Emerging mechanisms of elastin transcriptional regulation

Abstract

Elastin provides recoil to tissues that stretch such as the lung, blood vessels, and skin. It is deposited in a brief window starting in the prenatal period and extending to adolescence in vertebrates, and then slowly turns over. Elastin insufficiency is seen in conditions such as Williams–Beuren syndrome and elastin-related supravalvar aortic stenosis, which are associated with a range of vascular and connective tissue manifestations. Regulation of the elastin (ELN) gene occurs at multiple levels including promoter activation/inhibition, mRNA stability, interaction with microRNAs, and alternative splicing. However, these mechanisms are incompletely understood. Better understanding of the processes controlling ELN gene expression may improve medicine’s ability to intervene in these rare conditions, as well as to replace age-associated losses by re-initiating elastin production. This review describes what is known about the ELN gene promoter structure, transcriptional regulation by cytokines and transcription factors, and posttranscriptional regulation via mRNA stability and micro-RNA and highlights new approaches that may influence regenerative medicine.

INTRODUCTION

Elastic fibers are macromolecular components of the extracellular matrix (ECM) that allow tissues such as blood vessels, lungs, and skin to stretch without being damaged and reduce mechanical work following stretch by returning to an entropically favored state during relaxation (1). Elastin, encoded by the ELN gene, makes up the amorphous core of these fibers, providing them with extensibility and recoil capability. Abnormalities of elastin quality or quantity occur in a range of health conditions, including rare diseases like elastin-related supravalvar aortic stenosis (SVAS), Williams–Beuren syndrome (Williams syndrome; WS), and elastin-related autosomal dominant cutis laxa (ADCL), named according to the dyadic principle (2). Understanding of the mechanisms of ELN transcript regulation, therefore, may suggest targets to reactivate elastin production and mitigate disease.

OVERVIEW OF ELASTIN

The ELN gene is located on chromosome 7 in humans. It consists of 34 in frame exons (Fig. 1) (for review, see Ref. 7) that permit alternative splicing of the ELN pre-mRNA without alteration of the open reading frame. This combination results in the generation of multiple mature mRNA isoforms (Fig. 1) (5). The canonical transcript lacks exon 22 and encodes an ∼70 kDa protein of repeating cross linking and hydrophobic domains.

Figure 1.

Elastin expression. The human elastin (ELN) gene is located on chromosome 7 at 7q11.23. It consists of 34 exons, by convention numbered 1–33 and 36 as most species contain an additional pair of exons, exons 34 and 35, that were lost in higher primates (3). The gene is transcribed as a large (47.7 kDa) pre-mRNA (NM_001278939, ENST00000692049.1) with a high intron/exon ratio (4). The exons are structured as alternating hydrophobic and crosslinking domains and are symmetric (in-phase) facilitating alternate splicing of exons. In microRNA (mRNA) studies, exon 22 (marked with an *) is almost uniformly removed, likely due to simple read-through (5). Other commonly spliced exons are marked with ^. There are also alternate splice donor and acceptor that can extend or shorten exons 8, 20, 24, and 26 (marked with + or −). Splice isoforms are present at varying relative ratios in different tissues (5) although mRNA and proteomic studies are not completely concordant (6). There are more than 30 splice isoforms identified to date. The five most common protein coding splice isoforms in human aorta are presented. Short isoforms using alternative transcription start sites and including intronic sequence (marked with an “i”) have also been identified through RNA studies (light blue with dashed border), but experimental confirmation of their presence and abundance is necessary. After splicing, mature tropoelastin coding isoforms are either translated into tropoelastin or degraded. mRNA degradation is an important pathway for regulation of tropoelastin expression. Tropoelastin monomers are able to undergo some self-assembly but the precise mechanisms underlying tropoelastin secretion and assembly into mature elastin fibers is unresolved, although crosslinking and assembly proteins such as lysyl oxidase (LOX) and FBLN4 and/or FBLN5 are involved. The mature elastic fiber consists of tropoelastin at the core within a microfibril (mainly consisting of FBN1 and FBN2) scaffold. While the ratio of elastin to microfibril changes through development; the mature elastin fiber contains approximately 90/10 elastin:microfibril content. This conformation allows for elastic fibers to stretch and relax without mechanical deformation. Created with BioRender.com.

Translation of the mature mRNA isoforms results in the production of tropoelastin (TE) monomers. TE is secreted into the ECM through a specialized acidic pathway (8). TE then multimerizes and is deposited into the microfibril-rich ECM as mature elastin (9, 10). As part of this process, the lysines in the cross-linking domains are cross linked with the help of the enzyme lysyl oxidase (LOX) resulting in a highly hydrophobic and insoluble protein.

The hydrophobic domains contribute to recoil capability as stretch exposes the hydrophobic domains to the hydrophilic environment and entropy drives them back (1, 11–13). The insoluble elastic fibers are highly stable, with a half-life of 74 years (14). Over time they can undergo glycosylation with advanced glycosylation end products (AGEs) that alter the recoil capability of the molecule making it brittle and increasing the rate of turnover (15–17).

ELASTIN IN DISEASE

Elastin is required for normal function of multiple tissues, with the most obvious effects seen in tissues that undergo recurrent deformations, such as the blood vessels, skin, and lungs. However, other tissues such as ligaments and elastic cartilages can also be impacted. Elastin-mediated disease can come about due to changes in either the quality or quantity of elastic fibers. Rare variation within ELN itself can cause ELN-related SVAS and ELN-related ADCL. SVAS is generally thought to result from reduced quantity of elastin while ADCL is thought to reflect the incorporation of abnormal elastic fibers into the ECM; however, this may be an over-simplification (18). In addition, common variation in ELN has been identified through multiple genome wide association studies (GWAS) to contribute to abnormalities of blood vessels (19–21) and other connective tissue functions (22–24) on a population basis. Given the long half-life of elastic fibers, environmental exposures such as UV radiation, cigarette smoke, and high glucose environments related to diabetes impact turnover rate in dramatic ways (25, 26).

Elastin Haploinsufficiency (ELN-Related SVAS and Williams–Beuren Syndrome)

ELN-related SVAS typically comes about due to nonsense or frameshift variants, generally within the front half of the molecule, that are expected to produce haploinsufficiency for elastin (27–31). Splice variants have also been described with less consistent genotype-phenotype implications (32–34). More prevalent than ELN-related SVAS, though, is Williams syndrome (OMIM No. 194050), which is caused by the contiguous deletion of 25 to 27 coding genes on chromosome 7, including ELN. In one epidemiological study, it was reported in 1:7,500 individuals and causes haploinsufficiency for elastin as well as for the other deleted genes (35).

People with either ELN-related SVAS or WS exhibit a characteristic vascular disease called supravalvar aortic stenosis (36, 37) (for review, see Ref. 38). SVAS manifests as an area of focal narrowing in the aorta, just above the aortic valve. It is reportedly more severe in affected males (39, 40), although the reasons for this are unknown. Stenoses are also possible in other large elastic vessels such as the descending aorta, pulmonary arteries, and renal arteries (41). Long segment narrowing, where the entire vascular segment exhibits diminutive caliber rather than being focally stenotic, is also observed in a sizeable subset (42). Sudden death is present at a rate of 25 to 100 times higher than the general population (43–45) with increased rates of complications and death associated with anesthesia—features that have been attributed secondarily to the vascular disease. In addition to stenosis, individuals with these conditions often have attendant arterial stiffness and hypertension (46–49).

More recent data have suggested that individuals with these conditions may also exhibit abnormalities of other connective tissues. Ligamentous laxity (50) is seen as well as early wrinkling and facial drooping of the skin (51). More recently, lung phenotypes have been described in these individuals and in mouse models that suggest air trapping and lung obstruction that increases with age (52–57). Because WS impacts 24 to 26 genes beyond ELN, individuals with this condition manifest additional features attributable to those genes such as developmental delays, a characteristic hyper-social personality, and hypercalcemia (for review, see Ref. 58).

Mouse models of elastin insufficiency contain a deletion of exon 1 of Eln and 4 kb of the Eln promoter (59, 60). Those mice exhibit overlapping vascular features with humans with SVAS/WS, including narrow caliber vessels, arterial stiffness, and hypertension. Hemizygotes do not, however, exhibit the focal SVAS commonly seen in humans. A newer conditional deletion model, the Tagln-Cre; Eln^f/f, in which Eln is deleted from both alleles in smooth muscle cells and displays a phenotype that is more similar to human SVAS, although the exact position of the stenotic lesion in the ascending aorta differs (61, 62).

Autosomal Dominant Cutis Laxa

Cutis laxa describes a heterogeneous class of disorders characterized by skin that is loose, redundant, and inelastic. It can be broadly divided into autosomal recessive or autosomal dominant forms although others, including acquired and X-linked CL, have also been characterized. Most of the genes whose variation imparts the cutis laxa phenotype impact components of the elastic fiber (ELN, EFEMP2, FBLN5, LTBP1, and LTBP4) or decrease the production of mechanically competent elastic fibers by modifying relevant amino acid concentrations, glycosylated side chain addition, or efficient transit through the specialized acidic secretory pathway (ATP6V1A, ALDH18A1, ATP6V0A2, ATP6V1E1, and PYCR1) (63).

The multiple forms of cutis laxa cannot be easily distinguished from one another based on skin presentation, though each carries a varying set of systemic manifestations (63). At the microscopic level, all forms are characterized by a decrease in the number of elastic fibers and fragmentation of those existing fibers (64). On an ultrastructural level, biopsies from patients with ADCL show bare microfibrils with irregular clumps of elastin and can display vacuolization, possibly due to increased fragmentation and/or turnover of the elastic fiber (64).

Unlike other ELN-related disorders, ADCL is thought to be caused by production of poor-quality elastin. The elastic fibers of patients with ADCL have reduced elastin content and are abnormally branched and fragmented (64, 65). As opposed to SVAS and WS that are caused by variants that cause ELN haploinsufficiency, most cases of ADCL result from frameshift variants in the last four exons (human) of elastin. Stable mRNA expression has been detected in human skin fibroblasts for multiple variants (64) and disrupted protein has been detected in the matrix. These frame shift variants are largely expected to extend the C-terminus of TE with disruption of the C-terminal structure and function. Callewaert et al. (65) demonstrate that the elastin in ADCL has impaired deposition into microfibrils and some variants can lead to increased TGFβ signaling and endoplasmic reticulum (ER) stress. Variation impacting exon 32 has been reported to yield less severe outcomes, potentially due to alternative splicing of this exon (64)

In addition to the characteristic skin features of ADCL, affected individuals are also at increased risk of pulmonary and vascular manifestations. Risk for aortic dilation/dissection, for example is 35%, and may or may not be accompanied by bicuspid aortic valve (65). Of particular importance is emphysema, which occurs in 55% of affected individuals and can be mild or severe and contributes significant morbidity and mortality to the condition (65, 66).

Diseases of Elastin Protein Dysfunction

In addition to elastin insufficiency (SVAS) and abnormal elastin deposition (ADCL), elastin turnover and protein dysfunction are involved in a number of common disease processes. Changes in elastin are seen with chronic obstructive pulmonary disease (COPD) (67) and rare ELN variants have been implicated in severe, early onset COPD (68). Exposure to environmental agents such as UV radiation in normal aging (69) and exposure to environmental toxins such as cigarette smoke and air pollution (70) can also disrupt elastin function. Elastin is also susceptible to glycation and disruptions in calcium homeostasis—processes important in the development of cardiovascular disease associated with diabetes (17). Diabetes is also associated with accelerated vascular aging and vascular ECM remodeling, leading to increased levels of elastin degradation peptides (EDP) that may lead to feedback inhibition of elastin synthesis (25, 71). Although conditions such as these are primarily caused by dysfunctional protein and increased rates of turnover, it is possible that the development of therapeutics that increase or re-initiate transcription outside of the usual developmental period could also correct tissues with protein destruction/dysfunction as well.

MECHANISMS OF ELASTIN TRANSCRIPTIONAL REGULATION

Overview

Elastin production is tightly temporally and spatially regulated, with elastin protein deposition largely restricted to fetal and early postnatal life. Although transcriptional controls direct where and when elastin assembly is initiated, fine tuning of elastic fiber quantity and quality is most heavily influenced by posttranscriptional mechanisms. Consequently, control of elastic tissue design is complex, and the balance of these pathways must be considered when estimating the elastogenic potential of a tissue.

ELN Promoter

The ELN promoter (Fig. 2) is GC rich, lacks a conventional TATA box, and utilizes multiple transcription start sites (72). The high GC content and lack of strong TATA box are features of conventional “housekeeping” genes that are ubiquitously and continuously transcribed. Compared with other structural matrix proteins, the coding region of ELN has relatively low sequence homology among mammalian species (73). However, there is a high degree of sequence homology in the 5′ flanking sequence, with 94% homology in the “proximal” promoter from −1 to −192 and 86% homology in the “distal” promoter from −193 to −588, suggesting that this region has an important functional role (74). However, no large-scale genomic studies have been completed on this region to investigate its impact.

Figure 2.

Elastin (ELN, 2.2 kb) gene promoter structure. The human ELN gene lacks a conventional TATA box but contains at least two CAAT boxes and multiple transcription start sites. Several cytokines are known to affect ELN promoter activity. Cytokines that are generally ELN transcription are in green whereas cytokines that generally lead to decreased ELN transcription are in magenta. Arrows represent interactions that promote transcription factor binding to the promoter whereas inhibitory interactions are denoted by arrows with flat heads. Transcription factors that promote ELN transcription are purple, those that inhibit transcription are orange. There are multiple GC boxes that enable Sp1 binding clustered close to the initiation methionine. SP1 interacts with activated Rb to promote transcription in response to TGFb1 and insulin like growth factor-1 (IGF1). Sp1 binding to the promoter is disrupted by IL1b and PDGF. IL1b also promotes binding of C/EBPb to the proximal CAAT box, inhibiting transcription. There are also multiple AP1 binding sites. A proximal AP1 binding site inhibits ELN transcription through the binding of JUN homodimers in response to TNFa while a more distal AP1 site inhibits ELN expression in response to multiple cytokines. There are three glucocorticoid response elements (GRE) that promote transcription in the presence of glucocorticoids such as dexamethasone or hydrocortisone although this appears to be cell type and state specific. Evidence also suggests that insulin can promote ELN transcription through release of inhibitory FoxO1 binding. Created with BioRender.com.

Initial inquiries to identify the basic ELN promoter elements utilized fragments of the first 2.2 kb of the 5′ flanking sequence cloned into a chloramphenicol acetyltransferase (CAT) reporter plasmid and measured CAT activity as a surrogate for promotor activity (4, 72, 75) (summarized in Ref. 76). Using this technique, the basic promoter necessary for ELN transcription was defined as the first 128 bp 5′ to the initiation methionine. Multiple up and down regulatory elements were broadly identified using this technique with subsequent analysis further defining the sequences and transcription factors that modulate ELN promoter activity. The area beyond 2.2 kb upstream of ELN has yet to be studied.

Sequence analysis of the ELN 5′ flanking sequence yields multiple postulated transcription factor binding sequences including multiple Sp1, AP1, AP2, C/EBP binding sites, and TGFβ response elements that may be responsible for the positive and negative promoter activity identified by CAT reporter assay (4, 72, 75). However, modern sequencing and genomic techniques such as chromatin immunoprecipitation (ChIP) sequencing data available through the ENCODE Project (77) and predicted transcription factor binding sites available through the JASPAR CORE collection database (78) indicate a multitude of other transcription factors including WT1, transcription factors that interact with retinoblastoma, the multifunctional CTCF transcription factor, RAS responsive transcription factors, and members of histone modifying complexes may bind to the ELN promoter region.

TGFβ1 signaling is involved in numerous disease processes and is important in the regulation of the ECM. With regard to ELN, TGFβ1 can affect TE production through several mechanisms including regulation of transcriptional initiation, posttranscriptional mRNA processing and stability, and even mature elastin turnover. On a transcriptional level, TGFβ1 increases ELN mRNA transcription (79–83), potentially through the phosphatidylinostitol 3-kinase (PI3K)/AKT/p38 signaling pathway (84). However, the effect of TGFβ1 on transcription is cell type and cell state specific. For example, in the skin, exogenous administration of TGFβ1 in a transgenic mouse model leads to increased expression of a reporter construct (85), but in cultured human skin fibroblasts, a direct impact on transcription is no longer appreciated (80). Similarly, supplementation with TGFβ1 yields increased the levels of Eln mRNA and soluble TE in rat neonatal lung fibroblasts, but no change was observed in either mRNA or soluble TE in adult rat lung fibroblasts or smooth muscle cells (neonatal or adult) (81).

Insulin-like growth factor-I (IGF1) is also a positive regulator of Eln transcription in chicken and in rats (86–88). This effect also appears to be cell type specific, with no change in Eln mRNA levels in pulmonary fibroblasts despite an increase in mRNA levels in neonatal aortic smooth muscle cells (86, 89). IGF1 regulation of ELN mRNA levels is mediated via binding of Sp1 binding to the proximal Eln promoter in rats (88, 90). This effect is mediated by promotion of retinoblastoma (Rb) threonine-821 phosphorylation via a cyclinE-cyclin-dependent kinase 2 complex (91, 92).

Inhibition of ELN transcription is achieved through binding of the AP1 complex, composed of Fra1 and Jun heterodimers or Jun homodimers. There are two proposed AP1 binding sites, a site located more proximally at −229 to −223 and a more distal AP1 binding site at −564 to −558 (93). Binding of the AP1 complex is mediated by several cytokines and the proximal and distal site appear to act independently. Basic fibroblast growth factor (bFGF) decreases Eln mRNA levels in rat aortic smooth muscle cells (94, 95) and in rat pulmonary fibroblasts (96) via binding of the AP1 complex to the distal AP1 binding site. TNF-α similarly inhibits ELN mRNA via AP1 complex but acts via the more proximal AP1 binding site (93). Interestingly, stretch has been shown to activate Jun, but in rat pulmonary arterial adventitial fibroblasts, stretch increases Eln mRNA levels, an effect that may be attributed to increased activation of PKC having a stimulatory effect on Eln (97).

Glucocorticoids such as dexamethasone have long been used in cell culture to promote TE protein production (98). Del Monaco et al. (99) demonstrated the presence of three potential glucocorticoid response elements in the distal ELN promoter (at positions −1,018 to −1,023, −1,310 to −1,315, and −1,432 to −1,437) and demonstrated glucocorticoid receptor binding. They further demonstrated that those elements, when cloned into a CAT reporter construct and transfected into skin fibroblasts, increased CAT activity, thereby suggesting they are capable of initiating transcription. However, the effect of glucocorticoid on TE protein production is influenced by cell state (100) and that glucocorticoids also affect mRNA stability (101) and may involve transcriptional repression by microRNA (102).

Additional cytokines and signaling pathways have also been described, including insulin releasing FoxO1 mediated promoter inhibition and IL1β-mediated C/EBPβ binding to the proximal promoter (for review, see Ref. 103); more are likely yet to be discovered. These studies highlight the complexity of the elastogenic response depending on cell type and experimental model. The mechanisms that confer specificity to those responses have yet to be elucidated. Furthermore, although these studies demonstrate the effect of cytokines and promoter elements on ELN transcription, the “master switch” responsible for inducing ELN transcription in late gestation has yet to be determined. Studies of the ELN promoter to date have been limited to a region within 2.2 kb of the initiation methionine. There is a large (>150 kb) intragenic region between ELN and its closest 5′ genes (TMEM270, METTL27, and CLDN4) that has yet to be explored but may yield clues to the tissue and temporal specific regulation of ELN expression (104).

ELN mRNA Stability

Although a transcriptional switch is likely necessary to initiate TE production in elastogenic tissues, ELN pre-mRNA continues to be transcribed long after TE protein production ceases (105–108). This effect is seen across species (109). The decrease in protein production is secondary to a rapid decline in the half-life of mature Eln mRNA in the postadolescent period (110) (Fig. 1). The mechanisms that regulate mature ELN mRNA stability and the purpose of continuous transcription in the absence of TE production remain unclear.

In addition to its role in transcriptional regulation, TGFβ1 is also one of the best-studied factors influencing ELN mRNA stability. Its proelastogenic effect has been shown in multiple elastic cell types, including fibroblasts from lung (111) as well as vascular smooth muscle cells (79) and skin fibroblasts (80, 83). Data from these studies showed mature ELN mRNA levels increased and TE protein was produced even after the addition of a transcriptional inhibitor. The effect of TGFβ1 on ELN mRNA stability requires the protein kinase C, p38, and active SMADs (112).

In addition to TGFβ1, other cytokines have also been demonstrated to affect ELN mRNA stability. Glucocorticoids, for example, decrease ELN mRNA stability in some experiments (101). This may contribute to skin atrophy and loss of elastic fibers with prolonged topical glucocorticoid therapy (113). Transforming growth factor-α may have similar dual role in regulating Eln promoter activity and mRNA stability in rats (114).

The mechanisms by which ELN mRNA stability and turnover is regulated are incompletely understood. Zhang et al. (110) identified a sequence in exon 30 of Eln that interacts with a 50-kDa cytosolic protein in rats. Binding of the protein is associated with increased destruction of the mature TE mRNA. TGFβ₁ decreases binding of the 50-kDa protein to the TE mRNA, thereby lengthening its half-life. Likewise, Hew et al. (108) identified a sequence in the 3′ untranslated region (UTR) of chicken Eln that has also been demonstrated to bind both nuclear and cytosolic proteins and is associated with increased tropoelastin mRNA stability. However, neither the exon 30 nor the 3′ UTR binding proteins have been identified; subsequent studies transitioned away from protein-mediated (in)stability mechanisms and instead focused on the impact of miRNAs that bind close to the 3′ protein-binding sequence.

Role of microRNA

MicroRNAs (miRNAs) are short (average of 22 nucleotide), noncoding RNA molecules that have been implicated in the regulation of a vast number of biologic processes and diseases. Their most well-established function is regulation of gene expression through binding to the 3′ UTR of target mRNA, thereby leading to posttranscriptional repression and via the miRISC complex leading to cleavage of the target mRNA (115, 116). MicroRNAs have long been implicated in a number of cardiovascular diseases caused or impacted by changes in ECM/elastic fibers including arterial stiffness (117), abdominal aortic aneurysms (AAA) (118), and AAA dissection (119, 120). The 3′ UTR of the ELN gene contains multiple miRNA binding sites and available evidence suggests that miRNA play an active role in the modulation of ELN mRNA steady-state concentrations and TE protein production.

The effect of the miR-29 family of microRNA is most well established but other microRNAs have also been linked to elastin regulation (118, 119). The miR-29 family consists of three miRNAs expressed from two bicistronic clusters where miR-29a and miR-29b1 are co-expressed from one cluster and miR-29b2 and miR-29c are expressed from a second (121). miR-29 family members have a conserved seed sequence (or DNA target) and their predicted targets are enriched for ECM proteins, including elastin (122–124), COL1A1, COL3A1, and VEGF-A (117, 121).

Experimental studies have validated the role of miR-29 family members in the regulation of elastin mRNA stability and TE protein production. Transfection of dermal fibroblasts and or pulmonary artery smooth muscle cells (SMCs) with miR-29a, miR-29b, and miR-29c mimics leads to decreased mature elastin mRNA levels (125). Conversely, treatment with miR-29a antigomers increases elastin mRNA levels (miR-29b and miR29c antigomers were not studied). The authors found that dermal fibroblasts from two patients with WS treated with anti-miR29a had an increase in ELN mRNA and TE protein (125).

Although miRNA is an attractive therapeutic target, the utility of miRNA mimetics or anti-miRNA is limited by the inherent lack of specificity. miRNA target sequences are short, and a single miRNA species can target multiple genes. This may enable coordinated modulation of the cumulative elastogenic machinery during development but limits specificity when applied to elastin-specific regenerative medicine strategies. Moreover, while in vitro studies suggest that miRNAs may orchestrate large amplitude changes in the concentration of their target mRNAs, the effect of miRNA in vivo is thought to produce more incremental changes in gene expression and thus may not be expected to account for the near total loss of TE protein production in adult tissues. Likewise, inhibiting miRNA may not be adequate to restore ELN temporally specific expression in cases of elastin insufficiency.

ELN Splicing

Across species, the ELN gene has significant variation in nucleotide sequence but maintains a characteristic pattern where every exon contains a multiple of three nucleotides, with the exception of exon 1 which contains 3n + 1 nucleotides (126–128). This facilitates alternative splicing of exons whereby an entire exon can be removed from the ELN pre-mRNA without changing the open reading frame, leaving the amino acid sequence following the splice unchanged. In addition to simple exon excision or alternative splicing, ELN isoforms also utilize alternate donor or acceptor sites that lengthen or shorten exons (specifically exons 8, 20, 24, and 26) while also retaining the open reading frame.

To date, more than 30 human isoforms have been identified through RNA studies [GTEx (https://www.gtexportal.org/home/), Ensembl https://useast.ensembl.org/Homo_sapiens/Info/Index, RefSeq (https://www.ncbi.nlm.nih.gov/refseq/)] (examples shown in Fig. 1). The relative quantity of each isoform varies across tissues, although exon 22 is almost uniformly removed (129). It has long been speculated that isoforms created by alternate splicing change the physical and mechanical properties of the elastic fiber (127, 129, 130). However, the functional consequences of the multiple isoforms, and whether each isoform is translated into protein at all, remains unclear.

In vitro studies have demonstrated alteration of the physical properties of elastin monomers and elastic fibers (131–134). Although not typically spliced in vivo, deletion of exon 30 alters elastin microassembly and can affect the stiffness and viscoelastic properties of elastic fibers (131, 132). Similarly, deletion of exon 36 of bovine Eln (analogous to exon 34 in humans) leads to elastic fibers with decreased desmosine crosslinks (132, 135). The importance of splicing is also illustrated by naturally occurring human splice site variants (mutations) associated with SVAS. Splice site variants, in particular those in splice donor sites, may lead to intron readthrough and creation of an alternate transcript with a premature stop codon, which would then be predicted to be destroyed by nonsense-mediated decay. For some variants, the impact is less clear. For example, there is evidence that the c.800-3C>G variant alters splicing, leading to production of a transcript that lacks exon 16 and 17 (exons that are not typically subjected to alternative splicing) and exhibits impaired incorporation into elastic fibers (34). Interestingly, work from another laboratory, using a different RT-PCR primer set suggests that this variant leads to the generation of a transcript that would undergo nonsense-mediated decay (32). There is also an ultrarare single nucleotide variant (ELN c.1358-3G>A) that alters the splice acceptor site in exon 22, which is almost ubiquitously removed (spliced out) that experimentally increases retention of exon 22. Inclusion of this exon increases the hydrodynamic radius of TE and results in decreased stress relaxation (134). Although numerous splice variants have been associated with SVAS, there is evidence of reduced penetrance with this variant type and overall experimental evidence is lacking. Furthermore, although they seem to be expressed, the correlation between differences in RNA species and actual differences in deposited TE is unclear (6).

Unfortunately, the existing studies largely fail to address the effect of splicing on those exons that most commonly undergo alternative splicing in vivo (exons 3, 5, 11, 13, 22, 25, and 32), whose impact on tissue mechanics and matrix associated protein interactions may be more subtle. Nor has any study investigated the role of short transcripts such as those identified in the GTEx project as having high expression in the adult human aorta (136). Given the continual transcription of the ELN gene once the gene is activated in a tissue, it is intriguing to consider a nonstructural role for these alternative transcripts. The effect of isoform expression and the effect on matrix associated proteins has not been studied, but must also be considered.

Emerging Mechanisms of ELN Regulation

ELN gene transcription occurs in tissue-specific and temporally regulated patterns, but the factors that give rise to that specificity remain elusive. Additional knowledge in this area is critical for the development of new therapies for both rare diseases of elastic fibers and more common conditions like emphysema, atherosclerosis, or even typical aging where elastic fibers are destroyed. To address these goals directly, we propose the need for additional investigation in the following areas.

Transcriptional switch.

Why the immediate ELN promoter is structured like an always on, “housekeeping” gene, but the mature ELN mRNA and TE protein production are tightly regulated remains uncertain. Targeted studies have identified cytokines and their downstream transcription factors that modulate ELN promoter activity, but the master regulator(s) that control the “on switch” timing and tissue specificity of elastin expression are unknown. There is emerging evidence that epigenetic modifications and noncoding variants play a role. To answer this question, it may be necessary to utilize new technologies and evaluate the genome beyond the established ELN promoter.

There is a large (roughly 160 kb) intergenic region upstream of ELN that has yet to be interrogated. ELN is embedded in a differentially methylated region (DMR) extending ∼61 kb upstream that has increased methylation in aged, atherosclerotic vessels (137). In addition, GWAS studies have identified single nucleotide polymorphisms (SNPs) from the 3′ UTR of ELN to 124 kb upstream of the ELN locus that are associated with vascular and connective tissue outcomes (19, 20, 22–24). Although the impact of the specific variants is unknown, linkage to SNPs at such a distance suggests that complex long-range modulation of ELN is likely. It is also possible that this intergenic large region contains important enhancer elements that control three-dimensional folding of chromatin that affects ELN expression. Focused genomic interrogation of the region surrounding elastin as well as ChIP assays and other modern modalities may shed light on the effect of noncoding variants on ELN expression and, importantly, may help us understand key drivers of ELN-mediated disease (138).

RNA stability.

Once the ELN gene is turned on, it is transcribed at a near constant rate despite a lack of TE protein production after adolescence due to rapid turnover of ELN mRNA. There is evidence that microRNAs play a role in targeting ELN mRNA for degradation, but it is unclear if microRNAs alone are capable of reducing the level of mature ELN mRNA to sufficiently inhibit production of TE protein. Sequences within the ELN gene have been identified that bind to cytosolic or nuclear proteins and confer stability or increase turnover, however those proteins have not been identified (108, 110). It is possible that one or more of these proteins is somehow related to the microRNA degradation machinery. However, this has not been investigated, nor have other mRNA stability/degradation pathways. RNA purification followed by mass-spectroscopy or other techniques such as in cell protein-RNA interaction (incPRINT) may help identify RNA binding proteins responsible for determining the fate of ELN mRNA transcripts (139, 140). Elucidation of these proteins and pathways may further identify possible targets to increase TE protein production in diseases of elastin insufficiency.

Splice variants.

The ELN gene structure allows for the production of a multitude of alternatively spliced isoforms. Indeed, over 30 splice isoforms have been described to date (5) including short splice isoforms and those with retained introns. It is also clear that alternative splicing changes with development and that deregulation of splicing occurs in numerous diseases (141, 142). However, ELN splicing across tissues and development has not been extensively studied. Comprehensive studies of ELN mRNA and TE protein splicing patterns using modern technologies such as long-read RNA sequencing are necessary to understand the tissue and temporal patterns of ELN splicing and the impact alternative splicing has on elastic tissue structure. Furthermore, additional studies are needed to evaluate the direct effect of various isoforms on tissue mechanics. These studies may be particularly informative for regenerative medicine projects aiming to create precisely mechanically tuned structures. Likewise, investigations of alternatively spliced isoforms may illustrate differential interactions with matrix-associated proteins.

Interestingly, there are multiple short splice isoforms transcribed, some at high levels, the function of which is entirely unknown. Emerging studies have highlighted the noncoding function of splice isoforms (141, 143, 144). Given the dichotomy of a TE protein that is only produced during a short temporal burst but ongoing gene transcription, it is possible that these short highly transcribed ELN isoforms are performing alternate functions within the cell. Modern techniques such as single molecule mRNA fluorescence in situ hybridization (FISH) may shed light on noncoding functions of the short splice isoforms observed in ELN.

Conclusions

Earlier studies provide a glimpse at the complex processes that underlie ELN regulation, but many questions remain and advances in genetic and molecular techniques are poised to further elucidate these mechanisms. Further understanding the systems that regulate ELN expression may help develop therapeutics to treat both rare ELN-mediated diseases such as WS, SVAS, and ADCL and may even help treat common conditions such as solar elastosis (aging skin), atherosclerosis, and wound healing.

GRANTS

This study was funded by National Heart, Lung, and Blood Institute Grant HL006212 (to B.A.K.) and Postdoctoral Fellowship T32 HL125241 (to S.S.P.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

S.S.P. and B.A.K. analyzed data; S.S.P. and B.A.K. interpreted results of experiments; S.S.P. prepared figures; S.S.P. and B.A.K. drafted manuscript; S.S.P. and B.A.K. edited and revised manuscript; S.S.P. and B.A.K. approved final version of manuscript.