American Journal of Medical Genetics Part C (Semin. Med. Genet.) 121C:18 – 31 (2003) A R T I C L E Genetic Heritage of the Old Order Mennonites of Southeastern Pennsylvania E.G. PUFFENBERGER* The Old Order Mennonites of southeastern Pennsylvania are a religious isolate with origins in 16th-century Switzerland. The Swiss Mennonites immigrated to Pennsylvania over a 50-year period in the early 18th century. The history of this population in the United States provides insight into the increased incidence of several genetic diseases, most notably maple syrup urine disease (MSUD), Hirschsprung disease (HSCR), and congenital nephrotic syndrome. A comparison between the Old Order Mennonites and the Old Order Amish demonstrates the unique genetic heritage of each group despite a common religious and geographic history. Unexpectedly, several diseases in both groups demonstrate allelic and/or locus heterogeneity. The population genetics of the 1312T ! A BCKDHA gene mutation, which causes classical MSUD, are presented in detail. The incidence of MSUD in the Old Order Mennonites is estimated to be 1/358 births, yielding a corrected carrier frequency of 7.96% and a mutation allele frequency of 4.15%. Analysis of the population demonstrates that repeated cycles of sampling effects, population bottlenecks, and subsequent genetic drift were important in shaping the current allele frequencies. A linkage disequilibrium analysis of 1312T ! A mutation haplotypes is provided and discussed in the context of the known genealogical history of the population. Finally, data from microsatellite marker genotyping within the Old Order Mennonite population are provided that show a significant but modest decrease in genetic diversity and elevated levels of background linkage disequilibrium. ß 2003 Wiley-Liss, Inc. KEY WORDS: Old Order Mennonite; Old Order Amish; maple syrup urine disease; founder effect; genetic drift; linkage disequilibrium INTRODUCTION The terms Mennonite, Amish, and especially Anabaptist as used in genetic studies are often too broad to be meaningful. These terms denote shared social history that does not necessarily imply shared ancestry. The genealogical histories of many of these Plain sects are complex. Even within Pennsylvania, genealogical (and thus genetic) differences exist between neighboring groups of the same religious sect. The Clinic for Special Children, a nonprofit pediatric metabolic disease center, has patients from no fewer than E.G. Puffenberger, Ph.D., is Laboratory Director at the Clinic for Special Children, with special interest in molecular biology and population genetics. *Correspondence to: E.G. Puffenberger, Clinic for Special Children, 535 Bunker Hill Rd., Strasburg, PA 17579. E-mail: email@example.com DOI 10.1002/ajmg.c.20003 ß 2003 Wiley-Liss, Inc. The terms Mennonite, Amish, and especially Anabaptist as used in genetic studies are often too broad to be meaningful. These terms denote shared social history that does not necessarily imply shared ancestry. six different genetic isolates in Pennsylvania and Maryland; three are Old Order Amish and three are Old Order Mennonite. In some cases, these groups have distant genealogical ties to one another, but the proportion of shared ancestry varies widely. This paper will focus on the Old Order Mennonites of southeastern Pennsylvania and their derivative settlements in other states, with occasional reference to the Old Order Amish for comparative purposes. History of the Old Order Mennonites of Southeastern Pennsylvania The Old Order Mennonites of southeastern Pennsylvania are a religious isolate with origins in 16th-century Switzerland. In 1525, a small group of Protestants founded the Anabaptist movement. A major point of contention for these followers was the belief in adult baptism, a practice not followed by the Roman Catholic Church or the emerging Protestant churches. They held that baptism should be the voluntary act of an adult believer. In addition, they believed strongly in the separation of church and state. These and other theological issues placed the Anabaptists in direct conflict with both the church and state leaders [Redekop, 1989]. In the 1640s, forced exile from Canton Zurich and Schaffhausen resulted in the relocation of many Anabaptists to Canton Bern and Alsace. By the 1650s, many Anabaptists were moving into the Palatinate from Alsace and the cantons of Switzerland. By the 1670s, refugees ARTICLE from Bern were joining their brethren in the Palatinate [Davis, 1995, 1997]. Although the Palatinate provided a certain degree of tolerance, there were still restrictions placed upon the Anabaptists, including excessive taxes and limitations on worship and population size. The Anabaptist leaders of the time were searching for a place where they could live and worship freely. In the late 17th century, William Penn actively recruited Germanic peoples to immigrate to the New World. The first Germanic settlers arrived in Germantown, Pennsylvania, in 1683. During 1707–1757, a large migration of Swiss Mennonites took place [Redekop, 1989]. This included the ancestors of the present-day Old Order Mennonites of southeastern Pennsylvania. In the early years of the Swiss Mennonite migrations, individuals settled in the counties adjacent to Philadelphia, especially Bucks and Montgomery Counties. These Mennonites formed the core population of the Franconia Conference Mennonites. As tillable land became more scarce, immigrants progressively began settling farther west. The first Mennonite settlers in presentday Lancaster County (then known as the Conestoga settlement) arrived in 1710 [Davis, 1995]. Although no specific Mennonite census records exist for this period, it is estimated that several hundred Mennonite families settled in Lancaster County prior to the Revolutionary War. The Mennonite churches of the county were organized into districts under the leadership of a single governing body, the Lancaster Conference [Ruth, 2001]. Over the next 150 years, most growth in these communities was through reproduction, not migration, so many of these districts became isolated. The late 19th century was a turbulent time in Mennonite history. Mennonite conferences in several states were experiencing discontent among the laity. Several issues, such as increased evangelism, missionary work, Sunday school, English services, and higher education were polarizing Mennonite communities. This discontent culminated in AMERICAN JOURNAL OF MEDICAL GENETICS (SEMIN. MED. GENET.) schisms within several Mennonite conferences. The adherents to the old ways split away from the larger conference bodies and formed new Old Order conferences. This movement occurred first in Indiana and Ohio in 1872, followed by a schism in Ontario, Canada, in 1889. Eventually, this movement spread to the Lancaster Conference, the largest Old Mennonite Conference in the United States at the time. The schism in Lancaster County was principally led by Bishop Jonas Martin. The main issues were the increasing use of Sunday schools, English singing and preaching, and the marriage of couples who were not both members of the church. These changes were opposed by the conservative members of the Lancaster Conference. These troubles culminated at the fall conference of Lancaster bishops in 1893. Bishop Jonas Martin unambiguously expressed his opinion of the contentious issues. The other seven bishops conferred and decided that Bishop Martin had been too harsh and critical of the Lancaster Conference. They demanded that Bishop Martin recant, which he refused to do. In response, the bishops revoked his ministry and suspended his membership in the Lancaster Conference [Ruth, 2001]. Bishop Martin then established a new Old Order group called the Weaverland Conference. Following this 1893 split, the new Weaverland Conference Mennonites enjoyed an era of peace and growth. However, the introduction of the telephone, the automobile, and electricity strained the fledgling conference. The more conservative-minded members wished to ban automobile use, but there were an equal number who disagreed. In 1927, the Weaverland Conference officially split into the Weaverland and Groffdale Conferences. The former permitted the use of automobiles (blackbumper1 Old Order Mennonites), and the latter did not (horse-and-buggy Old Order Mennonites) [Scott, 1996]. 1 The term black bumper is used because the chrome trim of automobiles was considered an unnecessary adornment and was therefore painted black. 19 Old Order Mennonite Population Size and Growth Approximately 8,000 persons of SwissSouth German Mennonite background crossed the Atlantic from 1683–1880. This number includes about 3,000– 5,000 Swiss Mennonites who immigrated to Pennsylvania from 1707–1756 [Redekop, 1989]. As shown in Figure 1, a small percentage (probably several hundred) of these immigrants settled in Lancaster County. This provided the first significant bottleneck for the eventual Old Order Mennonites. For nearly 150 years, the church districts grew modestly in size, but regional genetic isolation arose due to low migration rates. The Lancaster Conference Mennonite population demonstrated steady growth during the latter half of the 18th century and the 19th century. By 1880, the Mennonite membership in the Lancaster Conference was estimated at 3,300 [Kraybill, 1987], and in 1892, the membership count was estimated at 6,500 [Ruth, 2001]. While neither number is wholly accurate, it is assumed that there were approximately 5,000 baptized members (adults) in the Lancaster Conference during this era. The Old Order movement in Lancaster County resulted in the formation of the Weaverland Conference in 1893. An estimate of the number of members who left the Lancaster Conference with Bishop Jonas Martin can be derived from the genealogies. By analyzing over 500 genealogies of extant members of the Weaverland and Groffdale Old Order Mennonites, it is estimated that about 125 families contributed most of the genetic diversity to the Weaverland Conference, half of which is attributable to 30 nuclear families. The newly formed conference was not a random sampling from the larger population. Bishop Martin and his followers were clustered in three adjoining church districts in northeastern Lancaster County, namely Bowmansville, Churchtown, and Groffdale. This is significant in that many of Bishop Martin’s followers were related due to regional and religious isolation. This nonrandom sampling of the Lancaster Conference resulted in another 20 AMERICAN JOURNAL OF MEDICAL GENETICS (SEMIN. MED. GENET.) ARTICLE Figure 1. Flow diagram depicting a brief population history of the Old Order Mennonites of southeastern Pennsylvania. bottleneck for the Old Order Mennonite population. As noted above, the Groffdale Conference split from the Weaverland Conference in 1927. This split was roughly even, although exact population numbers are not known. By 1940, there were roughly 1,078 members of the Weaverland Conference [A.B. Hoover, personal communication]. If estimates of the 1927 split are accurate, the total adult membership in 1940 for both conferences was about 2,160. Both conferences flourished in the 20th century. The Weaverland Mennonites established new settlements in five other states (Virginia, Missouri, New York, Wisconsin, and Iowa). The Groffdale Mennonites formed derivative settlements in Iowa, Ohio, New York, Missouri, Wisconsin, Michigan, Indiana, and Kentucky. Based on the Groffdale [Shirk and Shirk, 2002] and Weaverland [Wise and Martin, 2000] Conference directories, it is estimated that there are 26,500 Old Order Mennonites in Lancaster County and the distant settlements. Many late 17th- and 18th-century Mennonite settlements elsewhere in the United States were established by individuals migrating from established Mennonite settlements and Europe. Thus, these communities must be viewed as unique genetic isolates. The Old Order Mennonites in southeastern Pennsylvania comprise the oldest and largest Old Order Mennonite settlement in the country. It has been relatively immune to immigration and emigration for nearly 250 years. However, as the population in Lancaster County increased, families did migrate westward. During the mid and latter half of the 20th century, several new settlements were established exclusively by the Weaverland and Groffdale Mennonites. These new settlements had little or no admixture with other Mennonite demes. They were established in distant states (see above) and were formed by the movement of a large group of Old Order Mennonite families from the Lancaster area. Due to sampling effects and drift, these derivative settlements may experience different frequencies of genetic disease than the parent population. The Old Order Mennonites of the Weaverland and Groffdale Conferences are themselves not a homogeneous population. The Groffdale Conference membership directory includes a settlement in Indiana that is genetically and/ or genealogically distinct from the rest of the conference membership. This settlement contains the remains of the Ohio-Indiana Old Order Mennonites who split from their parent conference in 1872. Subsequently, a more conservative branch split off in 1907 under the leadership of Bishop John W. Martin. By 1970, the Ohio groups were extinct. In 1973, the remaining Indiana members officially merged with the Groffdale Conference, but still remain geographically and genetically isolated from the rest of the Groffdale Conference. There are subpopulations of individuals derived from the Weaverland and Groffdale Conferences, who have separated from the larger group. One notable example of this separation in Lancaster County is the Reidenbach Mennonites. Owing to disputes over the use of new technology and participation in Civilian Public Service camps during World War II, a conservative group of 35 individuals from the Groffdale Conference organized a separate church in 1946. In the mid-1990s, the group was estimated to contain 300 members [Scott, 1996]. ARTICLE Genetic Research in Isolated Populations Although the history of these groups is complex, an understanding of the structure of the population facilitates research into genetic diseases in these groups. The genetic mapping of disease loci within isolated populations offers many advantages not found in outbred populations, notably small population sizes that allow efficient data and sample collection as well as nearly complete ascertainment. The genetic mapping of disease loci within isolated populations offers many advantages not found in outbred populations, notably small population sizes that allow efficient data and sample collection as well as nearly complete ascertainment. Large nuclear families are frequent, which provides adequate numbers of affected and unaffected siblings within a sibship for sampling. Many isolates also keep excellent historical and genealogical records. Due to their sociologic and/or geographic isolation, there is usually little or no migration into the group. Finally, the members of the group exhibit relatively homogeneous lifestyles. The primary genetic advantage, however, results from the interaction of two overlapping phenomena: the founder effect and inbreeding. The genesis of an isolated population frequently entails a severe reduction in size, otherwise known as a bottleneck. Genetic drift may increase the frequency of one or more mutations introduced into the population by heterozygous founders. These phenomena, a population bottleneck accompanied by random genetic drift, are known as the founder effect. The founder effect is the major factor responsible for the relatively high inci- AMERICAN JOURNAL OF MEDICAL GENETICS (SEMIN. MED. GENET.) dence of genetic disease in isolated populations. This effect is exacerbated by inbreeding, which is an unintended consequence of genetic isolation. Many successful mapping studies have been performed in isolated and semi-isolated populations throughout the world. Some of the most well-known examples include both religious isolates, such as the Old Order Amish, Hutterites, and Ashkenazi Jews [Arcos-Burgos and Muenke, 2002], and geographic isolates, such as the populations of Finland [de la Chapelle and Wright, 1998; Peltonen et al., 1999] and Iceland [Jorde et al., 1982]. Genetic Research in the Old Order Mennonites of Southeastern Pennsylvania Although the Old Order Mennonites of southeastern Pennsylvania are not as well known as the Old Order Amish, they have similar population structures and genetic disease incidences. Early studies had identified an increased incidence of maple syrup urine disease (MSUD) [Marshall and DiGeorge, 1981] and Hirschsprung disease (HSCR) [Cohen and Gadd, 1982] in this population. The first mutation identified in this population was the 1312T !A mutation in the BCKDHA gene causing MSUD [Zhang et al., 1989]. This was followed by the mutations for HSCR [Puffenberger et al., 1994a], glycogen storage disease type 6 [Chang et al., 1998], and congenital nephrotic syndrome [Bolk et al., 1999]. Over the past several years, the Clinic for Special Children has elucidated the molecular basis of additional metabolic disorders and conducted research to better understand the behavior of these mutations in isolated populations such as the Plain sects of southeastern Pennsylvania. MATERIALS AND METHODS Samples All samples used for sequencing and microsatellite marker analyses were ac- 21 quired from patients and their families at the Clinic for Special Children. The clinic follows 56 Mennonite MSUD patients (44 Old Order Mennonites). In order to perform the microsatellite marker studies, DNA samples were collected from 24 Old Order Mennonite probands from 24 separate sibships. In addition, 27 unrelated (greater than second-degree relatives to probands) 1312T !A heterozygotes were also genotyped. In order to set phase and provide control (i.e., untransmitted or U) allele frequencies, at least one parent was genotyped for all affected and heterozygous individuals. We have ascertained several Old Order Mennonite probands in Ontario, Canada, a settlement founded by multiple migrations from Lancaster, Bucks, Berks, and Montgomery Counties in Pennsylvania [Bergey, 1992]. One affected MSUD individual from this group was included in our analysis. Genealogies Detailed genealogies for all mutation heterozygotes and homozygotes were prepared from private and published family records. The initial genealogical research was performed utilizing telephone interviews and several exhaustive family histories. Additional research was performed at the Lancaster Mennonite Historical Society, Lancaster, PA. All genealogical data were entered into the MennGen database, which utilizes the computer program Reunion 7.0 (Leister Productions, Inc.). This database was part of the effort to map the gene(s) for HSCR in the Old Order Mennonites [Puffenberger et al., 1994a, 1994b] and originally contained about 8,000 Old Order Mennonite individuals. The database has been expanded to include all individuals in the latest editions of the Weaverland and Groffdale Conference directories [Wise and Martin, 2000; Shirk and Shirk, 2002] and the Stauffer Mennonites [Sensenig and Sensenig, 1998] and now contains over 90,000 individuals. Data were exported to a HewlettPackard workstation for statistical manipulations. 22 AMERICAN JOURNAL OF MEDICAL GENETICS (SEMIN. MED. GENET.) ARTICLE Genotyping Sequencing Genomic DNA was extracted from 13 ml of whole blood using the Puregene DNA Isolation Kit (Gentra Systems). Microsatellite marker loci were amplified by polymerase chain reaction (PCR) using 30–50 ng of DNA and fluorescently labeled forward primers. Reaction volumes were 10 ml and included 1 unit of Taq polymerase (Qiagen), 200 mM each of dATP, dCTP, dGTP, and dTTP, and 1 ml of 10 incubation buffer. Samples were amplified in a PerkinElmer 480 thermocycler for 30 cycles (30 sec at 968C, 10 sec at 558C, 30 sec at 728C), followed by an 8-min incubation at 728C. Genotypes were determined using an ABI 310 Genetic Analyzer and the GeneScan software package. Allele sizes were determined from Centre d’Etudes du Polymorphisme Humaine (CEPH) control individual 1347-02 (when genotype known) and GeneScan-500 DNA size standards. Mutation analysis was performed for the following candidate genes: ACADM, HPD, HSD3B2, MTHFR, MVK, PAH, PCCB, SLC3A1, SLC7A9, and TJP2. Targeted mutation detection was performed for all other mutations listed in Tables I and II. The exons of the target genes were amplified by PCR using specific oligonucleotide primers and 30–50 ng of genomic DNA. Reaction volumes were 25 ml and included 1 unit of Taq polymerase (Qiagen), 200 mM each of dATP, dCTP, dGTP, and dTTP, and 2.5 ml of 10 incubation buffer. Samples were amplified in a PerkinElmer 480 thermocycler for 30 cycles (30 sec at 968C, 10 sec at 608C, 30 sec at 728C) followed by an 8-min incubation at 728C. The PCR products were then sequenced using a fluorescence-based cycle sequencing protocol (BigDye Terminator, Applied Biosystems). The extension products were subsequently size-fractionated on an ABI 310 Genetic Analyzer. The sample sequence was compared to the normal mRNA and genomic sequence for each gene from GenBank in order to identify sequence variants. For this analysis, exonic sequence (from the initiator codon to the termination codon) as well as the splice donor, splice acceptor, and branch point sites were screened for mutations. These analyses did not screen for mutations in the promoter region, the 30 untranslated region, or the introns of these genes (except as described above). Assessment of Linkage Disequilibrium In this inbred Mennonite pedigree, it is difficult to accurately assess the control allele frequencies for the population, and the allele frequencies in Genome Database (GDB) (http://www.gdb.org/) may not be applicable. The presumably TABLE I. Molecular Lesions Identified in the Old Order Mennonites of Southeastern PA Disease Congenital nephrotic syndrome Cystinuria Crigler-Najjar syndrome Fragile X syndrome Glycogen storage disease, type 6 Hirschsprung disease Maple syrup urine disease Medium chain dehydrogenase deficiency Mevalonate kinase deficiency Phenylketonuria Propionic acidemia Spinal muscular atrophy Tyrosinemia, type 3 3-methylcrotonylglycinuria a Gene Mutation Mutation reference ID in Mennonitesa NPHS1 NPHS1 SLC3A1 SLC3A1 SLC7A9 SLC7A9 UGT1A1 FMR1 PYGL EDNRB BCKDHA ACADM 1481delC 3250delG IVS6 þ 2T ! C 1354C ! T 200C ! T 1166C ! T 222C ! A (CGG)n expansion IVS13 þ 1G ! A 828G ! T 1312T !A 985A ! G Bolk et al.  Bolk et al.  Novel mutation Endsley et al.  Putative mutation? Novel mutation Kadakol et al.  Fu et al.  Chang et al.  Puffenberger et al. [1994a] Zhang et al.  Matsubara et al.  Sameb Sameb Clinic for Special Children Clinic for Special Children Clinic for Special Children Clinic for Special Children Clinic for Special Children Clinic for Special Children Sameb Same Same Clinic for Special Children ACADM MVK MVK PAH PAH PCCB SMN1 HPD MCCB IVS4-30A ! G 803T ! C 1174G ! A 782G ! A IVS10-11G ! A 1606A ! G exon 7 deletion 85G ! A 518insT Novel mutation Hinson et al.  Novel mutation Abadie et al.  Dworniczak et al.  Gravel et al.  Lefebvre et al.  Novel mutation Baumgartner et al.  Clinic for Special Children Sameb Clinic for Special Children Clinic for Special Children Clinic for Special Children Clinic for Special Children Clinic for Special Children Clinic for Special Children Sameb Mutation identification and/or verification in the Old Order Mennonites of southeastern Pennsylvania. Collaborative study with the Clinic for Special Children, Strasburg, PA. b ARTICLE AMERICAN JOURNAL OF MEDICAL GENETICS (SEMIN. MED. GENET.) 23 TABLE II. Molecular Lesions Identified in Old Order Amish Demes of Pennsylvania Disease Aldosterone deficiency Amish microcephaly Byler diseasec Cartilage-hair hypoplasia Crigler-Najjar syndrome Ellis-van Creveld syndrome Familial hypercholanemia Galactosemiad Gutaric aciduria, type 1 Homocystinuria McKusick-Kauffman syndrome Nemaline rod myopathy Osteogenesis imperfecta Phenylketonuria Propionic acidemiae Pyruvate kinase deficiencye Sitosterolemia 3-ß-OH-steroid dehydrogenase deficiency 3-methylcrotonylglycinuria Gene Mutation ID in Amisha Mutation reference CYP11B2 SLC25A19 ATP8B1 RMRP UGT1A1 EVC TJP2 BAAT GALT GCDH MTHFR MKKS TNNT1 COL1A2 PAH PAH PCCB PKLR ABCG8 HSD3B2 5 bp deletion 530G ! C 923G ! T 70A ! G 222C ! A IVS13 þ 5G ! T 143T ! C 226A ! G 563A ! G 1262C ! T 1129C ! T [250C ! T þ 724G ! T] 505G ! T 2098G ! T 280-282delATC 782G ! A 1606A ! G 1436G ! A 1720G ! A 35G ! A Mitsuuchi et al.  Rosenberg et al.  Bull et al.  Ridanpaa et al.  Kadakol et al.  Ruiz-Perez et al.  Carlton et al.  Carlton et al.  Reichardt et al.  Biery et al.  Goyette et al.  Stone et al.  Johnston et al.  McBride et al.  Caillaud et al.  Abadie et al.  Gravel et al.  Kanno et al.  Berge et al.  Novel mutation Sameb Sameb Sameb Clinic for Clinic for Same Sameb Sameb NeoGen Sameb Clinic for Same Sameb Same Clinic for Clinic for Clinic for Same Same Clinic for MCCB 295G ! C Baumgartner et al.  Sameb Special Children Special Children Special Children Special Children Special Children Special Children Special Children a Mutation identification and/or verification in Old Order Amish demes of Pennsylvania. Collaborative study with the Clinic for Special Children, Strasburg, PA. c Patients identified in the Old Order Amish of Lancaster County, PA and the Old Order Amish of Mifflin and Juniata Counties, PA. d No mutation homozygotes identified yet. e Patients identified in the Old Order Amish of Mifflin and Juniata Counties, PA. b normal, untransmitted parental chromosomes were used as controls. To test for linkage disequilibrium arising from descent from a common ancestor, we determined whether marker alleles had a higher frequency on mutant gene-bearing (transmitted, T) chromosomes vs. normal (untransmitted, U) parental chromosomes. A 2 2 contingency X2 statistic was employed to test the null hypothesis of equal allele frequency on T and U chromosomes for each allele at a marker, as previously described (Puffenberger et al., 1994a, 1994b). For estimating the average number of ancestral recombinants (y) between a mutation and marker locus, we assume that marker haplotypes evolve by recombination. If the frequency of the associated allele (AA) is y on mutant (T) chromosomes and x on normal (U) chromosomes, and a is the proportion of mutant chromo- somes attributable to a specific mutation, then y ¼ x þ að1xÞey The average number (y) is estimated from this equation by using the estimated values of y and x and assuming homogeneity (all mutations at a locus of single origin, a ¼ 1); conversely, for tight linkage (y ¼ 0), a can be estimated. The number of generations from the common founder (g) was estimated using a modified version of the previous formula: y ¼ x þ að1xÞð1yÞg The recombination fraction (y) for the estimation of g was based on the Marshfield integrated genetic map of chromosome 19 (http://research. marshfieldclinic.org/genetics/). Estimates of the distance from the mutation (in megabases, Mb) for each marker were derived from contig physical maps at National Center for Biotechnology Information (NCBI) (http://www.ncbi. nlm.nih.gov/) and the University of California, Santa Cruz (UCSC) Genome Builds (http://genome.ucsc.edu/). In addition to allele frequency data, haplotypes were constructed and haplotype frequencies were compared between affected individuals and untransmitted haplotypes from their parents. The association tests for allele and haplotype frequency distortions were performed on genotype data from either one homozygote or one heterozygote per nuclear family. In multicase families, only one affected individual was utilized in the enumeration of alleles and haplotypes. This procedure was implemented to avoid allele and haplotype frequency inflation. All allele and haplotype 24 AMERICAN JOURNAL OF MEDICAL GENETICS (SEMIN. MED. GENET.) analyses were performed on a Power Macintosh G3 using the custom program MacAllele (developed by Kashuk, Puffenberger, and Chakravarti, 1994). RESULTS AND DISCUSSION Molecular Lesions in the Old Order Mennonites of Southeastern Pennsylvania As described elsewhere in this issue, the Clinic for Special Children has been serving the Old Order Amish and Old Order Mennonite communities since 1989. Recently, the clinic has developed a program to identify mutations segregating in these populations for use in diagnostic and carrier testing. Collaborative efforts between the clinic and other researchers have led to the identification of 12 mutations segregating in these two isolated populations [Biery et al., 1996; Bull et al., 1998; Chang et al., 1998; Bolk et al., 1999; Hinson et al., 1999; Johnston et al., 2000; Baumgartner et al., 2001; Rosenberg et al., 2002; Carlton et al., 2003]. In addition, the clinic has independently identified 7 novel mutations and 20 previously published mutations in 20 additional disorders found in the Plain sects of Pennsylvania (Tables I and II). It is commonly assumed that small genetic isolates have little or no mutation heterogeneity. Due to the founder effect, a single mutation is postulated to segregate in the population and account for the increased incidence of disease. As Tables I and II demonstrate, molecular data show that this may not be correct. Five of 14 disorders in the Old Order Mennonites show mutation heterogeneity (Table I). This is Five of 14 disorders in the Old Order Mennonites show mutation heterogeneity. most striking for cystinuria, where allelic and locus heterogeneity has been identified by sequencing the SLC3A1 and SLC7A9 genes in five patients. In the Old Order Amish, two disorders demonstrate mutation heterogeneity, one of which also demonstrates locus heterogeneity. The assumption of mutation homogeneity can hinder mapping studies of disease loci. While attempting to identify the gene for HSCR among the Old Order Mennonites, we identified a cluster of HSCR cases in an associated Mennonite settlement in Ontario, Canada [Puffenberger et al., 1994a]. This group, the MarkhamWaterloo Conference Mennonites, was formed by multiple migrations from Lancaster, Montgomery, and Bucks Counties, Pennsylvania, during the early 19th century. We surmised that the gene(s) for HSCR would be the same in both groups owing to shared ancestry. However, upon identification of the causative mutation, it was found that the Canadian patients did not harbor the same mutation as the Weaverland and Groffdale Mennonites. Genealogical analysis of this Canadian group shows that the genetic contribution from the Lancaster Mennonites was about 40%. The molecular basis of four genetic diseases that occur in both Old Order Amish and Old Order Mennonites has been elucidated. Prior to identification of the causative mutation, it was widely believed that these disorders would have a common origin in the two populations. This was based on the knowledge that the Old Order Amish were a splinter group from the Swiss Mennonites and shared some ancestors with the Old Order Mennonites. This was partially correct as Crigler-Najjar syndrome and propionic acidemia are caused by the same mutation in both populations. In contrast, 3-methylcrotonylglycinuria involves a different mutation for each group. Finally, phenylketonuria is common to both groups with one shared mutation and two population-specific mutations. The identification of causative mutations in these populations does not imply that the mutation is in high frequency. Most of the early mapping studies focused on disorders that demonstrated a high incidence in these populations. In recent years, the Clinic for Special Children has attempted to define ARTICLE the molecular bases of disease in these populations regardless of the frequency. Thus, some mutations are represented by a single patient, as is the case for tyrosinemia type 3. Pennsylvania state newborn screening identified the patient, who subsequently was seen at our clinic at 26 days of age. Amino acid analysis confirmed the diagnosis of tyrosinemia, although the biochemical and clinical phenotype was inconsistent with type 1 or type 2 disease. By sequencing genomic DNA, the patient was found to be homozygous for a novel mutation in the HPD gene. The identification of these mutations regardless of their frequency permits the effective use of diagnostic molecular genetic testing in these populations. For several of these disorders, the differential diagnosis is large, requiring expensive testing and sometimes hospitalization. The ability to accurately diagnose the patient in a timely manner is invaluable for preventing the devastating effects of metabolic disease. Population Genetics of MSUD MSUD was the first published disease phenotype reported in the Old Order Mennonites of southeastern Pennsylvania [Marshall and DiGeorge, 1981]. Eight years later, the causative mutation, 1312T !A, was identified [Zhang et al., 1989]. The unique features of the Old Order Mennonite population, the high disease incidence, and the extensive carrier testing performed at the Clinic for Special Children afforded the opportunity to study the population genetics of the mutation in detail. Although disease incidence and mutation allele frequency are correlated, the mechanism(s) by which mutation frequencies change may be obscure. Figure 2 presents a flowchart of the MSUD mutation allele frequency in the parental Swiss-German population, the founding Mennonite population, and the extant Old Order Mennonite population. As the first panel demonstrates, the estimated incidence of MSUD in the general population is 1/200,000 births, yielding a heterozygote frequency of roughly 1/225 and a mutation allele frequency of 0.22%. ARTICLE AMERICAN JOURNAL OF MEDICAL GENETICS (SEMIN. MED. GENET.) Figure 2. Flow diagram depicting historical phenomena that influenced the 1312T !A mutation allele frequency in the Old Order Mennonites of southeastern Pennsylvania. It is hypothesized that two related events affected the MSUD mutation allele frequency in the fledgling Lancaster Mennonite settlement. First, a random sampling effect increased the mutation allele frequency relative to the parental population. This occurred through the introduction of a single, rare mutation by a heterozygous ancestor into the small founder group. Second, the sampling effect dramatically decreased mutation heterogeneity such that only one MSUD mutation (1312T !A) was found in the derivative group. It is widely believed that the phenomenon with the greatest effect on the mutation allele frequency in small genetic isolates is the founder effect. The effects of genetic drift are most dramatic when the population is small. As the population size increases, the ability of genetic drift to affect mutation allele frequency diminishes. Thus, the first few generations of Mennonites in Lancaster County likely experienced the greatest changes in mutation allele frequency. Importantly, the Old Order Mennonite population has experienced several significant bottlenecks in the past 300 years (Fig. 1). Thus, repeated cycles of sampling effects, population bottlenecks, and subsequent genetic drift were undoubtedly important in shaping the current allele frequencies within the population. Incidence of MSUD The Clinic for Special Children undertook an effort to update and confirm the reported incidence of MSUD in the Old Order Mennonites for the period 25 1985–1994. During this 10-year interval, there were 6,810 total Old Order Mennonite births. Among this cohort, there were 19 MSUD children born. This yielded an incidence of 19/6,810 (0.28%), or 1/358. Interestingly, the incidence of MSUD in the two conferences differs. Of the 19 MSUD children enumerated above, 15 were Groffdale Mennonites and 4 were Weaverland Mennonites. This yields conferencespecific MSUD incidence rates of 1/ 271 and 1/686, respectively. These estimates were calculated from the birth records for the entire Weaverland and Groffdale Conference populations. There were seven additional cases of MSUD in Mennonite children during this period, but they were excluded because their families belong to Mennonite conferences where population size, birth rates, and admixture are unknown. In some derivative Weaverland and Groffdale settlements, the incidence may be higher or lower based on sampling effects and genetic drift. An example is the Missouri settlements, where the incidence was 4/603 births or 0.66% [Love-Gregory et al., 2002]. This population-based estimate is less than a previous calculation by a factor of two [Marshall and DiGeorge, 1981]. However, this earlier calculation was performed before conference directories were available, thus making it difficult to assess the total population size or the birth rate. It is also probable that some Mennonite patients were enumerated whose families are not, or never have been, members of the Old Order churches. We are aware of at least 16 additional Mennonite MSUD patients whose families are not affiliated with the Old Order Mennonite churches. Allelic and Genotypic Frequencies of the 1312T !A Mutation By using the Hardy-Weinberg equilibrium and the incidence estimate, we can calculate the mutation and carrier frequency in this population. The mutation allele frequency in this population (q) is estimated from the square root of the incidence (q2). Thus, the 1312T !A 26 AMERICAN JOURNAL OF MEDICAL GENETICS (SEMIN. MED. GENET.) allele frequency is approximately 5.28%. The heterozygote (i.e., carrier) frequency is then easily calculated as 2(1q)(q), which yields an estimate of 10.01%. This estimate of the heterozygote frequency assumes random mating and no inbreeding within the population. However, pedigree analysis of Old Order Mennonite families clearly demonstrates significant levels of inbreeding. The principal consequence of inbreeding within a population is to increase the proportion of homozygous genotypes and reduce the fraction of heterozygous genotypes. Thus, the heterozygote frequency within an inbred population will be lower than the heterozygote frequency in a population with the same disease incidence, but no inbreeding. Thus, the previous calculation of the 1312T !A heterozygote frequency is undoubtedly an overestimate. Based on the MSUD incidence and the average inbreeding coefficient for MSUD families, a more precise estimate of the 1312T !A heterozygote frequency in the Old Order Mennonites can be derived. Genotype frequencies in the presence of inbreeding [Wright, 1921] were calculated based on the formulas in Table III. Although a population-wide average inbreeding coefficient has not been calculated for the Old Order Mennonites, inbreeding coefficients were calculated for all Old Order Mennonite MSUD sibships. The average inbreeding coefficient was 2.19%. Although a population-wide average inbreeding coefficient has not been calculated for the Old Order Mennonites, inbreeding coefficients were calculated for all Old Order Mennonite MSUD sibships. The average inbreeding coefficient was 2.19%. ARTICLE TABLE III. Effect of Inbreeding (F) on the 1312T !A Mutation Frequency Estimate F¼0 Mutation homozygotes Mutation heterozygotes Wild-type homozygotes Mutation allele frequency 2 q 2pq 2 ppﬃﬃﬃﬃﬃ q2 0.28% 10.01% 89.71% 5.28% F ¼ 2.19% 2 q (1F) þ qF 2pq (1F) 2 p (1F) þ pF ﬃ pﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ q2 ð1FÞ þ qF 0.28% 7.96% 91.87% 4.15% For the period 1985–1994, there were 19 Old Order Mennonite children born with maple syrup urine disease. The total number of Old Order Mennonite births for this same period was 6810; thus, the incidence of MSUD was 19/6810 ¼ 0.2790% (or roughly 1/ 358 births). This inbreeding coefficient is a reasonable approximation for the Old Order Mennonite population as a whole. By using this approximation of inbreeding, the corrected population heterozygote frequency was 7.96%. Genetic Drift Analogy: Surname Frequency Distribution Table III presents an example of random genetic drift in the Old Order Mennonite population. A surname frequency distribution was assembled using the two conference directories. The most common surname in both conferences is Martin, comprising 20% of Old Order Mennonite households. There is no known selective advantage to carrying this surname, but its frequency has risen from about 1–2% in the early 18th century. The surname distribution further illustrates the differential effects of random genetic drift in the two conferences. The second most common surname, Zimmerman, demonstrates a striking frequency difference between the two conferences. This difference may have occurred through random genetic drift, nonrandom segregation of the surname during the 1927 split, or a combination of the two factors. The distribution of several genetic diseases within the two conferences demonstrates this phenomenon as well. In addition to MSUD, the incidence of congenital nephrotic syndrome is decidedly unequal between the two conferences. The incidence in the Groffdale Mennonites is roughly 1/500 [Bolk et al., 1999]. To date, only one affected child has been born in the Weaverland Mennonites. Although separated for only three generations and similar genealogically, random sampling and genetic drift have operated differently in each subpopulation. These data show that the incidence of a trait (i.e., surname) may increase in the absence of inbreeding (Table IV). However, within natural populations, random genetic drift and inbreeding are overlapping phenomena that act together to cause the increased disease incidence in genetic isolates. Random genetic drift may lead to an increased mutation allele frequency within a population, as shown for MSUD in the Old Order Mennonites; conversely, it is equally probable for an allele to decrease in frequency or become extinct. Within the Old Order Mennonites, no cases of cystic fibrosis (CF) have ever been reported. In a population of 27,000, we would expect to find roughly 11 CF patients based on an incidence of 1/2,500 births. There are three possible explanations for this deficiency: 1) mutations in CFTR may never have been introduced into the population, 2) genetic drift has resulted in the extinction of any CFTR mutations, or 3) the frequency of CFTR mutations is so low that affected individuals have not been ascertained. Genealogical Analysis of the 1312T !A Mutation The MennGen genealogical database (see Materials and Methods) was used ARTICLE AMERICAN JOURNAL OF MEDICAL GENETICS (SEMIN. MED. GENET.) TABLE IV. Surname Distribution in the Weaverland and Groffdale Conference Mennonites Surname Total Weaverland Groffdale Martin Zimmerman Hoover Nolt Weaver 19.6% 14.1% 7.1% 6.0% 5.9% 20.0% 9.5% 5.8% 4.8% 6.7% 19.3% 17.8% 8.0% 7.0% 5.2% The top five surnames account for 52.7% of households. There were 92 additional surnames with frequencies <5.0%. Totals were based on head-of-household counts from the Weaverland [Wise and Martin, 2000] and Groffdale [Shirk and Shirk, 2002] Conference Directories. to perform pedigree analysis of 366 Mennonite 1312T !A heterozygotes to identify the putative common ancestor. Two ancestral couples were identified: Abraham Herr (1672–1725) and his wife, Anna Bear, and Hans Groff (1661–1746) and his wife, Susanna Kendig. Assuming a single origin for the mutation in this population, one of these four individuals was the original carrier of the 1312T !A mutation. It is remarkable that analysis of 366 separate pedigrees does not allow conclusive identification of the common ancestor for the MSUD mutation. This underscores the interrelatedness of extant Old Order Mennonites. Linkage Disequilibrium Analysis of the 1312T !A Mutation Most genetic mapping studies rely on linkage analysis to localize the altered gene. This parametric approach is particularly useful when mapping rare diseases in unrelated, outbred families. Since mutation heterogeneity is nearly certain, this method is preferred. It provides the ability to analyze cosegregation of marker alleles with the phenotype within sibships and calculate a test statistic based on the cumulative contribution of each family. In isolated populations, an alternative method is linkage disequilibrium analysis. This nonparametric technique involves the assessment of association between a phenotype and alleles or haplotypes at genetic marker loci. Association analysis detects those genomic regions that are identical by descent in affected individuals due to common ancestry and is ideally suited to mapping studies in genetic isolates. It is also technically and computationally simpler than linkage analysis. Ironically, the merits of linkage disequilibrium analysis have been touted for years, yet only two published mapping studies performed in the Old Order Amish and Old Order Mennonite populations from Lancaster County have utilized this technique [Puffenberger et al., 1994a, 1994b; Carlton et al., 1995]. Linkage disequilibrium analysis can also provide information about the behavior of mutations within populations. It can be used to examine the age of mutations within a population, the evolution of haplotypes, and the segregation of those haplotypes in the population. The haplotypes provide a historical record of recombinational events and thus provide interesting data on population age and structure. In order to examine these issues, a study of linkage disequilibrium surrounding the 1312T !A mutation in the Old Order Mennonite population was performed. Twelve microsatellite markers flanking the BCKDHA locus on human chromosome 19q13 were chosen for their proximity to the gene and heterozygosity values. Twenty-six separate mutation homozygotes and 25 heterozygotes were genotyped. When avail- 27 able, one or more parents were also genotyped in order to set phase and provide necessary population-specific allele frequencies. Multiple sampling of nearly identical haplotypes was minimized by analyzing one 1312T !A haplotype per sibship (n ¼ 51). The total number of independent haplotypes sampled was 77. While none of the haplotypes were truly independent due to identity by descent for the 1312T !A mutation, samples were excluded from analysis when a first- or second-degree relative of the proband or their parents was already present in the genotyping panel. A recent analysis determined the frequency of alleles for 5 microsatellite markers on 46 Old Order Mennonite and 10 non-Mennonite 1312T !A haplotpyes [Love-Gregory et al., 2002]. Unfortunately, these data suffer from multiple sampling, predominantly from only four sibships, of identical or nearly identical haplotypes. Thus, those allele or haplotype frequencies could not be used for this study of linkage disequilibrium. To analyze linkage disequilibrium, it was determined whether alleles on the transmitted, mutation-bearing haplotypes (T) were present at higher frequency than on presumably normal, untransmitted haplotypes (U) in the parents, as previously described [Puffenberger et al., 1994a, 1994b]. A Chisquare test was employed to test the null hypothesis of equal allele frequencies on T and U chromosomes for each allele at the microsatellite marker locus. The associated allele (AA) at each marker locus is presented with the corresponding frequency on T and U chromosomes (Table V). As expected, the markers physically closest to the BCKDHA gene demonstrated the greatest association, and D19S198 was the only locus where no recombinants were detected. This marker locus is physically the closest to the 1312T !A mutation, roughly 426 kb distal to BCKDHA. Although no recombinants were detected with the 145-bp allele at this locus on 77 mutant haplotypes, that allele is the most common allele in control chromosomes as well. Thus, historical recombinational events may 28 AMERICAN JOURNAL OF MEDICAL GENETICS (SEMIN. MED. GENET.) ARTICLE TABLE V. Association of Chromosome 19q13 Markers and the BCKDHA 1312T !A Mutation on Old Order Mennonite Haplotypes Marker D19S431 D19S248 D19S713 D19S421 D19S718 D19S223 D19S400 1312T !A D19S198 D19S420 D19S178 APOC2 D19S246 AA (bp) T% U% T(n) U(n) X2 y Mb g cM g 292 129 218 187 368 233 188 A 145 261 152 147 231 82 62 72 88 94 95 93 100 100 99 77 74 61 52 3 16 56 16 0 3 0 34 31 7 15 36 76 71 74 73 72 73 76 77 77 67 74 70 36 80 69 80 71 76 77 78 82 80 64 76 68 14 14.8 55.4 48.1 17.6 92.1 134.8 127.5 159.0 77.0 65.7 77.7 49.5 2.6 0.491 0.497 0.219 0.332 0.068 0.056 0.070 0.000 0.000 0.022 0.278 0.359 0.929 9.60 7.95 4.14 3.06 0.55 0.51 0.38 0 0.25 1.91 2.51 3.55 8.94 4.9 6.0 5.2 10.7 12.5 11.0 18.5 0.0 0.0 1.1 10.9 9.9 9.9 10.66 6.65 2.24 3.22 1.33 1.18 0.64 0 0.43 0.96 2.74 4.16 12.74 4.4 7.2 9.7 10.1 5.1 4.7 10.8 0.0 0.0 2.3 10.0 8.5 6.8 have occurred, but these events had a 33.8% chance of replacing the 145-bp allele with another 145-bp allele. The data demonstrate the recombinational decay of the ancestral haplotype. As distance from the mutation increases, association with a particular allele at each marker locus decreases. However, the size of the region that demonstrates statistically significant linkage disequilibrium is large, spanning roughly 14.8 cM around the 1312T !A mutation. A prior study of HSCR in the Old Order Mennonites identified a similarly sized region of 12.9 cM on chromosome 13 [Puffenberger et al., 1994a]. Allelic associations were detected over large genomic regions because the Old Order Mennonite population is relatively young. The extant population can trace their genealogies back to a small group of founding families roughly 10–12 generations ago. If this population were older (e.g., 50 generations), the shared region would be considerably smaller. The microsatellite marker data provide a related insight into allelic associations (Table V). The greatest statistical significance occurs when the AA is in high frequency on transmitted haplotypes (T) and low frequency on untransmitted haplotypes (U). The most extreme example of this phenomenon is the mutation itself, where the fre- quency is 100% on T chromosomes (by definition) and 0% on U chromosomes. The microstallite marker D19S223 is remarkable in that the associated 233-bp allele is found on 94.5% of T chromosomes, but was not detected on any U chromosomes. A similar association between this microsatellite marker and 1312T !A mutation haplotypes was shown previously [Love-Gregory et al., 2002]. These data allow us to calculate the average number of recombinational events (y) that have occurred between the mutation and the marker locus since the introduction of the mutation into the population. This estimate is based on the frequency of the AA on T and U chromosomes (see Materials and Methods). This calculation estimates the average number of recombinational events that likely occurred to produce the allele frequency distribution observed on T and U chromosomes. It accounts for the fact that the identical allele might occasionally be transferred through recombination onto the haplotype. This occurs with a frequency equal to the allele frequency on U chromosomes. This estimation of q provides a relative measure of the distance between the genetic marker and the mutation. A close examination of the data in Table V shows a discrepancy between the orders of the markers. The physical map places D19S713 proximal to D19S421, while the genetic map has the order reversed. The estimation of y is in agreement with that of the genetic map. When the allele frequencies for T and U chromosomes as well as the distance from the mutation are known, the average number of generations back to the common ancestor can be calculated. This calculation was performed using the T and U AA frequencies and the distance between the marker and the mutation. This was calculated for both physical distance (Mb) and genetic distance (cM). Using the physical distances (Mb), the estimation of g for each locus yields an average of 9.2 generations for all nonzero g values. The same estimate using the genetic distances (cM) yields a generation time of 7.2. These estimates are slightly low, but are not inconsistent with the known population history. This low estimate probably arose due to the occurrence of a bottleneck since the founding of the Mennonite settlement in Lancaster County. The 1893 Old Order schism likely reduced the diversity of recombinant 1312T !A mutation-bearing haplotypes in the derivative Old Order population. This sampling effect would provide an underestimate of the generation time to the common ancestor since some rare recombinants would not be found in the splinter group. ARTICLE The 1312T !A Mutation in Non-Old Order Mennonite MSUD The Clinic for Special Children has been performing MSUD mutation identification as a clinical service for several years. We have sequenced the BCKDHA, BCKDHB, and/or DBT genes in 22 MSUD patients and have identified the causative mutation on 43 of 44 alleles. We have found two patients who were heterozygous for the 1312T !A mutation in BCKDHA on chromosome 19q13. This yields a 1312T !A allele frequency of 4.5% for non-Old Order Mennonite MSUD (2/44). In these two 1312T !A heterozygotes, we confirmed a common origin for the 1312T !A mutation. By direct sequencing of the BCKDHA gene, we identified and analyzed 19 intragenic single nucleotide polymorphisms (SNPs). These SNPs revealed identical intragenic haplotypes for the 1312T !A mutation in homozygous Old Order Mennonite patients and the two heterozygous non-Mennonite patients (data not shown). In order to estimate the generation time from the common ancestor for the 1312T !A mutation for non-Old Order Mennonite patients, we calculated g using our two patients and the eight probands from the article by Love-Gregory et al. . The estimate of g is based on only two microsatellite markers that were genotyped in both data sets, D19S223 and D19S178. However, both loci demonstrated significant linkage disequilibrium in our study, and the AA was in low frequency on control (U) chromosomes. The combined data for D19S223 and D19S178 revealed an AA frequency of 41.7% for both loci (5/12) on nonOld Order Mennonite haplotypes. The estimates of g using D19S223 were 171.6 and 73.5 generations (average ¼ 122.6) for physical and genetic distances, respectively. Likewise, the estimates for D19S178 were 38.6 and 35.3 generations (average ¼ 37.0). While the sample sizes are small, these data are consistent with a common ancestor for non-Old Order Mennonite haplotypes that predates the formation of the Lancaster AMERICAN JOURNAL OF MEDICAL GENETICS (SEMIN. MED. GENET.) Mennonite settlement. This does not preclude the possibility that one or more of these haplotypes were derived from Mennonite ancestors, but the majority of these haplotypes demonstrate recombination events consistent with a deeper pedigree than the Lancaster Mennonite population can provide. Genetic Diversity There is widespread expectation that genetic variation is reduced in any genetically isolated population, particularly those of recent origins such as the Old Order Mennonites. One measure of genetic diversity was studied in Old Order Mennonite families with children affected with HSCR by analysis of 515 autosomal and 24 X-linked microsatellite markers in 40 chromosomes [Chakravarti et al., 1997]. These data were compared to 40 chromosomes from the Utah set of CEPH families for control purposes. Genetic variation, as measured by heterozygosity (h) at microsatellite marker loci, is significantly reduced in the Old Order Mennonites on autosomes (Mennonite h ¼ 74.9% 0.4% vs. CEPH h ¼ 76.8% 0.4%), but not on the X chromosome (Mennonite h ¼ 78.2% 1.6% vs. CEPH h ¼ 77.4% 1.9%). This represents a 1.9% reduction in heterozygosity for microsatellite markers on Old Order Mennonite autosomes. This figure closely approximates the calculated inbreeding coefficient for Old Order Mennonite HSCR families (1.2%). Since inbreeding creates a fractional reduction in heterozygous genotypes within a population, the correlation between the reduction in heterozygosity and inbreeding was anticipated. The additional reduction in heterozygosity is likely due to random genetic drift, which has undoubtedly led to the extinction of some alleles in the population. The loss of these alleles resulted in less diversity and lower heterozygosity values. Although the reduction in heterozygosity is statistically significant, the decrease in genetic variation was modest, and thus the population remains genetically diverse. 29 CONCLUSIONS The history of Old Order Mennonites of southeastern Pennsylvania provides insight into the increased incidence of genetic disease within isolated populations. It is commonly assumed that genetic isolates are founded by a small number of individuals and the population size increases over subsequent generations with few perturbations. The population history of the Old Order Mennonites does not fit such a simple model. An analysis of their history It is commonly assumed that genetic isolates are founded by a small number of individuals and the population size increases over subsequent generations with few perturbations. The population history of the Old Order Mennonites does not fit such a simple model. in the United States indicates multiple population bottlenecks, long-term genetic and geographic isolation, and uneven growth over the past 250 years (see Fig. 1). Due to the small number of founders and the rarity of most genetic diseases, it was assumed that isolated populations would show mutation homogeneity. The data presented here indicate that this assumption is not valid. This has important implications for the mapping of complex traits in isolated populations: the allelic diversity of these populations should not be underestimated. Although genetic drift and inbreeding are overlapping phenomena that increase the frequency of genetic disease, it is a common and popular belief that inbreeding is the major reason for elevated levels of genetic disease in isolated populations. The data on the 1312T !A mutation and the Old Order Mennonite surname distribution show that random 30 AMERICAN JOURNAL OF MEDICAL GENETICS (SEMIN. MED. GENET.) genetic drift had a major effect on the incidence of genetic disease. Inbreeding The data on the 1312T !A mutation and the Old Order Mennonite surname distribution show that random genetic drift had a major effect on the incidence of genetic disease. exacerbated this phenomenon by decreasing the frequency of heterozygous genotypes (and increasing homozygous genotypes). These two factors worked in concert to produce the increased incidence of specific diseases. While genetic diversity is postulated to be dramatically reduced in genetic isolates, the genotype data show that the reduction was modest. Much of the decrease in heterozygosity within the population can be explained by inbreeding. Since inbreeding only affects genotype frequencies, the allelic diversity is not truly lost, but rather partitioned into homozygous genotypes at the expense of heterozygous genotypes. Finally, the utility of linkage disequilibrium analyses to detect disease genes in isolated populations has been affirmed by this work. It was possible to detect association between a phenotype and marker alleles over large genomic regions in a young genetic isolate such as the Old Order Mennonites. This nonparametric technique is computationally simple, requires no assumptions about mode of inheritance, and obviates the need for detailed and comprehensive genealogical analyses. REFERENCES Abadie V, Lyonnet S, Maurin N, Berthelon M, Caillaud C, Giraud F, Mattei JF, Rey J, Rey F, Munnich A. 1989. CpG dinucleotides are mutation hot spots in phenylketonuria. Genomics 5:936–939. Arcos-Burgos M, Muenke M. 2002. Genetics of population isolates. Clin Genet 61:233– 247. Baumgartner MR, Almashanu S, Suormala T, Obie C, Cole RN, Packman S, Baumgartner ER, Valle D. 2001. The molecular basis of human 3-methylcrotonyl-CoA carboxylase deficiency. J Clin Invest 107:495–504. Berge KE, Tian H, Graf GA, Yu L, Grishin NV, Schultz J, Kwiterovich P, Shan B, Barnes R, Hobbs HH. 2000. Accumulation of dietary cholesterol in sitosterolemia caused by mutations in adjacent ABC transporters. Science 290:1771–1775. Bergey LL. 1992. The early settlement of Waterloo Township, Ontario, Canada. Pennsylvania Mennonite Heritage 15:9–20. Biery BJ, Stein DE, Morton DH, Goodman SI. 1996. Gene structure and mutations of glutaryl-coenzyme A dehydrogenase: impaired association of enzyme subunits that is due to an A421V substitution causes glutaric acidemia type I in the Amish. Am J Hum Genet 59:1006–1011. Bolk S, Puffenberger EG, Hudson J, Morton DH, Chakravarti A. 1999. Elevated frequency and allelic heterogeneity of congenital nephrotic syndrome, Finnish type, in the Old Order Mennonites. Am J Hum Genet 65:1785–1790. Bull LN, van Eijk MJ, Pawlikowska L, DeYoung JA, Juijn JA, Liao M, Klomp LW, Lomri N, Berger R, Scharschmidt BF, Knisely AS, Houwen RH, Freimer NB. 1998. A gene encoding a P-type ATPase mutated in two forms of hereditary cholestasis. Nat Genet 18:219–224. Caillaud C, Lyonnet S, Rey F, Melle D, Frebourg T, Berthelon M, Vilarinho L, Vaz Osorio R, Rey J, Munnich A. 1991. A 3-base pair inframe deletion of the phenylalanine hydroxylase gene results in a kinetic variant of phenylketonuria. J Biol Chem 266:9351– 9354. Carlton VE, Knisely AS, Freimer NB. 1995. Mapping of a locus for progressive familial intrahepatic cholestasis (Byler disease) to 18q21-q22, the benign recurrent intrahepatic cholestasis region. Hum Mol Genet 4:1049–1053. Carlton VEH, Harris BZ, Puffenberger EG, Batta AK, Knisely AS, Robinson DL, Strauss KA, Shneider BL, Lim WA, Salen G, Morton DH, Bull LN. 2003. Complex inheritance of familial hypercholanemia with associated mutations in TJP2 and BAAT. Nat Genet 34:91–96. Chakravarti A, Kashuk C, Puffenberger EG. 1997. Genetic variation in an isolated population: the Old Order Mennonites of eastern Pennsylvania. Am J Hum Genet 61:7971A. Chang S, Rosenberg MJ, Morton H, Francomano CA, Biesecker LG. 1998. Identification of a mutation in liver glycogen phosphorylase in glycogen storage disease type VI. Hum Mol Genet 7:865–870. Cohen IT, Gadd MA. 1982. Hirschsprung’s disease in a kindred: a possible clue to the genetics of the disease. J Pediatr Surg 17:632–634. Davis RW. 1995. Emigrants, refugees, and prisoners: I. Provo, UT: Richard Warren Davis. 427 p. Davis RW. 1997. Emigrants, refugees, and prisoners: II. Provo, UT: Richard Warren Davis. 446 p. de la Chapelle A, Wright FA. 1998. Linkage disequilibrium mapping in isolated populations: the example of Finland revisited. Proc Natl Acad Sci USA 95:12416–12423. ARTICLE Dworniczak B, Aulehla-Scholz C, Kalaydjieva L, Bartholome K, Grudda K, Horst J. 1991. Aberrant splicing of phenylalanine hydroxylase mRNA: the major cause for phenylketonuria in parts of southern Europe. Genomics 11:242–246. Endsley JK, Phillips JA 3rd, Hruska KA, Denneberg T, Carlson J, George AL Jr. 1997. Genomic organization of a human cystine transporter gene (SLC3A1) and identification of novel mutations causing cystinuria. Kidney Int 51:1893–1899. Fu Y-H, Kuhl DPA, Pizzuti A, Pieretti M, Sutcliffe JS, Richards S, Verkerk AJMH, Holden JJ, Fenwick RG Jr, Warren ST, Oostra BA, Nelson DL, Caskey CT. 1991. Variation of the CGG repeat at the fragile X site results in genetic instability: resolution of the Sherman paradox. Cell 67:1047–1058. Goyette P, Christensen B, Rosenblatt DS, Rozen R. 1996. Severe and mild mutations in cis for the methylenetetrahydrofolate reductase (MTHFR) gene, and description of five novel mutations in MTHFR. Am J Hum Genet 59:1268–1275. Gravel RA, Akerman BR, Lamhonwah AM, Loyer M, Leon-del-Rio A, Italiano I. 1994. Mutations participating in interallelic complementation in propionic acidemia. Am J Hum Genet 55:51–58. Hinson DD, Ross RM, Krisans S, Shaw JL, Kozich V, Rolland MO, Divry P, Mancini J, Hoffmann GF, Gibson KM. 1999. Identification of a mutation cluster in mevalonate kinase deficiency, including a new mutation in a patient of Mennonite ancestry. Am J Hum Genet 65:327–335. Johnston JJ, Kelley RI, Crawford TO, Morton DH, Agarwala R, Koch T, Schaffer AA, Francomano CA, Biesecker LG. 2000. A novel nemaline myopathy in the Amish caused by a mutation in troponin T1. Am J Hum Genet 67:814–821. Jorde LB, Eriksson AW, Morgan K, Workman PL. 1982. The genetic structure of Iceland. Hum Hered 32:1–7. Kadakol A, Ghosh SS, Sappal BS, Sharma G, Chowdhury JR, Chowdhury NR. 2000. Genetic lesions of bilirubin uridine-diphosphoglucuronate glucuronosyltransferase (UGT1A1) causing Crigler-Najjar and Gilbert syndromes: correlation of genotype to phenotype. Hum Mutat 16:297–306. Kanno H, Ballas SK, Miwa S, Fujii H, Bowman HS. 1994. Molecular abnormality of erythrocyte pyruvate kinase deficiency in the Amish. Blood 83:2311–2316. Kraybill DB. 1987. At the crossroads of modernity: Amish, Mennonites, and Brethren in Lancaster County in 1880. Pennsylvania Mennonite Heritage 10:2–12. Lefebvre S, Burglen L, Reboullet S, Clermont O, Burlet P, Viollet L, Benichou B, Cruaud C, Millasseau P, Zeviani M, Le Paslier D, Frezal J, Cohen D, Weissenbach J, Munnich A, Melki J. 1995. Identification and characterization of a spinal muscular atrophy-determining gene. Cell 80:155–165. Love-Gregory LD, Grasela J, Hillman RE, Phillips CL. 2002. Evidence of common ancestry for the maple syrup urine disease (MSUD) Y438N allele in non-Mennonite MSUD patients. Mol Genet Metab 75: 79–90. ARTICLE Marshall L, DiGeorge A. 1981. Maple syrup urine disease in the old order Mennonites. Am J Hum Genet 33:139A. Matsubara Y, Narisawa K, Miyabayashi S, Tada K, Coates PM, Bachmann C, Elsas LJ 2nd, Pollitt RJ, Rhead WJ, Roe CR. 1990. Identification of a common mutation in patients with medium-chain acyl-CoA dehydrogenase deficiency. Biochem Biophys Res Commun 171:498–505. McBride D, Streeten EA, Mitchell BD, Shuldiner AR. 2002. Variable expressivity of a COL1A2 gly-610-cys mutation in a large Amish pedigree. Am J Hum Genet 71(Abstr 1047):351. Mitsuuchi Y, Kawamoto T, Miyahara K, Ulick S, Morton DH, Naiki Y, Kuribayashi I, Toda K, Hara T, Orii T, Yasuda K, Miura K, Yamamoto Y, Imura H, Shizuta Y. 1993. Congenitally defective aldosterone biosynthesis in humans: inactivation of the P-450(C18) gene (CYP11B2) due to nucleotide deletion in CMO I deficient patients. Biochem Biophys Res Commun 190:864–869. Peltonen L, Jalanko A, Varilo T. 1999. Molecular genetics of the Finnish disease heritage. Hum Mol Genet 8:1913–1923. Puffenberger EG, Hosoda K, Washington SS, Nakao K, deWit D, Yanagisawa M, Chakravart A. 1994a. A missense mutation of the endothelin-B receptor gene in multigenic Hirschsprung’s disease. Cell 79:1257–1266. AMERICAN JOURNAL OF MEDICAL GENETICS (SEMIN. MED. GENET.) Puffenberger EG, Kauffman ER, Bolk S, Matise TC, Washington SS, Angrist M, Weissenbach J, Garver KL, Mascari M, Ladda R, et al. 1994b. Identity-by-descent and association mapping of a recessive gene for Hirschsprung disease on human chromosome 13q22. Hum Mol Genet 3:1217– 1225. Redekop C. 1989. Mennonite Society. Baltimore: Johns Hopkins University Press. 397 p. Reichardt JK, Packman S, Woo SL. 1991. Molecular characterization of two galactosemia mutations: correlation of mutations with highly conserved domains in galactose1-phosphate uridyl transferase. Am J Hum Genet 49:860–867. Ridanpaa M, van Eenennaam H, Pelin K, Chadwick R, Johnson C, Yuan B, vanVenrooij W, Pruijn G, Salmela R, Rockas S, Makitie O, Kaitila I, de la Chapelle A. 2001. Mutations in the RNA component of RNase MRP cause a pleiotropic human disease, cartilage-hair hypoplasia. Cell 104:195–203. Rosenberg MJ, Agarwala R, Bouffard G, Davis J, Fiermonte G, Hilliard MS, Koch T, Kalikin LM, Makalowska I, Morton DH, Petty EM, Weber JL, Palmieri F, Kelley RI, Schaffer AA, Biesecker LG. 2002. Mutant deoxynucleotide carrier is associated with congenital microcephaly. Nat Genet 32:175–179. Ruiz-Perez VL, Ide SE, Strom TM, Lorenz B, Wilson D, Woods K, King L, Francomano 31 C, Freisinger P, Spranger S, Marino B, Dallapiccola B, Wright M, Meitinger T, Polymeropoulos MH, Goodship J. 2000. Mutations in a new gene in Ellis-van Creveld syndrome and Weyers acrodental dysostosis. Nat Genet 24:283–286. Ruth JL. 2001. The earth is the Lord’s: a narrative history of the Lancaster Mennonite Conference. Scottdale, PA: Herald Press. 1390 p. Scott SE. 1996. An introduction to old order and conservative Mennonite groups. Intercourse, PA: Good Books. 252 p. Sensenig P, Sensenig E. 1998. Records of the members of the Stauffer Mennonite church. Shirk LN, Shirk BN. 2002. Directory of the Groffdale Conference Mennonite churches. Stone DL, Slavotinek A, Bouffard GG, BanerjeeBasu S, Baxevanis AD, Barr M, Biesecker LG. 2000. Mutation of a gene encoding a putative chaperonin causes McKusickKaufman syndrome. Nat Genet 25: 79–82. Wise RA, Martin LH. 2000. Directory of the Weaverland Conference Mennonite churches. Wright S. 1921. Systems of mating, I–V. Genetics 6:111–178. Zhang B, Edenberg HJ, Crabb DW, Harris RA. 1989. Evidence for both a regulatory mutation and a structural mutation in a family with maple syrup urine disease. J Clin Invest 83:1425–1429.