INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sarcoidosis is a systemic inflammatory disease of unknown etiology. It is characterized by high activity of macrophages and CD4⁺ helper T cells and by the formation of noncaseating granulomas in a variety of organs, predominantly the lung. The cause of this remains obscure, and intensive efforts so far have failed to conclusively identify a pathogenic agent in the granulomas (1, 2).

Familial clustering of sarcoidosis is well documented (3), and the existence of a genetic susceptibility to this disorder is widely accepted. Data from a nationwide chest radiography screening program indicate a recurrence risk in siblings of affected patients of approximately one in 100, which is 20 times higher than the prevalence rate of sarcoidosis in Germany (8). These numbers clearly exclude a simple mode of inheritance of sarcoidosis, but they are compatible with (1) single gene effects with greatly reduced penetrance, (2) polygenic inheritance, and (3) a multifactorial background with interaction of a genetic predisposition and environmental agents.

In order to localize predisposing genes for sarcoidosis, arbitrary polymorphic DNA sites can be used to label each chromosome and to test for cosegregation with sarcoidosis in families. Two principal statistical approaches are established: parametric and nonparametric linkage calculations. Parametric linkage analysis requires a model of inheritance of the trait under study and subsequently estimates the probability of the observed segregation of polymorphism with this trait. Negative results indicate that polymorphism and the trait locus reside in different parts of the genome and/or that the model is not correct. If models with single gene effects are assumed, then a greatly reduced penetrance must be taken into account to match the observed recurrence risk of sarcoidosis in first-degree relatives of affected patients.

In the absence of any knowledge about the mode of inheritance of sarcoidosis, a nonparametric linkage (NPL) analysis (i.e., operating without a model of inheritance) is appropriate. This analysis is based on the premise that affected siblings more often share parental chromosomal regions that harbor gene variants involved in the pathogenesis of sarcoidosis. The excess of shared parental alleles depends on the distance between the polymorphism and the site of the susceptibility gene.

Linkage calculations can be based on a single polymorphism or on a series of polymorphic sites along a chromosome. Linkage calculations of the latter type, known as multipoint linkage analysis, facilitate chromosomal mapping of sites showing positive linkage results in relation to the known positions of the polymorphic sites. The computer program GENEHUNTER (9) has been designed to perform single and multipoint linkage calculations with or without models and is widely used in genome-wide mapping studies.

The genome contains a vast number of highly polymorphic short tandem DNA motifs (so-called microsatellites) that are widely used to label chromosomes (10). We used the strategy of microsatellite genotyping and multipoint linkage analysis in a panel of 63 families with affected siblings to search the whole genome for chromosomal candidate regions of predisposition to sarcoidosis.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A sample of 63 German families (white individuals) with siblings suffering from sarcoidosis was drawn from a DNA bank of families affected by sarcoidosis. Details of these families are listed in Table 1. Most of the families had been recruited by inquiries among members of the German support group Deutsche Sarkoidose-Vereinigung (5). Sixteen families were identified from the files of Prof. K. Wurm of Höchenschwand, Germany, who for many years stressed the importance of family studies in sarcoidosis (3, 11). All patients and their physicians were interviewed by telephone or questionnaire. 101 of 138 patients had reportedly undergone biopsies of affected organs, and in 68 cases, the reports were available. The remaining patients showed characteristic radiologic signs of sarcoidosis in combination with clinical symptoms, laboratory parameters, and a clinical course consistent with the diagnosis of sarcoidosis. The parents of affected siblings were included in the study whenever possible (Table 1). In families missing one or both parents, one or two unaffected siblings were analyzed if available, in order to reconstruct the parental genotypes. A total of 95 first-degree relatives of affected patients were included in the study.

Two hundred and six microsatellite markers were chosen from a commercially available microsatellite genotyping set (MapPairs, version 8; Research Genetics, Huntsville, AL [ftp://ftp.resgen.com/pub/ mappairs]) that was developed for genome-wide mapping strategies. The distance between markers ranged from 6 to 29 centimorgan (cM; equivalent to 1 recombination in 100 meioses), with an average spacing of 19.6 cM. A further 19 markers from the MapPairs set or from the Genome Database (http://www.gdb.org) were included to gain a higher density of markers on the short arms of chromosomes 3 and 6 and throughout chromosome 16. (The positions of all markers are shown in Figure 2.)

View larger version (44K): [in this window] [in a new window]   View larger version (39K): [in this window] [in a new window]   Figure 2.   Genome-wide multipoint NPL results based on 63 families with siblings suffering from sarcoidosis (138 patients).

Genotypes for the microsatellite markers were determined through the polymerase chain reaction (PCR) amplification technique with M13 tailed primers and reaction details from the data bases given earlier. Fluorescent PCR products were separated according to size and detected on an automated sequence analysis apparatus (12). Allele frequencies were calculated on the basis of all genotyped individuals, and Mendelian inheritance of microsatellite alleles was controlled with the use of PEDCHECK (13).

Parametric and nonparametric linkage analysis and microsatellite haplotype construction were performed with the GENEHUNTER 2.0 linkage calculation program (9). The principle of linkage analysis was briefly outlined earlier. In detail, parametric linkage analysis calculates and compares the probability of the observed segregation of marker alleles in families: (1) under the hypothesis of close proximity to a susceptibility gene (cosegregating according to an assumed mode of inheritance) on the same chromosome and (2) in the case of independent segregation of a marker and susceptibility gene located on different chromosomes or in distant chromosomal regions from one another. In genome-wide linkage studies, a likelihood of 4,000 to 1 in favor of cosegregation is considered significant indication of a susceptibility gene close to the marker (14).

The NPL score is equivalent to the number of standard deviations by which the observed average proportion of parental alleles shared by sib pairs in the panel exceeds the null expectation of 0.50. It is shown together with the corresponding p value, which indicates the probability of the observed NPL score under conditions of independent assortment. We report nominal p values (i.e., values not corrected for the multiple testing problem inherent in multiple marker genotyping of whole genome search strategies). Usually, nominal p values on the order of 10⁴ to 10⁵ or smaller, depending on the family structure tested, are taken as indicative of significant linkage (14). In what follows, we report p values of lesser importance to enable others to follow up on this information.

The positions of candidate genes were taken from the Genetic Location Database (LDB, http://cedar.genetics.soton.ac.uk/pub).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We genotyped 138 affected siblings and 95 first-degree relatives from 63 families for 225 microsatellite DNA polymorphisms. The number of alleles observed for single markers ranged from four to 24, with an average of 8.6 alleles. This high degree of variability provided a high content of genotype information for the construction of haplotypes of adjacent polymorphisms. Thus, meiotic recombination of chromosomes could be determined and the segregation of chromosomal segments could be followed in families. As an example, the segregation of chromosome 16 markers in a family with four affected siblings is given in Figure 1.

View larger version (32K): [in this window] [in a new window]   Figure 1.   Microsatellite genotypes and haplotypes in a family with four siblings with sarcoidosis, showing nine of 16 markers analysed on chromosome 16. Paternal haplotypes could be reconstructed, and meiotic recombinations are indicated (x).

Figure 1 shows a varying degree of marker allele sharing in siblings along chromosome 16. Although all the different combinations of parental D16S516 alleles are present, the four siblings share the maternal allele of D16S769. This pedigree constellation is equivalent to NPL scores of 1.01 (p > 0.99) for D16S516 and 0.51 (p = 0.23) for D16S769. NPL scores from different families can be considered jointly, and can be displayed as curves along the chromosome. Results from all autosomes and the X chromosome are compiled in Figure 2 and give an overview of the whole genome (with the exception of the Y chromosome) for all 63 families of the panel.

For the complete panel, the most prominent peak was found on the short arm of chromosome 6 (6p21 to 6p22), with an NPL score of > 1.67 (p < 0.05) for six adjacent markers spanning a region of 16 cM, including the major histocompatibility complex (MHC). The highest NPL score, of 2.99 (p = 0.001), was obtained with marker D6S1666, which resides in the MHC Class III gene region.

Six other chromosomal regions showed minor peaks (p < 0.05). An NPL score of 2.39 (p = 0.009) was obtained for D3S1766 on the short arm of chromosome 3 (3p21), an NPL score of 1.87 (p = 0.03) for D1S1665 on the short arm of chromosome 1 (1p22), an NPL score of 1.82 (p = 0.034) for D9S934 on the long arm of chromosome 9 (9q33), and an NPL score of 1.64 (p = 0.047) for DXS6789 on the long arm of the X chromosome. Two markers on the long arm of chromosome 7 (D7S821 at 7q22 and D7S3070 at 7q36) showed NPL scores of 1.92 (p = 0.027) and 1.86 (p = 0.031), respectively.

In a subset of nine families with more than two affected siblings, the results were greatly different. This subset contributed little to the peak at the MHC. Actually, two of the six markers from that region (D6S299 and D6S2439) had negative NPL scores, and the highest NPL score was 0.79 (p = 0.21), for D6S1666. A negative NPL score was also obtained for D7S3070. On the other hand, five markers showed minor peaks (p < 0.05) in the subset. The highest NPL score, of 2.74 (p = 0.007), was obtained for D15S822, on the long arm of chromosome 15. Two adjacent markers on the long arm of chromosome 9 (D9S2157 and D9S1838 at 9q34) showed NPL scores of 2.32 (p = 0.017) and 2.15 (p = 0.023), respectively. Again, pronounced differences were found between results from the subset and results from the complete panel.

Parametric linkage calculations with models of dominant and recessive inheritance, frequencies of a predisposing gene allele ranging from 1 in 33 to 1 in 10,000, and different rates of penetrance and phenocopying gave no significant results.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sarcoidosis shows a wide range of organ involvement and widely variable course. However, a uniform basic event at the beginning of its pathogenic processes seems possible (1, 2). Events at this early stage include a characteristic distortion of the immune balance in favor of macrophage stimulation and CD4⁺ helper T-cell accumulation that leads to the formation of typical noncaseating granulomas. The components of the cytokine network that control the immune response are plausible candidates for creating susceptibility to sarcoidosis, and many of them are polymorphic. Numerous studies have been conducted to correlate the genetic variability of immunorelevant genes with the risk of developing sarcoidosis, but no major susceptibility gene could be identified.

It can be assumed that affected siblings tend to share predisposing gene variants and that significant allele sharing in a large panel of affected sib pairs points to the location of susceptibility genes for sarcoidosis. However, with the great number of markers tested in a genome-wide scan, the number of false-positive results for single markers increases. In our study of 225 markers, 11 peaks of p < 0.05 would have been expected to occur by chance alone. According to widely accepted definitions, a significance level of p < 0.0007 can be regarded as suggestive evidence of linkage, and an NPL score of > 3.6, followed by confirmation in an independent study, is required to conclusively map a susceptibility gene.

The lesson learned from other multifactorial disorders is that it most probably will take repeated genome-wide linkage studies to uncover the complex genetics of sarcoidosis. For instance, in inflammatory bowel disease (IBD; i.e., Crohn's disease and ulcerative colitrates of prevalence and sibling recurrence risk, repeated whole genome scans for susceptibility genes have identified significant evidence of linkage for defined regions on chromosomes 1, 3, 6, 12, and 16 (15).

This first genome-wide search for predisposing genes for sarcoidosis detected seven chromosomal regions of increased (p < 0.05) allele sharing in the complete panel of 63 families, and five different loci in the subset of nine families with more than two affected siblings. The most prominent peak in the complete panel, at the MHC (six adjacent markers, including D6S1666; NPL score 2.99; p = 0.001), is consistent with numerous reports of associations between gene variants in this region and the risk of sarcoidosis. These findings (from the first 55 families of the complete family panel examined in this study), together with genotypes of the candidate gene HLA DPB1, have been communicated in a previous publication (18).

The next prominent peak, labeled by the marker D3S1766, was found on the short arm of chromosome 3 (3p21). Of the genes from this region, the chemokine receptor genes CCR2 and CCR5 are especially intriguing, since the gene products are expressed on the surface of CD4⁺ T cells and are involved in resistance to human immunodeficiency virus infection (19). Two association studies (20, 21) suggested a relevance of these genes in sarcoidosis as well. However, a distance of 8 cM between the CCR genes and D3S1766 requires the investigation of additional markers in the region of 3p21. Other candidate genes at 3p21 encode macrophage stimulating protein-1 (MST1; 6 cM from D3S1766) and its receptor (MST1R; 5 cM from D3S1766), which contribute to the host defense system of bronchial epithelia (22). Interestingly, D3S1766 is close to D3S1076 (3 cM) and D3S1573 (6 cM), which showed linkage to IBD (16). The assumption of common genetic factors in the pathogenesis of IBD and sarcoidosis is supported by the coexistence of both disorders in the families of 66 of 1,026 Dutch patients with sarcoidosis (7). In this context, it is remarkable that D1S552 (cytogenetic localization 1p36), with an NPL score of 1.62 (p = 0.051) in our study, has been linked to IBD susceptibility in another genome scan (17).

The gene coding for the interleukin-2 receptor (IL-2R) gamma chain (IL2RG) is located between the peak at DXS6789 (17 cM from IL2RG) and DXS7127 (4 cM from IL2RG), the next microsatellite marker in our study. The product of IL2RG is the common gamma chain of several interleukin receptors (interleukin [IL]-2, IL-4, IL-13, IL-15, and others). Mutations resulting in loss of function of the gene product cause severe combined immunodeficiency (SCID) (23), and it seems possible that less severe variations in IL2RG are also involved in altered immune responses. An association between serum levels of soluble IL-2R and the prognosis of sarcoidosis has been shown (24). X chromosomal inheritance, irrespective of dominance and penetrance, excludes father-to-son transmission of IL2RG. If variations in IL2RG are relevant in a major fraction of sarcoidosis patients, one would expect a scarcity of affected father/son pairs. However, four father/son pairs and no father/ daughter pair were found in a questionnaire study of sarcoidosis patients (5), and our DNA bank includes eight father/son pairs and five father/daughter pairs.

Among the remaining minor peaks, the long arm of chromosome 9 attracts attention because of three adjacent markers with increased (p < 0.05) allele sharing in the complete panel (D9S934) or in the subset (D9D2157 and D9S1838). Possible immunorelevant candidate genes from this chromosomal segment encode orosomucoid (ORM1 and ORM2, 3 cM from D9S934) and transforming growth factor- receptor 1 (TGFBR1, 4 cM from D9S934). Although the allele ORM*1 has been suggested to confer a risk of sarcoidosis (25), no association studies of TGFBR1 have been reported so far. The ABO blood group gene locus, located close (1 cM) to D9S1838, was analyzed as a genetic marker in one of the first association studies in sarcoidosis (10), and showed overrepresentation of blood group A in a sample of 518 patients with sarcoidosis.

On the other hand, several suggested candidate susceptibility gene regions did not show significant allele sharing. The angiotensin converting enzyme (ACE) gene is located on the long arm of chromosome 17 (17q23) and carries a polymorphic site at intron 16, depending on the presence or absence (deletion) of a 287-base-pair DNA insertion. It has been suggested that the deletion is associated with an increased risk of sarcoidosis (26). However, the markers D17S2193 (1 cM from ACE) and D17S1301 (19 cM from ACE) that flank the ACE locus showed no evidence of linkage, with negative NPL scores of 0.71 (p = 0.76) and 0.67 (p = 0.75) in the complete panel, and similar results for the subset of nine more complex families.

Another major candidate region is the centromeric part of chromosome 16. The IL-4R gene (IL4R), a major IBD susceptibility gene (IBD1) (15), and the gene causing Blau syndrome (BLAU) (29) have been mapped to this region. Blau syndrome is a multiorgan granulomatous inflammatory disease that often starts in childhood and resembles sarcoidosis in many aspects (30). It is caused by a single gene and is inherited as a dominant trait. We found no significant evidence for linkage for the markers D16S769 (6 cM from BLAU) and D16S753 (1 cM from BLAU), which flank the Blau syndrome gene locus, nor for 14 other microsatellite polymorphisms throughout chromosome 16. The results of parametric linkage analysis (i.e., comparison of observed marker allele segregation in affected families with an assumed model of inheritance of the disease) ruled out a major contribution of the Blau syndrome region to the genetic predisposition to sarcoidosis. Even allowing for heterogeneity of the disorder, the best fit with dominant models of inheritance was found for D16S769, assuming that a fraction of 0.2 of the panel contributed to an odds ratio of less than 1 in 2 in favor of linkage for this subgroup. These results are consistent with a study of African-American sib pairs suffering from sarcoidosis that excluded a major effect of the Blau syndrome gene in sarcoidosis (31).

Clonality of T-cell populations in sarcoidosis has been detected by analysis of T-cell receptors (TCR), and could be linked to clinical presentation and prognosis of the disorder (32, 33). The TCRA and TCRD gene clusters are located at chromosome 14q11, TCRB at 7q35, and TCRG at 7p15. Of these loci, TCRB is close (6 cM) to D7S3070, with an NPL score of 1.86 (p = 0.031) in the complete family panel, but a negative NPL score of 0.34 in the subset of more complex families. Clonality of T-cell populations is a consequence of antigen contact and HLA gene interaction rather than an independent product of genetic variability at the TCR gene cluster. No association with sarcoidosis was found in a study of African-American patients for 14p11, but TCRB has not been tested (34). In any case, the marked difference between our results for the complete family panel and those for the subset of more complex families, at both 7q35 and the MHC locus, do not support the hypothesis that an inherited disturbance of T-cell selection plays a role in the latter sample.

Variants of the natural resistance-associated macrophage protein gene (NRAMP1) are linked to the risk of tuberculosis (35) and may influence the risk of sarcoidosis as well (36). NRAMP1 resides on the long arm of chromosome 2 (2q35), between the markers D2S1384 (15 cM from NRAMP1) and D2S1363 (14 cM from NRAMP1) used in our study. Neither marker gave evidence of a sarcoidosis susceptibility gene in the vicinity, but they are too far from NRAMP1 to uncover a minor effect of that gene. The same is true for the gene encoding IL-1 (IL1A) on the same chromosome (2q13). Divergent allele frequencies in African-American sarcoidosis patients have been reported for IL1A (34), but the gene is located in the middle of a 19-cM interval between D2S2972 and D2S1328, and minor effects could therefore have remained undetected in our study.

In this first genome-wide search for genes predisposing to sarcoidosis, we used 206 evenly spaced markers for an initial scanning of all autosomes and the X chromosome. The average distance of 19.6 cM between markers is too great to guarantee recognition of all sites of significant allele sharing. Indeed, we would have missed the peak at the MHC with this basic marker set, including D6S2439 (NPL = 1.43, p = 0.075) and D6S1017 (NPL = 1.55, p = 0.06) in the MHC region. The analysis of additional neighboring markers on chromosome 3 increased the calculated multipoint linkage score at D3S1766 from NPL = 2.19 (p = 0.014) to NPL = 2.39 (p = 0.009).

But the detection of predisposing genes would not be guaranteed even with a much higher density of markers. Depending on the number of genes involved in the predisposition to sarcoidosis, and on the degree of etiologic heterogeneity of the disorder, many more markers must be tested. Genetic components with minor effects require a much larger sample of affected sib pairs. The marked difference between our results for the complete panel of families and for a subset of families with a more complex inheritance pattern might be an indicator of heterogeneity of the genetic contribution to sarcoidosis. However, one should be aware that the number of families in the subset of families with a complex pattern of inheritance was nine, a number that is in general not indicative of a very high power either for the detection of linkage or for the detection of heterogeneity. Thus, these differences between the two samples might also be spurious.

On the other hand, very close markers do not produce much more linkage information, owing to the lack of recombinations between the markers within small families. However, the resulting linkage disequilibrium between markers and haplotype sharing would provide a powerful basis for fine mapping of susceptibility genes in isolated populations or extended families, as described in the Northern districts of Sweden (4). Although this approach might disclose predisposing genes relevant only for the population under study, a more universal strategy of NPL analysis and subsequent fine mapping through marker allele association studies has recently been introduced (37).

Microsatellite genotyping and the mapping of putative predisposing gene loci is only one step toward unraveling the complex genetics of sarcoidosis. The next step would be to identify candidate susceptibility genes in chromosomal regions with significant linkage to sarcoidosis, using additional methods, such as transmission disequillibrium tests in families, or extended association studies. Once predisposing genes are identified, an attempt can be made to estimate the effects of single predisposing genes and environmental triggers of a multifactorial pathogenic process in large case-control studies. However, this task could be complicated by the limited number of patients and by difficulties in defining an appropriate control group in the absence of knowledge of pathogenetic determinants. At that advanced stage of research, another type of family study could be advantageous, using related control subjects (38). The design of this last type of family study includes siblings without the disorder, classified according to the number of predisposing gene variants they share with their affected siblings. Careful analysis of these groups with respect to other, eventually interacting genes or environmental agents could help to understand the puzzle that causes sarcoidosis.

In any case, the results of the first genome-wide search for genes predisposing to sarcoidosis, presented here, require repetition with independently selected family panels.