|
|
| |||||||||||||||||||||||||||||||
Background The fibrillin gene encodes a protein in the extracellular matrix, and this protein is widely distributed in elastic tissues. The fibrillin gene is the site of mutations causing Marfan's syndrome. This disorder shows a high degree of clinical variability both between and within families. Each family appears to have a unique mutation in the fibrillin gene, which precludes the routine use of mutation screening for presymptomatic diagnosis of the disorder. The goal of this study was to develop a widely applicable method of molecular diagnosis.
Methods We used three newly characterized intragenic sites of normal DNA repeat-sequence variation (i.e., polymorphisms) as markers to follow the inheritance pattern of specific copies (alleles) of the fibrillin gene in multiple kindreds with various clinical features of Marfan's syndrome.
Results The polymorphic markers allowed identification of the particular copy of the fibrillin gene that cosegregated with Marfan's syndrome in 13 of the 14 families tested. In 11 families a definite presymptomatic diagnosis of Marfan's syndrome could be made in family members who had only equivocal manifestations of the disorder. In two other families, some family members demonstrated either classic Marfan's syndrome or a milder but closely related phenotype. The copy of the fibrillin gene that cosegregated with classic Marfan's syndrome was not inherited by family members with the latter, atypical, form of the disease. These milder phenotypes, previously diagnosed as Marfan's syndrome, were not associated with aortic involvement.
Conclusions These results document the usefulness of novel polymorphic DNA repeat sequences in the presymptomatic diagnosis of Marfan's syndrome. Our findings also demonstrate that the various clinical phenotypes seen in selected families may be due not to single fibrillin mutations, but rather to different genetic alterations. These findings underscore the need for a modification of the current diagnostic criteria for Marfan's syndrome in order to achieve accurate risk assessment.
-adrenergic-blocking agents has led to a dramatic decrease in morbidity and mortality from cardiovascular causes over the past two decades3,4. The effective treatment of patients with Marfan's syndrome relies heavily on early and accurate diagnosis. Until recently, such efforts were hampered by the lack of a sensitive and specific diagnostic test for the disorder. In 1991 a cause-and-effect relation was established between mutations in the gene on chromosome 15 encoding fibrillin (FBN1), a glycoprotein component of the extracellular microfibril, and the Marfan's phenotype5. In addition, linkage between FBN1 and the classic phenotype of Marfan's syndrome has been observed in all families tested, suggesting that FBN1 defects are the predominant, if not the sole, cause of the disorder6,7,8,9,10,11. It therefore seemed likely that molecular genetic analysis would prove a valuable adjunct to clinical diagnosis for purposes of risk assessment. Unfortunately, the rate of detection of mutations in Marfan's syndrome has been extremely slow, and with one exception, all mutations identified to date have been unique to a single family5,12,13,14,15,16,17,18. These factors have precluded the widespread use of direct analysis of mutations for presymptomatic diagnosis.
As an alternative, we have explored the use of linkage analysis in family screening. Our intention was to devise a method that would allow discrimination between the many normal copies of the fibrillin gene and the single abnormal copy that is inherited by the affected members of a given family. To this end, we have identified a panel of highly informative intragenic microsatellite polymorphisms (genetic markers) that are widely interspersed in FBN1. Microsatellites consist of repeating sequences of a short block of nucleotides, and the number of repeats may vary from person to person. Typing each family member for these polymorphic markers (i.e., determining the number of repeats) and observing the inheritance pattern of these sequence variants allow the determination of haplotypes (combinations of polymorphic variants that are inherited as a unit and that therefore define single copies of the fibrillin gene). Haplotypes that carry disease-producing mutations vary from family to family. It is therefore necessary to determine which haplotype cosegregates with the disease in each kindred. In this article we demonstrate the usefulness of this method, called haplotype-segregation analysis, for a variety of clinical and diagnostic purposes in families affected by Marfan's syndrome.
Methods
Study Subjects
All the study subjects either were followed at the Medical Genetics Clinic of Johns Hopkins Hospital or were referred for presymptomatic diagnosis after evaluation by a medical geneticist. In addition, all the subjects were studied by echocardiography, angiography, or both and by slit-lamp examination, and those considered to be affected satisfied the international diagnostic criteria for Marfan's syndrome19.
Identification of Novel FBN1 Microsatellite Polymorphisms
A phage library was derived from a single yeast artificial-chromosome library clone (YAC-15) that contains the entire fibrillin gene, including the 5' and 3' untranslated regions. DNA isolated from overlapping phage clones was cleaved to completion with a panel of restriction enzymes, fractionated by electrophoresis through an agarose gel, and transferred to a nylon membrane. The resulting filters were hybridized to a labeled poly(dA-dC) • (dT-dG) probe (Pharmacia) as described elsewhere20. Hybridizing DNA fragments were purified from agarose gels, subcloned into the pUC19 plasmid vector, and sequenced by a modification of the dideoxy-chain-termination procedure21.
Typing of Family Members and Control Subjects
Polymerase-chain-reaction (PCR)22 primers were fashioned from the sequences flanking each of the short-sequence tandem-repeat regions, which were designated mts-1, mts-2, and mts-4. A fourth intragenic microsatellite polymorphism (mts-3) was evaluated with primers that have been previously described23. The sequence and relative position of each primer pair are shown in Figure 1. Each PCR was performed in a total volume of 25 microl, including 500 ng of template genomic DNA, 1.25 units of Taq polymerase (Cetus), 2.5 microl of 10 x PCR buffer (Cetus; final concentration of magnesium chloride, 1.5 micro M), 400 nM final concentration of unlabeled primer, 66 nM final concentration of end-labeled primer (one primer in each set), and deoxynucleotide triphosphates (Cetus; 400 micro M each). The primers were end-labeled in a reaction volume of 6.8 microl containing 4.4 microl of primer (10 micro M stock), 0.5 microl of T4 polynucleotide kinase and 1.4 microl of 5x kinase buffer (BRL), and 0.5 microl of 
-32P-labeled ATP (0.15 micro Ci per microl; NEN). The mixture was incubated at 37 °C for one hour and diluted with 20 microl of distilled water, and 1 microl was used per template for PCR amplification.
|
Allele frequencies, marker heterozygosity, and haplotype heterozygosity were determined by analyzing genomic DNA from 50 unrelated and unaffected control subjects (100 chromosomes in all). The largest marker allele found in this test population for a given polymorphic locus was assigned the number 1. Successively smaller marker alleles were assigned successively larger numbers (Table 1). Genomic DNA was isolated from whole blood or immortalized lymphoblasts by methods that have been previously described24.
|
Four Novel FBN1 Microsatellite Polymorphisms
Southern blot analysis of overlapping phage DNA clones that spanned the entire fibrillin gene with a probe as described in the Methods section revealed four positively hybridizing genomic DNA fragments. One short sequence repeat (mts-3) had been previously identified. Subsequent mapping of the four repetitive sequences revealed that they were widely spaced throughout the 100-kb expanse of FBN1, with localization to introns 1, 5, 28, and 43. Direct sequencing of the tandem repeats and flanking regions revealed that mts-1, mts-2, and mts-4 are composed of (CA)n blocks (Figure 1). Mts-3 was previously known to be a pentanucleotide repeat -- (TAAAA)n.
To determine whether any of these repetitive sequences showed polymorphic variation in the general population, we generated fragments of DNA that spanned each repeat with flanking primers and fractionated them according to size. The number of distinct marker alleles for each microsatellite ranged from 4 (mts-3) to 15 (mts-2). The use of template genomic DNA from 50 unrelated and unaffected control subjects (100 chromosomes in all) allowed the determination of allele frequencies and marker heterozygosity in the population (Table 1). These data predict a minimum of 44 distinct haplotypes in the general population, with the haplotype 5/11/2/7 (ordered 5' to 3', from mts-1 to mts-4) being the most common and the only one seen in homozygous subjects. Genotyping for these four polymorphic markers identified two distinct haplotypes in 86 percent of those tested. Informativeness, a measure of the ability to determine which allele is passed from a parent to a child in haplotype-segregation analysis, is therefore predicted to be quite high.
Haplotype-Segregation Analysis
The parents of a four-month-old infant (Subject III-1, Family A) (Figure 2) requested presymptomatic diagnosis. The father (Subject II-1) and the paternal grandmother (Subject I-1) had classic features of Marfan's syndrome in the ocular, skeletal, and cardiovascular systems. The infant had mild pectus carinatum, but no specific signs of the disorder. Haplotype-segregation analysis suggested that the disease was passed from Subject I-1 to Subject II-1 on the 5/11/2/7 allele but that Subject III-1 inherited the 1/1/2/7 allele from her affected father. These data predict that this infant is unaffected.
|
Although the particular Marfan's mutation in Family B has not yet been identified, this hypothesis is strengthened by our analysis of another family (Family C) (Figure 2). The proband in this family was found to have a clinical variant that shows involvement of the mitral valve, aorta, skin, and skeleton (which we have termed the MASS phenotype), but no ectopia lentis or progression of aortic dilatation to dissection25. This patient had extreme dolichostenomelia (excessive growth of long bones), mitral-valve prolapse, early myopia, striae atrophicae, and an aortic root that was at the upper limit of the normal size when standardized with body-surface area. All the other family members were tall, but selected members also had early myopia, mitral-valve prolapse, scoliosis, or a combination of these. This milder connective-tissue phenotype segregates with the 9/3/4/9 haplotype, an allele that the proband did not carry. In fact, the occurrence of a novel frame-shift mutation in the fibrillin gene has been identified in the proband (Subject III-2)18.
Haplotype-segregation analysis cannot be used in diagnosis when there are not enough family members available for study to determine which allele is segregating with the disease. The specific number of family members required will vary with family structure and the informativeness of the marker alleles. Presymptomatic diagnosis by haplotype-segregation analysis was attempted successfully in 10 of 11 additional families tested (data not shown). Although we have not used linkage methods for prenatal diagnosis, such a use would not differ from presymptomatic diagnosis with regard to the theory, execution, or predicted informativeness of the technique.
Discussion
Fibrillin is a major glycoprotein component of the extracellular microfibril, a structure with wide distribution in elastic tissues26. Although its function remains to be fully elucidated, it is believed that fibrillin participates in determining the structural integrity of tissues, partly by guiding the organization of tropoelastin molecules into maturing elastic fibers. The identification of new polymorphic variants in the gene encoding fibrillin should facilitate the study of Marfan's syndrome and other disorders in which there is primary disruption of elastic tissues.
To date, 14 different FBN1 mutations causing Marfan's syndrome have been reported, thus limiting the use of direct mutation analysis for presymptomatic diagnosis12,13,14,15,16,17,18. In contrast, FBN1 haplotype-segregation analysis is applicable to all subjects with a positive family history and a sufficient number of participating family members to determine which allele is segregating with the disease. All families with classic Marfan's syndrome that have been studied with flanking markers, intragenic markers, or both have shown a consistent linkage between FBN1 and the disease phenotype6,7,8,9,10,11. This observation suggests that fibrillin-gene defects account for the vast majority of cases of classic Marfan's syndrome, if not all. It seems appropriate, therefore, to counsel small families on the basis of haplotype segregation even though statistically significant evidence of linkage may be unattainable. More caution is warranted when atypical phenotypes are considered.
Fibrillin is encoded by a large (approximately 110 kb) and highly fragmented (65 exon) gene27,28. Previous work identified a number of intragenic markers, including one microsatellite polymorphism (mts-3) and two single-nucleotide polymorphisms, which are all located in the central portion of the gene5,23,29. The combined informativeness of these markers is low, precluding their use for linkage analysis in many families5,11. To increase our ability to counsel families with Marfan's syndrome, we searched for additional highly informative intragenic microsatellite polymorphisms.
The data presented here clearly show that newly identified intragenic markers can be used to follow the segregation pattern of FBN1 alleles in most families with Marfan's syndrome. Although the 5/11/2/7 haplotype was observed in homozygosity in approximately 14 percent of unrelated persons in the general population, haplotype-segregation analysis was informative in 13 of the 14 families tested in this study. In addition, this type of analysis facilitates the detection of intragenic recombination, a potential confounding factor in any linkage study.
Recent work has demonstrated the usefulness of -adrenergic blockade in slowing the rate of aortic-root dilatation in Marfan's syndrome4. It has also been suggested that the earlier this therapy is instituted in the course of the disease, the greater the benefit. This is one example of how presymptomatic diagnosis may have a direct effect on medical management. In addition, early exclusion of the diagnosis would avert the need for the routine use of expensive diagnostic testing in the follow-up of clinically equivocal cases.
Intrafamiliarol l clinical variability is a hallmark of Marfan's syndrome1. A previous study demonstrated that such variability can be seen among family members carrying an identical FBN1 mutation, predicting the existence of genetic or environmental modifiers (or both) to phenotypic expression12. This study offers a second molecular explanation. In two of the families shown in Figure 2, some family members have specific features of Marfan's syndrome, whereas others have nonspecific connective-tissue manifestations that are commonly seen as components of the Marfan's phenotype. Interestingly, all these family members satisfy the international diagnostic criteria for Marfan's syndrome and would therefore be assigned to affected status19. In the presence of an unequivocally affected first-degree relative, these guidelines do not require one of the major manifestations that are relatively specific for Marfan's syndrome (ectopia lentis, aortic-root dilatation, and dural ectasia), but only abnormalities of two organ systems that are consistent with the diagnosis. Haplotype-segregation analysis clearly demonstrates that not all affected members of Families B and C are segregating the same mutant fibrillin allele. It remains to be determined whether the milder disease phenotypes that indicate dominant mendelian inheritance in these families are caused by defects of the fibrillin gene or by mutations in other genes.
Many isolated features of Marfan's syndrome, including early myopia, mitral-valve prolapse, scoliosis, and joint hypermobility, have a high prevalence in the general population. Patients with mitral-valve-prolapse syndrome, the MASS phenotype, or variants of other common connective-tissue disorders (e.g., the Ehlers-Danlos syndrome) can have a clustering of these relatively mild features of Marfan's syndrome. It is not unexpected, but in fact predictable, that a subgroup of families with Marfan's syndrome will also segregate a clinically related but etiologically distinct connective-tissue disorder. This could explain why a large French kindred with incomplete, atypical, and variable manifestations of Marfan's syndrome, in which members with abnormal phenotypes were considered to be affected with the same disorder, did not have linkage to FBN130. Determining the precise prevalence of this phenomenon awaits molecular analysis of the majority of families with Marfan's syndrome that have wide variability in tissue involvement and severity of disease.
In the cases presented here, it is clear that not all family members who satisfy the diagnostic criteria for Marfan's syndrome are at equal risk for life-threatening cardiovascular manifestations. The implications include the need for individualized counseling and management and for appropriate caution when studies of natural history and therapeutic outcomes are interpreted. The requirement of a major manifestation of Marfan's syndrome for the clinical diagnosis, even when there is an unequivocal family history, complemented by the use of molecular analysis for those with atypical disease, should help optimize the assessment of cardiovascular risk in families segregating the Marfan phenotype.
Note added in proof: We have identified an additional family in which a person with atypically mild disease previously diagnosed as Marfan's syndrome did not inherit the FBN1 allele that cosegregates with the more severe Marfan's phenotype seen in the extended family. We have now identified the precise mutation, substitution of a cysteine residue in an epidermal growth factor-like domain, that is associated with classic Marfan's syndrome in this family. The person with milder manifestations, including skeletal features and mitral-valve prolapse, does not carry this mutation. This is consistent with the findings in this report.
Supported by grants (HL-02815 and HL-35877) from the National Institutes of Health, by a General Clinical Research Center grant (RR-00722), and by grants from the National Marfan Foundation, the American Heart Association (New York City affiliate), the Smilow Family Foundation, the Dr. Amy and James Elster Research Fund, and the British Heart Foundation.
We are indebted to Mrs. Evelyn Bull for expert technical assistance.
Source Information
From the Brookdale Center for Molecular Biology, Mount Sinai School of Medicine, New York (L.P., O.L., F.R.); the Institute of Molecular Medicine, University of Oxford, John Radcliffe Hospital, Headington, United Kingdom (J.R.L., B.S.); the Department of Human Genetics, Allegheny-Singer Research Institute, Pittsburgh (R.E.P.); and the Division of Pediatric Cardiology and the Center for Medical Genetics, Johns Hopkins University School of Medicine, Baltimore (H.C.D.).
Address reprint requests to Dr. Dietz at the Division of Pediatric Cardiology, Ross 1170, Johns Hopkins Hospital, 720 Rutland Ave., Baltimore, MD 21205.
References
-adrenergic blockade in Marfan's syndrome. N Engl J Med 1994;330:1335-1341. -globin genomic sequences and restriction site analysis for diagnosis of sickle cell anemia. Science 1985;230:1350-1354.
| |||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
