Alu Elements

The four non-reference MEI event lists (Table 1) were submitted to the 1000 Genomes Structural Variation subgroup for validation experiments to assess false detection rates. 200 loci from each list were randomly selected for primer design and subsequent PCR validation. Primers were designed as described previously [32] (link), [36] to span across the insertion breakpoint and to guarantee unique mapping to build 36.3. In addition to the estimation of the false detection rates, validation genotypes were derived from gel-band size comparison for each sample and site tested by PCR. We also used the validation data to estimate detection sensitivity based on the overlap of events called between the two independent sequence data platforms and algorithms.
For loci with ambiguous PCR results, no amplification, or amplification of only the empty insertions site, a second primer pair was designed. For the primer design, 600 bp of flanking sequence on either side of the insertion site was retrieved from genome.ucsc.edu using Galaxy. Alu elements within the flanking sequence were masked to “N” using RepeatMasker (repeatmasker.org). Primers were designed with BatchPrimer3 v2.0 in the flanking sequence, leaving at least 100 bp before and after the predicted insertion site. Next, all primers were tested with BLAT to determine the number of matches in the human genome. If one primer of a primer pair matched several times and the other primer was unique, a virtual PCR was performed. Primer combinations with one predicted PCR product were tested on our panel. Otherwise primers were designed manually (if possible) after repeat-masking the flanking sequence with the complete repeat library.
In addition, for L1 and SVA loci without unambiguous PCR amplification, primers were designed, placing one primer within the 3′ end of the mobile element sequence [75] (link). The primers were designed to match the consensus sequences of the youngest L1 and SVA sub-families. All PCR primers were ordered from Sigma Aldrich, Inc. (St. Louis, MO). All LSU-designed PCR primer sequences used in this project can be found at http://batzerlab.lsu.edu.

RNA editing candidates were detected using a pipeline based on our REDItools suite20 (link)56 . In particular, we applied REDItoolDnaRna.py to each BAM file obtained by the mapping of RNA-Seq data from human tissues using GSNAP (see the “Alignment of RNA-Seq reads” paragraph above). Initially, nucleotide changes were called using loose parameters (-c 1,1 -m 20,20 -v1 -q25,25 -s2 -g2 -S -e -n0.0 -N0.0 -u -l). Then, read pairs harbouring nucleotide changes were realigned onto the human genome (hg19 assembly version) using Blat and only uniquely mapping pairs were retained. Base changes residing in not unique read alignments were discarded. In addition, we removed positions surrounding genomic regions (+/− 10 bases) in which the multiple alignment of reads was not optimal by the presence of indels.
Resulting tables were subsequently updated adding genomic and exomic information by means of the REDItoolAddGenome.py script. Individual positions in updated table were finally annotated using the AnnotateTable.py script and the following databases: RepeatMask, dbSNP (v. 138) and RetroposedGenes from UCSC. For each table, we separated positions residing in Alu elements, repetitive non-Alu regions and non-repetitive regions using custom bash scripts. During the split, we retained only positions supported by at least 10 genomic reads and completely homozygous. DNA-RNA changes in retroposed genes were eliminated as well as sites in which the alternative nucleotide was supported by less than 2 RNA reads. RNA editing candidates in repetitive non-Alu regions and non-repetitive regions underwent more stringent filters. Indeed, we excluded positions in the first and last 6 bases of reads, with quality score less than 30, with DNA-RNA frequency change lower than 0.1, in homopolymeric regions longer than 5 bases, in which the alternative nucleotide was supported by less than 3 RNA reads. In addition, we removed duplicated reads and applied again the Blat filter. Resulting tables were used for downstream analyses by means of custom scripts.
RNA editing sites in hyper-edited reads were detected using the pipeline described in Porath et al.18 (link).
All detected positions were finally annotated by ANNOVAR26 (link).
The comparison between our detected editing sites and available RNA editing databases, such as DARNED and RADAR, was performed by a custom script. DARNED annotations for human were downloaded from http://beamish.ucc.ie/. Version 2 RADAR annotations were downloaded from http://rnaedit.com/.

Several TE families, including L1, Alu and SVA (SINE-VNTR-Alu) elements, remain active in the human genome [33 (link)]. In the integration process of an Alu element, a duplicated sequence is copied and inserted at the flanking site, which is called the target site duplication (TSD) [34 (link)]. This is the hallmark of retrotransposition. By surveying Alu insertions accompanied by TSDs, we analyzed TSD characteristics. We made a non-redundant Alu insertion set by merging the results of the NA18943 and NA19240 (merging condition: pairwise distance < 500 bp). We then used the MEME Suite [35 (link)] to find motifs at the first and second nicking sites. We also analyzed the length distribution of the TSDs of Alu elements.

To gain insight into the intronic determinants regulating the biogenesis of the circRNAs whose expression is altered upon the FUS mutation, we focused on the 1 kb-long regions upstream and downstream the circRNAs tested for differential expression; among the deregulated circRNAs, we analyzed only those whose linear counterpart does not vary significantly in the same direction. To assess FUS binding in the intronic regions, we used the publicly available FUS^WT and FUS^P525L PAR-CLIP data produced from in vitro-derived human MNs [44 (link)]. Flanking intronic regions containing at least one T to C transition were marked as bound by FUS; the overlap between intronic regions and PAR-CLIP transition events was assessed using the BEDTools intersect tool [105 (link)]. To evaluate whether FUS binding events were enriched in the intronic regions flanking deregulated circular RNAs, we performed one-sided Fisher’s exact tests comparing the proportions of upregulated and downregulated circular RNAs with at least one binding site in the intronic regions to the proportion of non-deregulated circRNAs with such binding.
To evaluate the presence of inverted Alu repeats in the circRNA intronic regions, we first extracted the coordinates of Alu elements from the UCSC genome browser [111 (link)] hg38 RepeatMasker [112 (link)] annotation. BEDTools intersect was employed to identify the circRNAs carrying at least two Alu elements with opposite orientation, one in the upstream intronic region and the other in the downstream intronic region. The significance of the overlap between the circRNAs with FUS binding sites in the intronic regions and the circRNAs with inverted Alu repeats was assessed by performing hypergeometric tests.
To calculated the Alu editing index, preprocessed FASTQ files produced from Total RNA-Seq analysis of MN FUS^WT and FUS^P525L samples [46 (link)] were given as input to the PRINSEQ software [113 (link)] in order to remove exact duplicates and reverse complement exact duplicates; reads where then aligned to the hg38 genome using STAR [114 (link)] with parameters --outSAMtype BAM SortedByCoordinate --outFilterType BySJout --outFilterMultimapNmax 1 --alignSJoverhangMin 8 --alignSJDBoverhangMin 1 --outFilterMismatchNmax 999 --outFilterMismatchNoverLmax 0.04; finally, the RNAEditingIndexer tool [115 (link)] was employed, providing it with the alignment files and the previously computed gene-level FPKM values [46 (link)] also used to evaluate the expression of ADAR enzymes). RNAEditingIndexer was also employed to calculate the editing index only for the Alus localized within the circRNA flanking intronic regions. Paired Student’s t-tests were conducted to evaluate the Alu editing index difference between deregulated and non-deregulated circRNAs.

Large F8 gene deletions were identified by consistent failure of PCR amplification of a single exon or adjacent F8 exons, as indicated by missing or altered bands upon electrophoresis of PCR products. At least three separate attempts to amplify missing fragments from the subject’s genomic DNA were performed using the same primers and PCR conditions, alongside successful amplification and sequencing of the exons flanking the suggested deletion. For detection of deletion breakpoints, we carried out LD-PCR using primers for amplification of the exons framing the deletion and GoTaq^® Long PCR Master Mix following the manufacturer’s protocol. We were able to identify deletion breakpoints for two patients. To accomplish this, we designed primers flanking the copies of the Alu element that could be involved in the deletion formation. For patient A375, we used primers F8-delF (5′-GTTTGTTTACATTTGTCCCAACT-3′, c.787+2045_2067, intron 6) and F8-delR (5′-TGCAACTCAAAGGACTAAACA-3′, c.1903 +1569_1589, intron 12). For patient A469, we used primers F8-5D [32 (link)] and Del6R (5′-CAGTTGACTCTTGAACAATACA-3′, c.787+2976_2995, intron 6).
Large duplications that could not be identified using routine PCR-based analysis methods were tested with multiplex ligation-dependent probe amplification (MLPA). MLPA was carried out using the F8 SALSA MLPA kit P178 (MRC Holland, Amsterdam, The Netherlands) according to the manufacturer’s instructions. Exon dosage was calculated using Coffalyser.Net software (MRC Holland).