual disability (ID), mesomelic dysplasia and horseshoe kidney, caused by de novo variants in the degron of AFF3. Mouse knock-ins and overexpression in zebrafish provided evidence for a dominant-negative mode of action, wherein an increased level of AFF3 resulted in pathological effects.

Methods: Evolutionary constraints suggest that other modes-of-inheritance could be at play. We challenged this hypothesis by screening ID cohorts for individuals with predicted-to-be damaging variants in AFF3. We used both animal and cellular models to assess the deleteriousness of the identified variants.

Results: We identified an individual with a KINSSHIP-like phenotype carrying a de novo partial duplication of AFF3 further strengthening the hypothesis that an increased level of AFF3 is pathological. We also detected seventeen individuals displaying a milder syndrome with either heterozygous Loss-of-Function (LoF) or biallelic missense variants in AFF3. Consistent with semi-dominance, we discovered three patients with homozygous LoF and one compound heterozygote for a LoF and a missense variant, who presented more severe phenotypes than their heterozygous parents. Matching zebrafish knockdowns exhibit neurological defects that could be rescued by expressing human AFF3 mRNA, confirming their association with the ablation of aff3. Conversely, some of the human AFF3 mRNAs carrying missense variants identified in affected individuals did not rescue these phenotypes. Overexpression of mutated AFF3 mRNAs in zebrafish embryos produced a significant increase of abnormal larvae compared to wild-type overexpression further demonstrating deleteriousness. To further assess the effect of AFF3 variation, we profiled the transcriptome of fibroblasts from affected individuals and engineered isogenic cells harboring + / + , KINSSHIP/KINSSHIP, LoF/ + , LoF/LoF or KINSSHIP/LoF AFF3 genotypes. The expression of more than a third of the AFF3 bound loci is modified in either the KINSSHIP/KINSSHIP or the LoF/LoF lines. While the same pathways are affected, only about one third of the differentially expressed genes are common to the homozygote datasets, indicating that AFF3 LoF and KINSSHIP variants largely modulate transcriptomes differently, e.g. the DNA repair pathway displayed opposite modulation.

Conclusions: Our results and the high pleiotropy shown by variation at this locus suggest that minute changes in AFF3 function are deleterious.

Keywords: Horseshoe kidney; Intellectual disability; Mesomelic dysplasia; Transcriptome; Zebrafish model.

Fig. 1

AFF3 variants. A Schematic protein structure of AFF3 (NM_002285.3) with its N-terminal homology domain (NHD, cyan), the AF4-LAF4-FMR2 (ALF, pink) domain [2, 3, 5] containing the degron motif, a Serine-rich transactivation domain (TAD, yellow) [6], a bipartite nuclear/nucleolar localization sequence (NLS, green), and the C-terminal homology domain (CHD, blue) [4] showing positioning of all AFF3 coding variants mentioned in the text. Missense and truncating variants are shown above, while extents of microdeletions and microduplications are depicted below the structure with continuous and dashed lines, respectively. The variants are color coded: loss-of-function (truncation and deletion) in red, biallelic missense outside the degron in blue and KINSSHIP-associated missense variants, deletion, and duplication in purple. The de novo missense identified in individuals M1 and M2 are shown in green. While the p.(Arg947Pro) shown in black was shown to segregate with isolated syndactyly [12], we have also identified it in individuals with no digit abnormalities. The 500 kb KINSSHIP deletion was originally wrongly defined as encompassing the entirety of the gene [13] but was later shown to only remove exons 4 to 13 of AFF3 [11, 14, 15]. B UCSC genome browser snapshot of the Chr2 99.5 to 101.4-Mb region showing the genes mapping to this interval. The extent of the duplication identified in the DUP1 individual that encompasses exon 10 to exon 24 of AFF3 and exon 1 to 3 of REV1 is indicated by the black bar and the light blue shadow. C Examples of pedigrees of transmitting affected families suggesting semi-dominance

Fig. 2

A Variant-specific pathophysiological mechanisms of AFF3. Schematic summary of the different type of AFF3 identified variants, their mode of inheritance, postulated mechanisms, and associated clinical phenotypes (see text for details). The number of patients we identified in each category and their identifiers are indicated. All symptoms are reported in Tables S1 and S2. Filled form = KINSSHIP syndrome, Half-filled form = milder syndrome, Form with included filled disk = more severe syndrome associated with semi-dominance, Question mark = possible association warranting further investigation. B Summary of in vivo and in vitro experiments. Schematic summary of traits associated with diminished expression and increased stabilization of AFF3 in mouse, zebrafish, HEK293 human cell, and affected individuals. The results previously published in Voisin et al., AJHG 2021 [11] are in blue, while results of this report are in black. The p.(A233T) variant is the most common de novo KINSSHIP variant [11], while the p.(M238T) and p.(M238V) variants described in this report present a milder phenotype (see text for details). Abbreviations: DD, developmental disorder; del, deletion; dup, duplication; ex, exon, ID, intellectual disability; LoF, loss-of-function; + , wild-type allele

Fig. 3

aff3 knocked down zebrafish larvae display altered behavior and morphological anomalies. The conditions analyzed are the following: Uninjected (Un), Mock-injected (M), and aff3 knockdown (KD). A Proportions of normal and developmentally defective 5 dpf embryos. In 10% of aff3 KD zebrafish, we identified several morphological anomalies such as head malformations, belly and heart edema, skeleton-muscular dysmorphologies, and alteration of eye pigmentation. B Alcian blue staining at 5 dpf revealed jaw malformation in 57% of aff3 KD zebrafish. C Visualization of morphological inter-ocular distance (IOD) and head width (HW) measurements from dorsoventral zebrafish image. Quantification of IOD (D) and HW (E) indicates a significative decrease in IOD and HW in aff3 KD larvae; p* < 0.04; p** < 0.0049. F Touch test response assay at 3 dpf. Upon a touch stimulus, we classified the larvae swimming behavior in «normal swimming», «pause», «looping swimming», «pinwheel swimming», or «motionless» due to malformations. G Swimming global velocity analysis at 5dpf in the dark of Un, M, and aff3 KD and aff3 KD co-injected with human AFF3 (hAFF3) mRNA wild-type (Wt), which recovered 57% of the locomotion function compared to aff3 KD, or harboring the indicated missense variant, i.e., the KINSSHIP variants Ala233Thr and Val235Gly or the biallelic variants identified in this report in a healthy (Gln179Glu) or affected individuals (Lys528Arg and Thr592Ser). H Proportions of normal and developmentally defective 5 dpf embryos uninjected (Un), injected with water as control (H₂0) or with 360 ng of hAFF3 mRNA Wt or the indicated missense variant. Larvae were cataloged as described: (i) normal phenotype, (ii) Class 1 with skeletomuscular dysmorphology and/or small dimension, (iii) Class 2 with a more severe phenotype including at least three of the following characteristics: skeletomuscular dysmorphology, small dimensions, head malformations, eyes’ alteration, pericardial edema, and lateral belly edema or (iv) dead. Injections of 180 and 720 ng of hAFF3 mRNA showed similar results

Fig. 4

Transcriptome profiles of engineered isogenic HEK293T cells. A Four-way Venn diagram of differentially expressed genes (DEGs) in biallelic loss-of function (LoF/LoF) AFF3 lines and biallelic KINSSHIP/KINSSHIP (DN (dominant negative)/DN) AFF3 lines upon comparison with unmutated wild-type lines. DEG counts are stratified in genes up- (UP) and downregulated (DOWN). B Volcano plots of DEGs in biallelic loss-of function (LoF/LoF) AFF3 lines (left panel) and biallelic KINSSHIP/KINSSHIP (DN/DN) AFF3 KINSSHIP lines (right panel) upon comparison with unmutated wild-type lines. The top 30 most significant DEGs in LoF/LoF that are dysregulated in an opposite manner in KINSSHIP/KINSSHIP (DN/DN) are indicated, together with some of the most differentially expressed genes (-log10(Padj) > 20 and abs(log2FoldChange) > 0.5). C Four-way Venn diagram of differentially expressed genes (DEGs) in biallelic loss-of function (LoF/LoF) AFF3 lines and biallelic KINSSHIP/KINSSHIP (DN/DN) AFF3 KINSSHIP lines upon comparison with unmutated wild-type lines and AFF3 ChIP-seq peaks identified in HEK293T cells (HEK293T) and in Mus musculus ES cells (mmES). DEGs bound by AFF3 discussed in the text are indicated. D Gene set enrichment analysis (GSEA) for hallmark pathways of DEGs in biallelic loss-of function (LoF/LoF) AFF3 lines (left panel) and biallelic dominant-negative KINSSHIP/KINSSHIP (DN/DN) AFF3 KINSSHIP lines (right panel) upon comparison with unmutated wild-type lines. E Examples of DEGs NRC31 (top) and DDX17 (bottom) loci bound by AFF3. UCSC genome browser snapshot showing from to top to bottom AFF3 ChIP-seq HEK293T results, UCSC and REFSeq curated gene structure and vertebrate PhyloP conservation scores (left panels). Expression level of NRC31 (top) and DDX17 (bottom) in + / + (blue), LoF/LoF (yellow), and KINSSHIP/KINSSHIP (DN/DN; green) HEK293T engineered lines (right panels)