Population Mutation Scanning of Human GHR by meltMADGE and Identification of a Paucimorphic Variant

Abstract

Current studies of human genetic diversity are focused in two areas: first, detection of rare mutations in highly selected clinical cases; and second, in common single-nucleotide polymorphism (SNP) and haplotype effects in the general population. Less frequent SNPs and “paucimorphisms” remain underexplored, although lower frequency coding SNPs are more likely to have functional impact. We have developed a cost-efficient mutation scanning technology, meltMADGE, for population mutation scanning. Previous research in GHR has explored its role in extreme (−3 SD) growth retardation and, subsequently, “moderate” (−2 SD) growth retardation cases. Here, we describe meltMADGE assays for the entire coding region of GHR. As a first step we have established long polymerase chain reaction subbanks for GHR from 2423 unselected subjects and have applied meltMADGE scanning assays of exons 4 and 5 to these subbanks. A novel paucimorphism present at 439+30A>C (allele frequency: 0.0021) in intron 5 (location chr5:42,695,221 in GRCh37/hg19) was identified in 10 individuals, confirmed by sequencing and analysis made for major phenotypic effects. This approach is relevant to the deep sampling of populations for less frequent sequence diversity, some of which is expected to exert significant phenotypic effects.

Introduction

So far most research into human genetic diversity has been aimed at detecting rare mutations in highly selected clinical cases or common single-nucleotide polymorphisms (SNPs) and haplotypes in the general population (Hampl et al., 1998; Gastier et al., 2000; Lucci-Cordisco et al., 2001; Majka et al., 2001; Nasioulas et al., 2001; Mozas et al., 2004). We argued the case for an intermediate possibility, that is, “paucimorphic” alleles (arbitrarily, 0.0005<q<0.05) may represent a significant contribution to the population burden of genetically driven diseases and risk traits (Day et al., 2004). Others have also argued that the allelic spectra of common diseases may resemble the allelic spectrum of the full genome. One study resequenced three genes affecting high-density lipoprotein (HDL)-cholesterol level in the top and bottom 5% of subjects for HDL-cholesterol levels (128 individuals) from a Dallas population survey (Cohen et al., 2004). It was concluded that nonsynonymous sequence variants were significantly more common (16% vs. 2%) in the bottom quintile, compared with the top quintile. However, this study effectively represents a case study (of trait extremes) rather than a full-population study. A principal problem for mutation identification at the whole population level, both in research and in diagnostic screening possibilities, is the high cost and limited throughput of resequencing or mutation scanning. For example, one resequencing reaction and capillary electrophoresis costs 3-5 USD and prescreening by dHPLC (wave) costs 2-4 USD. Next-generation sequencing methods are enabling genome-wide resequencing, but there remains a need for approaches offering large-scale scanning of selected targets. This is analogous with the use of SNP typing for single SNP replication follow-up studies based on genome-wide association SNP chip surveys (for example, Loos et al., 2008). We have developed two techniques, melt-MADGE (Alharbi et al., 2005; Alharbi et al., 2007) and endo VII-MADGE (Smith et al., 2005), for such scanning. Both methods have a scanning cost of ∼1-2 cents per template-amplicon, that is, 1/5th-1/10th of polymerase chain reaction (PCR) cost instead of a multiple of PCR cost. We have applied the melt-MADGE approach to all of MC4R in a study of 1100 participants of a population-based cohort, which resulted in the discovery of a novel functional variant (Alharbi et al., 2007), and to parts of the LDLR in studies of up to 9000 subjects in relation to body mass index and serum cholesterol, respectively (Alharbi et al., 2005). Future developments to fully explore complete allelic spectra in the general population in relation to disease traits will continue to require considerable reduction in sequencing and/or scanning cost, high throughputs, and very large DNA banks such as the ongoing national biobanks in several countries.

GHR is located on chromosome 5p (Barton et al., 1989), spans 85 kp, and includes nine coding exons. Exon 2 encodes an 18-amino-acid signal sequence, exons 3-7 encode the extracellular domain, exon 8 encodes the transmembrane domain, and exons 9 and 10 encode the intracellular domain (Meacham et al., 1993), including a proline-rich region.

Over 50 mutations have been reported (Milward et al., 2004). These mutations include large and small deletions, splice site mutations, and nonsense and missense single-base codon mutations. Earlier literature focused on classical Laron syndrome inherited in an autosomal recessive fashion. In general, this spectrum of mutations (Amselem et al., 1993) results in absence of receptor function. Haploinsufficiency in heterozygotes for such mutations does not produce a clinically recognizable phenotype. Studies of heterozygous Ecuadorians for splice mutation E180, which has been shown to be functionally null, did not identify any phenotypic effects in heterozygotes (Rosenbloom et al., 1998). However, expression of a defective receptor may exert dominant negative effects (Ayling et al., 1997; Dos et al., 2004). The growth hormone receptor is expressed on the cell surface and receptor molecules dimerize in the process of binding GH and inducing signal transduction. Defective receptor molecules can thus interfere with wild-type molecules when defective and wild-type molecules dimerize. Additionally, if defective molecules have long half-lives at the cell surface, this will further contribute to dominant negative effects. A mutation (871-1G>C) leading to exon 9 splice skipping exemplifies these effects (Ayling et al., 1997). Further, presence of variant forms of or differential release of the extracellular domain of the growth hormone receptor into the circulation might also have quantitative effects. A common polymorphism involves the presence or absence of (all of) GHR exon 3, which encodes part of the extracellular domain of the receptor. Compared with wild type, heterozygotes show a different responsiveness to exogenous growth hormone, representing another example of the “dominant” or “quantitative” effects that expressed GHR can exert. Most mutation studies have been undertaken in extremely growth-retarded children (>3 SD below the mean), although one study demonstrated the occurrence of amino acid mutation in children selected for moderate (−2 SD) growth retardation (Goddard et al., 1995). However, no studies of the full spectrum of sequence variation, which may be present in the general population, or studies of their potential quantitative effects have ever been conducted.

In this article, we describe development of mutation scanning assays of GHR using a system that we developed to be capable of high throughput and economy, which are necessary for large-scale population-based mutation scanning. We present scans for exon 4 and exon 5, representing a functionally important part of the receptor, in a population sample of 2577 anthropometrically characterized subjects.

Methods

Participants and ethics statement

The genomic DNA banks represented previously described cohorts recruited in Hertfordshire, United Kingdom. The first DNA bank represented 663 men and 449 women, 1112 subjects in total (from the North and East regions). The second bank represented a further 750 men and 690 women, 1440 subjects in total (from the East region). These cohorts and their clinical characterization have been previously described (Syddall et al., 2005 and references therein). Ethical approval was from the Hertfordshire and Bedfordshire Local Research Ethics Committee (reference EC98248) for the second bank and from East and North Hertfordshire Health Authority for the first bank (1980s). Written informed consent was obtained for the second bank (1990s-2000s); verbal informed consent was obtained for the 1980s bank, with explanation of the study in initial contact by letter, followed by interview at home of those consenting. Participants were then invited (apart from a few who were housebound or on warfarin therapy) to attend a local clinic for venesection for hematological, biochemical, and genetic studies, which most did attend. All research studies were conducted on anonymized data. Parameters relevant to the present study were anthropometric measures as older adults.

DNA bank and GHR long-range PCRs

Six pairs of long-range primers for GHR were designed using the Primer3 program (Table 1). Pairs of primers were optimized to get clear, low-background bands on agarose checking gels. Long PCR subbanks were then prepared for the regions around exons 4 and 5 by amplifying six sets of 7×384-well plates with long-range primers. Two microliters of 1/1000 dilution of long PCR product was used as template in a 10 μL melt MADGE PCR.

Table 1.

Long-Range Polymerase Chain Reaction of GHR Gene for Subbanks

Sequences of primers	Regions
L2-F 5′-GGTCGTTTGCTGTGAGGTGTTTC-3′	Exon 2, intron 2
L2-R 5′-GGGTATATTCAAGGGTAGACTG-3′
L3-F 5′-GGGACTTGCAGGTCATAGTTTGG-3′	Exon 3, intron 3
L3-R 5′-GGGCCAGATACTATTCTAACTAC-3′
L45-F 5′-GCTGACAATGGCAAGCAACAGGCACT-3′	Exon 4, exon 5
L45-R 5′-GGGAGAGTCCAATTCCAGTTCAGA-3′	Intron 4, intron 5
L6- F 5′-GGGCAACAGCATGCTCCTCCACC-3′	Exon 6, intron 6
L6-R 5′-GCTGGGCATCACCTGATGAGTGG-3′
L78-F 5′-GGGTTGTGAGACCTAGATGTTTCTT-3′	Exon 7, exon 8
L78-R 5′-CTAGCGTCCCAATAATTCTTCTTC-3′	Intron 7, half intron 8
L910-F 5′-TCCAGTGTGGGCAATGAAGTAAGA-3′	Exon 9, exon 10
L910-R 5′-GCTTTCCTTCCTTCCAGACAACAT-3′	Half intron 8, intron 9

Amplification of GHR exons

We adopted the set of primer sequences used in DGGE analysis of GHR (Amselem et al., 1993), but GC clamps were lengthened to be consistent with our own laboratory protocols. For each exon, we also designed a pair of artificial positive control primers with a single-base mismatch near the 3′ end of the shorter primer. Each pair of primers was optimized to get clear, low-background DNA bands on MADGE PAGE nondenaturing checking gels. Reactions were then run using the relevant long PCR subbank of 7×384-well plates.

Melt-MADGE analysis

Equipment

Melt-MADGE analysis is based on a real-time variable-temperature electrophoresis system essentially as previously described (Alharbi et al., 2005). In brief, it includes a 10×10×15 cm electrophoresis tank with system for ramping the buffer temperature in real time. The tank accommodates up to 11 urea MADGE gels poured in a purpose-built box.

Optimization of melt-MADGE assay for each exon

First, the melt temperature (T_m) of each exon was calculated using an in-house program acting as a windows interface to Melt87 (Lerman and Silverstein, 1987). Then, three constant temperature electrophoreses were conducted on 5% polyacrylamide gels containing 4 M urea for each exon at the temperatures of T_m−5°C, T_m, and T_m+5°C. Finally, the velocity-temperature charts were drawn for each exon and potentially suitable temperature ramps for melt-MADGE were determined.

Artificial, positive “heterozygote” controls were generated by mixing 4 μL of wild-type PCR product and sample and 4 μL of mutated primer-PCR product (single-base mismatch to wild type). The mixtures were denatured at 95°C for 5 min, put on ice for 2 min, and then held at room temperature for 1 h.

Wild-type and artificial “heterozygote” controls were electrophoresed on a 5% polyacrylamide gel containing 4 M urea in the melt-MADGE system using the most appropriate thermal ramp. According to these results, some small refinements of temperature, concentration of polyacrylamide, urea, or electrophoresis time might be made.

Melt-MADGE scanning of GHR in population samples

For each exon, PCR products from seven 384-well subbank plates were run in a melt-MADGE analysis system. The PCR products of each 384-well plate were loaded on four 96-well urea MADGE gels along with artificial heteroduplex control samples. Band detection was performed by Vistra Green staining and Fluorimager 595 scanning.

Sequence analysis

Each sample with variant DNA patterns (more than the one wild-type band) in the melt-MADGE assay was sequenced using a standard shrimp alkaline phosphatase and exonuclease I and Bigdye Sequencing Kit (ABI) protocol. Sequencing products were purified by ethanol precipitation and were redissolved in 25 μL template compression reagent (ABI) prior to analysis on an ABI 310 automated sequencer (ABI). Sequences were aligned and compared using Lipman and Pearson's Align Program (http://molbiol.soton.ac.uk/compute/align.html).

Results

The development of melt-MADGE assays for GHR mutation scanning

For each exon of GHR, a temperature-velocity chart was drawn according to three electrophoresis results at predicted T_m+5°C, T_m, and T_m−5°C. The temperature range with rapid mobility change was thus determined. The plot for exon 5 is shown as an example (Fig. 1). The predicted T_m was 61°C for this fragment after correcting for use of the 4 M urea. Therefore, the temperatures for the three electrophoreses are 56°C, 61°C, and 66°C. An obvious mobility shift was observed (Fig. 1) between 56°C and 66°C, with gel migration velocity reduced from 7 to 3 mm/h. The result showed that this temperature range was sensitive for melting analysis of exon 5, and it was trialed as the initial temperature ramp. Using an artificially constructed “heterozygote” PCR co-annealing, it was found that a 5°C-7°C ramp was the most effective in achieving good resolution, typically from about T_m−5°C to T_m+1°C. Most artificial positives resolved to a four-band pattern (some to three if the homoduplexes did not resolve), whereas wild-type produced one band. The optimized melt-MADGE assay conditions for exons 2-9 inclusive of overlapping for three amplicons representing the coding region of exon 10 are given in Table 2. For each exon, good band patterns and resolution were obtained for both wild-type amplicons and an artificial positive “heterozygote.”

FIG. 1.

Velocity-temperature chart of exon 5 amplicon of GHR gene in melt-MADGE assay.

Table 2.

Parameters of Optimized melt-MADGE Assays for GHR

Exon	Fragment (bp) ^a	Urea (mol/L)	PA (%)	Predicted T_m (°C)^b	RAMP (°C)	Time (h)
2	246	4	5	64	56-62	3
3	215	4	5	61.6	56-61	3
4	261	4	5	63.2	58-64	3
5	334	4	5	59.3	55-60	3
6	314	4	4.5	62	55-62	3
7	320	4.5	5	60.4	56-62	3
8	241	4.5	5	58	53-59	3
9	237	4.5	5	58.2	53-59	3
10A	443	4	4.5	62.8	55-62	3
10B	454	4	4.5	64.4	60-66	3
10C	549	4	4.5	64.4	58-64	3

Including GC clamps.

Corrected for 4 M urea.

PA, polyacrylamide gel.

Variant band patterns for exon 5 amplicon on melt-MADGE gels

Ten individual subjects with the same variant pattern were identified on melt-MADGE gels. Wild-type samples presented one-band pattern (homoduplex) on the melt-MADGE gels, but these 10 DNA samples showed a four-band pattern. An example is shown in Figure 2. All 10 variant amplicons were subjected to sequence analysis.

FIG. 2.

Image of one array for GHR exon 5 amplicon melt-MADGE analysis. a, wild-type pattern; b, artificial positive “heterozygote”; c, variant (natural) heterozygote. Most tracks represent pattern a including the two top rightmost tracks (in the column outside the 8×12 array), which were from a sequence-proven wild-type template. The two lower rightmost tracks represent pattern b. Pattern c is observed in track B6, where the 8×12 array is labeled A-H 1-12, with track A1 being the topmost in the figure.

Sequence analysis of the 10 DNA samples with variant band patterns

Direct sequencing showed that all 10 DNA samples possessed the same A>C base at 439+30A>C in intron 5 (location chr5:42,695,221 in GRCh37/hg19) (Fig. 3B). This paucimorphism has not been previously described and has an allele frequency of 0.0021.

FIG. 3.

Sequencing results for one sample with wild type (A) and for one variant sample in intron 5 of GHR (B). Base N represents either A or C in panel B.

Phenotypic characteristics of 439+30A>C carriers

Sex-specific Z-scores were derived for weight, height, waist, waist-hip ratio, and body mass index in relation to the relevant cohort. For height, these were −1.39, −0.78, −0.93, −0.62, −0.93, 1.96, −0.10, −0.64, −0.40, and −0.53. For weight, these were −1.06, −0.28, 1.73, −0.65, −0.82, 1.04, −0.017, −0.40, −1.09, and 0.11. Note that values between −1.96 and 1.96 represent 95% of observations, that is, these mutation carriers are not outliers for either trait, and although there was a slight bias toward negative values for height, there was no significant difference for this or the other traits (data not shown).

Discussion

We have developed melt-MADGE assays for mutation scanning of the entire GHR coding region. These assays have the characteristics of high throughput, low cost, and easy operation, which will be suitable for population mutation scanning without requirement for high capital expenditure. Scanning of amplicons spanning exons 4 and 5 (where mutations are prevalent in Laron Syndrome) identified a novel paucimorphism in 10 subjects in a population sample of 2423 subjects.

The characteristics of melt-MADGE assays have been previously examined in studies of LDLR and BRCA1 (Alharbi et al., 2005) and MC4R (Alharbi et al., 2007). In brief, most single-base mutations and small indels are detected by these assays. In contrast to denaturing HPLC or direct sequencing, overall cost is a fraction (e.g., 1/7th) of PCR cost rather than a multiplex (e.g., 30 times) of it (Alharbi et al., 2005). This lends the technique well to scoping the full range of sequence diversity through to singleton, paucimorphic, or low-frequency polymorphic alleles in the general population. Our identification here of a previously unknown GHR intron 5 paucimorphism 439+30A>C (location chr5:42,695,221 in GRCh37/hg19) in 10 of 2423 subjects (allele frequency 0.0021) exemplifies the potential.

In this development we used a systematic approach to assay optimization. In every case, an artificial (primer generated) positive control was established by co-annealing “wild-type” and “artificial” amplicons. Three constant temperature runs were then tried, T_m−5°C, T_m, and T_m+5°C. Usually, this confirmed that a ramp T_m−5°C to T_m+1°C was suitable for resolution of heteroduplexes.

For large-scale scans for rare sequence variants in finite DNA banks, DNA template consumption will be a significant problem. Possible conservative approaches include degenerate oligoprimer preamplification (Zhao et al., 1998), phi 29, or Rubicon preamplification (www.genengnews.com/gen-articles/whole-genome-amplification-technologies/854/). For focused studies of specific genes, the establishment of long PCR subbanks is an alternative. In this study, we developed five long PCR subbanks from 2423 subjects, giving complete tiling of the gene from exon 2 to exon 10. At 1/100-1/1000 dilution, these banks would support any future scanning or resequencing of exons or introns.

Scanning of populations for rare sequence variants, potentially to describe the full depth of sequence diversity and its contributions to the full range of phenotypic diversity, has so far received little attention. The 1000 Genomes project (www.1000genomes.org) will soon change this for discovery of less frequent sequence variation genome wide, but the study of phenotyped collections presents limitless potential. Selection of trait extremes for HDL-cholesterol from a population sample followed by directed sequencing of three genes relevant to HDL-C has been undertaken (Cohen et al., 2004) and exemplifies one approach to relating intermediate traits to their underlying mutational spectrum. However, the costs and time for direct resequencing even by automated multicapillary systems somewhat preclude full-population studies. This leaves scope for efficient scanning methodologies that can identify rare variants for sequencing while evading resequencings of the wild type; and until whole genome or exome sequencing is routinely affordable on a cohort-scale basis, focus on specific genes of major interest (as for SNPs identified in genome-wide association studies for complex traits) will continue to be necessary. It has been noted from a combination of theoretical and empirical data that missense alleles with a frequency of 6% or less are more likely to exert phenotypic effects than common protein polymorphisms (Wong et al., 2003). It also appears that synchronous duplex and multiplex point mutations within close syntenic range are quite common both somatically and in the germ line (Wong et al., 2003). Syntenic paucimorphisms or rare mutations can occur in complete linkage disequilibrium (Alharbi et al., 2005 and references therein). Rare alleles, therefore, merit greater focus of attention.

GHR intron 5 paucimorphism 439+30A>C occurs in a generally %AT rich and T-motif-rich region. Such regions have been shown to increase efficiency of expression in both mammals and plants, for example, in HBB in man (Bharadwaj et al., 2003) and maize (Clancy and Hannah, 2002). However, anthropometric phenotypes in these subjects were not significantly different from wild type; therefore, this particular paucimorphism appears to be silent or of unmeasurable minor effect. The mutation is in intronic sequence near the exon but not in a location predicted to influence pre-mRNA splicing. Although the assay focus was on exonic sequence and splice sites, a small amount of intronic sequence is necessarily included in the PCR amplicons. It is notable that no coding sequence variation was identified in 5154 alleles in this study. For LDLR, we previously estimated that 4/5 of all amino acid changes do not cause the severe hypercholesterolemia of familial hypercholesterolemia (Day et al., 1997). A comparable overrepresentation of codon mutations relative to stop codons appears to occur in Laron syndrome (Amselem et al., 1993). However, for any given exon, amino acid changes of moderate, no, or protective effect are only predicted to occur at a frequency of one per few thousand alleles. Very large population studies will be needed to define the “reference range” of such sequence variation, which will be valuable in the interpretation of missense mutations in case diagnostics.

In conclusion, we have demonstrated the potential of melt-MADGE assays of GHR to explore the full sequence diversity of this gene in the general population.

Footnotes

Acknowledgments

G.H. was the recipient of a Government of China Scholarship. The Hertfordshire cohort studies and original development of meltMADGE were funded by grants from the Medical Research Council of the United Kingdom.

Disclaimer

The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Disclosure Statement

No competing financial interests exist.

References

Alharbi

, Spanakis

, Tan

et al. 2007. Prevalence and functionality of paucimorphic and private MC4R mutations in a large, unselected European British population, scanned by meltMADGE. Hum Mutat, 28:294-302.

Alharbi

, Aldahmesh

, Spanakis

et al. 2005. Mutation scanning by meltMADGE: validations using BRCA1 and LDLR, and demonstration of the potential to identify severe, moderate, silent, rare, and paucimorphic mutations in the general population. Genome Res, 15:967-977.

Amselem

, Duquesnoy

, Duriez

et al. 1993. Spectrum of growth hormone receptor mutations and associated haplotypes in Laron syndrome. Hum Mol Genet, 2:355-359.

Ayling

, Ross

, Towner

et al. 1997. A dominant-negative mutation of the growth hormone receptor causes familial short stature. Nat Genet, 16:13-14.

Barton

, Foellmer

, Wood

, Francke

. 1989. Chromosome mapping of the growth hormone receptor gene in man and mouse. Cytogenet Cell Genet, 50:137-141.

Bharadwaj

, Trainor

, Pasceri

, Ellis

. 2003. LCR-regulated transgene expression levels depend on the Oct-1 site in the AT-rich region of beta -globin intron-2. Blood, 101:1603-1610.

Clancy

, Hannah

. 2002. Splicing of the maize Sh1 first intron is essential for enhancement of gene expression, and a T-rich motif increases expression without affecting splicing. Plant Physiol, 130:918-929.

Cohen

, Kiss

, Pertsemlidis

et al. 2004. Multiple rare alleles contribute to low plasma levels of HDL cholesterol. Science, 305:869-872.

Day

, Alharbi

, Smith

et al. 2004. Paucimorphic alleles versus plymorphic alleles and rare mutations in disease causation: theory, observation and detection. Curr Genomics, 5:431-438.

10.

Day

, Whittall

, O'Dell

et al. 1997. Spectrum of LDL receptor gene mutations in heterozygous familial hypercholesterolemia. Hum Mutat, 10:116-127.

11.

Dos Santos

, Essioux

, Teinturier

et al. 2004. A common polymorphism of the growth hormone receptor is associated with increased responsiveness to growth hormone. Nat Genet, 36:720-724.

12.

Gastier

, Berg

, Vesterhus

et al. 2000. Diverse deletions in the growth hormone receptor gene cause growth hormone insensitivity syndrome. Hum Mutat, 16:323-333.

13.

Goddard

, Covello

, Luoh

et al. 1995. Mutations of the growth hormone receptor in children with idiopathic short stature. The Growth Hormone Insensitivity Study Group. N Engl J Med, 333:1093-1098.

14.

Hampl

, Hampl

, Plaschke

et al. 1998. Evidence that TSG101 aberrant transcripts are PCR artifacts. Biochem Biophys Res Commun, 248:753-760.

15.

Lerman

, Silverstein

. 1987. Computational simulation of DNA melting and its application to denaturing gradient gel electrophoresis. Methods Enzymol, 155:482-501.

16.

Loos

, Lindgren

, Li

et al. 2008. Common variants near MC4R are associated with fat mass, weight and risk of obesity. Nat Genet, 40:768-775.

17.

Lucci-Cordisco

, Rovella

, Carrara

et al. 2001. Mutations of the ‘minor’ mismatch repair gene MSH6 in typical and atypical hereditary nonpolyposis colorectal cancer. Fam Cancer, 1:93-99.

18.

Majka

, Pogorzelski

, Mlynarczyk

et al. 2001. Effect of genotype on selected clinical features of Polish cystic fibrosis adults. J Appl Genet, 42:367-377.

19.

Meacham

, Brown

, Murphy

et al. 1993. Characterization of a noncontiguous gene deletion of the growth hormone receptor in Laron's syndrome. J Clin Endocrinol Metab, 77:1379-1383.

20.

Milward

, Metherell

, Maamra

et al. 2004. Growth hormone (GH) insensitivity syndrome due to a GH receptor truncated after Box1, resulting in isolated failure of STAT 5 signal transduction. J Clin Endocrinol Metab, 89:1259-1266.

21.

Mozas

, Castillo

, Tejedor

et al. 2004. Molecular characterization of familial hypercholesterolemia in Spain: identification of 39 novel and 77 recurrent mutations in LDLR. Hum Mutat, 24:187.

22.

Nasioulas

, Jones

, St John

et al. 2001. Profuse familial adenomatous polyposis with an adenomatous polyposis coli exon 3 mutation. Fam Cancer, 1:3-7.

23.

Rosenbloom

, Guevara-Aguirre

, Berg

, Francke

. 1998. Stature in Ecuadorians heterozygous for growth hormone receptor gene E180 splice mutation does not differ from that of homozygous normal relatives. J Clin Endocrinol Metab, 83:2373-2375.

24.

Smith

, Pante-de-Sousa

, Alharbi

et al. 2005. Combination of His-tagged T4 endonuclease VII with microplate array diagonal gel electrophoresis for high-throughput mutation scanning. Clin Chem, 51:1043-1046.

25.

Syddall

, Sayer Aihie

, Dennison

et al. 2005. Cohort profile: the Hertfordshire cohort study. Int J Epidemiol, 34:1234-1242.

26.

Wong

, Yang

, Passey

et al. 2003. A population threshold for functional polymorphisms. Genome Res, 13:1873-1879.

27.

Zhao

, Molnar

, Zhang

et al. 1998. 3′-end cDNA pool suitable for differential display from a small number of cells. Biotechniques, 24:842-850852.