Novel LTBP3 mutations associated with thoracic aortic aneurysms and dissections

Background Thoracic aortic aneurysm and dissection (TAAD) is a hidden-onset but life-threatening disorder with high clinical variability and genetic heterogeneity. In recent years, an increasing number of genes have been identified to be related to TAAD. However, some genes remain uncertain because of limited case reports and/or functional studies. LTBP3 was such an ambiguous gene that was previously known for dental and skeletal dysplasia and then noted to be associated with TAAD. More research on individuals or families harboring variants in this gene would be helpful to obtain full knowledge of the disease and clarify its association with TAAD. Methods A total of 266 TAAD probands with no causative mutations in known genes had been performed wholeexome sequencing (WES) to identify potentially pathogenic variants. In this study, rare LTBP3 variants were the focus of analysis. Results Two compound heterozygous mutations, c.625dup (p.Leu209fs) and c.1965del (p.Arg656fs), in LTBP3 were identified in a TAAD patient along with short stature and dental problems, which was the first TAAD case with biallelic LTBP3 null mutations in an Asian population. Additionally, several rare heterozygous LTBP3 variants were also detected in other sporadic TAAD patients. Conclusion The identification of LTBP3 mutations in TAAD patients in our study provided more clinical evidence to support its association with TAAD, which broadens the gene spectrum of LTBP3. LTBP3 should be considered to be incorporated into the routine genetic analysis of heritable aortopathy, which might help to fully understand its phenotypic spectrum and improve the diagnostic rate of TAAD.


Background
Thoracic aortic aneurysms are often asymptomatic, as they progressively enlarge and are finally aware of being diagnosed clinically when dissections or ruptures occur, which are life-threatening and can cause sudden death in up to 50% of patients [1,2]. Therefore, early diagnosis and timely treatment of thoracic aortic aneurysm and dissection (TAAD) are very important for reducing mortality.
Many factors contribute to TAAD development, while genetic defects play a major role. Marfan syndrome (MFS) is the most well-known heritable aortic disease, and other syndromic connective diseases involving the aorta, such as Loeys-Dietz syndrome (LDS) and vascular Ehlers-Danlos syndrome (EDS), are less commonly seen. An increasing number of causative mutations have also been identified in nonsyndromic TAAD patients, but they can only explain a small proportion (approximately 20%) of TAAD  [3], which suggests that there are many unrecognized genes to be explored.
LTBP3 belongs to the latent TGF-ß binding protein (LTBP) family and contains several modules including TGF-β-binding (TB) domain, epidermal growth factor (EGF), 4 Cys, hybrid, and calcium-binding EGF domains [13], which closely resembles that of fibrillin 1. Studies have shown that LTBP3 is assisted by fibrillin-1 to incorporate into the extracellular matrix, which can be hindered in the absence of fibrillin-1 microfibrils in vivo or in vitro [12]. Meanwhile, it has an essential role in regulating the TGF-β signaling pathway [5,6]. Therefore, LTBP3 might be involved in TAAD. Nevertheless, only a few cases with LTBP3 mutations have been reported and we may not understand the full clinical spectrum of the disease. Aortic deformity might be an important characteristic that occurs more frequently but has been neglected in the past.
In this study, we reported two compound heterozygous LTBP3 mutations in a TAAD patient along with DASS and several rare heterozygous LTBP3 variants in sporadic TAAD patients. Our results provide more clinical evidence supporting that LTBP3 mutations might be responsible for heritable TAAD.

Patients
Patients with TAAD from the Center of Vascular Surgery in Fuwai Hospital were referred to our Diagnostic Laboratory Service for genetic testing. The targeted sequencing contained 15 genes (FBN1, TGFB2, SMAD3, TGFBR1, TGFBR2, ACTA2, MYH11, SMAD4, MYLK, NOTCH1, PRKG1, SKI, COL3A1, SLC2A10, FBN2) associated with aortic disease and was performed as previously reported [14]. Among them, 266 patients who did not have an identified causative mutation in the panel testing were recruited in our study. Wholeexome sequencing was then performed to identify potentially pathogenic mutations.

Whole exome sequencing (WES)
Genomic DNA was extracted from EDTA-anticoagulated whole blood of patients and their relatives using a QIAamp DNA Blood Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions [15,16]. The DNA samples and sequencing data were processed as previously described [17]. WES was performed on an Illumina HiSeq2500 platform (Illumina Inc., San Diego, CA, USA) using the TruSeq Rapid PE Cluster kit V2 or TruSeq Rapid SBS kit V2 -HS (Illumina Inc., San Diego, CA, USA).

Bioinformatics analysis
The process of WES bioinformatics analysis can be roughly divided into three modules: data preprocessing, variant detection, and annotation. FastP software was used to filter the low-quality sequence reads [18], and sequence reads were aligned to the reference genome hg19 using BWA (Burrows-Wheeler Aligner) MEM software [15,19,20]. Picard (MarkDuplicates) software (https:// github. com/ broad insti tute/ picard) was used to label repetitive reads, and GATK3.7 software was used to correct systematic errors in the sequencing process. The GATK Haplotypecaller (https:// gatk. broad insti tute. org) was used in variant calling, and Annovar software (https:// annov ar. openb ioinf ormat ics. org/ en/ latest/) was used for annotation analysis. High quality variants (mean sequencing depth >=20×, alternation frequency (AltFeq) between 0.3 and 0.6 or > 0.95, strand bias (STB) between 0.5 and 0.7) detected by WES were filtered for further analysis. Variants based on the < 1% minor allele frequency (MAF) in any gnomAD (version 2.1) population (gnomAD_popmax) and location in the coding regions and variable splicing sites (canonical ±1 or 2 splice sites or splice sites of either dbscSNV_ADA or dbscSNV_RF > 0.6 were retained. Then LTBP3 rare variants with a MAF < 0.01 and < 0.00005 in gnomAD were focused separately for autosomal recessive and dominant conditions. The variant interpretation was then performed according to ACMG guidelines [21,22]. Variants detected in our study were written according to the HGVS nomenclature rules.

Sanger sequencing
All the mutations detected by WES were verified by Sanger sequencing. According to the reference genomic sequences of the Human Genome from GenBank in NCBI [22], Primer3 Input(version 0.4.0) was used to design primer pairs. Sanger sequencing was performed with the following primers:

Results
Homozygous or compound heterozygous LTBP3 mutations were first reviewed in our 266 WES data, with a MAF< 0.01 in gnomAD. Two heterozygous mutations, c.625dup (p.Leu209fs) and c.1965del (p.Arg656fs), were detected in patient AD2002. These two variants were not recorded in the population database, and both of them were predicted to undergo nonsense-mediated decay. They were both confirmed by Sanger sequencing. After detecting his parents' genotype, we found that one of them (c.1965del, p.Arg656fs) was inherited from his mother, while the other (c.625dup, p.Leu209fs) was de novo. To determine whether they were in trans or cis, the genotypes of their offspring were also detected. His twin sons carried only the variant c.625dup, which highly suggested that these two variants were on opposite chromosomes. Therefore, these compound heterozygous variants could be interpreted as pathogenic according to the ACMG guidelines (Table 1) and were causative for his disease. Patient AD2002 was a 42-year-old man, 165 cm/63 kg, with no hypertension history. He had sudden back pain during a business trip and went to the local hospital. Echocardiography showed dilatation of the aortic sinus (53 mm) and ascending aorta (38 mm), and thoracic computed tomography (CT) revealed aortic dissection in the descending aorta (DeBakey type III). TEVAR surgery was then performed. Two months later, he came to Fuwai Hospital for reexamination and a genetic test. After recognizing his two LTBP3 mutations, we further asked about his symptoms and family history. He also had spinal stenosis and dental anomalies, for which his teeth were all replaced at the age of 38. Nevertheless, his mother and sister, as well as his two sons, who all harbored only one heterozygous variant, had no abnormalities (Fig. 1A). No other homozygous or compound heterozygous mutations in this gene were detected in our cohort.
Subsequently, rare heterozygous LTBP3 mutations were filtered in our dataset, with a MAF< 0.00005 in gno-mAD. A total of four missense mutations were detected in four individuals affected with early-onset thoracic aortic dissection ( Table 2). None of these patients had significant systemic abnormalities other than the aortic  disease (details in Table 2). None of the mutations were found in TAAD genes that had been approved by Clin-Gen in these patients, except for a variant of unknown significance (VUS) in the FBN1 gene. Patient AD820 harbored the FBN1 variant c.994C>T (p.Arg332Cys) and the LTBP3 variant c.1588A>G (p.Thr530Ala), both of which remained uncertain significance.

Discussion
Disturbed TGF-β signaling has been extensively implicated in thoracic aortic aneurysm and dissection. Mutations in FBN1 lead to defective ECM microfibrils and dysregulation of TGF-β signaling, both of which play a vital role in the development of aortopathy in MFS patients [23][24][25][26][27]. Several genes in the TGF-β signaling pathway, such as TGFBR1, TGFBR2, SMAD2, TGFB2, TGFB2, and SMAD2, have been proven to be causative for syndromic TAAD [28,29]. The LTBP3 gene encodes latent transforming growth factor-beta binding protein 3, which has a similar structure to FBN1. Both of them have multiple EGF-like repeats and unique 8-cys domains that are interspersed with TB domains [30]. LTBP3 regulates TGF-β activity by enabling its secretion, directing it to specific sites in the ECM, and participating in its activation [31][32][33][34]. It is widely expressed, including in the skeleton, tooth, heart, and aorta, which suggests that mutations in this gene may lead to a broad range of phenotypes.
Ltbp3-deficient mice were initially documented to have small body sizes, unique craniofacial malformations, bone abnormalities, and dental anomalies [35][36][37][38][39], which were consistent with the clinical features in individuals with LTBP3 mutations, recessive-pattern DASS, and dominant-pattern ACMCID and GPHYSD3 patients. In earlier reports, cardiovascular deformities were not described in either mouse model or patient. Subsequently, Zilberberg et al. [34] have shown that lack of Ltbp3 could attenuate the aneurysmal phenotype and prevent premature death of Fbn1 mgR/mgR mice, probably by the reduced activation of TGF-β signaling. However, Guo et al. [10] pointed out that when aortic diameters were normalized by body mass, the diameter of the aortic root and ascending part in Ltbp3 −/− Fbn1 mgR/mgR mice was similar to that in the Fbn1 mgR/mgR group, and Ltbp3 deficiency resulted in spontaneous aortic dilation. These data suggested that Ltbp3 was associated with the formation and progression of thoracic aortic aneurysms, although the exact effect and mechanism were not well understood. It was reasonably assumed that LTBP3 mutations might contribute to TAAD by disturbing the TGF-β signaling pathway and extracellular matrix assembly. More studies are needed to determine its specific role in TAAD.
Regarding the clinical data, there were only a few reports supporting the role of LTBP3 in the cardiovascular system. Dugan et al. reported that two sisters in a single family with a homozygous truncated mutation of LTBP3 both had mitral valve prolapse (MVP) [7]. Guo et al. reported that biallelic null mutations in LTBP3 could predispose individuals to thoracic aortic aneurysms and dissections, and heterozygous rare LTBP3 variants might be related to an early onset risk of acute aortic dissection [10]. This was the first and to date, the only report that supported LTBP3 mutations that might be responsible for TAAD. Our study identified two compound heterozygous variants in LTBP3 in an aortic dissection patient, along with short stature and dental problems. They were both frameshift mutations, which were predicted to result in nonsense-mediated mRNA decay (NMD). This was consistent with the inherited pattern and mutation types in DASS, and it offered more clinical evidence showing that LTBP3 mutations could cause TAAD.    When reviewing the previously reported cases with homozygous or compound heterozygous LTBP3 null mutations, it was not hard to find that the limited numbers of affected individuals were all diagnosed with DASS at an early age (Table 3). Cardiovascular manifestations were not observed and mentioned specifically. Therefore, we could not exclude the possibility that they might have aortic deformities for the rest of their life. It was reasonable and advisable to follow up these patients to see whether they would have aortic or valvular problems. As LTBP3 was not included in the reevaluation list of the ClinGen expert group on familial TAAD genes [1], it was not incorporated into routine targeted genetic analysis and was mostly not taken into account as a key candidate even in WES. Thus, there was a need to reanalyze the existing WES data on TAAD patients, which might reveal more novel mutations in this gene. More data are needed to assess the frequency of aortic or valvular abnormalities in DASS, which might be more frequent than we thought.
A comprehensive summary of all reported LTBP3 mutations is demonstrated in Fig. 2. It was observed that the majority of bi-allelic loss-of-function LTBP3 mutations, which contributed to DASS, were present in the highly conserved EGF-like calcium-binding domain, while most of the rare monoallelic variants associated ACMICD, GPHYSD3, and thoracic aortic dissection (TAD) were missense mutations located in the same region, indicating that different inheritance modes and natures of LTBP3 might lead to different diseases. Nevertheless, currently reported monoallelic variants in this gene are so limited that further investigations are needed to assess their pathogenicity and apply thorough evaluations to the patients.
The relationship between rare LTBP3 heterozygous variants and aortic dissection risk also remains unclear. Most of the rare LTBP3 variants detected in early-onset of aortic dissection patients by Guo et al. [10] were in the key EGF-like calcium-binding domains, while ours were not, which were less likely to be causative. However, it could not be ruled out that these variants had a mild effect on the development and/or severity of the disease, due to their high CADD predictive scores. Patient AD820 was identified with a variant with unknown significance (PM1, PP3) in the FBN1 gene, c. 994C>T (p.Arg332Cys), which was reported at a low frequency (0.000008808) in the gnomAD database. At the same time, he was identified to carry an LTBP3 heterozygous variant. There was a possibility that these rare variants contributed together to the disease. More cases are needed to fully describe the disease map and genotype-phenotypic association.

Conclusions
In summary, we identified the first case of a TAAD patient with bi-allelic LTBP3 frameshift mutations in an Asian population, as well as several rare LTBP3 variants in affected individuals with early aortic dissection. This expanded the gene spectrum of LTBP3 and provided more support for its role in TAAD. Our data show the necessity of incorporating this gene into the routine genetic analysis of aortic aneurysms and dissections, which would help to fully understand its phenotypic spectrum and raise the diagnosis rate of TAAD.