¹Department of Psychiatry, University of Pittsburgh, Pittsburgh, PA, USA

²Department of Statistics, Carnegie Mellon University, Pittsburgh, PA, USA

³Department of Psychiatry, University of Utah Medical Center, Salt Lake City, Utah, USA

⁴Department of Psychiatry, University of California, Irvine, CA, USA

⁵Department of Genetics, University of Pittsburgh, Pittsburgh, PA, USA

⁶Belau National Hospital, Korror, Palau

Correspondence to: W Byerley, Department of Psychiatry, University of California, Irvine, CA 92697-4260, USA. E-mail: wbyerley@uci.edu

The oceanic nation of Palau has been geographically and culturally isolated over most of its 2000 year history. As part of a study of the genetic basis of schizophrenia in Palau, we genotyped five large, multigenerational schizophrenia pedigrees using markers every 10 cM (CHLC/Weber screening set 6). The number of affected/unaffected individuals genotyped per family ranged from 11/21 to 5/5. Thus the pedigrees varied in their information for linkage, but each was capable of producing a substantial LOD score. We fitted a simple dominant and recessive model to these data using multipoint linkage analysis implemented by Simwalk2. Predictably, the most informative pedigrees produced the best linkage results. After genotyping additional markers in the region, one pedigree produced a LOD = 3.4 (5q distal) under the dominant model. Seven of nine schizophrenics in the pedigree, mostly 3rd-4th degree relatives, share a 15-cM, 7-marker haplotype. For a different pedigree, another promising signal occurred on distal 3q, LOD = 2.6, for the recessive model. For two other pedigrees, the best LODs were modest, slightly better than 2.0 on 5q and 9p, while the fifth pedigree produced no noteworthy linkage signal. Similar to the results for other populations, our results suggest there are multiple genes conferring liability to schizophrenia even in the small population of Palau (roughly 21 000 individuals) in remote Oceania.

Molecular Psychiatry (2002) 7, 689-694. doi:10.1038/sj.mp.4001056

remote Oceania; haplotypes; genome scan; genetic isolate

Introduction

Schizophrenia is a complex neuropsychiatric disorder that causes untold suffering and disability. Remarkably, despite differing cultures and environments around the world, schizophrenia consistently afflicts roughly one percent of all sizeable populations.^1,2 Results of family, twin, and adoption studies suggest that genetic factors account for at least 60-70% of liability. These studies also indicate that schizophrenia is a complex genetic disease, characterized by reduced penetrance, non-Mendelian transmission, phenocopies and locus heterogeneity. Indeed, by studying populations of European origins,^{3,4,5,6,7,8,9,10,11,12,13,14,15,16,17} linkage analyses of schizophrenia pedigrees yield evidence for loci predisposing to schizophrenia on chromosomes 1q, 5q, 6p, 6q, 8p, 10p, 13q and 22q.^{9,10,11,12,13,14,15,16,17}

Selection of pedigrees from highly-structured populations is another important tool for mapping genes affecting liability to schizophrenia. In some instances, genes having only a small population attributable risk in open-mating populations will confer a much greater risk in highly-structured populations. In addition, the structure itself, in the form of linkage disequilibrium, can be exploited to map and fine-map disease genes.^18,19 Other advantages of isolated populations include the frequent presence of large, extended pedigrees and centralized treatment facilities for finding such pedigrees.

The peoples of the Republic of Palau present a good example of a small, somewhat isolated population. The islands comprising Palau are scattered over 125 miles of the South Pacific, quite far from any large land mass. Carbon dating suggests Palau was first populated about 2000 years ago.²⁰ By 230 years ago, the population had grown to about 20 000.²¹ Then, by 100 years ago, epidemics originating from American and European contact reduced the population to 4000. Now the population stands at about 21 000, more than half of whom are over 15 years old. Linguistic analyses and ethnographic studies suggest the Palauan population developed in relative isolation, even from other Micronesian populations;²² nonetheless, this population shows evidence of immigration from surrounding populations.^23,24

As part of our ongoing study of Palauan schizophrenia, 154 of the 156 schizophrenics living in Palau were ascertained along with key relatives. From this unique population we also gathered extensive genealogical records. Previously²⁵ we reported genome-wide linkage results for one, large multiplex schizophrenia Palauan pedigree. Assuming autosomal dominant inheritance and a strict definition of illness, the highest two-point lod scores were found for loci mapping to 2p13-14, with a maximum lod score of 2.17 near D2S441. Genome-wide multipoint linkage analysis was not performed on this pedigree because, at that time, genome-wide multipoint analyses of large pedigrees were too challenging.

In this report we perform genome-wide multipoint linkage analysis on this previously reported kindred, K1583, and four other large multiplex Palauan kindreds: K1584, K2034, K2238c K2364a. To implement the multipoint analyses, we used Simwalk2.

Methods

Following informed consent, subjects were interviewed by one psychiatrist (William Byerley), who used a modified version of the SADS-L semi-structure interview.^26,27 Medical records were available for all cases and they were reviewed and summarized. Information from Schedule for Affective Disorders and Schizophrenia - Lifetime version (SADS-L) interviews, hospital records, and family histories were used to derive consensual Research Diagnostic Criteria diagnoses by two psychiatrists (Fred Reimherr and Paul Wender) blind to the genotypic status of individuals. Only cases with chronic schizophrenia or chronic schizoaffective disorder, mainly schizophrenic course (ie, schizophrenia only), were considered for linkage. All other cases were coded unaffected.

DNA was isolated using phenol/chloroform extraction methods as previously described.²⁸ Genotyping was performed using the Cooperative Human Linkage Center²⁹ microsatellite markers (Weber Version 6) spaced, on average, every 10 cM throughout the genome. For three regions showing promising LOD scores for two-point analyses of K1583, namely 2p13-14, 5q telomeric, and 1q, we genotyped additional microsatellite markers, creating a roughly 2 cM grid in those regions.

PCR was carried out using standard cycling conditions and the resulting fragments were size fractionated using 6% acrylamide sequencing gels. Genotypes were recorded by two technicians without knowledge of phenotypic status. Genotypes that could not be resolved by consulting a third reader were recorded unknown, or the sample was re-genotyped if it were critically informative to a region of interest.

We evaluated markers and pedigrees for Mendelian errors using the Pedcheck program.³⁰ Genotyping errors were set to missing. Allele frequencies were estimated by direct count over kindreds, ignoring familial relationships.

We chose two simple models for these analyses: a dominant model in which disease allele frequency was 0.01 and penetrances were 0.05, 0.5 and 0.5 for 0, 1 and 2 disease alleles; and a recessive model in which disease allele frequency was 0.12 and penetrances were 0.05, 0.05 and 0.75. By these models, about 1/20 schizophrenics in a family trace the origins of their disease to sources other than the specified liability locus, and about 45% of those carrying a dominant or two recessive disease alleles will be affected. Because almost all schizophrenics were affected as young adults, age-dependent penetrances were ignored, as were sex-dependent effects.

To lend intuition regarding the informativeness of these pedigrees, we also performed a very simple analysis to approximate the maximum achievable LOD, by pedigree, for an affected-only analysis. To make this calculation, we included only the affecteds who were genotyped, and assumed they all shared an allele identical-by-descent (IBD) at a disease locus, which was completely linked to a fully informative marker. Linkage analysis of these modified pedigrees yielded the maximum achievable LOD score for an affected-only analysis.³¹

Linkage analyses were implemented using Simwalk2,³² version 2.60, which performs multipoint analysis of highly complex pedigrees by using stochastic techniques based on descent graphs.^32,33 Before beginning the linkage analyses we studied the convergence properties of the software as applied to complex pedigrees, by using multiple starting configurations and various numbers of repetitions of the Monte Carlo Markov Chain procedure to determine how many iterations were necessary to obtain reliable results (unpublished data). Larger chromosomes were broken up into units of 6-8 markers to promote faster convergence and thus acceptable computing times. Although it was potentially feasible to analyze full chromosomes simultaneously, we observed that the amount of information gained by analyzing whole chromosomes was small compared to the computational effort required to obtain satisfactory convergence results for larger chromosomal segments. To refine LOD score traces in specific regions of interest, we ran up to 15 markers in and surrounding the region. Simwalk2 analyses were performed on two Beowulf clusters, each with 16 processors running at 500 (University of Pittsburgh) or 600 (Carnegie Mellon University) MHz.

LOD scores were computed only for individual pedigrees for several reasons: we expect some linkage heterogeneity, even in the Palauan population; we analyze only a small subset of the Palauan families; and the performance of linkage heterogeneity models for a small number of pedigrees and a complex disease is unpredictable.

Results

Table 1 gives the attributes of these pedigrees, including the median kinship of the schizophrenics in the pedigrees and the maximum achievable LOD score for an affected-only analysis (see Methods). After genotyping additional markers in the 5q distal region, K1583 yields a LOD = 3.4 under the dominant model (Figure 1). Seven of nine schizophrenics in the pedigree, mostly 3rd-4th degree relatives, share a 15-cM, 7-marker haplotype near UT6397. No other kindreds show a strong linkage signal in this exact region, although the best linkage signal for K2238c is also on 5q near D5S1480, LOD = 2.1 (Figure 1). This signal is slightly more proximal than the signal from K1583 and the shared haplotypes within each family do not overlap. Thus, there could be multiple liability loci on 5q.

The second largest signal for K1583 occurs on Chromosome 2p, LOD = 2.2 near D2S393, for the recessive model. This signal corresponds to that reported in Coon et al²⁵ for two point analysis. Again this signal reflects extensive haplotype-sharing, in this case more than half of the schizophrenics share a 25 cM haplotype.

For K2364a, the best signal is found on distal 3q, at D3S1311, LOD = 2.6 for the recessive model (Figure 1). For the same region, the dominant model produces a LOD = 2.2. Both of these results are obtained from markers in the original 10 cM set, and additional markers have not yet been genotyped in this region. No other signals are especially noteworthy.

For the remaining two pedigrees, the best LODs are modest (Figure 1). The best signal for K2024 occurs on 9p at D9S921, LOD = 2.0 for the dominant model, while the highest LOD for K1584 is 1.18 on 12q, near D12S1294.

We also examine the entire distribution of LOD scores to determine if there were any evidence for too much mass in the upper tail, a feature indicative of invalid model assumptions (eg, affected individuals substantially more related than that specified by the pedigree, as might be expected for a population isolate). There is no evidence for excess mass (Figure 2) over that expected if the pedigree contains zero or one schizophrenia-liability genes. On the other hand, the differential informativeness of the pedigrees stands out: under the assumed dominant model, linkage for a substantial portion of the genome is excluded for K1583 but little of the genome is excluded for K1584 or K2024.

Discussion

Our most promising LOD score, 3.4 (Figure 1), was obtained from K1583, and occurred on Chromosome 5q in the telomeric region near UT6397 (194.9 cM, according to the Marshfield map (http://research.marshfieldclinic.org/genetics/)). Kindred 2238c produced a weaker signal somewhat more proximal on 5q, LOD = 2.1 (Figure 1), near D5S1480 (147.5 cM). The haplotypes shared among affected individuals within each kindred do not overlap across kindreds; therefore, using reasoning for simple genetic disorders, the distinct haplotype-sharing implies there are distinct schizophrenia genes (if any) for the two kindreds. On the other hand, given the complexity of the disease and the possibility of phenocopies, this conclusion could be premature.

The implicated region for K2238c, with its peak at D5S1480, overlaps with that reported previously by Straub et al⁷ and Schwab et al.¹⁰ Although Straub et al⁷ did not report haplotype data, positive LOD scores were observed from D5S815 (101.0 cM) to D5S658 (142.9 cM). The highest LOD scores (approximately 3-3.4)¾depending on the genetic model¾were observed with D5S393 (140.7 cM) and D5S804 (133.7 cM). Schwab et al¹⁰ reported positive linkage data from D5S666 (135.3 cM) to D5S658 (142.9 cM). The most positive scores were with D5S399 (140.7 cM), which yielded a LOD score of 1.8. Unfortunately none of the markers we used coincided with the markers yielding peak LOD scores for the Straub et al⁷ or Schwab et al¹⁰ studies. Recently, Paunio et al³⁴ and Gurling et al³⁵ reported positive linkage data (LOD scores of approximately 3-3.6) in this chromosomal area as well. Because several groups using unique pedigrees have implicated this region, it should receive priority for candidate gene mutation searches.

In an earlier report, Coon et al²⁵ identified two promising regions on the basis of two-point linkage analysis of K1583 under a dominant model, namely 2p and 1q. Our multipoint linkage analysis did not support the 1q finding: LOD scores were less than 0.3 everywhere on 1q. Multipoint linkage analysis continued to lend some support to 2p, with LOD = 2.2 near D2S393 (80.2cM) under a dominant model. In the earlier analysis,²⁵ the maximum two-point LOD score (LOD = 2.17) occurred at D2S441, roughly 6 cM away from D2S393.

If the other four pedigrees we analyzed shared susceptibility alleles at the same locus or loci as K1583, then linkage signals should be amplified by evaluating these pedigrees. For Chromosome 5q, the LOD scores (dominant model) summed over families showed a weak peak at UT6397 (LOD = 0.94) and were strongly negative at D5S1480 (LOD = -4.8). For either marker, no family other than K1583 and K2238c, demonstrated a LOD score exceeding 0.2.

For Chromosome 2p region, three out of four LOD scores for families other than K1583 were negative, with the one positive score being 0.2. This result stands in contrast to the results of Camp et al,³⁶ who reported a LOD = 4.8 for an overlapping set of seven Palauan pedigrees. It is worth noting, however, that Coon et al²⁵ analyzed three STR markers in this region from 17 Palauan families, which includes all the families analyzed by Camp et al³⁶ and this study, and found only modest support for linkage, even under a heterogeneity model. Thus, while the 2p region in the vicinity of D2S393 merits further study, a common disease mutation in this region shared by most Palauan families is unlikely.

Putting our results in perspective, the best linkage signals occur for the most informative families (Table 1), namely K1583, K2364a and K2238c, but only K1583 produced a very strong linkage signal. Furthermore, the overlap of noteworthy multipoint LOD scores across kindreds is not substantial. These observations suggest the origins of schizophrenia in Palauans are heterogeneous. Such complexity, even in a relatively isolated island population, is not especially surprising when viewed in the historical perspective of schizophrenia research (Hovatta et al 1999)³⁷ and that of other complex diseases. In fact, three genes for the rare recessive disorder known as Bardet-Biedl syndrome have been found in the isolated Bedouin-Arab population.^38,39

Our long-term plans are to use the unique population structure of Palau to explore the genetic etiology of this devastating illness. Because the population of Palau is of recent origin, has undergone at least two substantial bottlenecks, has had a small effective population size throughout its history, and has apparently experienced extensive male-biased gene flow,²⁴ linkage disequilibrium among alleles on the same chromosome is detectable even at substantial distances.²⁴ Thus, in our estimation, the Palauan population is amenable to both linkage and disequilibrium mapping, which we hope will yield some of the genetic mutations conferring liability to schizophrenia in this Oceanic population. This research will be facilitated by a 10 cM genome scan of 154 Palauan schizophrenics, as well as additional family members, which is being performed by the Center for Inherited Disease Research.

Acknowledgements

Research supported by NIMH grants MH57881 to BD and KR and MH56098 to BB. We thank Marina Myles-Worsley for some genealogical information.

1 Gottesman II. Schizophrenia Genesis: the Origins of Madness WH Freeman: New York, 1991.

2 Jablensky A, Sartorius N, Ernberg G, Anker M, Korten A, Cooper JE et al. Schizophrenia: manifestations, incidence and course in different cultures. Psychol Med 1992; 22 [Monograph Suppl 20]: 1-97.

3 Coon H, Jensen S, Holik J, Hoff M, Myles-Worsley M, Reimherr F et al. Genomic scan for genes predisposing to schizophrenia. Am J Med Genet 1994; 54: 59-71. MEDLINE

4 Pulver AE, Karayiorgou M, Wolyniec PS, Lasseter VK, Kasch L, Nestadt G et al. Sequential strategy to identify a susceptibility gene for schizophrenia: report of potential linkage on chromosome 22q12-q13.1: Part 1. Am J Med Genet 1994; 54: 36-43. MEDLINE

5 Pulver AE, Lasseter VK, Kasch L, Wolyniec P, Nestadt G, Blouin JL et al. Schizophrenia: a genome scan targets chromosomes 3p and 8p as potential sites of susceptibility genes. Am J Med Genet (Neuropsych Genet) 1995; 60: 252-260.

6 Straub RE, MacLean CJ, O'Neill A, Burke J, Murphy B, Duke F et al. A potential vulnerability locus for schizophrenia on chromosome 6p24-22, evidence for genetic heterogeneity. Nat Genet 1995; 11: 287-293. MEDLINE

7 Straub RE, MacLean CJ, O'Neill FA, Walsh D, Kendler KS. Support for a possible vulnerability locus in region 5q22-31 in Irish families. Mol Psychiatry 1997; 2: 148-155. MEDLINE

8 Straub RE, MacLean C, Martin RB, Yunlong M, Myakisehev V, Harris-Kerr C et al. A schizophrenia locus may be located in region 10p15-p11. Neuropsychiatric Genet 1998; 81: 296-301.

9 Schwab SG, Albus M, Hallmayer J, Honig S, Bormann M, Lichtermann D et al. Evaluation of a susceptibility gene for schizophrenia on chromosome 6p by multipoint affected sib-pair linkage analysis. Nat Genet 1995; 11: 325-327. MEDLINE

10 Schwab SG, Eckstein GN, Hallmayer J, Lerer B, Albus M, Borrmann M et al. Evidence suggestive of a locus on chromosome 5q31 contributing to susceptibility for schizophrenia in German and Israeli families by multipoint affected sib-pair linkage analysis. Mol Psychiatry 1997; 2: 156-160. MEDLINE

11 Schwab SG, Hallmayer J, Albus M, Lerer B, Hanses C, Kanyas K et al. Further evidence for a susceptibility locus on chromosomse 10p14-p11 in 72 families with schizophrenia by nonparametric analysis. Neuropsychiatric Genet 1998; 81: 302-307.

12 Karayiorgou M, Morris MA, Morrow B, Shprintzen RJ, Goldberg R, Borrow J et al. Schizophrenia susceptibility associated with interstitial deletions of chromosome 22q11. Proc Natl Acad Sci USA 1995; 92: 7612-7616. MEDLINE

13 Moises HW, Yang L, Kristbjarnarson H, Wiese C, Byerley W, Macciardi F et al. An international two-stage genome-wide search for schizophrenia susceptibility genes. Nat Genet 1995; 11: 321-324. MEDLINE

14 Wang S, Sun C, Walczak CA, Ziegle JS, Kipps BR, Goldin LR et al. Evidence for a susceptibility locus for schizophrenia on chromosome 6pter-p22. Nat Genet 1995; 10: 41-46. MEDLINE

15 Faraone SV, Matise T, Svrakie D, Pepple J, Malaspina D, Suarez B et al. Genome scan of European-American schizophrenia pedigrees: Results of the NIMH Genetics Initiative and Millennium Consortium. Neuropsychiatric Genet 1998; 81: 290-295.

16 Blouin JL, Dombroski BA, Nath SK, Lasseter VK, Wolyniec PS, Nestadt G et al. Schizophrenia susceptibility loci on chromosomes 13q32 and 8p21. Nat Genet 1998; 20: 70-73. Article MEDLINE

17 Brzustowicz LM, Hodgkinson KA, Chow EW, Honer WG, Bassett AS. Location of a major susceptibility locus for familial schizophrenia on chromosome 1q21-q22. Science 2000; 288: 678-682. Article MEDLINE

18 Jorde LB. Linkage disequilibrium as a gene-mapping tool. Am J Hum Genet 1995; 56: 11-14. MEDLINE

19 Jorde LB. Linkage disequilibrium and the search for complex disease genes. Genome Res 2000; 10: 1435-1444. Article MEDLINE

20 Takayama J. Early pottery and population movements in Micronesian prehistory. Asian Perspectives 1981; 24: 1-10.

21 Parmentier RJ. The Sacred Remains, Myth, History and Polity in Belau Univ of Chicago Press: Chicago, 1987.

22 Simmons R, Graydon J, Gajdusek D, Brown P. Blood group genetic variations in natives of the Caroline Islands and in other parts of Micronesia. Oceania 1965; 36: 132-170.

23 Lum JK, Cann RL. mtDNA lineage analyses: origins and migrations of Micronesians and Polynesians. Amer J Phys Anthro 2000; 113: 151-168.

24 Devlin B, Roeder K, Otto C, Tiobech S, Byerley B. Genome-wide distribution of linkage disequilibrium in the population of Palau and its implications for gene flow in remote Oceania. Hum Genet 2001; 108: 521-528. MEDLINE

25 Coon H, Myles-Worsley M, Tiobech J, Hoff M, Rosenthal J, Bennett P et al. Evidence for a chromosome 2p13-14 schizophrenia susceptibility locus in families from Palau, Micronesia. Mol Psychiatry 1999; 3: 521-527.

26 Endicott J, Spitzer RL. A diagnostic interview: the schedule for affective disorders and schizophrenia. Arch Gen Psychiatry 1978; 35: 837-844. MEDLINE

27 Spitzer RL, Endicott J, Robins E. Research diagnostic criteria. Arch Gen Psychiatry 1978; 35: 773-782. MEDLINE

28 Bell G, Karam J, Rutter W. Polymorphic DNA region adjacent to the 5' end of the human insulin gene. Proc Nat Acad Sci USA 1981; 78: 5759-5763. MEDLINE

29 Cooperative Human Linkage Center (CHLC): Genethon, University of Utah, Yale University, and Centre d'Etude du Polymorphisme Humain. A comprehensive human linkage map with centiMorgan density. Science 1994; 265: 2049-2054.

30 O'Connell JR, Weeks DE. PedCheck: a program for identification of genotype incompatibilities in linkage analysis. Am J Hum Genet 1998; 63: 259-266. Article MEDLINE

31 Ott J. Analysis of Human Genetic Linkage The Johns Hopkins University Press: Baltimore, 1991.

32 Sobel E, Lange K. Descent graphs in pedigree analysis: applications to haplotyping, location scores, and marker-sharing statistics. Am J Hum Genet 1996; 58: 1323-1337. MEDLINE

33 Thompson EA. Monte Carlo likelihood in genetic mapping. Stat Sci 1994; 9: 355-366.

34 Paunio T, Ekelund J, Hovatta I, Varilo T, Terwilliger JD, Meyer J et al. Genome wide scan of an extended Finnish schizophrenia study sample. Am J Med Genet 2000; 6: 460.

35 Gurling HM, Kalsi G, Brynjolfson J, Sigmundsson T, Sherrington R, Mankoo BS et al. Genomewide genetic linkage analysis confirms the presence of susceptibility loci for schizophrenia, on chromosomes 1q32.2, 5q33.2, and 8p21-22 and provides support for linkage to schizophrenia, on chromosomes 11q23.3-24 and 20q12.1-11.23. Am J Hum Genet 2001; 68: 661-673. MEDLINE

36 Camp NJ, Neuhausen SL, Tiobech J, Polloi A, Coon H, Myles-Worsley M. Genomewide multipoint linkage analysis of seven extended Palauan pedigrees with schizophrenia, by a Markov-Chain Monte Carlo method. Am J Hum Genet 2001; 69: 1278-1289. MEDLINE

37 Hovatta I, Varilo T, Suvisaari J, Terwilliger JD, Ollikainen V, Arajarvi R et al. A genomewide screen for schizophrenia genes in an isolated Finnish subpopulation, suggesting multiple susceptibility loci. Am J Hum Genet 1999; 65: 1114-1124. Article MEDLINE

38 Sheffield VC, Stone EM, Carmi R. Use of isolated inbred human populations for identification of disease genes. Trends Genet 1998; 14: 391-396. Article MEDLINE

39 Sheffield VC, Nishimura D, Stone EM. The molecular genetics of Bardet-Biedl syndrome. Curr Opin Genet Dev 2001; 11: 317-321. MEDLINE

Figure 1 The best LOD scores for five Palauan multiplex schizophrenia pedigrees. Because the signals for K1583 (Chromosome 5), K2364 and 2024 occur in telomeric regions, the LOD scores are strongly positive on one end of their respective graphs.

Figure 2 The genome-wide distribution of LOD scores, under a dominant model, for five Palauan multiplex schizophrenia pedigrees.

Table 1 Characteristics of Palauan pedigrees containing schizophrenics (SCZ)