INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

After reporting a dominant inheritance pattern for atopy in seven extended UK pedigrees (1), linkage to the microsatellite marker D11S97 on 11q13 and atopy was described with a maximum lod score of 5.58 at a recombination fraction of 10.5% (2). Atopy was broadly defined as one or more positive skin-prick tests, a total serum immunoglobulin (Ig) E > 2 SD above the mean for the normal population, or one or more positive allergen-specific radioallergosorbent tests (RASTS). Linkage to the same marker was replicated by the same group in a second larger sample of 64 nuclear families (3) with a lod score of 3.8 achieved at a recombination fraction of 0.067. Subsequent work suggested that atopy was inherited at the 11q13 locus via the maternal line (4). There have since been a number of further independent publications claiming either positive (5) or negative (8) findings for linkage to this chromosomal region. Initially, in a randomly ascertained population of 131 families recruited without reference to asthma or atopy, we were unable to find evidence for linkage of atopy or asthma to markers on 11q13 (14), although we reported allelic association with the polymorphic marker D11S527 and bronchial hyperresponsiveness (BHR) in the same population (6). This failure to reach consensus on a chromosomal region highlights many of the problems that beset asthma genetics; small population samples, failure to agree on phenotype definition, and multiple hypothesis testing generating type 1 errors.

The  subunit of the high affinity receptor for IgE (Fc RI-), which is expressed in mast cells, basophils, eosinophils, and dendritic cells (15), was localized to the chromosome 11q13 region and proposed as a candidate gene (16). The complete high affinity IgE receptor consists of one  chain, one  chain and two  chains. The  chain functions as an amplifier of the -mediated activation of a tyrosine kinase (15). The process of tyrosine phosphorylation of cross-linked Fc R1 subunits leads to the assembly of a "signaling complex" of receptor-associated proteins. This in turn leads to the production of inositol phospholipids, which signal a rise in intracellular calcium essential for the release of preformed, granule associated, and newly generated mediators (17). Determination of the coding sequence of the Fc RI- gene initially identified two amino acid substitutions in exon six: isoleucine to leucine at residue 181 and valine to leucine at residue 183, designated I181L and V183L, respectively. In 163 randomly selected subjects and 60 nuclear families ascertained through an asthmatic proband, the I181L variant was reported to be strongly associated with high serum total IgE levels, and in families the I181L variant was strongly associated with atopy in the children (p < 0.0001), and was maternally inherited (18). An additional substitution, designated E237G, has been identified in exon seven of the Fc RI- gene and is associated with diagnosed asthma, but in the absence of maternal transmission (19). The relevance of the I181L/V183L polymorphisms remains unclear. A functional analysis of histamine release from basophils and mast cells in subjects reported to express the I181L variant of the Fc RI- gene has failed to provide supportive evidence of differences between atopic subjects with the mutation and those without (20).

Because of continued uncertainty regarding the Fc RI- gene as a candidate gene for asthma and atopy, the aim of the present study was to investigate linkage to Fc RI- using non-parametric tests in two large well-characterized populations from the Wessex region of the United Kingdom. The frequencies of the coding variants reported within the Fc RI- gene, I181L/V183L and E237G, were then determined for both populations.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

With approval from the Southampton University Hospitals Joint Ethical Committee, two distinct family population samples were recruited from the Wessex region of the United Kingdom in and around Southampton. The first sample comprised 685 subjects from 131 nuclear families, with three or more children recruited without reference to asthma from Primary Health Care registers in the Southampton area (21) and are referred to as the random sample. The second sample comprised 354 subjects from 60 extended families recruited through an asthmatic proband attending Child Health Clinics in Southampton, with at least one other family member affected by asthma, and will be referred to as the asthma multiplex sample. A further 49 families were recruited on the basis of a single asthmatic proband from Primary Health Care registers in the Southampton area and are referred to as the simplex sample. The children gave informed verbal and the parents informed written consent. The total sample yielded 240 families comprising 626 sib pairs from 237 informative families in which at least two sibs were genotyped, and 114 affected sib pairs from 81 informative nuclear families.

Description of the Phenotypes

Each family member completed a structured written questionnaire on atopic symptoms and diseases derived from the International Union Against Tuberculosis and Lung Disease (IUATLD) questionnaire (22). This was supplemented by a video questionnaire with questions specifically on wheeze during exercise, nocturnal symptoms, and symptoms occurring at rest (23). Skin-prick testing to 14 common allergens was performed: Dermatophagoides pteronyssinus, Dermatophagoides farinae, cat, dog, feathers, egg white and yolk, cow's milk, mixed grass, mixed trees, Alternaria, horse, Aspergillus, and Cladosporium (Bayer Corporation, Spokane, WA), with a negative (saline) and a positive (histamine) control using the procedure described by Pepys (24). The perpendicular diameters of each wheal were recorded.

Bronchial reactivity to histamine was measured using the handheld DeVilbiss nebulizer technique described by Yan and colleagues (25) using a dose range from 0.03 to 4.4 µmol. This technique was chosen as the portability of the equipment and the brevity of the protocol makes it ideal for large epidemiologic surveys. The output of the DeVilbiss no. 40 nebulizers (DeVilbiss Health Care, Heston, UK) was calculated by measuring the weight of the nebulizer containing 1 ml normal saline in grams before and after 100 squeezes. The final weight was subtracted from the initial weight and divided by 100 to give the output per squeeze. The nebulizers were recalibrated for each sample collection. Bronchial challenge testing for all three samples was carried out by two research nurses, each of whom had a dedicated set of nebulizers. The challenge ended when either the patient's FEV₁ had fallen to less than 80% of the postsaline value or the highest dose of histamine had been administered.

Blood samples were taken for serum total IgE and specific IgE levels: house dust mite (Der p1, Der p2), dog (can f1), cat (fel d1), alternaria (Alt a1) and timothy grass (Phl p5). The serum total IgE was measured by enzyme-linked immunosorbent assay (ELISA). Specific IgE antibodies were measured using the Magic Lite SQ system (Ciba-Corning, ALK, Copenhagan, Denmark; kindly supplied by Dr. Carsten Schou).

Molecular Methods

DNA was extracted from peripheral blood leukocytes collected in EDTA and the following microsatellites amplified using the polymerase chain reaction; Fc RI-, D11S527, D11S534, and D11S480. The first two of these were genotyped in both the random and the multiplex samples (191 families) and the second two in the random sample only (131 families). The oligonucleotide primers were synthesized commercially (OSWEL, Southampton, UK) with the forward primer from each pair fluorescently labeled to allow the product to be visualized on a urea/polyacrylamide gel in a Model 373A automated DNA sequencer (Perkin Elmer, Medical Instruments, Branchburg, NJ) and analyzed using Genescan 672 and Genotyper software. The frequency of the I181L/V183L substitutions was determined in the parents of both the random and the multiplex samples by direct cycle sequencing of a 426 bp PCR product spanning exon six, amplified with the primers 5'-GTCAGATGGTAGGGAGATGA and 5'-CAGTCCTATTATCCCAACCC. Using a "hot start" procedure, PCR reactions were performed in 50-µl volumes, containing 100 ng genomic DNA, 100 ng each primer, 200 µM each dNTP, 1.5 mM magnesium chloride, and 1 unit AmpliTaq DNA polymerase (Perkin Elmer). Thermocycling was carried out in a Perkin Elmer 9600. PCR products were purified using Wizard preps (Promega, Southampton, UK) and 100 ng of the eluted DNA product used as template for Dye Terminator cycle sequencing using the Ready Reaction Mix containing AmpliTaq FS (Perkin Elmer). After the ethanol precipitation, denatured extension products were separated on a urea/polyacrylamide gel in a Model 373A automated DNA sequencer (Perkin Elmer). Sequencing reactions were performed with both original PCR amplification primers so that the sequence of both strands was read to give an unambiguous consensus.

The frequency of the E237G substitution was determined in all three population samples using an ARMS procedure described previously (19), but with a reduction in the amount of genomic DNA template to 30 ng (M. R. Hill, personal communication). PCR-generated products of 446 bp (control band), 280 bp (wild type) and 238 bp (E237G) were separated by agarose gel electrophoresis in a 3% (wt/ vol) Metaphor agarose gel (Flowgen) and visualized by staining with ethidium bromide (1 µg/ml). A summary of the markers used in the population samples can be found in Table 1.

Statistical Methods

Quantitative phenotype scores were derived for asthma and atopy using methods previously described (26). Briefly, each written questionnaire response was coded as an integer, monotonic on expected risk for asthma. The video questionnaire data were already in this format. The questionnaire data were then reduced to two variables by taking the first principal component of the written (abbreviated as wheeze or WZ) and video (VID) questions. The histamine challenge data (BHR) were summarized as the slope of the dose-response curve. Spirometry was expressed as the ratio of predicted to observed FEV₁ based on height and age (RFEV). The four summary variables (WZ, VID, BHR, RFEV) were then age- and sex-adjusted and normalized by rank transformation (27). An asthma score (AS) was defined as the first principal component of the four summary variables.

Total serum immunoglobulin E, specific IgE, skin-prick, and written questionnaires on hay fever and eczema were used as indicators of atopy (Lawrence and colleagues, 1994 [21]; Doull and colleagues, 1996 [6]). Questionnaire responses were coded as integers monotonic on expected risk for eczema and hay fever, respectively. The questionnaire data were then reduced to two variables by taking the first principal component of the eczema (EZ) and hay fever (HF) questions. IGE is the natural logarithm of the total serum IgE. RAST is the natural log of the specific IgE summed across each allergen. SP is the principal component of the size of the wheals to 14 allergens, which were measured as major and minor axes. The five summary atopy variables (IGE, RAST, SP, EZ, HF) were age- and sex-adjusted within ascertainment, as for asthma variables, and then rank-transformed. IGE was selected as the best atopy indicator on the basis of the highest value for heritability estimated by the POINTER program that takes account of incomplete ascertainment (28). The few missing observations for both asthma and atopy scores were imputed using PRINQUAL, a SAS/STAT procedure (SAS Institute, Cary, NC). As we were looking for disease-specific genes, we subsequently covariance-adjusted the asthma score (AS), so that its value was linearly independent of IGE.

For nonparametric (sib pair) linkage analysis we applied the BETA program (29), which obtains allele sharing proportions for 0, 1, and 2 alleles identical by descent (IBD) in a pair of sibs as a function of a single parameter (). This approach has been shown to be more powerful than the alternatives (30, 31). Single point analyses estimate  for a single marker, whereas multipoint analysis estimates  and location S using genotypic information from markers in a genetic map. Multipoint lods with 2 degrees of freedom can be transformed to a ²(2 df) multiplying by 2 log 10. A lod with 1 degree of freedom is then obtained from the corresponding probability (31). Allelic association testing was carried out using NOPAR (nonparametric testing) (32).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results of the nonparametric single locus tests using the BETA program are shown in Table 2 for AS (asthma score), asthma affection (defined as a positive response to the question "Have you ever had asthma?"), WZ (wheeze) and IGE. Lods greater than one were achieved for E237G (frequency of 3.6% in the random sample and 3.4% in the multiplex sample) with AS and asthma affection (lods 1.078 and 1.522, respectively). None of the other markers on 11q13 showed evidence for linkage to asthma or IGE. Results using the multipoint program are shown in Table 3 as lods with 2 degrees of freedom (lod₂) when  and S are estimated, and 1 degree of freedom (lod₁). A maximum lod score of 1.428 with 2 degrees of freedom was achieved at location 70.068 for AS, the closest marker in this region being D11S480.

The presence of the I181L/V183L polymorphisms within exon 6 of the Fc RI- gene was investigated in both parents of the 60 probands in the asthma multiplex sample and in both parents of 90 nuclear families from the random sample. In all 300 subjects the wild type sequence was always seen at the positions of the three possible nucleotide substitutions.

When tested in the combined population sample, none of the microsatellite markers showed significant positive allelic association.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study focusing on 11q13 revealed that the largest single point lod scores were achieved for E237G for AS (asthma score) and asthma affection (lods 1.078 and 1.522, respectively). The maximum lod score achieved using multipoint linkage analysis was 1.428 using AS at location 70.068. No evidence for linkage was found for the other microsatellites, including the intronic marker within the Fc RI- gene. Weakly positive results are common when looking for genes in complex disease, where genes are likely to be many but individually of small effect. A standard level of significance has not been established for lod scores in complex disease, and their interpretation remains controversial. This is an important issue to address in order to avoid the publication of a large number of false linkage claims that cannot be replicated. A classification scheme for reporting linkage results in complex disease has been proposed (33). Suggestive linkage, i.e., linkage that would be expected to occur one time at random in a genome scan, corresponds to a lod score of 2.2, whereas significant linkage corresponds to a lod score of 3.6. The lod score achieved in this analysis for E237G provides weak evidence for linkage. In the absence of clear linkage to the more informative marker, Fc RI-, it would seem likely that the linkage result for E237G has occurred as a result of a type 1 error.

The E237G substitution occurs at a slightly lower frequency of approximately 3.5% in both our random and our asthma multiplex UK populations than in either the Australian or the Japanese populations (approximately 5% in both populations) (19, 34). Unlike the Japanese study, which was based on unrelated participants, there was no increase in frequency of E237G between our random and asthma-enriched population samples. Even accounting for any possible bias introduced by the extended families in the multiplex sample, by looking at the frequency in the parents (of each proband in the case of the asthma multiplex sample), there was no significant increase.

We also found no evidence of the I181L/V183L polymorphism in our population sample. The original report identified the I181L polymorphism alone using the ARMS (amplification refractory mutation system) assay (18), and found a strong association with high total serum IgE and skin test reactivity to grass pollen, when transmission was through the maternal line. In contrast, using a revised ARMS assay, the combined I181L/V183L polymorphism was found in 4.5% of a Western Australian population (35) and was highly predictive of asthma and atopy when maternally inherited. Two subsequent studies have reported an increased frequency of the combined polymorphism; 34% in a second Australian (aboriginal) population (36) and 48% in asthmatics and 31% of nonasthmatics in a Kuwaiti population (37). Four studies have failed to identify the polymorphism using the ARMS assay (34, 38). To overcome the problems associated with the ARMS assay in our population, the frequency of the I181L/ V183L polymorphism was determined by direct cycle sequencing. This provides complete information on the nature, location, and predicted amino acid changes of any mutations present. In all of the 300 subjects investigated the wild-type sequence was observed in every case. Other workers have also failed to identify I181L/V183L using sequencing (9, 41). Turner and Kinet (42) have investigated whether mutant Fc RI-, containing the I181L and V183L substitutions singly or doubly or the E237G single substitution, functioned differently from the wild-type receptor using the single-amplifier effect on Ca²⁺ mobilization and the trafficking-based effect of the  chain on receptor surface expression. They concluded that none of the  chain polymorphisms tested enhanced or diminished Fc RI- function. However, they make the important point that polymorphisms in FC R1- may, in fact, be in linkage disequilibrium with another gene(s) that is phenotypically associated with atopy.

Several groups have looked for evidence of linkage to asthma and atopy and the intronic marker within the Fc RI- gene, and a set of results have emerged that are difficult to explain. Using a stringent definition of atopy and including only four families from an original sample of 270 in the final analysis, Shirakawa and colleagues (43) found linkage to D11S97 and Fc RI- on 11q13 in a Japanese population with evidence of a maternal effect. Other workers have demonstrated linkage to the original D11S97 marker but not to Fc RI- (5), or linkage to Fc RI- and asthma and BHR but not atopy (7). Significant association but no linkage has been found between serum IgE and D11S97 but not with Fc RI- (38), and in an Italian study (44), an excess of maternal allele sharing was observed at the Fc RI- locus, but this failed to achieve statistical significance. In a whole genome screen conducted in 80 nuclear families, the Fc RI- marker was linked to the atopy phenotypes, whereas in a second confirmatory sample, the same marker was linked to asthma but not to atopy (45). In a recent study of 83 families from Northern Holland, no evidence for linkage was found to Fc RI- and IgE, BHR, or skin-test reactivity (9).

The original data on chromosome 11 suggested that atopy was inherited at the 11q13 locus via the maternal line (4). Two possible explanations were offered for this; paternal genomic imprinting or maternal modification of the developing immune response. Paternal genomic imprinting is not supported by expression studies of homologous regions in the mouse (46), although this does not necessarily exclude imprinting as a cause of the proposed maternal effect. A maternal effect may occur through intrauterine or early postnatal environmental influences. Intrauterine environmental factors may in part account for a maternal influence over the progeny's immune response in favor of atopy, although in the developing foetus it is difficult to see how this could be allele-specific, for which there is no biologic precedent. Studies reporting positive linkage results and association to 11q13 vary in their detection of the maternal effect. Neither van Herwerden and colleagues (7), who reported linkage to Fc RI- and BHR, nor Collee and colleagues (5), who found linkage to D11S97, could find any maternal effect. No maternal effect has been reported for the E237G polymorphism, the only marker showing suggestive linkage in this study (19). In the present study, parent of origin was not taken into account in the BETA program, so we are unable to report whether any maternal effect existed.

Recently, Laing and colleagues (47) have reported an association between asthma and a polymorphism of CC16 (CC10), A38G, which is close to the Fc RI- gene on chromosome 11q13. This gene encodes uteroglobin, a 10-kD protein secreted by Clara cells with anti-inflammatory properties. The same group measured plasma CC16 levels and identified the A38G genotype in a population of 130 asthmatic and nonasthmatic children. The plasma CC16 levels were lower in subjects homozygous for the A38G polymorphism, and those with lower levels of CC16 were more likely to have asthma (48). It is possible that genetic linkage or association between markers close to or including Fc RI- and clinical variants of asthma could be the result of linkage disequilibrium with the CC16 or other gene.

Our study failed to strengthen the case for a candidate gene on chromosome 11q13, but it does attempt to address some important methodologic issues. The use of quantitative phenotype scores for asthma and atopy offered several advantages over other phenotype definitions. An arbitrary dichotomization of the data is avoided. The lack of consensus over asthma definition is obviated by treating asthma as a quantitative trait, and the problem of multiple hypothesis testing is also reduced. The use of scores for asthma and atopy could also provide the way forward for meta-analysis in order that data from many studies can be combined and analyzed. If studies reporting weak or no linkage or association with a particular chromosomal region or candidate gene are not reported, then this could lead to distorted values when data from positive studies are pooled in meta-analysis (49). It is now well recognized that these problems of replication are unlikely to be resolved until results are pooled and the true significance of a positive linkage result can be evaluated in a larger data set.