Loss of function mutation in LOX causes thoracic aortic aneurysm and dissection in humans

Significance

The mechanical integrity of the arterial wall is dependent on a properly structured ECM. Elastin and collagen are key structural components of the ECM, contributing to the stability and elasticity of normal arteries. Lysyl oxidase (LOX) normally cross-links collagen and elastin molecules in the process of forming proper collagen fibers and elastic lamellae. Here, using whole-genome sequencing in humans and genome engineering in mice, we show that a missense mutation in LOX causes aortic aneurysm and dissection because of insufficient elastin and collagen cross-linking in the aortic wall. These findings confirm mutations in LOX as a cause of aortic disease in humans and identify LOX as a diagnostic and potentially therapeutic target.

Abstract

Thoracic aortic aneurysms and dissections (TAAD) represent a substantial cause of morbidity and mortality worldwide. Many individuals presenting with an inherited form of TAAD do not have causal mutations in the set of genes known to underlie disease. Using whole-genome sequencing in two first cousins with TAAD, we identified a missense mutation in the lysyl oxidase (LOX) gene (c.893T > G encoding p.Met298Arg) that cosegregated with disease in the family. Using clustered regularly interspaced short palindromic repeats (CRISPR)/clustered regularly interspaced short palindromic repeats-associated protein-9 nuclease (Cas9) genome engineering tools, we introduced the human mutation into the homologous position in the mouse genome, creating mice that were heterozygous and homozygous for the human allele. Mutant mice that were heterozygous for the human allele displayed disorganized ultrastructural properties of the aortic wall characterized by fragmented elastic lamellae, whereas mice homozygous for the human allele died shortly after parturition from ascending aortic aneurysm and spontaneous hemorrhage. These data suggest that a missense mutation in LOX is associated with aortic disease in humans, likely through insufficient cross-linking of elastin and collagen in the aortic wall. Mutation carriers may be predisposed to vascular diseases because of weakened vessel walls under stress conditions. LOX sequencing for clinical TAAD may identify additional mutation carriers in the future. Additional studies using our mouse model of LOX-associated TAAD have the potential to clarify the mechanism of disease and identify novel therapeutics specific to this genetic cause.

Thoracic aortic aneurysms and dissections (TAAD) comprise a large group of heterogeneous conditions with substantial phenotypic diversity. Individuals presenting with an inherited form of TAAD may have evidence of a defined genetic syndrome (e.g., Marfan syndrome, Loeys–Dietz syndrome, Ehlers–Danlos syndrome type IV, or a TGF-β–related vasculopathy) or simply, a family history of TAAD [termed familial thoracic aortic aneurysms and dissections (FTAAD)]. Anatomically, FTAAD can affect various segments of the arterial system from the aortic root and ascending aorta to more distal arterial segments, such as the hepatic or pulmonary arteries. In addition, FTAAD is characterized by significant locus heterogeneity with mutations in a diverse group of genes reported to date (1–10). Despite significant progress in defining the genetic basis of FTAAD, the molecular etiology of many cases remains enigmatic.

Elastin and collagen are two of the major structural components that comprise the arterial wall. Lysyl oxidase (LOX) and its related gene family members are a group of copper-dependent oxidodeaminases that cross-link lysyl residues on these structural proteins in the process of forming proper elastic lamellae and collagen fibers (11). The homozygous knockout (KO) of the murine Lox gene results in perinatal death from aortic aneurysm and spontaneous dissection, and mutant Lox null mice exhibit highly abnormal aortic histology characterized by fragmented elastic fibers and aberrant smooth muscle cell layers (12, 13). These mouse model findings strongly support the functional significance of LOX in the maintenance of normal arterial wall integrity.

Here, we report a LOX missense mutation discovered through whole-genome sequencing in a family with autosomal dominant TAAD, where prior clinical genetic testing was unrevealing. Using genome engineering techniques, we created a mouse model of the specific human mutation to study its in vivo effects. In the heterozygous state, we found that mice with the human mutation displayed abnormal aortas with disorganized assembly of elastic lamellae in the aortic wall, whereas mice homozygous for the human mutation died from perineonatal aortic aneurysm and spontaneous hemorrhage. Together, our data identify another genetic etiology for autosomal dominant TAAD with important implications for clinical patient care.

Results

Characterization of a Family with FTAAD.

The proband (Fig. 1A, individual III-1) presented to Brigham Genomic Medicine as a 35-y-old Caucasian male for evaluation of his personal and family history of TAAD. His history included a surgical repair of pectus excavatum at two years of age and the diagnosis of a large ascending aortic aneurysm at 19 years old. The aneurysm, which did not involve the aortic root and extended to the brachiocephalic artery, measured 10.5 cm in diameter and was discovered on chest computed tomography imaging as part of an evaluation for complaints of chest pain and cough. The patient underwent a valve-sparing aortic root replacement. The histopathological analysis of his resected aortic tissue found a contained posterior rupture with evidence of cystic medial necrosis and fragmented external elastic lamella. Based on his medical history and suggestive physical features (which included tall stature, high arched palate, and dental crowding), he was given a clinical diagnosis of Marfan syndrome. However, genetic testing, which included fibrillin-1 (FBN1) gene sequencing and multiplex ligation-dependent probe amplification, transforming growth factor beta receptor 1 (TGFBR1) gene sequencing, and transforming growth factor beta receptor 2 (TGFBR2) gene sequencing, was negative. His physical examination was notable for pectus excavatum, presence of venous varicosities in his right lower extremity, positive thumb but not wrist signs, and skin striae on his flanks. A review of his family history suggested an autosomal dominant disorder (Fig. 1A).

Fig. 1.

FTAAD associated with LOX mutation. (A) The family’s pedigree shows autosomal dominant inheritance of FTAAD. Black symbols indicate affected individuals with arterial dissection or aneurysm. Gray symbols indicate individuals affected with arterial tortuosity. White symbols indicate unaffected individuals; − indicates presence of the mutation. (B) Contrast-enhanced axial computed tomography image from individual II-1 showing ascending aortic aneurysm (arrow) with dissection and an intimal flap separating the false and true lumens. (C) Histologic evaluation of aortic tissue resected from individual II-1. H&E staining shows abnormal architecture and a dissection tear. Masson’s trichrome staining for collagen (blue) reveals disorganization of the collagen fibers and disruption of the medial architecture. Verhoeff–van Gieson staining for elastin (dark-purple fibers) illustrates disarray and fragmentation of elastic fibers. (D) The location of the M298R missense mutation is depicted in relation to the domains of the LOX protein. Methionine at position 298 is highly conserved as shown by the homologous protein sequences from multiple organisms shown below. Histidine positions that are essential for copper binding and/or LOX catalytic activity (14) are highlighted in blue.

The mother of the proband (Fig. 1A, individual II-1) was also evaluated in our clinic because of her history of acute ascending aortic dissection with repair at 52 years old (Fig. 1B), abdominal hernia repair at 34 years of age, and myopia. The histopathological analysis of her resected aortic tissue also showed cystic medial necrosis with a disorganized appearance of collagen and elastic lamellae (Fig. 1C). She underwent expanded genetic testing, which failed to reveal a causal mutation in TGFBR1, TGFBR2, ACTA2, COL3A1, MYH11, SLC2A10, SMAD3, or MYLK. A variant of unknown significance c.703G > C (p.V235L) was found in the TGFB1 gene. Clinical features of other family members are provided in Table S1.

Table S1.

Clinical characteristics of family members from the study

Individual	Age, y	Clinical features
I-1	Died at 60	Height 183 cm; emphysema in the setting of tobacco use; pectus excavatum and scoliosis
I-2	Died at 92	Height 158 cm; normal screening echocardiogram; no history of aortic aneurysm or dissection
II-1	61	Height 179 cm; ascending aortic aneurysm and dissection at the age of 52 y old
II-2	76	Height 162 cm; normal echocardiogram; total body magnetic resonance angiography (MRA) shows no evidence of arterial tortuosity (Fig. S4) but reveals an incidental finding of a 2-mm brain aneurysm
II-3	73	Height 175 cm; normal serial echocardiograms; total body MRA shows arterial tortuosity (Fig. S4); however, this finding is also compatible with aging
III-1	39	Height 203 cm; 10.5-cm ascending aortic aneurysm with contained rupture diagnosed at the age of 19 y old requiring valve-sparing aortic root replacement; pectus excavatum, scoliosis, positive thumb sign, high arched palate, dental crowding, and skin striae
III-2	35	Height 173 cm; normal serial echocardiograms
III-3	46	Aortic arch dissection at the age of 41 y old; hepatic artery aneurysm
III-4	41	Height 196 cm; infrarenal abdominal aneurysm

Identification of a Missense Mutation in LOX Associated with FTAAD.

In November of 2013, we performed whole-genome sequencing in two first cousins with TAAD (Fig. 1A, individuals III-1 and III-3) to identify the causal gene and mutation underlying the disease in this family. We required putative causal mutations to (i) be shared in a heterozygous state between these two individuals; (ii) be rare, with a minor allele frequency in all populations from the National Heart Lung and Blood Institute Exome Sequencing Program and Exome Aggregation Consortium of 0.01% or less, and not be observed in a local database of individuals sequenced for other rare, nonvascular Mendelian disorders; and (iii) exert a functional impact on the gene’s product, restricting our analysis to missense, nonsense, frameshift, or splice site variants. This analysis resulted in a total of seven candidate mutations (Table S2); six of these were eliminated because of lack of cosegregation with disease (i.e., either not present in other affected individuals from the family or present in unaffected individuals in the family) or other considerations (details are in Table S2).

Table S2.

Candidate mutations identified from whole-genome sequencing

Chr	Position	Ref/alt	Gene	Amino acid change	Affected without mutation	Unaffected with mutation	Notes
1	51871629	T/C	EPS15	p.H542R	III-4
2	215813765	C/T	ABCA12	p.G2321R		I-2, II-2
5	121409850	A/C	LOX	p.M298R			Lox null mice develop aortic aneurysm and spontaneous hemorrhage (12, 13)
7	19156851	C/T	TWIST1	p.G32S	III-4
11	65299132	T/C	SCYL1	p.M365T			Scyl1 null mice display neuromuscular disease similar to amyotrophic lateral sclerosis (45)
12	32480750	A/G	BICD1	p.K454R			Bicd1 is expressed exclusively in the developing nervous system of the mouse, and depletion seems to affect neurotrophin trafficking (46)
19	9066474	G/A	MUC16	p.S6991F			Encodes Ovarian Cancer Antigen CA125, a large gene with numerous rare missense mutations in ESP and ExAC; primarily expressed in surface epithelium (47)

Affected indicates individuals from the pedigree with evidence of TAAD. Unaffected indicates individuals from the pedigree without evidence of TAAD. Alt, alternate allele; Chr, chromosome; Ref, reference allele.

The remaining candidate mutation was a missense substitution (c.893T > G encoding p.Met298Arg) (Fig. S1) in LOX, which was considered a strong candidate because of the gene’s known role in arterial wall biology (12, 13). LOX, an enzyme that requires copper for its activity, catalyzes the cross-linking of collagen and elastin by deaminating side chains of specific lysine and hydroxylysine residues (15). Methionine at amino acid 298 is highly conserved and located within the copper binding domain of LOX (Fig. 1D), suggesting that this missense mutation might disrupt normal LOX function.

Fig. S1.

Identification of a heterozygous p.Met298Arg LOX mutation in affected individuals. Sequence chromatograms of genomic PCR products are shown. The sequence of the normal allele from the unaffected individual II-2 is shown in Upper. The sequence of the mutant c.893T > G (p.Met298Arg) LOX allele from the affected individual II-1 is shown in Lower.

Mice Heterozygous for the Lox Missense Variant Have Longer Ascending Aortas with Fragmented Elastic Fibers.

To explore the functional relevance of the LOX p.M298R mutation in terms of vascular disease and determine how the mutation affects LOX function, we used clustered regularly interspaced short palindromic repeats (CRISPR)/clustered regularly interspaced short palindromic repeats-associated protein-9 nuclease (Cas9) genome editing to introduce the human mutation into the homologous site in the mouse genome that corresponds to amino acid 292 (Lox p.M292R) (Fig. S2). Animals heterozygous for the mutation (Lox^+/M292R; hereafter referred to as Lox^+/Mut) appeared grossly normal and showed no increased mortality through 6 mo of age. Aortic diameter in Lox^+/Mut animals was normal, but ascending aortic length measured from the aortic root to the brachiocephalic artery was 10% longer in the Lox^+/Mut animals (3.28 ± 0.05 mm; n = 9) compared with Lox^+/+ littermate controls (2.94 ± 0.06 mm; n = 7) (Fig. 2A). Compared with Lox^+/+ littermate controls, the Lox^+/Mut animals did not have significantly different systolic (110.71 ± 9.96 vs. 118.22 ± 9.46 mmHg in Lox^+/+ and Lox^+/Mut animals, respectively; P = 0.15) or diastolic (75.57 ± 8.30 vs. 77.56 ± 4.19 mmHg in Lox^+/+ and Lox^+/Mut animals, respectively; P = 0.87) blood pressure, although we cannot exclude the possibility of a small difference (Fig. 2B). Circumferential vessel wall stiffness, extrapolated from pressure/diameter measurements (Fig. 2 C and D), showed a stiffer carotid and ascending aorta at high pressures compared with in Lox^+/+ animals but normal circumferential stiffness at lower pressures. These findings indicate that the Lox^+/Mut animals have altered vessel wall material properties but normal vessel wall mechanics at physiologic pressures.

Fig. S2.

Sequence verification of mouse genotypes. Sequence chromatograms are shown for mice that are (A) Lox^+/+, (B) Lox^+/Mut, and (C) Lox^Mut/Mut.

Fig. 2.

Lox^+/Mut animals have altered aortic wall dimensions and normal blood pressure. (A) The length of the ascending aorta from the heart to the brachiocephalic artery is significantly longer in 3-mo-old Lox^+/Mut animals compared with Lox^+/+ littermate controls. ****P = 0.0005. (B) Blood pressure measurements using an arterial catheter found no difference in systolic (P = 0.15) or diastolic (P = 0.87) blood pressure between Lox^+/+ and Lox^+/Mut mice. n.s., not significant. Pressure–diameter responses collected for (C) the left common carotid artery and (D) ascending aorta show that both vessels in Lox^+/Mut animals are slightly stiffer than controls at high pressure but are not different in the physiological pressure range. Data are from n = 9 Lox^+/Mut and n = 7 Lox^+/+ animals. *P = 0.02; **P = 0.001; ***P = 0.0002.

Ultrastructural analysis of the unloaded Lox^+/Mut aorta showed a thicker arterial wall with an appropriate number (7 to 8) of elastic lamellae and smooth muscle cell layers (Fig. 3A). Although there were regions of the wall that were morphologically normal, the majority of areas in the aortic tissue of Lox^+/Mut mice had discontinuous elastic lamellae compared with the normal lamellae seen in Lox^+/+ mice (Fig. 3A). The abnormal lamellae observed in the aortic walls of Lox^+/Mut mice were similar to those in Lox^+/− mice. Autofluorescence of elastin in aortic tissue (Fig. 3B) showed that these breaks were present at significantly higher density throughout the aorta of Lox^+/Mut mice compared with Lox^+/+ littermate controls (29.9 vs. 11.9 breaks per 1 mm, respectively; P = 0.0006).

Fig. 3.

Lox^+/Mut animals have altered ultrastructural properties of the aortic wall. (A) EM of the ascending aorta from Lox^+/+, Lox^+/Mut, and Lox^+/− mice. The aortic walls of Lox^+/+ animals showed smooth and continuous elastic lamellae (white arrowhead), whereas aortas from both Lox^+/Mut, and Lox^+/− mice were found to be thicker with fragmented (red arrowheads) and disorganized elastic lamellae. (Scale bar: 10 µm.) *Aortic lumen. (B) Autofluorescence of elastin in aortic tissue showed that this was not an isolated finding and found that Lox^+/Mut animals had a significantly higher density of elastic lamellae breaks (red arrowheads) compared with Lox^+/+ mice (P = 0.0006). (Scale bar: 20 µm.) *Aortic lumen.

Mice Homozygous for the Lox Missense Variant Die Shortly After Birth Because of Ruptured Aortic Aneurysms.

Mice homozygous for the mutation (Lox^Mut/Mut) were born alive but did not survive more than a few hours. These animals were similar in size to their Lox^+/+ and Lox^+/Mut littermates (Fig. 4A), although cranial, thoracic, and abdominal hemorrhages associated with internal bleeding were frequently observed (Fig. 4B). Some animals had severe kyphosis (Fig. 4A) and ruptured diaphragms. All Lox^Mut/Mut animals had highly tortuous vessels with aneurysms in the ascending aorta and/or aortic arch as well as frequent aneurysms in the descending abdominal aorta near the renal artery branches (Fig. 4C).

Fig. 4.

Lox^Mut/Mut exhibit normal size, kyphosis, hemorrhages, and arterial tortuosity, and they die perinatally of aortic aneurysm/dissection. (A) Lox^Mut/Mut animals were born in the expected Mendelian numbers but died within a few hours of birth. Body size of the mutant animals was comparable with WT and heterozygous littermates. Some Lox^Mut/Mut animals had severe kyphosis (arrowhead in A), and cranial, thoracic, and abdominal hemorrhages (arrows in B) were common. Arterial architecture visualized by yellow latex injection into the left ventricle showed that Lox^Mut/Mut animals have highly tortuous vessels together with ascending and abdominal aortic aneurysms (arrows in C). Vessel tortuosity or aneurysms were not observed in Lox^+/+ or Lox^+/Mut littermates. Blood clots around blood vessels (arrowheads in C) indicated that aneurysmal rupture was a frequent occurrence in Lox^Mut/Mut mice. (Scale bar: 1 mm.)

Lox Missense Mutation Does Not Decrease mRNA Expression or Protein Synthesis.

Gene array analysis showed that all Lox family members (Lox, Loxl1, Loxl2, Loxl3, and Loxl4) are expressed in the developing aorta, with Lox having the highest expression level at every developmental time point (Fig. S3). Loxl1 is the second most highly expressed member, with an expression pattern similar to Lox but at levels two- to fourfold less. Both Lox and Loxl1 reach their highest expression levels during the late fetal period followed by relatively constant expression until decreasing rapidly around postnatal day 30 (P30). Expression of Loxl3, in contrast, is highest postnatally (P0–P21). Loxl2 and Loxl4 have low but detectable levels that remain relatively unchanged from the embryonic period through adulthood.

Fig. S3.

Lox is the predominant Lox family member expressed in developing mouse aorta. Data from our previous microarray study (44) were used to ascertain Lox family member expression levels in developing mouse aorta. The data showed that Lox was the LOX gene family member with highest expression levels at all developmental time points. E, embryonic day; P, postnatal day.

To determine if the Lox mutation alters expression of the mutant gene, quantitative real-time PCR was performed on mRNA from aortic tissue of newborn animals (P0). We found that Lox and other Lox isoforms were expressed in aortic tissue from Lox^+/Mut and Lox^Mut/Mut animals at the same levels as littermate controls (Fig. 5A). In addition, the mutation did not block pro-Lox protein synthesis (Fig. 5A, Inset).

Fig. 5.

Mutant Lox is expressed and synthesized but lacks enzymatic activity. (A) Quantitative RT-PCR analysis of mRNA from the aortas of Lox^+/+, Lox^+/Mut, and Lox^Mut/Mut P0 animals showed normal expression levels of all Lox family members in all three genotypes (note that the Lox primer/probe recognizes both Lox⁺ and Lox^Mut alleles). Two-way ANOVA P = 0.63 for Lox^+/+ vs. Lox^+/Mut; P = 0.99 for Lox^+/+ vs. Lox^Mut/Mut; and P = 0.63 for Lox^+/Mut vs. Lox^Mut/Mut. (Inset) Lox protein immunoblotting confirmed that the mutant Lox protein is produced by Lox^Mut/Mut cells at levels similar to the WT (Lox^+/+) protein. The absence of a band in extracts of Lox^−/− cells confirms the specificity of the Lox antibody. (B) Lox activity in the presence and absence of β-aminopropionitrile (BAPN) was assayed in conditioned culture medium from Lox^+/+ and Lox^Mut/Mut MEFs at 30-min time intervals. Medium incubated without cells served as the baseline control. MEFs cultured from two Lox^+/+ and three Lox^Mut/Mut embryos were tested in duplicates for n = 4 and n = 6, respectively, and plotted according to genotype at each time point (mean ± SEM). Lox activity, which is the difference between activity with and without BAPN, is expressed in relative fluorescent units (RFUs). ****P < 0.0001.

Lox Missense Mutation Decreases Enzymatic Activity.

Given the evidence that the Lox mutation did not decrease gene expression or protein synthesis, we next investigated the effect of the mutation on protein function. To determine the enzymatic activity level of the mutant Lox protein, we cultured primary mouse embryonic fibroblasts (MEFs) from Lox^+/+ and Lox^Mut/Mut animals. Lox secreted into the culture medium was collected, and enzyme activity was measured by the production of fluorescent resorufin using the Amplex Red assay at 0, 30, 60, 90, 120, and 150 min. Lox activity in MEFs cultured from Lox^+/+ animals was significantly higher than that from Lox^Mut/Mut animals beginning at 60 min (Fig. 5B). There was no significant difference in Lox activity between conditioned media from Lox^Mut/Mut MEFs and cell-free media samples.

Discussion

Through whole-genome sequencing in two affected individuals from a family exhibiting autosomal dominant TAAD, we identified a missense mutation in the copper binding region of LOX as the most likely causal variant. To further evaluate the significance and potential impact of this variant, we used genome engineering techniques to insert the human mutation into the orthologous position in the mouse genome. Mice bred to homozygosity for the human mutation recapitulated the human phenotype, supporting the likely causal role of this mutation in the human disease. This conclusion is further bolstered by the recent identification of additional FTAAD probands harboring LOX mutations (16).

Several conclusions emerge from our results. First, we report another successful use of genome-scale sequencing in mapping a causal gene underlying human disease. With the advent of next generation DNA sequencing technology (17), the cost and time of genome-scale sequencing have both decreased substantially. In small kindreds, such as the one presented in this study—in which the traditional mapping tools of linkage and directed sequencing are unlikely to provide a definitive result—genome-scale (either whole-exome or whole-genome) sequencing provides a rapid and cost-effective alternative means of identifying the causal variant and gene (18).

Second, although the historically accepted criteria for considering a novel gene causal in Mendelian disease include identifying additional kindreds with the same phenotype harboring independent variants in the same gene (19), these standards are likely to be increasingly difficult to meet in the future. After sequencing 410 unrelated FTAAD probands, Guo et al. (16) identified five possibly causal mutations in LOX, suggesting that this gene is probably responsible for ≤1% of FTAAD. As gene mapping studies continue to identify novel genes underlying FTAAD and other Mendelian diseases, new gene discoveries will invariably represent increasingly smaller proportions of the inherited basis of disease. Thus, new techniques and criteria for proving causality in Mendelian disease should be considered. Genome engineering techniques in animal models to study the in vivo effects of human alleles represent one path forward in the absence of identifying additional humans with disease.

Third, the generation of an animal model specific to this genetic defect provides insight into the potential mechanism underlying the disease. LOX is the major ECM cross-linking enzyme in blood vessels, and loss of Lox activity through either enzyme inhibition or gene inactivation leads to vascular dilatation and rupture (12, 13). Although mutations in LOX had not hitherto been associated with human disease, alterations in LOX expression levels and functional activity have been observed in diverse genetic disorders, with some having vascular involvement. For example, reduced LOX activity has been reported in two X-linked recessively inherited disorders, Menkes disease and occipital horn syndrome (20), which are caused by mutations in the ATP7A gene encoding a copper-transporting ATPase. Moreover, mottled blotchy mice, which carry mutations in Atp7a, also have high incidence of aortic aneurysms and exhibit disrupted elastic fibers (21). Functional inhibition of Lox by the toxin β-aminoproprionitrile, which is present at high levels in peas and lentils, has been shown to cause osteolathyrism (22), a connective tissue disease characterized by skeletal abnormalities and aortic dissections. The vascular phenotype of the mutant mice, including fragmented elastic lamellae in Lox^+/Mut animals and tortuous and aneurysmal vessels in animals homozygous for the mutation, resembles vascular changes seen in Lox KO animals (12, 13), implying that the arginine for methionine substitution leads to loss of Lox function. In fact, we identified similar abnormalities of the aortic wall architecture in both Lox^+/Mut and Lox^+/− mice, suggesting that the p.M298R missense change results in a functionally null enzyme. This conclusion was further corroborated by the observation that fibroblasts derived from Lox^Mut/Mut animals did not secrete Lox with enzymatic activity that was detectable above baseline. While this result is in contrast to a recent study suggesting that FTAAD was associated with missense mutations in LOX which only partially decreased protein function (16), methodological differences may account for this discrepancy. For example, our assay tested the function of Lox that was secreted from cells that were only producing mutant protein, whereas the other study tested the function of LOX present in the cell lysates of cultured human cells (which express normal LOX) that were overexpressing the mutant protein.

There are several mechanisms by which the missense change that we studied may result in loss of function. The introduction of an arginine residue within the copper binding domain may reduce the ability of the enzyme to bind copper, leading to loss of catalytic activity. Alternatively, it might disrupt normal LOX protein processing, or it could produce a dominant negative effect from decreased substrate binding. It is also possible that the mutant LOX protein is unable to interact with fibulin-4 (23) or components of the TGF-β signaling pathway (24)—both of which are known functions of LOX—thereby leading to aortic disease. It should be noted that, although Loxl1 is expressed at appreciably high levels in the aorta, it cannot compensate for loss of Lox activity and that mice lacking Loxl1 do not have a vascular phenotype (25). Future studies will be needed to clarify the exact mechanism by which this mutation leads to loss of Lox function.

Similar to other genetically engineered mouse models of human aortic disease (26, 27), we did not observe any aneurysm or other arterial disease in Lox^+/Mut mice. There are many reasons why the severity of phenotypes observed in murine models might differ from that seen in humans (28), and for reasons that are not completely understood, mutations causing human disease in an autosomal dominant manner often reproduce disease in mouse models only when present in a homozygous state (29). Despite the lack of overt disease in the heterozygous mouse, however, the increased ascending aortic length and fragmented elastic fibers in Lox^+/Mut mice suggest that mutant animals may be predisposed to vascular diseases because of weakened vessel walls under stress conditions. This reasoning may explain why some individuals in the family (for example, individual II-3) do not appear to develop aneurysm or dissection and are only affected with arterial tortuosity (Fig. S4). Additional studies in which heterozygous mice are subjected to hemodynamic stress conditions may provide additional insight into this question.

Fig. S4.

Representative images taken from the 3D reconstructions magnetic resonance angiograms. (A) Individual II-3 ,who inherited the LOX p.M298R mutation, was found to have severely tortuous aortic branch vessels, including the common carotid arteries (arrow). (B) Individual II-2, who did not inherit the LOX p.M298R mutation, was found to have normal carotid arteries without tortuosity (arrow).

Finally, our genome engineering approach has created a framework in which therapeutic hypotheses relevant to this genetic form of disease may be directly tested. For example, the observations that LOX activity is directly correlated with dietary copper and increases even when dietary copper supplementation is above what is needed for normal growth and development (30, 31) promote the hypothesis that copper supplementation in humans with TAAD caused by LOX mutations might augment the LOX enzymatic function of the normal allele and prevent vascular disease (although this approach is complicated by the fact that supplementation would need to be initiated early in embryonic development before a genetic diagnosis could easily be made). Regardless, hypotheses such as this are now directly testable using our mouse model of the human mutation. More generally, beyond individuals with TAAD caused by LOX mutations, there is evidence that LOX may also play a role in modifying other forms of TAAD (32–34), which suggests that therapeutic manipulation of LOX activity may prove beneficial in other inherited aortopathies.

In summary, the discovery of LOX mutations underlying TAAD in this family and others (16) suggests that this gene plays an important role in disease. Sequencing this gene may be useful in identifying the genetic basis for additional FTAAD probands and families. Future mouse model system studies of Lox mutations may provide additional mechanistic insights into disease pathogenesis and potential therapeutic options specific to this genetic cause.

Methods

Family Recruitment and Genome Sequencing.

The family examined in this study was referred to our clinic because of the two-generational history of aortic dissection. Eight members of the family were recruited for the study (Fig. 1A and Table S1). The Partners Healthcare Institutional Review Board approved the human research portion of the study, and informed consent was obtained from all study participants. Individuals I-2, II-1, II-2, II-3, III-1, and III-2 (Fig. 1A and Table S1) were examined in our clinic. Individuals III-3 and III-4 were evaluated at another hospital, and their records were subsequently reviewed. Whole-genome sequencing was performed by the Illumina Clinical Services Laboratory (Illumina, Inc.) as described in SI Methods.

Generation of Animals Harboring the Lox Mutation.

CRISPR/Cas9 genome editing technology was used in collaboration with the Washington University School of Medicine Hope Center Transgenic Vectors Core to generate mice harboring the Lox mutation identified in our family at the homologous murine position (c.857T > G encoding p.M292R) as detailed in SI Methods. All animal studies were performed according to protocols approved by the Animal Studies Committee of the Washington University School of Medicine.

Blood Pressure and Vessel Mechanical Property Measurements.

Blood pressure, ascending aortic length, and compliance of the ascending aorta and left common carotid arteries were measured in Lox^+/+ (n = 7) and Lox^+/Mut (n = 9) animals at 3 mo of age as described in SI Methods. The differences in ascending aortic length and blood pressure between Lox^+/+ and Lox^+/Mut animals were assessed using a t test. The difference in compliance between Lox^+/+ and Lox^+/Mut animals was tested using a two-way ANOVA.

Internal Aortic Latex Casting.

Because animals homozygous for the missense mutation (Lox^Mut/Mut) die shortly after birth, newborn (P0) pups from Lox^+/Mut × Lox^+/Mut breeders were used to visualize the vascular architecture of Lox^+/+ (n = 7), Lox^+/Mut (n = 24), and Lox^Mut/Mut (n = 5) animals. After euthanasia, PBS was flushed through the left ventricle of the heart followed by injection of 200 μL yellow latex diluted in deionized water in a 1:1 ratio. The latex was allowed to polymerize at 4 °C for 3 h before fixing the entire animals in 10% (vol/vol) buffered formalin overnight at 4 °C. The formalin was replaced with 70% (vol/vol) ethanol 2 h before dissection.

Quantification of Lox mRNA Levels.

Aorta and lung tissue were collected from Lox^+/+ (n = 3), Lox^+/Mut (n = 3), and Lox^Mut/Mut (n = 4) P0 animals and stored in RNAlater Solution (ThermoFisher Scientific) at −20 °C. Total RNA was collected using TRIzol (ThermoFisher Scientific) following the manufacturer’s protocol, and real-time PCR was performed using techniques described in SI Methods.

MEF Culturing, Immunoblotting, and Lox Activity Assay.

Primary MEF cultures were established from embryonic day 14.5 embryos harvested from Lox^+/Mut × Lox^+/Mut breeders as previously described (35) and maintained as described in SI Methods. Lox protein immunoblotting and enzymatic activity were performed as described in SI Methods.

Ultrastructural Analysis and Elastin Fluorescence Imaging of the Aortic Wall.

For EM, aortas from 3-mo-old Lox^+/+ (n = 3), Lox^+/Mut (n = 3), and Lox^−/− (n = 3) animals were prepared and imaged as described in SI Methods. To visualize elastin lamellae in the aortic wall, the ascending aortas from 3-mo-old Lox^+/Mut (n = 3) mice and their Lox^+/+ (n = 3) WT littermate controls were prepared and imaged as described in SI Methods. The difference in the number of breaks per millimeter between Lox^+/+ and Lox^+/Mut animals was assessed using a t test.

SI Methods

Whole-Genome Sequencing.

Briefly, genomic DNA was randomly fragmented and then, sequenced using 100-bp paired end reads on an Illumina HiSeq 2000 Sequencer to an average depth of 30× across the genome. The resulting output was converted to FASTQ format. The paired end FASTQ files were aligned to the human reference sequence (UCSC HG19 build) using the Burroughs–Wheeler Alignment tool (36) in paired end mode followed by base quality recalibration and targeted local realignment focused around known short insertions and deletions (indels) using the Genome Analysis Toolkit (GATK) (37). Duplicated reads from sequencing the same DNA fragment were discarded. Single-nucleotide substitutions and indels were identified for both samples simultaneously using the UnifiedGenotyper tool from the GATK in multisample calling mode. Variant quality score recalibration was performed using the GATK to identify a set of high-confidence variants. The functional consequence of the resulting set of variants was predicted using the Variant Effect Predictor from Ensemble (38).

Generation of Animals Harboring the Lox Mutation.

To design the CRISPR/Cas9 construct, Lox single guide RNA (sgRNA) (5′-cattagacattaccacagca-3′) was cloned into BbsI-digested plasmid pX330 (42230; Addgene). sgRNA activity was validated in vitro by transfection of NIH 3T3 cells using Roche Xtremegene HP followed by T7E1 assay (New England Biolabs). T7 sgRNA template was PCR amplified, gel purified, and in vitro transcribed with the MEGAshortscript T7 Kit (Life Technologies). T7 Cas9 template was PCR amplified, gel purified, and in vitro transcribed with the T7 mMessage mMachine Ultra Kit (Life Technologies). After transcription, RNA was purified with the Megaclear Kit (Life Technologies). A 200-nt single-stranded oligodeoxynucleotide (ssODN) donor DNA with the mutation centered within the oligonucleotide was ordered from Integrated DNA Technologies as an ultramer oligo. Founders were identified using a Qiagen pyrosequencer and Pyromark Q96 2.5.7 software. B6CBA F1/J female mice (3–4 wk old; Jackson Laboratory) were superovulated and mated overnight with B6CBA F1/J male mice (>7 wk old). Zygotes were harvested from the ampullae of superovulated females and placed in potassium-supplemented simplex optimized medium (KSOM; MR-106D) before microinjection. Microinjection was performed in flushing holding medium (FHM; MR-024-D; EmbryoMax; Millipore). The Cas9, sgRNA, and ssDNA template were coinjected into the pronucleus of 204 zygotes. The final concentration of the mixture was 50 ng/µL Cas9 WT RNA, 25 ng/µL each sgRNA, and 20 ng/µL ssODN DNA. After injection, zygotes were incubated at 5.5% CO₂ at 37 °C for 2 h, and surviving embryos were transferred to ICR recipient mice by oviduct transfer. Subsequently, we identified four founder animals that were mosaic for the c.875T > G (p.Met292Arg) missense mutation. Additional breeding of the founders gave rise to animals heterozygous for the missense mutation. Using PCR amplification and Sanger sequencing of 600 bp genomic DNA surrounding the missense mutation, we screened a total of 178 offspring to identify five animals that were heterozygous for the desired mutation without the presence of any other genomic alterations. These five animals were bred for all of the studies presented here. The missense mutation was genotyped using Custom TaqMan SNP Genotyping Assays from ThermoFisher Scientific. The P0 and 3-mo-old animals used for this study were crossed with C57BL6/J (Jackson Laboratory) for one generation.

Blood Pressure and Vessel Mechanical Property Measurements.

Lox^+/+ (n = 7) and Lox^+/Mut (n = 9) animals at 3 mo of age were anesthetized with 1.5% inhaled isofluorane and placed on a 37 °C heated pad to maintain body temperature. A Millar pressure transducer was placed into the right common carotid artery and advanced to the ascending aorta, where systolic and diastolic blood pressures were recorded and analyzed as previously described (39). After blood pressure measurements, ascending aortic length was measured from the aortic root to the first branch point at the brachiocephalic artery using an electronic caliper. The ascending aorta and left common carotid arteries were then excised, and pressure–diameter curves were generated using a pressure arteriograph that measures the vessel’s outer diameter while gradually increasing the intravascular pressure by 25-mmHg increments from 0 to 125 mmHg. Using the SD of blood pressure observed in Lox^+/+ animals, we estimated (40) that our sample size provided 80% power (at P = 0.05) to detect a 15-mmHg difference in systolic blood pressure.

Ultrastructural Analysis and Elastin Fluorescence Imaging of the Aortic Wall.

For EM, aortas from 3-mo-old Lox^+/+ (n = 3), Lox^+/Mut (n = 3), and Lox^−/− (n = 3) animals were perfused with PBS through the left ventricle of the heart. The ascending aorta was excised from the aortic root to the brachiocephalic artery and fixed in 2.5% (vol/vol) glutaraldehyde and 0.1 M sodium cacodylate solution at 4 °C overnight. The fixed samples were then processed, and thin sections were prepared by the Washington University’s Electron Microscopy Core Facility and imaged using a Joel 1400 Electron Microscope.

The ascending aortas from 3-mo-old Lox^+/Mut (n = 3) mice and their Lox^+/+ (n = 3) WT littermate controls were excised, flushed with 1× PBS, and immediately frozen at −80 °C in optimal cutting temperature (OCT) compound (Fisher Scientific). The frozen OCT blocks were then cut into 3-µm sections and fixed in 4% (vol/vol) formaldehyde at 4 °C for 10 min. The sections were treated with 0.5% Chicago Sky Blue 6B (Sigma-Aldrich) and PBS solution for 5 min at room temperature and washed three times in PBS. Tissues were mounted with DAPI Fluoromount-G (Southern Biotech) and imaged using a Zeiss Axioskop Fluorescence Microscope and QCapture Pro software (Media Cybernetics, Inc.). The difference in the number of breaks per millimeter between Lox^+/+ and Lox^+/Mut animals was tested using a t test.

Quantification of Lox mRNA Levels.

One microgram RNA from aorta and lung tissue was reverse-transcribed to cDNA using a High-Capacity RNA-to-cDNA Kit. One microliter cDNA was used for real-time PCR using Taqman Fast Universal PCR Master Mix and Taqman assays primer/probes (Life Technologies) for Lox (Mm00495386_m1), Loxl1 (Mm01145738_m1), Loxl2 (Mm00804740_m1), Loxl3 (Mm01184865_m1), and Loxl4 (Mm00446385_m1); 10-μL reactions were performed in duplicate using a ViiA 7 Real-Time PCR System. Lox levels were normalized to Gapdh. Differences in expression levels between Lox^+/+, Lox^+/Mut, and Lox^Mut/Mut P0 animals were tested using two-way ANOVA.

MEF Culturing, Immunoblotting, and Lox Activity Assay.

MEFs collected from each embryo were cultured in a 10-cm dish using DMEM supplemented with 10% FBS, 1% penicillin/streptomycin, 1% l-glutamine, and 1% nonessential amino acids. Five days after reaching visual confluency, the cell layers were extracted using buffered 8 M urea and prepared for immunoblotting as described (41). Protein concentration in each sample was assessed using Bio-Rad Reagent (Bio-Rad), and 25 µg MEF extract were separated by SDS 10% polyacrylamide gel and transferred to ProBlott Membrane (Applied Biosystems). The blot was incubated in 5% milk overnight at 4 °C to block nonspecific binding sites followed by incubation with rabbit PP-Lox antibody targeting the propeptide portion of uncleaved pro-Lox (provided by Phillip Trackman, Boston University, Boston, MA) at 1:4,000 dilution for 1 h at room temperature (42). After washing, ECL Anti-Rabbit IgG HRP-Linked Secondary Antibody (GE Healthcare) was added at 1:4,000 for 1 h at room temperature. The blots were then washed, and the immunoreactive bands were detected using Immobilon Western Chemilluminescencent HRP Substrate (Millipore). The blot was then stripped of the PP-Lox antibody for immunodetection of β-actin (1:20,000; Sigma-Aldrich).

MEFs were derived as described above. After 5 d in culture, MEFs were passaged into a 15-cm dish and maintained as above until fully confluent. Twenty-four hours before collection, culture medium was replaced with DMEM lacking phenol red (ThermoFisher Scientific) supplemented with 50 μg/mL ascorbic acid and 0.1% BSA. Conditioned medium from each dish was collected and concentrated using Amicon Ultra-15 Centrifugal Filter Units (Millipore) to a final volume of 1 mL. Lox activity was assessed in 50 μL concentrated culture medium and incubated with and without the Lox inhibitor β-aminoproprionitrile (BAPN) using the Amplex Red assay as previously described (43). Resorufin fluorescence, the product of Amplex Red oxidation, was measured at excitation and emission wavelengths (540 and 600 nm, respectively) every 30 min for 150 min using a Biotek H4 Hybrid Reader. Lox activity was calculated as the difference between total activity and activity in the presence of BAPN. Differences in relative fluorescent units were tested using two-way ANOVA.