2.1 Exome capture and sequencing

They were performed by standard techniques according to the manufacturing instructions.

2.2 Sequencing data and relatedness analyses

Reads were mapped against the human reference genome assembly (hg19 genome, GRCH37Hg19), and variants were annotated using the Qiagen® Clinical Insight Translational™ software (Qiagen, Redwood City, CA, USA), and the Integrative Genomics Viewer (IGV).²¹

The pathogenicity of variants was estimated using the in silico prediction software Mutation Taster (http://www.mutationtaster.org), Polymorphism Phenotyping v2 (PolyPhen2, http://genetics.bwh.harvard.edu/pph2), Sorting Intolerant From Tolerant (SIFT, https://sift.bii.a-star.edu.sg/sift4g/), MaxEntScan (http://hollywood.mit.edu/burgelab/maxent/Xmaxentscan_scoreseq.html),²² and NetGene2²³ version 2.42 (DTU Technical University, Lyngby, Denmark). Relatedness analysis was performed using PLINK (v1.9)²⁴ and Peddy.²⁵

2.3 Sanger sequencing

Sanger sequencing was performed by standard technique. The GeneBank (National Center for Biotechnology Information, NCBI) sequence TSEN2 NM_001145394 (corresponding to Ensembl reference ENST00000454502.6) was used as reference sequence.

2.4 Alignments

Protein alignments were performed using Clustal Omega version 1.2.2 (https://www.ebi.ac.uk/Tools/msa/clustalo/) and visualized using Jalview 2.11.1.0.

2.5 Bulk RNA sequencing and analysis

Four affected individuals (PN556-II3, PN878-II2, PN1093-II1, PN1939-II3 in Figure S1) and one healthy mother of an affected individual (PN878-I2) were RNA sequenced obtained from peripheral blood cells using Novaseq 6000 (Illumina, Inc., San Diego, CA, USA) sequencer according to the manufacturer's recommendations. Details of the analyses are given in the Supplemental Data.

2.6 RNA Sanger sequencing

RNA sequencing was performed by standard techniques.

2.7 In vitro splicing assay

The exon trap vector pET01 (MoBiTec GmbH, Goettingen, Germany) was used for in vitro characterization of the splicing (Figure S2A). Details of the assay are given in the Supplemental Data.

2.8 Zebrafish experiments

Zebrafish experiments were approved by the University of Iowa's IACUC, PHS Assurance No. A3021-01, under animal protocol No. 5091513. The detailed experiments are described in the Supplemental Data.

3.1 An intronic variant in the gene TSEN2 causes a syndrome characterized by microcephaly, craniofacial malformations, growth and intellectual retardation, and aHUS

In a large cohort of patients affected with aHUS, we ascertained four individuals from two consanguineous families (pedigrees PN556 and PN878 in Figure S1) who showed additional similar extrarenal symptoms (Figure 1). Kidney pathology showed lesions of glomerular TMA (Figure 2). Detailed clinical data are available in the Supplemental Data. In addition to these symptoms, all four individuals presented microcephaly, multiple craniofacial malformations, and cognitive and developmental delay of variable severity (Figure 1 and Table 1). We excluded mutations in the genes CFH, CFI, CFB, C3, CD46 (MCP), CFHR1-5, DGKE, PLG, and THBD, known to be relevant for aHUS in all four affected individuals, by Sanger sequencing as well as the presence of anti-complement factor H antibodies by ELISA. We then performed WES of three affected and one unaffected individual as well as the healthy consanguineous parents of pedigree PN556, and of one affected and one unaffected sibling, and the consanguineous parents of family PN878 (Figure S1A,B). We applied a stepwise filtering strategy to prioritize clinically relevant variants among all the ones annotated by WES (Figure S3 and Supplemental Data). After stringent filtering, the only variant that was homozygous in all four affected individuals and segregated in the pedigrees compatibly with autosomal recessive inheritance was the intronic change c.914-5T>A in the gene TSEN2 (NM_001145394, Figure S4A,B). Both families came from the Southeastern Anatolia with no clear record of known relatedness. However, relatedness analysis showed relatedness values >0.5 indicating kinship in these two families (Table S1 and Supplemental Data). Moreover, haplotype analysis using WES data showed shared homozygosity of approximately 4.2 Mb between single nucleotide polymorphisms rs17222536 and rs2279017 on chromosome 3 in all affected individuals of families PN556 and PN878. This region contains seven genes (i.e., SLC6A1, TAMM41, TSEN2, RAF1, CAND2, NUP210, and XPC; Table S1). We excluded variations in all genes but TSEN2 based on their relatively high MAFs and high ratio of homozygosities in the Genome Aggregation Database (gnomAD; https://gnomad.broadinstitute.org) as well as in slico analyses with Mutation Taster, which predicted them to be polymorphisms (Table S1). Therefore, we considered the variation detected in TSEN2 as responsible for the phenotype. This variation was not present in our in-house database composed of 187 WES and 39 whole genome sequencing data from Turkish individuals without renal phenotype. Later, we identified from our large aHUS cohort two more individuals with aHUS, and similar craniofacial abnormalities from two consanguineous families (i.e., PN1093 and PN1939 in Figure S1 and Figure 1) in which disease-causing variations in complement and other relevant genes had been previously excluded (Table 1). Sanger sequencing of the 10 coding exons of TSEN2 identified the same intronic variant in both individuals (Figure S4C,D), indicating a founder effect and strongly supporting the causal role of the variant in determining the observed phenotype.

Cranial magnetic resonance imaging (MRI) of the six patients showed distinctive radiological abnormalities of the central nervous system, including hypoplasia of the anterior pituitary gland with preserved neurohypophysis and thin corpus callosum (Figure S5). Each patient's MRI findings are presented in the Supplemental Data.

3.2 The intronic variant c.914-5T>A affects TSEN2 mRNA splicing

The c.914-5T>A change is localized within the pyrimidine-rich region of intron 9 that is normally not evolutionary conserved. However, in the TSEN2 intron 9 the “T” and the surrounding bases are highly conserved in vertebrates (Figure S6). The c.914-5T>A change was also predicted to affect the splicing efficiency by MaxEntScan²² (score decrease of 31% from 11.34 to 7.71), to be deleterious by Combined Annotation Dependent Depletion (CADD),²⁶ and to be disease-causing by Mutation Taster. We reasoned that, since the c.914-5T is followed by a c.914-4G, the transition to A may give rise to an alternative AG acceptor splice site three base pairs before the canonical one, which could interfere with the normal splicing process. In silico splice-site analysis using NetGene2²³ supported this hypothesis (Figure S7). We, therefore, performed Sanger sequencing of cDNA from the patients. Detailed scrutiny of the chromatogram showed the presence of two transcripts: one in which exon 9 is normally spliced to exon 10, and the other in which exon 10 is skipped, and exon 9 is spliced to exon 11 (Figure 3). Furthermore, we also noticed the presence of a putative third transcript with a very low-intensity sequencing signal. For this reason, to further characterize the splicing defect, we performed a minigene-based splicing assay (Figure S2A–C). The results of the assay showed the coexistence of transcripts that are normally spliced from exon 9 to exon 10, but also a splicing variant in which exon 9 is spliced to the newly created alternative splice site (Figure S2D), and a transcript in which exon 9 is spliced to exon 11 (Supplementary Figure S2E). The splicing to the alternative splicing site gives rise to a TSEN2 transcript with an additional in-frame codon (CAG) between exon 9 and exon 10 that results in the inclusion of two extra amino acids, a proline and a valine, in the protein (Figure S8), whereas the splicing to exon 11 results in the skipping of exon 10, which is highly conserved in the evolution (Figure S9). Collectively, our results indicate that the intronic change c.914-5T>A results in the creation of alternative splice site that stochastically competes with the canonical one giving rise to a protein in which exon 10 is skipped or another protein with two additional amino acids. These results also suggest that the variant results in only partial loss-of-function of TSEN2 as a complete loss of function of any of the components of the TSEN complex is not expected to be compatible with life.¹³

3.3 TSEN2 exon 10 skipping in zebrafish causes vascular and cranial anomalies

Morphological analysis of the splice-block knockdown larva (tsen2sbMO) presented blood cells “pooling,” a phenotype compatible with vascular anomalies (Figure 4). The tsen2sbMO knockdown was performed in a transgenic zebrafish line that expresses enhanced GFP under the control of endothelial cell-specific fli1 promoter Tg(fli1:EGFP)y¹²⁷ and that allows the visualization of the vasculature in live embryos. We observed constriction of the Common Cardinal Vein, enlarged pericardium and extra numbers of vessel branching, and branches with no connections. We also observed smaller anterior cranium, forebrain and mid-brain, and kidney cysts and edema at 7dpf (Table S2).

3.4 tRNA composition is altered in the presence of the TSEN2 c.914-5T>A intronic variant

The human genome contains approximately 500 tRNA genes, which are transcribed as pre-tRNAs that undergo complex modification before becoming mature tRNAs, and 32 of these contain a short intron located within the anticodon that is cleaved out by the tRNA splicing endonuclease.^{28, 29} While no published in vivo data are available for individuals with mutations in one of the four TSEN components of the spliceosome, depletion of mature RNAs and accumulation of unspliced RNAs have been documented in patient-derived induced neurons.¹⁰ To establish if the c.914-5T>A variant affects the tRNAs biogenesis, we performed bulk sequencing of the RNA extracted from peripheral blood cells of four affected individuals and one of their heterozygous mothers (PN878-I2) and compared them to healthy children of similar ages. Four affected children shared five abnormal tRNA transcripts that were absent in the heterozygous mother and the healthy children (Table S3). These tRNAs were searched in the annotation files from GENCODE³⁰ and FASTA files for “High confidence tRNA sequences” from GtRNAdb.³¹ The tRNAs with gene ID 26988, 25052, 22262 and 13036 were found both in the FASTA file from GtRNAdb and GFF3 file from GENCODE, whereas tRNA with gene ID 8789 was found in GtRNAdb only. Interestingly, some of these tRNAs, such as tRNA^Tyr_GTA represent isodecoders as intron-containing precursors for which splicing has been reported to be essential for their production.³

Pathway enrichment analysis of downregulated genes shared by all four children identified 20 pathways in total; “Synaptic vesicle cycle” (pathway ID: hsa04721), “Glutamergic synapse” (pathway ID: hsa04724) and “Hippo signaling pathway” (pathway ID: hsa04392) as the top three most significantly enriched terms with high-fold value and lowest p-values (Figure 5). On the other hand, pathway enrichment analysis of upregulated genes from differential expression analysis from the mother resulted in 226 pathways, whereas 71 pathways were observed in her child. Ten pathways with the highest significance from mother and her child were compared. We showed that the mother and her child share the “Spliceosome” (pathway ID: hsa03040) and the “Ribosome” (pathway ID: hsa03010) terms in their top 10 pathways with the highest significance (Figure 6).