Cutis Laxa Arising from Frameshift Mutations in Exon 30 of the Elastin Gene (ELN)*

Congenital cutis laxa, a rare syndrome with marked skin laxity and pulmonary and cardiovascular compromise, is due to defective elastic fiber formation. In several cases, skin fibroblast tropoelastin production is markedly reduced yet reversed in vitro by transforming growth factor-β treatment. We previously showed that this reversal was due to elastin mRNA stabilization in one cell strain, and here this behavior was confirmed in skin fibroblasts from two generations of a second family. cDNA sequencing and heteroduplex analysis of elastin gene transcripts from three fibroblast strains in two kindreds now identify two frameshift mutations (2012ΔG and 2039ΔC) in elastin gene exon 30, thus leading to missense C termini. No other mutations were present in theELN cDNA sequences of all three affected individuals. Transcripts from both alleles in each kindred were unstable and responsive to transforming growth factor-β. Exons 22, 23, 26A, and 32 were always absent. Since exon 30 underwent alternative splicing in fibroblasts, we speculate that a differential splicing pattern could conceivably lead to phenotypic rescue. These two dominant-acting, apparently de novo mutations in the elastin gene appear to be responsible for qualitative and quantitative defects in elastin, resulting in the cutis laxa phenotype.

Elastic fibers are the extracellular matrix structures responsible for the properties of resilience and elastic recoil in all elastic tissues (1,2). There are two morphological elements in elastic fibers: the microfibrillar component and the amorphous component. The microfibrillar component is made up of 10 -12-nm microfibrils that are composed of at least seven different glycoproteins, including the two genetically distinct fibrillins, whose genes are the loci for Marfan's syndrome and congenital contractural arachnodactyly (3,4). Elastin is also present in the amorphous component as a cross-linked complex of hydrophobic proteins synthesized from the single copy, multi-exon elastin gene (ELN) by extensive alternate usage of several exons (1,(5)(6)(7).
There are several inherited disorders characterized by aberrant elastin synthesis or degradation. Abnormal elastic fibers are seen in Menkes' syndrome due to altered copper transport resulting in decreased activity of lysyl oxidase (8), whereas elastic fibers are prematurely degraded due to unregulated elastase activity in patients with ␣1-antitrypsin deficiency (9) and some forms of atrophoderma (10). Pseudoxanthoma elasticum and Buschke-Ollendorff syndrome are examples of heritable skin diseases in which increased deposition of cutaneous or vascular elastin has been demonstrated (11,12). In contrast, decreased or aberrant deposition of elastic fibers in certain tissues is characteristic of Marfan's syndrome (13), supravalvular aortic stenosis (SVAS) 1 (14 -16), and cutis laxa (6,17).
In some diseases of elastic tissue, mutations in genes for structural proteins have been demonstrated. Currently, there are at least three dominant disorders, Marfan's syndrome (18 -20), ectopia lentis (19,20), and congenital contractual arachnodactyly (19), that are caused by mutations in fibrillin genes. Most patients with Williams syndrome that have SVAS are heterozygous for deletions of ELN and presumably other contiguous genes on chromosome 7q (21,22). However, only in patients with SVAS have disruptions or point mutations within ELN itself been described (15,16,23). The connective tissue features of SVAS and Williams syndrome are consistent with, but not yet shown to be due to, functional hemizygosity at ELN.
We have previously shown in one cutis laxa cell strain that transcript instability was the basis of a defect in ELN mRNA accumulation and tropoelastin production (24). TGF-␤ was able to increase mRNA stability markedly and to stimulate production of immunoreactive tropoelastin protein in this cell strain. Since the metabolic and ultrastructural defect was confined to elastin, we hypothesized that a structural defect in the ELN transcript could be responsible for decreased mRNA stability. In this report, we describe heterozygosity for a frameshift mutation (2012⌬G) in ELN in this cutis laxa patient and a similar mutation in two generations of a second cutis laxa family (2039⌬C), which we propose to be responsible for defects in tropoelastin production. * This work was supported by Grant AR44431 (formerly Grant GM37387) from the National Institutes of Health and by the Department of Veterans Affairs. This work was reported in preliminary form at a meeting entitled "Elastin and Elastic Tissue" (Maratea, Italy, October 17-20, 1996)

EXPERIMENTAL PROCEDURES
Clinical Summary-Patient K.T. was the 3.8-kg full-term male product of an uncomplicated pregnancy, labor, and delivery to a 23-year old G 2 P 1 Caucasian mother. The parents denied consanguinity. Loose skin, stridor, and feeding difficulties were apparent from birth. Additional clinical findings ascertained during infancy included moderate subglottic stenosis with floppy airway structures, redundant mitral and tricuspid valves, mild dilatation of the proximal aorta and great vessels, and umbilical and inguinal hernias. He underwent inguinal herniorrhaphies at ages 7 months, 3 years, and 14 years.
At age 17 years, his height and weight are at the 75th percentile. His physical exam is significant for an aged appearance: smooth, loose skin lacking elastic recoil; tortuous, pulsatile external carotid arteries; and a hoarse voice. He complains of fatigue, dyspnea on exertion, and shortness of breath. His echocardiogram is relatively unremarkable except for minimal aortic root dilatation; an electrocardiogram reveals mild right ventricular hypertrophy. Pulmonary function testing shows reduced expiratory flow, suggestive of fixed or collapsible upper airway obstruction.
Histologic, ultrastructural, and biochemical analyses of skin and cultured fibroblasts from the patient have been reported previously (17,24). Briefly, dermal collagen fibers appear normal, but elastic fibers appear fragmented with a paucity of amorphous elastin in the matrix. Tropoelastin production in cultured fibroblasts from this patient was the lowest of six cutis laxa patients studied (17), and an apparent nonspecific increase in type VI collagen production was also noted (25).
Limited clinical information is available on the second family at this time. The female proband (WM) was ascertained in 1965 with classical cutaneous features of cutis laxa at 2 years. This individual gave birth to an affected son (WS) in 1991, and skin biopsies were cultured from both individuals in 1993 by Drs. J. Uitto and E. Tan (Jefferson Medical College). The father and maternal parents were reportedly unaffected, and the patients have been lost to follow-up.
Tissue Culture-Skin fibroblasts were grown from skin biopsies obtained after appropriate consent. Normal human skin fibroblasts (GM4390) were obtained from the NIGMS Human Genetic Mutant Cell Repository (Camden, NJ). Cells were grown to confluence in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (Atlanta Biologicals, Inc., Norcross, GA).
cDNA Synthesis-Confluent cultured cells were washed twice in phosphate-buffered saline and lysed in 2 ml of 4 M guanidine isothiocyanate containing 0.1 mM ␤-mercaptoethanol. DNA was sheared by three passages through a 22-gauge needle, and RNA was isolated by extraction in acid phenol/chloroform (26). Isolated RNA was stored at Ϫ70°C. cDNA was synthesized using 1-3 g of total cellular RNA with 200 units of Moloney murine leukemia virus reverse transcriptase and 0.75 mM gene-specific oligonucleotide primers in a total volume of 20 l for 60 min at 37°C.
PCRs were carried out in a Perkin-Elmer minicycler in 50-l volumes using 1-2 l of reverse transcription products, a 0.2 M concentration of each primer, a 0.4 mM concentration of each dNTP, and 2.5 units of Taq polymerase (Promega, Madison, WI). The reaction cycles consisted of an initial denaturation at 95°C for 3 min followed by 95°C for 1 min, annealing at 60°C for 1 min, and extension at 72°C for 1 min per cycle for 30 cycles.
Direct DNA Sequencing-PCR products amplified by different F and R primer combinations from both cutis laxa and control cell strains were separated by ethidium bromide-1% agarose gel electrophoresis. Amplimers of the expected sizes were excised and purified using Prep-A-Gene DNA purification systems (Bio-Rad). Sequencing was performed using an ABI PRISM TM 377 DNA Sequencer (Perkin-Elmer) or manual sequencing using the double-stranded DNA cycle sequencing system (Life Technologies, Inc.). For mutation confirmation, three different RT-PCR products were independently sequenced.
Cloning of RT-PCR Products-RT-PCR products of ϳ3 kilobases synthesized with primers 96F and He3ЈUTR741R were separated by electrophoresis, purified, and cloned into the pNoTA vector using the PRIME PCR CLONER TM cloning system (5 Prime 3 3 Prime, Inc., Boulder, CO) according to manufacturer's instructions. A total of 31 white colonies was picked from overnight cultures and propagated in LB medium for plasmid minipreps. The plasmid DNAs were digested with BamHI, and positive clones with correctly sized inserts (2854 bp) were identified by gel electrophoresis.
Restriction Endonuclease Digestion-The 2012⌬G deletion in exon 30 of K.T. created a novel Alw26I restriction site. RT-PCR products amplified by primers 11 and 12 (476 bp) from K.T. fibroblasts, a control cell strain, and genomic PCR products amplified by primers 23 and 24 (277 bp) from the proband, his parents, and 65 unaffected individuals were digested with Alw26I at 37°C for 60 min. The 2039⌬C deletion abolished a PflMI site in the same PCR product of the mutant allele in WM and WS. DNAs from an additional 48 individuals were also screened for the same mutation. The digests were electrophoresed through an 8% polyacrylamide gel and visualized by ethidium bromide/ UV fluorescence.
Heteroduplex Assay-PCR was performed with primers 1761F and 2292R in the presence of 0.3 Ci of [ 32 P]dCTP for 30 cycles. Equal amounts of PCR products from normal and cutis laxa fibroblasts were mixed, incubated at 95°C for 2 min, and gradually (20 -30 min) cooled to 37°C. After adding 1 l of loading buffer to 5 l of sample, the samples were separated by electrophoresis on a 0.5ϫ Mutation Detection Enhancement gel (FMC Corp. BioProducts, Rockland, ME).
RNA Stability-Elastin production and mRNA stability studies were carried out as described previously (17,24). RNA was isolated and purified using the QIAshredder TM and the RNeasy TM mini kit (QIA-GEN Inc., Chatsworth, CA) according to the manufacturer's instructions.

RESULTS
A Frameshift Mutation of ELN in Cutis Laxa Strain K.T.-A single base deletion (2012⌬G) in exon 30 of ELN was initially identified by direct sequencing of the RT-PCR products from patient fibroblasts (Fig. 1). This mutation created a novel Alw26I recognition site (![N] 5 GAGAC); heterozygosity for the deletion was confirmed in the RT-PCR product by the presence of diagnostic 244-and 132-bp bands after digestion of the 476-bp PCR product with Alw26I ( Fig. 2A). Furthermore, heterozygosity for the mutation was confirmed at the genomic DNA level in the patient, but the polymorphism was absent in DNA from both parents (Fig. 2B) and 65 unrelated control samples (data not shown).
Upon sequencing the entire ELN coding region as well as 240 bp of the 5Ј-UTR and 1.2 kilobases of the 3Ј-UTR, no other sequence variations could be found between the two cloned patient alleles and controls. Both alleles contained A at position 1313, the site of an A/G (Ser 422 /Gly) polymorphism (29). Several other presumably less significant base changes were found in the 1.2-kilobase ELN 3Ј-UTR of both patient and control strains that differed from or complemented GenBank TM entry M17282 (28). These included seven base changes (T 174 to A, T 196 to C, C 737 to G, A 1195 to C, C 1216 to A, T 1245 to A, and T 1258 to A), four insertions (C 233 , C 776 , C 788 , and C 861 ), three deletions (A 335 , A 724 , and A 1137 ), and 15 newly identified residues (G 857 , C 861 , C 863 , C 876 , C 924 , T 925 , C 926 , G 939 , C 946 , C 962 , C 963 , C 1018 , A 1227 , A 1228 , and C 1231 ). Although these ELN 3Ј-UTR sequences differ from the original GenBank TM data, they were identical in our patient and control fibroblast mRNAs. Further sequencing of elastin cDNA revealed that exons 22, 23, 26A, and 32 were spliced out of transcripts from both control and K.T. fibroblasts (Fig. 3), a feature that has been reported previously in normal cells (30).
A Related Mutation in a Second Family Is Alternatively Spliced-A second set of fibroblast strains from a kindred with an affected mother (WM) and son (WS) showed characteristic low tropoelastin production with high induction of the protein by TGF-␤ (Fig. 4). Elastin mRNA only accumulated in these strains in the presence of TGF-␤, and it rapidly degraded as soon as TGF-␤ was withdrawn (Fig. 4B). Heteroduplex analysis revealed a novel band mobility in WM and WS using primer pairs that spanned the exon 30 region (data not shown). Eight full-length cDNA clones were derived from WS mRNA and completely sequenced. Three of the cDNA clones lacked exon 30 (Fig. 5A). Of the five cDNA clones containing this exon, four showed a deletion of C 2039 (Fig. 5A), which predicts a frameshift mutation with consequences similar to those of strain K.T.
Heterozygosity of the mutation in WM and her son (WS) was established by direct sequencing of genomic DNA. To confirm the heterozygosity and to rule out a polymorphism, restriction digests were performed with Pf1MI (Fig. 5B). 50% of WS and WM DNAs were resistant to enzyme digestion, whereas DNAs from 50 other unrelated individuals were fully susceptible to Pf1MI enzyme digestion at the exon 30 locus.
TGF-␤ Does Not Change the Ratio of ELN mRNA Expressed from Two Alleles-As shown previously for K.T. (24) and above for WM and WS, TGF-␤ could in part restore tropoelastin expression in these cutis laxa fibroblasts by stabilizing ELN mRNA. To examine whether partial restoration of ELN mRNA stability was functioning through selective expression of ELN mRNA from the normal allele, RT-PCR products amplifying the region spanning nucleotides 1671-2292 from TGF-␤treated and untreated K.T. fibroblasts were analyzed under semi-quantitative conditions. Amplified products were digested with Alw26I. The intensity of the 244-bp mutant allele fragment relative to the normal 376-bp fragment did not change after TGF-␤ exposure (Fig. 6). Similar results were obtained with RNAs from WS and WM (data not shown). Transcripts of both alleles had equivalent instability.
Mutational Consequences-The two single base deletions in exon 30 each predict a frameshift in the coding region for the elastin carboxyl terminus. The predicted, truncated protein product would consist of 667 amino acids (Fig. 7). Since exon 32 was usually absent, the predicted open reading frame (ORF) of the mutant transcript would more frequently continue into the 3Ј-UTR to be translated as a missense structure of 713 amino acids lacking the distinctive carboxyl terminus of tropoelastin (Fig. 7). In the less likely event of inclusion of exon 32, the mutation would create a premature termination codon in exon 32, predicting a truncated, missense C terminus and a translation product of 667 amino acid residues. Among the 3Ј-UTR FIG. 1. A deletion (2012⌬G) in ELN in a cutis laxa patient. RT-PCR products from both patient (K.T.) and normal fibroblasts were generated by primers 1761F and 2292R (see "Experimental Procedures") and subjected to automated fluorescent sequencing. The location of the deletion is indicated by the arrow.

DISCUSSION
Cutis laxa is a relatively rare connective tissue disease characterized by genetic heterogeneity and clinical variability (6,(31)(32)(33). In all cases, the primary diagnostic feature is loose, hyperextensible skin with decreased resilience and elasticity, leading to a premature aged appearance. The skin changes are often accompanied by extracutaneous manifestations, including pulmonary emphysema, bladder diverticula, pulmonary artery stenosis, and pyloric stenosis. Histological examination of the skin in cutis laxa often reveals marked fragmentation or diminution of elastic fibers (6,17,32). This is a considerably different phenotype than that found in SVAS, a pathology arising from elastin mutations that lead to functional hemizygosity (15,16,23,34,35). Skin fibroblast cultures from many cutis laxa patients exhibit reduced ELN mRNA levels or tropoelastin production, but fibroblasts from other affected individuals exhibit normal levels of elastin production with abnormal elastic fiber morphology (17,36). At the time of submission of this manuscript, the molecular basis for cutis laxa had not

FIG. 4. TGF-␤ reversal of the elastin deficit in WM and WS is related to transcript instability.
A, fibroblasts from the control strain (GM4390), WS, and WM were analyzed for tropoelastin (TE) production (9) in the presence and absence of 10 ng/ml recombinant TGF-␤2 (a gift of Genzyme Corp., Cambridge, MA). Under basal (5% newborn calf serum) conditions, tropoelastin production (expressed as molecular equivalents (mol. eq.)/cell/h Ϯ S.E.) was near the base-line limits of the enzyme-linked immunosorbent assay in the cutis laxa strains, whereas production rose to nearly normal levels in the presence of TGF-␤. B, cells pretreated with TGF-␤2 for 48 h as described above were put into fresh medium lacking TGF-␤ and containing 75 M 5,6dichlorobenzimidazole riboside to block transcription. RNA was isolated immediately, 6, 12, and 24 h after treatment, and elastin transcripts were quantified in relation to cyclophilin transcripts by Northern blot hybridization. Consistent with the production data, elastin mRNA levels were lower in cutis laxa strains WM and WS and undetectable in the absence of TGF-␤ (data not shown). Elastin mRNA levels fell rapidly to base-line levels by 6 h. In control fibroblasts, the t1 ⁄2 of elastin mRNA was Ͼ15 h. been described. However, another group has recently reported autosomal dominant cutis laxa associated with a point deletion in exon 32 (37). This exon was absent from the cDNAs examined in this study. Unlike the biochemical phenotype reported here, elastin mRNA is expressed and apparently stable in this strain, and the data from that report are consistent with a dominant-negative effect at the level of elastic fiber formation. The fibroblasts in this study produced insignificant amounts of elastin mRNA and protein in the absence of TGF-␤ stimulation.
We identified two single base deletions (2012⌬G and 2039⌬C) in the coding region of ELN cDNA from two cutis laxa lineages. Analysis of the parental DNA indicated a de novo mutation in proband K.T., and limited familial information suggests that a de novo mutation in WM was passed to her son, WS. Both of the deletions would result in translation of missense protein sequence 3Ј to the mutation, and they predict subsequent loss of the functional carboxyl terminus in the tropoelastin molecule by missense and/or premature termination, depending on the more frequent splicing out or less frequent inclusion of exon 32 (Fig. 7). Thus, different tissues could produce defective elastin molecules with slightly different carboxyl termini, depending on the tissue-specific mRNA splicing pattern. Regardless of the status of exon 32 splicing, each of these point deletions dictates the loss of two conserved cysteine residues that create an intramolecular disulfide bridge and that are thought to be important, together with a basic tetrapeptide tail, for interaction of tropoelastin with other pro-teins present in the elastic fibers of microfibrils (38,39). Although little or no normal or mutant protein is expressed in fibroblast cultures, tissues in vivo (particularly under the influence of TGF-␤) might express significant levels of either defective gene product and lead to a dominant-negative effect at the level of protein assembly. We may be able to resolve this issue with C terminus-specific antibodies when tissue becomes available.
Analysis of the mutation in WM and WS confirmed ELN as the cutis laxa locus. Both the similar location in exon 30 and the predicted translational consequences of 2012⌬G and 2039⌬C mutations are intriguing, and they may indicate a susceptible site for production of dominant-negative, viable ELN mutations. Exon 30 has recently been identified, at least in the rat, as a site for a cis-acting mRNA stability element. 2 More important, our finding that exon 30 is alternatively spliced may be indicative of a phenotypic rescue mechanism (40) whereby cells or tissues could escape the effects of the mutations. The same possibility would apply for the recent mutation described in exon 32 (37). The fact that all three affected individuals are not severely incapacitated suggests that pulmonary and cardiovascular elastic tissues must be relatively spared. In contrast to the predominant effect of SVAS in the aortic root, both of the cutis laxa mutations appear to have a largely cutaneous phenotype. This could be due to tissue-specific or developmentally regulated differential expression of the two alleles, increased frequency of alternative splicing of exon 30 (or 32) in non-cutaneous tissues, or differential consequences of mutant RNA and protein in different cell types.
The low frequency of polymorphisms in ELN makes analysis of allelic usage problematic. Transcripts from both normal and mutant alleles were equally unstable, and both transcripts were equally stabilized by TGF-␤2. Both alleles were otherwise identical throughout the coding region, 223 bp of the 5Ј-UTR, and 1.2 kilobases of the 3Ј-UTR (data not shown). A comparable increase in the amount of both normal and mutant transcripts by TGF-␤ stimulation suggests that at least the proximal regulatory regions of both alleles are intact, including the response sites for TGF-␤ (41) and insulin-like growth factor-1 (42) in the elastin promoter. We have previously shown that ELN transcription rates are normal in K.T. (24). In addition, since the point deletion was passed from WM to WS, compound heterozygosity could only have persisted in WS by donation of an independent mutation from his paternal ELN allele. These findings, together with the absence of the deletions in DNAs from both parents of K.T. and unrelated controls, argue that K.T. and probably WM are the origins of new dominant mutations that account for the cutis laxa phenotype. The molecular data predict that if mutant tropoelastin were synthesized, at least in skin, it would contain defective carboxyl termini. The quantitative and qualitative defects in tropoelastin production could readily account for the patients' abnormal (cutaneous) phenotype. We have not determined the mechanism whereby the heterozygous mutation appears to exert a dominant effect on ELN mRNA stability. Nonsense-mediated mRNA decay is a protective cellular mechanism described for several genes; however, it acts in cis as a rule (43). Indeed, this mechanism appears to be operative in some forms of SVAS (16). A dominant, trans-acting effect might arise if mutant transcripts altered the availability of a rate-limiting factor for elastin mRNA stability. Elastin mRNA levels drop if secretion is perturbed (44), and translational or packaging defects may have a feedback effect on mRNA stability (45). To test these possibilities, 2 M.-C. Zhang and W. C. Parks, submitted for publication. Normal translation terminates at TGA in exon 36, producing a tropoelastin molecule of 688 amino acids (aa). Missense sequence following the frameshift is indicated by hatched areas in the two mutant ORFs. In the absence of exon 32 (as in skin fibroblasts), missense translation terminates at a stop codon (TAA) that is 75 bp 3Ј to exon 36, producing a defective protein of 713 amino acids. In the presence of exon 32, missense translation terminates at a premature stop codon (TAG) in exon 32, producing a defective protein of 667 amino acids. In transcripts lacking exon 30 (rescued ORF), the mutation is skipped, and a 664-residue tropoelastin is synthesized. kb, kilobases. current studies are directed at examining the effects of introducing a mutant ELN cDNA into elastin-expressing cells.
In conclusion, we have found two point deletions in ELN from individuals with the classical cutaneous phenotype of congenital cutis laxa. These mutations result in frameshifts that disrupt the sequence of the conserved carboxyl terminus of tropoelastin and lead to marked transcript instability. We speculate that variable splicing of the mutated exon may account for reduced severity and selective tissue effects. The elastin mutations in SVAS appear to involve functional hemizygosity that affects the elastin-rich aortic root (14 -16). Complementing these findings, autosomal dominant cutis laxa appears to arise from dominant-negative effects at the level of both RNA stability and protein assembly, with the prospect of a novel mechanism for variable tissue consequences. Identification of additional elastin mutations in both dominant and recessive forms of cutis laxa will enable us to understand better the structural biology and regulation of this important extracellular matrix component.