Genetic Variants Identified by Whole Exome Sequencing in a Large Italian Family with High Plasma Levels of Factor VIII and Von Willebrand Factor

High plasma levels of factor VIII (FVIII) and von Willebrand factor (VWF) have been indicated as independent risk factors for venous thromboembolism. However, the genetic factors responsible for their increase remain poorly known. In a large Italian family with high FVIII/VWF levels and thrombotic episodes, whole exome sequencing (WES) was performed on 12 family members to identify variants/genes involved in FVIII/VWF increase. Twenty variants spread over a 8300 Kb region on chromosome 5 were identified in 12 genes, including the low frequency rs13158382, located upstream of the MIR143/145 genes, which might affect miR-143/145 transcription or processing. The expression of miR-143/145 and VWF mRNA were evaluated in the peripheral blood mononuclear cells of six family members. Members with the variant (n = 3) showed lower levels of both miRNAs and higher levels of VWF mRNA compared to members without the variant (n = 3). An analysis of genetic and expression data from a larger cohort of individuals from the 1000 Genomes and GEUVADIS project confirmed a statistically significant reduction (p-value = 0.023) in miR-143 in heterozygous (n = 35) compared to homozygous wild-type individuals (n = 386). This family-based study identified a new genetic variant potentially involved in VWF increase by affecting miR-143/145 expression.


Introduction
Factor VIII (FVIII) is a plasma glycoprotein predominantly synthesized in the liver by sinusoidal endothelial cells that circulates in a stable complex with von Willebrand factor (VWF), a large multimeric glycoprotein synthesized by endothelial cells and megakaryocytes.It is constitutively secreted and released upon stimulation by endothelial cells and secreted by activated platelets [1].FVIII is involved in the intrinsic coagulation pathway by promoting the activation of factor X. VWF plays a dual role in hemostasis by supporting the adhesion and cohesion of platelets by acting as a ligand for their glycoproteins and preventing the degradation of plasma FVIII.The levels of FVIII and VWF in plasma are correlated, and both have been independently associated with a higher risk of venous thromboembolism (VTE) [2][3][4][5][6].
VTE is a multifactorial disease that develops as the result of three major mechanisms: blood stasis, endothelial damage, and hypercoagulability [7,8].Several environmental risk factors are known to predispose VTE [9], while genetic risk factors predisposing VTE are partially known.Inherited risk factors that cause a hypercoagulable state, commonly as thrombophilia abnormalities, include factor V Leiden (rs6025), the prothrombin variant rs1799963, and deficiencies of the natural anticoagulants antithrombin, protein C, and protein S. In addition, several genes mainly related to erythrocytes, platelets, and inflammation, but also including F8 and VWF, that code for FVIII and VWF have been associated with the risk of VTE in meta-analyses of genome-wide association studies [10,11].
Plasma FVIII coagulant activity (FVIII:C) >150 UI/dL has been associated with an increased risk of VTE in a dose-dependent manner [12,13].Moreover, high FVIII levels have been associated with an increased risk of VTE recurrence [2,14,15].Thus, plasma levels of FVIII are sometimes measured as part of thrombophilia screening.High levels of FVIII may be due to congenital or acquired conditions (e.g., systemic inflammation).Only a few family-based studies have investigated the genetic factors responsible for FVIII increase [16], and the mechanism of this association is not well characterized.Recently, a prospective study confirmed the association between an increased risk of unprovoked VTE and high plasma levels of VWF [6].In healthy individuals, plasma levels of VWF vary greatly, and approximately 65% of this variability is inherited [17] (with levels being regulated by several genetic loci) [18,19].In this study, whole exome sequencing (WES) was performed on a large Italian family with high/extremely high levels of FVIII and VWF and episodes of thrombosis to identify new genetic risk factors associated with FVIII/VWF increase.
In this family, additional disorders that may be associated with FVIII/VWF increase were reported, including the following: obesity (II-8, II-3, II-9), overweight (II-2, II-5, III-5), diabetes (II-8, II-2, II-3, II-5), dyslipidemia (II-8), atrial fibrillation (II-3), tumor and hypertension (II-5) (Table 1).Thus, the effect of these comorbidities, age, and blood group on high or extremely high FVIII and VWF levels was evaluated.For each variable, a positive relationship was found with extremely high levels of FVIII:C (range: 317-400%) and VWF:Ag (range: 295-390%) but not with high levels of FVIII:C (range: 180-200%) and VWF:Ag (range: 185-227%) (Figure S1).Thus, in the family, the presence of high levels of FVIII/VWF in (i) young subjects with O blood group and no/few comorbidities (Figure S1A-C) and in (ii) two consecutive generations suggested a possible genetic predisposition to high levels of both FVIII and VWF and an autosomal modality of inheritance of both phenotypes.Hence, a whole exome sequencing study was initiated.

WES Reveals Variants Associated with High Levels of FVIII and VWF
WES was undertaken and carried out on subjects with DNA available at the time of the analysis: the proband (II-8) and 11 family members (II-2, II-3, II-5, II-6, II-9, III-1, III-2, III-3, III-5, III-7, III-8) (Figure 1, Table 1).After quality control, data analysis revealed a total of 208,103 variants in 26,457 genes.Variants were filtered according to the plasma levels of FVIII:C or VWF:Ag.Two case-control association analyses were then performed, assuming an autosomal dominant inheritance model in both cases (high levels being present in two consecutive family generations).
The FVIII-based analytical approach, conducted in seven cases (II-2, II-3, II-5, II-8, II-9, III-3, III-5) with FVIII:C > 150% and five controls (II-6, III-1, III-2, III-7, III-8) with FVIII:C < 150% (Figure 1, Table 1), allowed for the identification of three intronic variants in the CYFIP2 gene and one synonymous variant in the FNDC9 gene (Table 2, part A) located in the CYFIP2 intron 23.All the variants found in the heterozygous state in the cases but absent in the controls were localized in a 56 Kb region on chromosome 5 and were common SNPs with a minor allele frequency (MAF) ranging between 8 and 46% (Table 2, part A).

In Silico Prediction of Variants Identified by WES
The PredictSNP2 consensus classifier, which combined the five best performing prediction methods (CADD, DANN, FATHMM, FunSeq2, and GWAVA) to provide a prediction on the pathogenicity of the variants, was assessed.Among the four variants identified via the FVIII:C approach, only the rs3734027 variant was predicted to be deleterious (Table 2, part A).This is an A to G transition localized at -28 nucleotides from the acceptor splice site (3 ss) of the CYFIP2 exon 21.This A>G nucleotide substitution was predicted by the Human Splicing Finder tool (http://www.umd.be/HSF3/,accessed on 1 March 2019) to reduce the branch point score from 95.75 to 66.12 (Figure S2A).Among the variants identified by the VWF:Ag approach, the deep intronic variant rs78112077 in LARP1 and both rs2270818 and rs2270819 localized in the 5 UTR of THG1L were predicted to be deleterious (Table 2, part B).The rs78112077 variant is a G to A transition that is localized in the exon 1 of the LARP1 alternative transcripts NM_001367718 and NM_033551 (Figure S3A), resulting in the synonymous Gly23 = substitution (Figure S3B), was also predicted via the use of the NNSPLICE tool (https://www.fruitfly.org/seq_tools/splice.html,accessed on 15 February 2022) to activate an intronic cryptic donor splice site (5 ss) (score 0.75) localized 369 nt upstream of the physiologic exon 1 5 ss (score 0.93) of the LARP1 alternative transcripts NM_001367718 and NM_033551 (Figure S3C).
No deleterious effect was predicted by PredictSNP2 for the two rare intronic variants in the ARHGEF37 and NDST1 genes (Table 2, part B).
The VarElect tool was also used to prioritize genes identified via WES analysis.A total of nine genes with a direct (seven) or indirect (two) link with thrombosis, inflammation, and VWF were found (Table 3).Among them, the MIR143, MIR145, and SPARC genes had the higher score (range: 5.38-8.45)(Table 3).No pathogenic role was predicted for variants rs729853 and rs2116780 in the SPARC gene (Table 2), and no correlation with the chosen phenotypes was found for the CTB gene, the putative target of the regulatory elements (promoter and enhancer) where these variants are localized (Table 3).Conversely, the rs13158382 variant, mapped via EnhancerAtlas in the common promoter of MIR143/145 genes (Table 2, part B), was localized at −122 nucleotides from the mature miR-143 and −7 nucleotides from the precursor of miR-143 with a possible role on miRNAs processing/expression.
Taken together, these findings prompted us to further investigate the predicted pathogenic role of the two intronic variants (rs3734027 and rs78112077) in CYFIP2 and LARP1, the two variants (rs2270818 and rs2270819) in the 5 UTR of THG1L, and the rs13158382 variant located upstream the MIR143/145 in vitro.

Analysis of Intronic Variants in CYFIP2 and LARP1
The effect of the CYFIP2 rs3734027 (A>G) and LARP1 rs78112077 (G>A) was evaluated via reverse transcription (RT)-polymerase chain reaction (PCR) on total RNA from peripheral blood mononuclear cells (PBMCs).An analysis of the CYFIP2 transcript in the heterozygous (AG) proband and two homozygous (AA and GG) subjects showed two amplicons of different intensity for all three analyzed samples (Figure S2B).Direct sequencing of the strong high band (399 bp) showed expected splicing, while the low-light band (199 bp) resulted from the skipping of exon 21.Quantitative polymerase chain reaction (qPCR) analysis showed no increase in the skipped isoform in individuals with the G allele (Figure S2B).Our analysis of LARP1 alternative transcripts in the heterozygous (GA) proband and a homozygous (GG) subject showed no amplification of LARP1 in both samples despite the amplification of the housekeeping GAPDH transcript, thus suggesting a possible lack of expression of the alternative transcripts in the analyzed cells.

Analysis of 5 UTR Variants in THG1L
Expression of THG1L was evaluated in a larger cohort of subjects with and without the rs2270818 (C>A) and rs2270819 (C>A) variants by retrieving RNA sequencing data available at the GEUVADIS consortium.No statistically significant difference in gene expression was found among the groups of compound heterozygotes (CA/CA) and homozygotes (CC/CC; AA/AA).

Analysis of the Variant in the Promoter of MIR143/145
The effect of the rs13158382 (C>T) variant on levels of miR-143 and miR-145 was first evaluated in three heterozygous (CT) (II-8, II-5, II-9) and three homozygous (CC) (III-4, III-6, III-9) family members.In them, levels of VWF and F8 mRNAs were also analyzed.A slight reduction in both miR-143 and miR-145 and a slight increase in VWF compared to homozygotes was observed in the heterozygous subjects (Figure 2A-C, respectively), while no difference regarding F8 was observed between the two groups (Figure 2D).The observed differences did not reach statistical significance, perhaps due to the very small number of analyzed individuals.To confirm the putative effect of rs3158382 on miR-143 and miR-145 expression in a larger cohort, the genetic and miRNA profiling data of 452 subjects from the GEUVADIS project were obtained and analyzed.A total of 31 subjects-27 homozygous (CC) and 4 heterozygous (CT)-who exceeded the first or the third interquartile were excluded from the analysis as outliers.Due to the non-normal distribution of miRNA expression values (Shapiro-Wilk: p-value = 0.03653), the nonparametric test (Mann-Whitney U) was performed and confirmed the statistical significance (p-value = 0.023) of the reduction in miR-143 expression among the heterozygous subjects compared to the homozygous wild-type ones (Figure 3A).The expression levels of miR-145 were too low to allow for a differential expression analysis.The mRNA data of VWF and F8 were also retrieved and analyzed using the same statistical approach; levels of VWF expression were too low (mean count of unnormalized matrix < 7) to allow a differential expression analysis, and no difference in the expression of F8 was found (Figure 3B).

Impact of the rs13158382 Variant on miRNA Synthesis and Processing
Additional investigations were performed in order to assess whether the reduction in mature miRNAs resulted from a lowered synthesis of primary miRNA (pri-miRNA) and/or from the aberrant processing of precursor miRNA (pre-miRNA).Levels of pri-miRNA-143/145, evaluated in heterozygous (CT) (II-5, II-9) and homozygous (CC) (III-4, III-6, III-9) family members were slightly reduced in heterozygotes compared to homozygotes (Figure S4), thus suggesting that the rs13158382 variant had an impact on miRNA synthesis.The impact of the rs13158382 variant on the secondary structure of the pre-miR-143 was also evaluated using the miRVaS tool.Three different representations of secondary structures (centroid, maximal expected accuracy, and minimal free energy) were accessed.All strategies predicted structural changes in the hairpin and flanking regions that

Impact of the rs13158382 Variant on miRNA Synthesis and Processing
Additional investigations were performed in order to assess whether the reduction in mature miRNAs resulted from a lowered synthesis of primary miRNA (pri-miRNA) and/or from the aberrant processing of precursor miRNA (pre-miRNA).Levels of pri-miRNA-143/145, evaluated in heterozygous (CT) (II-5, II-9) and homozygous (CC) (III-4, III-6, III-9) family members were slightly reduced in heterozygotes compared to homozygotes (Figure S4), thus suggesting that the rs13158382 variant had an impact on miRNA synthesis.The impact of the rs13158382 variant on the secondary structure of the pre-miR-143 was also evaluated using the miRVaS tool.Three different representations of secondary structures (centroid, maximal expected accuracy, and minimal free energy) were accessed.All strategies predicted structural changes in the hairpin and flanking regions that may affect miRNA processing, such as an alteration of Drosha cleavage site (Figure 4).The effect of rs13158382 on the secondary structure of miR-145 could not be predicted since this variant is located 1735 nucleotides upstream of the pre-miRNA sequence.may affect miRNA processing, such as an alteration of Drosha cleavage site (Figure 4).The effect of rs13158382 on the secondary structure of miR-145 could not be predicted since this variant is located 1735 nucleotides upstream of the pre-miRNA sequence.

Discussion
A large Italian family presenting with high/extremely high levels of both FVIII and VWF was investigated to identify new genetic factors predisposing to FVIII/VWF increase.Since, in the general population, VWF:Ag and FVIII:C levels are regulated by genetic (blood group) and acquired factors (comorbidities, body mass index, age) [20], the putative contribution of these variables on the levels of FVIII and VWF was evaluated in the family.The expected correlation was found in subjects with normal and extremely high levels of FVIII/VWF (>500% the lower limit of normal ranges).Conversely, no correlation was found in family members with high levels (up to 300% the lower limit of normal ranges) (Figure S1).These observations suggested the presence of a possible genetic predisposition to high FVIII/VWF levels, coupled with an additive effect of non-O blood group, elderly age, and other acquired variables on the additional FVIII/VWF increase in family members with extremely high levels of VWF.Levels (normal/high/extremely high) of FVIII and VWF were positively correlated in the majority (11) of family members; only 4 members (III-4, III-5, III-6, III-9) had FVIII levels slightly above the upper normal limit and VWF levels in the normal range.Since member III-3 had FVIII levels similar to those of member III-5 (200% vs. 198%) but higher levels of VWF (227% vs. 145%) (Figure 1, Table 1), the reduced levels of VWF found in III-5 could be assay-related or partially due to the different blood groups (non-O vs. O) that are known to contribute to ~15% of the VWF:Ag variance in plasma [21].In our family, high levels of FVIII and VWF were found in two consecutive generations, suggesting an autosomal dominant inheritance of both phenotypes (Figure 1).Hence, two (FVIII and VWF based) case-control association analyses were performed on WES data, leading to the identification of two overlapping regions on chromosome 5 (56 Kb at q32 and 8350 Kb at q32-q33.3, respectively) (Table 2).This result is not surprising because FVIII is bound by VWF, and the plasma levels of the two proteins are closely related.Indeed, VWF deficiency is accompanied by low levels of FVIII, and, in turn, increased VWF levels induce an increase in FVIII [22,23].In addition, previous genetic association studies identified and replicated few loci associated with FVIII plasma levels that were a subset of 15 loci spread on different chromosomes associated with VWF plasma levels [18,[24][25][26].Concerning chromosome 5, two genes (TMEM171 and TNPO1) localized at q13.2 have been functionally characterized in vitro, and their silencing was found to increase the release of VWF by HUVEC cells [18].
Among the genes identified by our analysis, only NDST1, CYFIP2, and MIR143/145 have a putative functional link to VWF.NDST1 is involved in the differentiation of embryonic stem cells into endothelial cells, with transcript and protein levels of NDST1 being directly correlated to the levels of VWF in endothelial cells [27].CYFIP2 protein is a component of the WAVE1 complex; the binding of WAVE1 to RAC1 mediates actin polymerization, and the depletion of RAC1 has been demonstrated to prevent VWF secretion in HUVEC cells [28].Concerning miR-143 and miR-145, their downregulation has been associated with the upregulation of the VWF protein in patients with acute coronary syndrome and of the VWF transcript during the endothelial differentiation of human adipose-derived stem cells (hADSCs) [29][30][31].It is worth noting that individuals in the family under study with reduced levels of miR-143/145 and high (II-9)/extremely high (II-3, II-8) levels of VWF suffer from a different type of obesity (Table 1).miR-143 and miR-145 have also been linked to thrombosis.In particular, the intravenous injection of miR-145-transfected endothelial progenitor cells (EPCs) facilitates the recanalization of arterial thrombi [32].Coagulation factor XI and tissue factor (TF) are targets of miR-145 [33,34]; in patients with VTE, both proteins were elevated, and an inverse correlation between miR-145 levels and TF levels was observed [34].Moreover, the restoration of miR-145 levels in thrombotic rats via miR-145 mimic delivery resulted in decreased TF level and activity, accompanied by reduced thrombogenesis [34].Since TF is a potent stimulator of VWF secretion [35], a possible link between the reduction in miR-143/145 and increase in VWF mediated by TF may be hypothesized, but TF measurements were not performed for our family members.
Low levels of miR-143 and miR-145 are also associated with metabolic diseases such as type 2 diabetes, atherosclerosis, obesity, and an increased inflammatory response [36][37][38].Recently, it has been reported that miR-145, which is secreted by mesenchymal-stemcells, packaged in exosomes, and delivered to endothelial cells, reduces the formation of atherosclerotic plaques in mice [39].It is worth noting that several diseases, such as diabetes, obesity, cancer, and hypertension, were present in our family members.
Very recently, experiments in HEK293 and HUVEC cell lines also demonstrated that miR-143 could target and inhibit VWF [40].
In vitro and in silico studies on the identified variants highlighted a possible causal role only for the low-frequency rs13158382 variant (located upstream of the MIR143/145 genes).The MIR143/145 genes clustered at 5q32 share a common promoter consisting of a conserved region of 4.2 Kb that regulates their expression.miR-143 and miR-145 are co-transcribed into a single bicistronic unit consisting of a primary transcript (pri-miRNA) [41] with a hairpin structure and are released into the nucleus as precursor miRNA (pre-miRNA) after Drosha and DGCR8 cleavage [42].The identified rs13158382 variant is a low-frequency SNPs (MAF 0.037) located 7-bp from pre-miRNA that has been previously predicted to modify a transcription factor binding site [43].Two other SNPs (rs4705342 and rs4705343), located in the promoter of MIR143/145 at −400 and −510 bp, respectively, have been functionally characterized via dual-luciferase reporter assays and have been found to increase and reduce luciferase activity respectively, thus suggesting altered promoter activity and miR-143/145 synthesis [44,45], as observed in our family (Figure S4).
Since genetic variants in or close to miRNA genes can affect structural modifications [46], we also assessed the impact of the rs13158382 variant on the secondary structure of pre-miR-143 using the mirVAS tool.A conformational change in the miRNA secondary structure was predicted, with potential consequences on mature miRNA processing.
To assess the (inverse) correlation between miR-143/145 and VWF transcript levels, we performed transcript analysis in the available cells (PBMCs) of family members with and without the rs13158382 variant.The expected trend was observed, even if differences were not statistically significant, perhaps due to the small number of analyzed samples and low (i.e., ectopic) expression of F8/VWF in PBMCs.To further tackle this issue, we analyzed miR-143, miR-145, F8, and VWF transcript data from a larger cohort of subjects [47].A differential expression analysis, feasible only for miR-143 and F8, confirmed a statistically significant reduction in miR-143 in heterozygous rs13158382 individuals, thus suggesting a possible role for the identified variant in miR-143 expression.We assessed BioGPS, Geo Profiles, and ImmGen Project databases containing expression data [48], but no genetic and transcriptional evidence concerning cells expressing the VWF were available for further analyses.
The limitations of this study are attributable to the following: (i) the unavailability of endothelial cells to evaluate the real expression of F8/VWF transcripts and (ii) the lack of in vitro studies to confirm the role of the rs13158382 variant on miR-143/145 transcription or processing.Hence, a luciferase assay should be performed to confirm our findings.In conclusion, notwithstanding the aforementioned limitations, there is evidence that the rs13158382 variant modulated miR-143/145 expression, even though deeper investigations are needed in order to understand the effective contribution of this genetic variant to miRNA expression and to elucidate new possible molecular mechanisms underlying VWF expression and VTE.If confirmed, the rs13158382 variant may become a new genetic risk factor of VTE and miR-143/145 potential biomarkers of VTE and a potential future therapeutic tool to reduce VWF expression.

DNA and RNA Extraction
Genomic DNA was extracted from PBMCs via the standard salting-out method [51].Total RNA, including miRNA, was extracted from whole blood using the PaxGene ® Blood miRNA kit (Preanalytix, Zurich, Switzerland), following manufacturer's protocol.The DNA and RNA samples were quantified using the NanoDrop 2000c spectrophotometer (Thermo Fisher Scientific, Wilmington, DE, USA).A total of 15 DNA samples (12 collected before and 3 after the whole exome sequencing) and 6 RNA samples were available for study.
High levels of FVIII and VWF were considered as two different phenotypes, and variant filtering was performed twice.Variants were filtered according to an autosomal dominant inheritance model of high levels of FVIII or VWF; variants observed to be in the heterozygous state in all family members with FVIII:C > 150% or VWF:Ag > 180% (i.e., cases) and absent in individuals with FVIII:C < 150% or VWF:Ag < 180% (i.e., controls) were further analyzed (FVIII and VWF cut-off values were chosen based upon literature data and the upper normal limits of FVIII/VWF measurement assays).Only variants with a call rate of 100% were included.

Prioritization of Variants and Genes
The effect of the identified variants was predicted in silico using the PredictSNP2 tool v2.1 [57].The localization of variants in predicted regulatory regions was accomplished using the Enhancer Atlas (http://enhanceratlas.org/,accessed on 17 September 2021) database [58].Genes were prioritized based on the phenotype of interest (i.e., thrombosis, VWF disease, and inflammation) using the VarElect tool (https://varelect.genecards.org/,accessed on 17 September 2021) [59].Genotyping of cases and controls was performed via the direct sequencing of the regions encompassing the identified variants using the BigDye Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems, Waltham, MA, USA) and an ABI PRISM 3130 Genetic Analyzer (Applied Biosystems).Primers and PCR conditions are available upon request.

First-Strand cDNA Synthesis and qPCR
First-strand cDNA synthesis was performed on total RNA by using random nonamers and the High-Capacity cDNA Reverse Transcription kit (Thermo Fisher, Waltham, MA, USA).miRNA-specific RT was performed on total RNA by using a specific RT primer and the TaqMan™ MicroRNA Reverse Transcription kit (Thermo Fisher).
The genotyping of factor V Leiden and the prothrombin variant was performed on 175 ng of DNA by using the Q-PCR Alert Kit (ELITechGroup, Berkhamsted, UK).
The fluorescence signals were monitored using the StepOnePlus Real-Time PCR system (Applied Biosystems) and StepOne software v2.3 (Applied Biosystems).Two duplicates of each sample were analyzed, and the relative amounts were determined using the 2−∆∆Ct method.Student's t-test was applied for the statistical analysis.

Secondary Structure Prediction
The impact of the genetic variant on miRNA secondary structure was predicted using the mirVAS tool (http://mirvas.bioinf.be/,accessed on 8 October 2021); the default setting (i.e., 100 nucleotides upstream and downstream of the pre-miR-143) was used for the analysis.

Figure 1 .
Figure 1.Pedigree of the family.Levels of FVIII coagulant activity (FVIII:C) and VWF antigen (VWF:Ag) are indicated below each symbol and reported as percentages (%) (FVIII:C normal range 51-147%; VWF:Ag normal range 40/55-165/169%).Values above the upper normal limits are indicated in bold.The arrow indicates the proband.Barred boxes refer to deceased family members, and gray boxes denote family members not tested by WES.na, not available.

Figure 1 .
Figure 1.Pedigree of the family.Levels of FVIII coagulant activity (FVIII:C) and VWF antigen (VWF:Ag) are indicated below each symbol and reported as percentages (%) (FVIII:C normal range 51-147%; VWF:Ag normal range 40/55-165/169%).Values above the upper normal limits are indicated in bold.The arrow indicates the proband.Barred boxes refer to deceased family members, and gray boxes denote family members not tested by WES.na, not available.

Figure 2 .
Figure 2. rs13158382 (C>T) variant and transcript levels in heterozygous and homozygous family members.Box plots show the relative quantitation (RQ) of (A) miR-143, (B) miR-145, (C), VWF, and (D) F8 transcripts in heterozygous (CT) and homozygous (CC) family members."n" denotes the number of analyzed individuals for each genotype.

Figure 2 .
Figure 2. rs13158382 (C>T) variant and transcript levels in heterozygous and homozygous family members.Box plots show the relative quantitation (RQ) of (A) miR-143, (B) miR-145, (C), VWF, and (D) F8 transcripts in heterozygous (CT) and homozygous (CC) family members."n" denotes the number of analyzed individuals for each genotype.

Figure 2 .
Figure 2. rs13158382 (C>T) variant and transcript levels in heterozygous and homozygous family members.Box plots show the relative quantitation (RQ) of (A) miR-143, (B) miR-145, (C), VWF, and (D) F8 transcripts in heterozygous (CT) and homozygous (CC) family members."n" denotes the number of analyzed individuals for each genotype.

Figure 3 .
Figure 3. rs13158382 (C>T) variant and transcript levels in heterozygous and homozygous subjects from the GEUVADIS project.Box plots show the number of reads of (A) miR-143 and (B) F8 evaluated in heterozygous (CT) and homozygous (CC) subjects from the GEUVADIS project."n" denotes the number of analyzed individuals for each genotype.

Figure 3 .
Figure 3. rs13158382 (C>T) variant and transcript levels in heterozygous and homozygous subjects from the GEUVADIS project.Box plots show the number of reads of (A) miR-143 and (B) F8 evaluated in heterozygous (CT) and homozygous (CC) subjects from the GEUVADIS project."n" denotes the number of analyzed individuals for each genotype.

Figure 4 .
Figure 4. Prediction of miR-143 secondary structures.(A) miRVaS visual output shows the predicted structure of pri-miR-143 with (Variant) and without (Reference) the rs13158382 variant.A structural change in the 5p flanking region upstream of the hairpin arm is shown.Functional regions within the pre-miRNA are differentially colored: seed sequence (orange), mature miRNAs (magenta), ter-

Figure 4 .
Figure 4. Prediction of miR-143 secondary structures.(A) miRVaS visual output shows the predicted structure of pri-miR-143 with (Variant) and without (Reference) the rs13158382 variant.A structural change in the 5p flanking region upstream of the hairpin arm is shown.Functional regions within the pre-miRNA are differentially colored: seed sequence (orange), mature miRNAs (magenta), terminal loop (blue), hairpin arm (cyan).The localization of the rs13158382 variant (red) and the nucleotides involved in putative structural changes (black) are highlighted.(B) miRVaS output shows the predicted most important region (arm) with a structural impact and the predicted conservation of the hairpin along with the number of changed bases within the hairpin.All predictions are based on minimal free energy (MFE), maximal expected base-pair accuracy (MEA), and the smallest total base-pair distance to the sampled structures of that ensemble (centroid).

Table 1 .
Clinical and laboratory data of analyzed subjects.
Abbreviations: F, female; M, male; BMI, body mass index; FVIII:C, factor VIII coagulant activity; FVIII:Ag, factor VIII antigen; VWF:Ag, von Willebrand factor antigen; VWF:RCo, von Willebrand Ristocetin cofactor.aSubjectsare named according to the pedigree depicted in Figure1; subjects not analyzed via WES are in italics.b Diseases that may be associated with an increase in FVIII/VWF levels.c Normal range for O blood type.d Normal range for non-O blood type.The "-" denotes unknown data.

Table 2 .
Variants identified via WES through (A) the FVIII-based approach and (B) the VWF-based approach.
Abbreviations: MAF, minor allele frequency; CADD, Combined Annotation Dependent Depletion; DANN, Deleterious Annotation of Genetic Variants using Neural Networks;
a MIR143 and MIR145 genes are localized in an exonic region and an intronic region of the CARMN gene, respectively.