I’m still not sure why Dick Cotton asked me to help co-ordinate a HUGO Mutation Detection Training Course back in 1997. It was at the HUGO Mutation Detection Meeting in Brno, and maybe it was because I said in my lecture, as an aside, "I never met a mutation detection method that I didn’t like". Someone told me straight after the lecture that they never met a mutation detection method that they did like. I feel that way too sometimes, but the ingenuity that scientists have brought to bear on mutation detection technology is fascinating in its own right (even if the techniques themselves don’t always work perfectly). So it was a great privelege, and quite a responsibility, to be asked to help develop a training counterpoint to the HUGO Mutation Detection Research meetings that have been taking place since 1993. Fortunately a group of senior research scientists was persuaded by Dick to act as the Scientific Advisory Panel to develop a suitable programme. The concept was to provide a training and educational forum enabling life and medical scientists to gain a broad familiarity with the theoretical and practical aspects of mutation detection. We were not aiming to develop an EMBO or Cold Spring Harbor course format of intense, focused activity, but to develop a critical overview of the tools available, their applications and their limitations. The first course ran at Hinxton Hall (the Sanger Centre campus) in 1998 and was well received by the participants. Feedback suggested that more wet work would have been well received, and some participants felt that the coffee breaks were too long! We have addressed both of those issues over the years. Subsequent course have run at Leeds University (2000) and in Montpellier (2002, organized by Professor Mireille Claustres). Each course has improved on its predecessor.

The words mutation, polymorphism, sequence variant and so on are all used with varying degrees of precision. For present purposes, mutation can be defined as a sequence change in a test sample compared with the sequence of a reference standard. This definition implies nothing about the phenotypic consequences (e.g. pathogenicity) of the mutation. A polymorphism may then be defined as a mutation that occurs in a substantial proportion (>1%) of a population. Although a polymorphism is tacitly assumed to be non-pathogenic the true pathogenicity may be unknown.

Mutation detection techniques can be divided into those which test for known mutations (genotyping) and those which scan for any mutation in a given target region (mutation scanning). Broader aspects of mutation detection include identification of gene dosage alterations, gross re-arrangements and methylation. The prime considerations in any approach to mutation detection are sensitivity (the proportion of mutations that can be detected) and specificity (the proportion of false positives). Cost per genotype and throughput are also important factors.

Is mutation detection now so trivial that one simply uses an off the shelf system? Not yet! Despite priomises of "$1,000" genome, we are some considerable distance away from such low cost analyses. Further, there are choices to be made regarding the type of study planned: would a simple genotyping technique suffice or is a full mutation analysis required? And for full analysis, is direct sequencing the preferred route, or can more be achieved by prescreeining samples with a scanning method?

Provided that the prescreening process is not unduly costly, any prescreening technique requiring no additional manipulation of PCR product must always cost less than a PCR-based sequencing method, because the sequencing reaction will as a minimum double the time and reagent cost. Our current experience is that sequencing costs 5 to 10 times more than scanning by HPLC or capillary electrophoresis. Multiplexing sample fragments at the pre-screening stage can make further gains in cost effectiveness. It goes without saying that the sensitivity of any pre-screening method must be comparable to sequencing itself. If this condition is met then pre-screening should be cost effective. The break-even point between pre-screening and sequencing depends on the relative cost of the two methods and on the number sequence runs generated by pre-screening. If the cost of pre-screening (p)=1 Unit and the cost of sequencing (s)=5 Units, then the break-even point is when a series generates less than 4 sequencing runs (e.g. by revealing a band mobility-shift on pre-screening) per 5 analyses [(s-p)/(s)]. In most cases, pre-screening will reveal sequence variations much less often than this, and provided that high batch sizes can be made available, pre-screening is likely to be more economical. As standards improve and costs fall, the expectation may develop that all mutation scanning tests, apart from gene dosage, should be by sequencing.

The naïve answer is that mutation scanning must reveal all sequence variants in the sample being tested as compared to a control sequence. The answer becomes more complex when the nature of the sample is considered in more detail. Even germ line mutations may occur as mosaics (in which not all of the cells carry a mutation); tumour samples will contain a mixture of germ line and tumour DNA. These factors, and the choice of genes or parts of genes that require scanning will influence the choice of method. The technical tools available to deliver mutation scanning are many and varied, including conventional di-deoxy sequencing, alternative sequencing chemistries, as well as enzymatic and physico-chemical scanning techniques, some requiring standard, low cost laboratory equipment, others requiring substantial capital investment.

Mutation detection techniques can be described in terms of sensitivity, specificity, throughput (including failure rate), hands-on time, reagent costs and equipment costs. Of these, sensitivity and specificity are the most important, because they set the limits on what an assay can report reliably. Sensitivity has two meanings: i) the proportion of true positive results that the test will report as positive (i.e. absence of false negatives) and ii) the ability to detect a small proportion of positive material in an excess of normal tissue (e.g. tumour DNA in an excess of normal DNA). It is the first meaning that we shall usually refer to when describing the performance of a mutation detection assay, and of course it is desirable that the sensitivity should be as close to 100% as possible, although it is not easy to establish this other than by empirical studies. Specificity is the absence of false positive results; only true positives are scored in a 100% specific assay. In mutation detection, this can be made more demanding by asking that only pathogenic mutations, not normal sequence variants, are reported. For example, the protein truncation test (PTT) will only report chain-terminating (nonsense) mutations. Since nonsne mutations are almost always pathogenic, the assay is very specific. However PTT does not report missense mutations so is less sensitive than denaturing HPLC. On the other hand DHPLC, whilst very sensitive, will not distinguish between pathogenic and non-pathogenic sequence variants, and in this regard is less specific. Of course, direct sequencing would report the presence and type of mutation at the same time.

Whilst it is true to say that the benchmark definition of a mutation is the description of the sequence change with respect to a common reference sequence, the idea that sequencing is the "Gold Standard" for mutation detection is something of a myth, seized upon with alacrity by manufacturers of sequencing kit and reagents. Conventional unidirectional sequencing of PCR products by fluorescent di-deoxy terminators probably has a sensitivity of about 95% (i.e. may miss 1 in 20 mutations) and a low but finite false positive rate due to base misincorporation and other sequencing, optical and polymer artefacts. Bidirectional sequencing (sequencing both strands of an amplicon) reduces the false positive rate and improves the sensitivity; however there are other scanning methods of comparable sensitivity and specificity. Sequencing is clearly an important tool for mutation detection and high quality sequencing should certainly be considered as an option for mutation scanning.

The course structure is divided into two main sections: morning lectures and afternoon lab demonstrations. Because it is easier to work in small groups in the labs, we have split the course into four colour-coded groups. Each lab session will last for 40 minutes and be run for each group. Some lab work is more "hands-on" than other, but given the range of material that we intend to cover, it will not be possible to run closely supervised practical work. This is a very busy course and you will be need to keep to the schedule to get the most out of it. However, please do not hesitate to ask if there is anything that is unclear at any time.

The most valuable part of the course is the contacts that you will make with the other participants, lecturers and demonstrators. So the social side of the course, even though it is quite limited, is just as important as the laboratory and lecture time.

Please help others who may come on courses like this after you by providing honest, constructive feedback based on your own experience. We will distribute feedback forms on the last day of the course.

The DNA of the human genome is subject to a variety of different types of heritable change (mutation). Any change in a specific DNA sequence is a mutation; however, the term "mutation" is generally used to mean disease-causing sequence change, while "allelic variant" means non-causing disease although it may modulate the phenotype of a disease. Allelic sequence variation is usually defined as a polymorphism if more than one allele at a locus occurs with a frequency greater than 0.01. The mean heterozygosity for human genomic DNA is in the order of 0.001-0.004 (that means that 1/250 to 1/1000 bases are different between allelic sequences and these changes are mostly inherited because the mutation rates are low.) Alternatively, single nucleotide polymorphisms (SNPs), which account for about 90% of polymorphisms and represent a natural genetic variability at high density in the human genome, can also convey enhanced or reduced susceptibility toward specific diseases by impairing or improving physiological function as well as modulating drug response. New mutations arise in single individuals, in somatic cells or in the germ-line where they can be transmitted to offspring. Mutations can be induced in our DNA by exposure to a variety of mutagens occurring in our external environment or to mutagens generated in the intracellular environment. However the greater source of mutations is spontaneous errors in DNA replication and repair. Coding and non-coding DNA are equally susceptible to mutation, however the consequences are restricted to the 3% of our 3x10⁹ base-pairs human genome which is coding DNA. Mutations either do not change the sequence of the gene product (silent or synonymous mutations) or result in an altered sequence in a polypeptide or functional RNA (nonsynonymous mutations, which can have no effect, a beneficial effect, or a deleterious effect on gene expression resulting in disease or lethality). The great majority of new nonsynonymous mutations in coding DNA reduce the fitness of their carriers. They are therefore selected against and removed from the population. Occasionally a new mutation may be advantageous in heterozygotes and spread in the population (∆F508 of the CFTR gene).

Large-scale chromosome abnormalities include loss or gain of chromosomes or breakage and rejoining of chromatids. They are rare as constitutional mutations but are often pathogenic. They are much more common as somatic mutations and often found in tumour cells. Such DNA rearrangements may be non-randomly distributed in the genome as a consequence of selection for growth advantage of the mutant cell. The location of these lesions is also influenced by the inherent potential of some DNA sequences to be involved in recombination (either homologous unequal recombination or nonhomologous recombination). To facilitate the search for sequence characteristics that render certain DNA regions prone to rearrangement, a special gross rearrangement breakpoint database (GraDB) has been established (Abeysinghe SS et al, 2004).

Smaller scale mutations mostly occur at three types of DNA sequence at a gene locus : i) the coding sequence of the gene, where the great majority of recorded pathogenic mutations have been identified, ii) intragenic noncoding sequences, especially those which are necessary for correct expression of the gene, and iii) regulatory sequences outside exons, notably promoter elements. Three classes of small-scale mutations can be distinguished : base substitutions, deletions or insertions.

-Base substitutions involve replacement of usually a single base ; in rare cases several clustered bases may be replaced simultaneously as a result of gene conversion event, which mostly occur at certain repeated sequences. Base substitutions are the most common sequence changes both in coding and noncoding DNA. Unexpectedly, transitions (pyrimidine C or T replaced by a pyrimidine, or purine A or G replaced by a purine) are commoner than transversions (pyrimidine replaced by a purine or conversely), especially in mitochondrial DNA. The excess of transitions over transversions is at least partly due to the high frequency of C->T transitions resulting from cytosine methylation and subsequent spontaneous deamination in the CpG dinucleotide, which is often a hotspot for mutation in human genes : its mutation rate is 8.5 times higher than that of the average dinucleotide mutation rate (Krawczak and Cooper 1995). Substitutions result into three different classes of effects on the genetic code: i) nonsense mutations (a codon specifying an amino acid is replaced by a stop codon), almost always associated with a dramatic reduction in gene function), ii) missense mutations (the altered codon specifies a different amino acid that is either chemically similar to it (conservative substitution) or chimically different (nonconservative substitution), and iii) synonymous (silent) substitutions, which do not change the sequence of the gene product as the new codon specifies the same amino acid; they often occur at the third base position of a codon (« wobble »); however in rare cases these substitutions may be not as neutral as expected, causing disease by activating a cryptic splice site or creating an unstable mRNA. Nucleotide substitutions occurring in noncoding DNA usually are neutral variations. Exceptions include certain changes in regulatory elements or in canonic intronic sequence positions, such as at splice junctions (donor or acceptor sites) or at splice branch site. Moreover, a proportion of base substitutions in intronic or even in exonic regions will cause failure of splicing or the use of an alternative illegitimate (cryptic) or natural splice site, resulting in some exon skipping or intron retention in the mRNA.

-Base insertions : one or a few nucleotides are inserted into a sequence ; in rare cases this involves transposition from another locus via an RNA intermediate. Large insertions and duplications are rare, as insertions of transposable elements.

-Base deletions : one or a few nucleotides are eliminated from a sequence.

Microinsertions and deletions are very common in noncoding DNA but rare in coding DNA where they produce frameshifts. Larger deletions are rarer and occur mostly at regions containing tandem repeats or between interspersed repeats. Slipped strand mispairing of short tandem repeat during DNA replication predisposes to frameshifting deletions and insertions.

Nonsense mutations, frameshifting deletions and insertions and splicing mutations skipping a single exon containing a number of nucleotides that cannot be divided by 3, introduce a premature termination codon in the mRNA. The chain-terminating mutations have three major possible consequences : i) unstable mRNA which may be degraded in vivo by a mechanism called nonsense-mediated decay, ii) truncated polypeptide, iii) skipping of constitutive exons.

Dynamic mutations (unstable triplets) These mutations can be stably transmitted from carriers to their offspring without any change through generations. Over the ten past years a new class of mutations has been discovered, which consists in rapid expansion of intragenic or extragenic triplet repeats that can affect gene expression and cause disease. The expanded triplet is unstable, expanding (most often) or contracting (rarely) the number of its repeats when transmitted from parent to child. There are 10 different trinucleotide repeats in the human genome. Most of these are known as useful polymorphic markers. However, close to or within a few number of genes, certain triplets show abnormal behaviour. Usually, their length (their number of repeats) are stable in mitosis and meiosis in the normal population; however, above certain threshold lengths (in certain cases, these alleles are considered as a premutation), the repeats become extremely unstable, depending on the sex of the transmitting parent as well as the length and composition of the repeat.

Genes showing unstable expanding trinucleotide repeats can be divided into two major classes :

i) modest expansions of (CAG)n repeats within the coding sequence, in the order of 30 repeats for the polymorphic alleles and 40-200 repeats for the pathological alleles (ex. Huntington disease); the resulting protein product with abnormal polyglutamine tract shows a gain of function 

ii) genes with very large expansions of a noncoding repeat that can be found in the promoter, the untranslated regions or intronic sequences; unstable pathological alleles have several hundreds (or thousands) of copies which inhibit gene expression, causing loss of function (ex. Fragile-X syndrome). Recently, other classes of expanded triplets have been discovered as causing disease, including polyalanine tracts.

A variety of mechanisms which affect the chromatin environment of a gene and hence its capacity for gene expression have been discovered in the last few years. They can be responsible for genetic changes which are heritable although they are not directly attributable to DNA sequence changes :

-DNA methylation is an important mammalian factor of gene control, maintaining repression of transcription. Pathogenic silencing by methylation occurs in many tumors and can also be triggered by a local expansion of a trinucleotide repeat as in the case of Fragile-X syndrome.

-Genomic imprinting of genes : a small number of genes are subjected to allelic exclusion according to parent of origin, which reflects a nonequivalence in expression of the maternal and paternal genomes at these loci. Disorders of imprinting may be caused by deletions or point mutations in imprinted genes themselves, if an individual inherits two copies of one gene from one parent and none from the other (uniparental disomy), or if the imprinting center is not functioning normally. There are now several examples where alterations of certain chromosomal regions produce different phenotype depending on the parent of origin. The best example is deletion of 15q12, which on the paternal chromosome produces Prader-Willi syndrome and on the maternal chromosome produces Angelman syndrome. The parental imprint is erased and reset between each generation during gametogenesis. The mechanisms that are able to distinguish between maternally and paternally inherited alleles are not deciphered, but allele-specific DNA methylation has been shown to be a key component in maintaining the imprinted status.

-Changes in large-scale chromatin configuration can also exacerbate or silence expression of intact genes. Examples include position effects in aniridia (silencing of PAX6 gene) or in Fascioscapulohumeral muscular dystrophy (probable gene silencing by a 3.2-kb deletion in 4q telomeric repetitive sequences).

Because of the very large size of the nuclear genome, most mutations occur in nuclear DNA sequences. However, the maternally-inherited mitochondrial genome (16,000 bases, about 1/200.000 of the size of the nuclear genome) is a hotspot for pathogenic mutations because of two main factors: i) 93% of mitochondrial DNA is coding, ii) the mtDNA has a higher mutation rate than the nuclear genome, being not protected by histones, undergoing many rounds of replication, and not being repaired as efficiently as the nuclear genome. Thus the mitochondrial genome is a significant cause of human genetic disease. A characteristic feature of mitochondrial diseases is their matrilineal inheritance (as sperm mitochondria are lost soon after fertilization, sperm do not contribute mitochondria in the zygote), affecting both sexes but transmitted only by affected mothers. The zygote receives its total complement of mtDNA (>10⁵ copies) entirely from the cytoplasm of the oocyte. If a portion of these mtDNA molecules contains a mutation, the random segregation of mutant and wild type mitochondria during mitosis will result in daughter cells with different proportions of mitochondrial mutations (heteroplasmy). Only a few mitochondrial disorders are characterized by a mtDNA mutation which is present in virtually all cells (homoplasmy).

The degree to which a pathogenic mutation results in an aberrant phenotype depends on several factors including i) the mutation class and the way in which the expression of the mutant gene is altered (loss of function, gain of function, dominant-negative effect, overexpression or ectopic expression…), ii) the degree to which the aberrant phenotype is expressed in the heterozygote, iii) the proportion and nature of cells in which the mutant gene is present (inherited or somatic mutations) and, for a few number of genes, the parental origin of the mutation (a few genes are imprinted). Since Haldane first observed that most de novo mutations resulting in hemophilia were generated in the male germline, direct observation of disease-causing mutations supports higher mutation rates in males, most likely to be related to the greater number of germ cell divisions in males.

Since 1979, where the first single-base pair disease-causing substitution was described for ßthalassemia, about 100,000 mutations have so far been reported in 2,000 genes (data extrapolated from locus specific mutation databases contents, see HUGO HGVS website). The spectrum of DNA-disease producing mutations range from large gene alterations (insertions, deletions, duplications, inversions, expansion of trinucleotide repeats), micro-deletions or -insertions and single-base substitutions. Nearly all common monogenic diseases (200-250) and a large number of rare ones (1000-1200) have been traced back to one or more defective genes (3000-4000). New genetic mechanisms have been uncovered, such as genomic imprinting, triplet-repeat expansion, deficiencies of DNA repair and methylation defects.

The number and frequency of disease-causing or predisposing alleles varies enormously from locus to locus. Rare diseases are generally caused by a wide range of different mutations, with each mutation constituting only a small fraction of the spectrum. Some disorders, such as cystic fibrosis, are associated with a small number of common mutations and a background of more than 1,200 very rare alleles already identified. The common disease/common variant (CD/CV) hypothesis, proposed several years ago (Lander et al, 1996; Cargill et al,1999; Chakravarti 1999) predicts that the genetic risk for common disorders will often be due to one or a few predominating disease alleles. Some prototypical examples include the apoE E4 allele in Alzheimer disease or Factor V Leiden in deep venous thrombosis. Reich and Lander (2001) provided a framework for understanding and predicting the allelic spectra of disease genes, assuming that the human population experienced a dramatic expansion from a small, ancestral population with a size on the order of 10.000 individuals to the modern population size of 6 milliards that began between 20.000-150.000 years ago. According to their hypothesis, the increase in allelic diversity will occur rapidly for the rare diseases but will require hundreds of thousands years after the expansion for the common diseases, which is rather optimistic in the issue of mutation detection.

Genetic mosaicism can be defined as the presence in a single individual of two genetically distinct populations of cells that differ from each other at the level of DNA sequence but that derive from a single zygote. Germ-line and somatic mosaicism have now emerged as important factors that contribute to phenotypic variability. There are increasing reports of multiple different types of mosaicism detected in patients with inherited and non-inherited disorders including chromosomal mosaicism, mitochondrial mosaicism (heteroplasmy), germ-line mosaicism (de novo mutation), somatic mosaicism (due to either de novo mutation during embryogenesis or to reversion to normal of inherited mutations), and mosaicism of neoplasia (where DNA changes occur in the tumour cells but not in the other cells of the body, such as loss of heterozygosity (LOH) by mitotic recombination or microsatellite instability in HNPCC syndromes) (Hirschhorn 2003, Youssoufian and Pyeritz, 2002). Germ-line mosaicism is suspected when several offspring of unaffected parent manifest the phenotype, which is otherwise transmitted as autosomal dominant or X-linked disorder. Phenotypically normal individuals might transmit several gametes that are clonal descendants of a single progenitor cell in which a de novo mutation occurred during the early development of a parent. There is now a long list of monogenic disorders in which somatic mosaicism has been shown including metabolic diseases, skeletal dysplasias, muscular dystrophy (DMD), clotting disorders and tumour-suppressor disorders. Mosaicism is underestimated, as genetic testing for mutations is made by analysing DNA in blood cells, in which a low level of mosaicism will remain usually undetected.

The first area benefiting directly from the discovery of causative mutations is that of molecular diagnostics in affected families. The ability to determine who is and who is not at risk for disease, sometimes before the onset of symptoms, is becoming essential for proper management of patients and their families. In individuals who inherit mutant genes, preventative measures might reduce morbidity and mortality and allow more thoughtful planning for the future. The benefits of genetic testing are equally important for those family members who are found not to carry the relevant mutation: these individuals are spared unnecessary medical procedures and tremendous anxiety.

In a few number of diseases there are sites within the gene that are mutated in virtually all cases. Examples include sickle cell disease, in which an A to T transversion at codon 6 of the ßglobin gene is ever-present, and Huntington’s disease, in which virtually all patients have an expanded tract of CAG trinucleotide repeats, creating a long polyglutamine stretch within the huntingtin protein. These mutations can be readily identified with assays designed to detect the specific alterations. However, the majority of inherited diseases are caused by diverse mutations scattered along the length of the affected gene. In non-genetically isolated populations without a history of severe bottlenecks, many highly penetrant disease genes have complex mutation spectra. The nature of mutations is diverse and the ease of detecting them stretches across a continum between « compliant mutations », which can be detected or identified by using routine PCR-based genetic tests, and « refractory mutations » (including deletions encompassing one or more exons, insertions, translocations or a variety of alterations that affect the expression of the gene at the RNA or protein level) which will not be detected by routine testing (Yan et al, 2000).

This has led to the development of two broad categories of mutation detection technologies: the first group, designed to scan for unknown mutations within a gene, are highly informative but are tedious and are incompatible with high throughput and low cost. In the second group, direct methods focused on specific mutations have been developed; together with robotics, these methods for direct mutation analysis have helped in reducing cost and increasing throughput, but only a limited number of mutations can be analyzed. Because of the extreme diversity of DNA alterations, no single method of detection is applicable for all situations, and clinical laboratories have often to combine several techniques in order to screen for all possible changes in the homozygous and heterozygous states.

Testing for the presence or absence of a known sequence change is a much simpler problem than scanning a gene for the presence of any mutation. Consequently, a number of specific mutation detection assays are commercially available. In one set of methods, mutations are analyzed after the target sequence has been amplified by PCR: they are detected by restriction digest, allele-specific hybridization, or by ligation or nonligation of adjacent probes. In a second set of methods, PCR is part of the detection system, and they rely on the selective extension of primers or on real-time quantitative analysis of the products. Non gel-based detection systems have been developed for most of the assays described, making these methods favorable for application in routine laboratories. For each technique, reaction conditions must be standardized and appropriate internal controls must be included. One must keep in mind that misleading results may be obtained because of unknown polymorphisms within the target region affecting either restriction enzyme recognition sequences or hybridization of probes and binding of primers.

-Restriction digestion analysis of PCR-amplified DNA has been the simplest way of testing for the presence or absence of a mutation when it creates or abolishes either a natural restriction site or one engineered by a form of PCR mutagenesis using a primer carefully designed to have a mismatcched nucleotide which together with the mutant sequence creates a restriction site not present in the normal allele. Gains in throughput can be achieved using high-density gels such as the MADGE (microtiter array diagonal gel electrophoresis) format (O’Dell et al, 2000).

A number of methods rely on hybridization with specific oligonucleotide probes that can effectively discriminate between the wild-type and variant sequences.

-Allele-specific oligonucleotide (ASO) hybridization. Using suitably stringent hybridization conditions, short allele-specific synthetic probes hybridize to the target sequence only when they match perfectly. Wild type or mutant ASO probes are usually designed to have differences in the central segment of the sequence in order to maximize thermodynamic instability of mismatched duplexes. The original dot blot or slot blot formats have evolved into nonradioactive detection in reverse dot blots and multiple solid-phase or other multiplex allele-specific diagnostic assays. ASO hybridization is also the principle on which the design of DNA chips was based.

-Allele-specific amplication (ASA). Complementary either to the wild type or mutant sequence, ASA has been assayed through several approaches, of which the most popular is the amplification refractory mutation system (Newton CR et al, 1989). In ARMS there is a lack of primer elongation due to a mismatch at the far 3’-end of the primer. Paired PCR reactions have to be carried out using a common primer and either the primer for the normal sequence or the primer for the mutated one. The technique can be multiplexed to type up to 20 to 30 mutations simultaneously (ex. Cystic fibrosis mutations), and high-throughput formats can be achieved using closed-tube assays. Other methods rely on a lack of primer annealing due to a mismatch located within the primer.

-Single nucleotide primer extension (PEX), or "minisequencing". Extension of the primer by one base occurs only when the labeled nucleotide is complementary to the nucleotide of the target DNA adjacent to the 3’-end of the primer. When incubated in the presence of dideoxynucleotide triphosphates labeled with different fluorophores, the allele-specific dye-labeled dNTP is linked to the primer in the presence of DNA polymerase and target DNA. Simply exciting the fluorescent dye in the reaction and determining the change in fluorescence can determine the genotype of the target DNA molecule. PEX is a very robust allelic discrimination method. It is highly flexible and requires the smallest number of primers/probes. Probe design and optimization of the assay are usually very straightforward. Minisequencing can be run on DNA sequencers, such as the "SnapShot" assay from Applied Biosystems. Mass spectrometry, in which molecules are ionized and volatilized in a vacuum and separated by their mass-to-charge ratios, offers an alternative highly automatable detection scheme using MALDI-TOF-MS, Matrix-assisted-laser desorption ionization- time-of-flight- mass spectrometry. A powerful multiplex SNP analysis using solid-phase-capturable biotinylated dideoxynucleotide terminators has been recently designed (Kim et al, 2003), where measuring the mass of purified primer-extension products identifies the nucleotide at the polymorphic site. Both homozygous and heterozygous genotypes for human hereditary haemochromatosis C282Y and H63D were clearly determined.

-Allele-specific ligation: Oligonucleotide ligation assay (OLA (Nickerson et al, 1990). Three oligonucleotides are used in OLA: two allele-specific probes (one specific for the wild-type allele and the other specific for the mutant allele) plus a fluorescent common probe. The forward primer has its 3'base placed at the position of the variation and located immediately adjacent to the 5’-end of the common probe. Under appropriate conditions, ligation of both primers by DNA ligase will take place only when the 3’-end of the primer matches perfectly with the target sequence. The first step is PCR, the second ligation. The ligation products are then separated by electrophoresis. Solid-phase formats, single-tube multiplex amplification and multi-colour fluorescence detection on automated sequencers have been described (by varying the combinations of colour dyes and probe lengths, multiple mutations can be detected in a single reaction, see for example the OLA test for 31 common CFTR mutations).

Many of the techniques used routinely in a wide range of molecular genetics rely on PCR amplification of a target molecule followed by a variety of post-amplification analyses to detect the specific analyte amplified. Development of thermocyclers that can take fluorescence measurements during PCR amplification have enabled genetic testing in single close-tube reactions, with the ability to amplify, detect and quantify DNA targets in the tube as PCR proceeds (reviewed by Foy and Parkes 2001). Advantages over traditional methods include speed, improved reproducibility, reduced risk of contamination, possible automation and high-throughput processing in microplates, and the ability to more accurately quantify the amount of starting material present as the methods allow for « real-time » monitoring of the entire PCR reaction. The signal detected directly correlates with the amount of PCR accumulating, which in the log-linear phase of the amplification process is dependent on initial target copy number. Performing measurements in the log-linear phase of amplification enables more accurate quantification than was previously possible with gel-based techniques that examine end-point reactions. Since its development in the 1990s, many different assay formats have been developed and the number of real-time PCR machines of different design is continuously increasing. A recent review provides a survey of the instruments and assay formats available and discuses the pros and cons of each (Wilhelm and Pingoud, 2003). The current chemistries used to monitor the amplification process include either non-specific double-stranded DNA-binding dyes such as SYBRGreen, which produces enhanced fluorescence signals upon binding to double strand DNA duplexes as PCR proceeds and is mostly used for gene dosage, or specific probe hybridization-based systems that rely on the principle of FRET (fluorescence resonance energy transfer) for signal generation and are mostly used for genotyping. FRET is a process in which energy from an excited donor fluorophore is transferred to a nearby acceptor dye. When two fluorophores whose excitation and emission spectra overlap are in close physical proximity, excitation of one fluorophore will cause it to emit light at wavelengths that are absorbed by and that stimulate the second fluorophore, causing it to fluoresce.

- The LightCycler system uses two adjacent probes that are labeled such that when both probes are hybridized to the target, the labels are brought close to each other; the donor fluorophore is excited by the light source of the instrument, and energy is transferred from the donor to the acceptor, producing an increase in fluorescence from the acceptor fluorophore. The system is programmed to monitor the melting curve analysis of allele-specific FRET probes after the PCR, allowing direct typing of sample without any further processing.

- TapMan probes, Molecular Beacons and Scorpion primers rely on the close proximity of donor fluorophore and non-fluorophore acceptor molecules (quenchers). In the unhybridized probe conformation, little or no signal is generated (the fluorophore of the donor is quenched). Upon hybridization to the target, the fluorophore and quencher become separated through either conformational changes that occur (Molecular Beacons and Scorpion primers) or enzymatic cleavage of the fluorophore from the quencher as a result of the 5’nuclease activity of Taq polymerase (TaqMan probes). The physical separation of the fluorophore and quencher moieties produces an increase in signal. Mutation typing is determined by measurement of the signal intensity of the two allele-specific reporter dyes after PCR amplification. Because of the presence of a hairpin-loop structure, molecular beacons and scorpion primers might have a higher specificity than TaqMan probes.

-Allele-specific nucleotide incorporation. Pyrosequencing is a real-time sequencing method in which DNA polymerase catalyzes the incorporation of a dNTP during the extension of the sequencing primer only if it is complementary to the base in the template strand (Ronaghi et al, 1999). Unlike traditional sequencing, it does not require any electrophoresis. The generation of pyrophosphate when the dNTP is incorporated is coupled to a luciferase-catalyzed reaction resulting in light emission and yielding a quantitative and distinctive pyrogram. Genotyping of 34 common mutations in the CFTR gene is commercially available and current instrumentation sold by Pyrosequencing AB (Uppsala, Sweden) has been adapted for high-throughput detection of SNPs.

-Hybridization chips contain hundreds of oligonucleotides matching normal or mutant sequences in a gene or in several genes, thus offering simultaneous analysis of many polymorphisms or mutations. The DNA sample of interest is PCR-amplified to incorporate fluorescently labeled nucleotides and then hybridized to the array. Each oligonucleotide in the high-density array acts as an allele-specific probe. Perfectly matched sequences hybridize more efficiently to their corresponding oligomers on the array and, therefore, give stronger fluorescent signals over mismatched probe-target combinations. The hybridization signals are quantified by high-resolution fluorescent scanning and analyzed by computer software.

-Minisequencing chips use arrayed oligonucleotide primers with free 3’-OH ends. Unlabeled PCR-amplified test DNA is hybridized to the array and acts as template for addition of a single labelled dNTP to each arrayed primer : extension will occur only if the 3’end of the primer perfectly matches the template.

Although DNA chip technology is in commercial use and has yielded vast amounts of genetic and cellular information, major improvements are needed to make this technology an accurate and inexpensive tool for the detection of multiple types of homozygous or heterozygous mutations in the diagnostic laboratories (Chen and Ren 2004; Larsen et al, 2001, Mantripragada et al, 2004).

Two methods are sensitive enough to work directly from genomic DNA.

-The Invader assay (Third Wave Technology) is a FRET-based method in which two oligonucleotides (a wild-type or variant signal probe) plus an upstream Invader probe are used in each reaction. The signal probe contains a portion perfectly matched to the target and a fluorescently labeled flap. These probe hybridize in tandem to a specific region of genomic DNA. When the 3’end of the Invader oligonucleotide overlaps the hybridization site of the 5’end of the dowstream signal probe by at least one bp (when the upstream probes "invades" at least one nucleotide into the downstream duplex), the unique structure will be recognized and cleaved by Cleavase enzymes. The cleaved signal probe dissociates, and a new signal probe anneals and is subsequently cleaved. The process repeats multiple times, resulting in an accumulation of reporter molecules and producing linear amplification of the fluorescent signal. A single nucleotide mismatch between the Invader probe and the template renders the conformation unrecognizable by Cleavase and there is no signal generated. Thus, instead of amplifying the target, the Invader procedure results in the accumulation of signal molecule only in case of perfect match between the downstream signal probe and the target.

-The rolling circle amplification (RCA) method is based on ligation-dependent circularization of allele specific DNA probes followed by exponential rolling circle amplification of the circularized probes (also called "padlock" probes). Mutations are interrogated with open circle probes that can be circularized by DNA ligase when the probe matches the genotype. An amplified detection signal is generated by exponential rolling circle amplification (ERCA) of the circularized probe. The ERCA products for normal and mutant alleles incorporate a different fluorescent primer.

Sequence-specific base pairing between the strands of DNA according to the Watson-Crick model forms the basis of many detection systems. The crucial specificity of this hybridization reaction in discriminating between single base variations may be enhanced by using synthetic PNA (peptide nucleic acids) and LNA (locked nucleic acid) profluorescent probes (Latorra et al, 2003; Igloi 2003).

Direct detection of mutations is not simple, for many reasons that are related both to genes themselves and to technologies. It is the scanning of large genes for private mutations that is the rate-limiting step. For a number of diseases, complete sequencing of large genes cannot be a routine procedure in clinical laboratories, not only because it is time-consuming, costly and tedious, but also because mutations can be missed. Although introduction of capillary electrophoresis in sequencing technology has improved and simplified dramatically the process, DNA sequencing, like any other technique, does not detect 100% of point mutations. Moreover, depending on the size of the genes, it may not be practical to carry out sequencing of vast stretches of DNA without knowing whether or not mutations exist and without knowing in advance the sites of mutations in the target genes. The alternative approach is to amplify exons and introns of interest by PCR from genomic DNA, scan the PCR products for the presence of DNA variants by a scanning method and then use limited sequencing to confirm and identify the mutation in the abnormal fragment (most scanning methods do not identify the precise nature of the mutant allele). The requirements for clinical routine diagnostic and for research use are rather different in terms of mutation detection sensitivity and specificity, speed and reproducibility of the assay. In the diagnostic lab, considerable time can be spent optimizing a test for a particular gene, provided the method once developed is quick and simple.

Conformation-based methods of mutation detection rely on the fact that DNA fragments containing a sequence alteration have altered mobility under certain conditions of gel electrophoresis. A shift in the position of the band representing the mutant DNA relative to the position of the band for normal DNA connotes the presence of a mutation somewhere within the test fragment, but it does not indicate the precise position or identity of the mutation. A single-base mismatch can produce conformational changes in the double helix that cause the differential migration of homoduplexes and heteroduplexes containing mismatches. Several methods (DGGE, HA, DHPLC) rely on formation of heteroduplexes. Most mutations occur in heterozygous form; for homozygous mutations or X-linked mutations in males, it is necessary to add some reference wild-type DNA.

- The first conformation-based procedure introduced for mutation detection was Denaturing Gradient Gel Electrophoresis (DGGE) (Myers et al, 1985,1987), which detects mutations as differences in partial melting behaviour of double strand DNA and hence, differences in migration distance when subjected to increasingly denaturing conditions, as regions of single stranded DNA within double-stranded fragments greatly retard the migration of DNA through acrylamide gels. Visual inspection of the gel immediately detects heterozygote samples (presence of homo- and heteroduplexes). Primers are made with GC clamps or with modifiers which cross-link the ends of the PCR product ; these clamps reduce the likehood of complete denaturation of the PCR product. Theoretical melting profiles of each DNA fragment can be predicted by appropriate computer programs (MELT87 and SQHTX). However running conditions must be defined for each PCR product and these laborious electrophoretic methods are evaluated by visual inspection, which makes standardization difficult. It is now well established that denaturing (DGGE) or temperature gradient gel electrophoresis (TGGE) (Henco et al, 1994) detect close to 98% of point mutations in fragments under 400 bp in size and under optimal conditions, so that it is appreciated by clinical laboratories. Some multiplexing of PCR products is possible, allowing the evaluation of several exons simultaneously. However the principal disadvantage of DGGE is that it is time-consuming and of low throughput.

The DGGE method has seen interesting recent adaptations such as constant denaturant capillary electrophoresis (CDGE) ( Hovig et al, 1991), and two-dimensional electrophoresis (this latter combines the high resolution of 2-D electrophoresis with DGGE) (Vijg and van Orsouw , 1999).

- Heteroduplex analysis (Het) or (HA) is a very simple method, performed in non-denaturing gels that carries the advantage that the assay conditions do not have to be determined for each DNA fragment. However, it detects lower mutation than DGGE. A variant of HA analysis is the Conformation sensitive gel electrophoresis (CSGE), where an appropriate system of mildly denaturing solvents can amplify the tendency of single-base mismatch to produce conformational changes (Ganguly 2002).

-Denaturing high-performance chromatography (DHPLC) (reviewed in Xiao and Oefner , 2001; Frueh and Noyer-Weidner 2003; Kwok and Chen, 2003) exploits the differential retention of homoduplex and heteroduplex molecules under partially denaturing conditions during reverse-phase ion-exchange HPLC. It combines the non-ionic size separation of DNA fragments with thermal denaturation. In addition to the fact that temperature conditions can be accurately reproduced, the main advantage over electrophoretic approaches is that the procedure can be automated for high-throughput mutation or SNP scanning. The wave DNA analysis system (commercialized from Transgenomic, Inc) uses temperature-modulated heteroduplex analysis (TMHA). The heteroduplexes can be separated from the homoduplexes by column chromatography at a temperature that partially denatures the mismatched DNA. Fully denaturing temperatures allow for separation of single stranded molecules in applications such as oligonucleotide purification or primer extension assays, whereas partially denaturing conditions permit accurate separation of sequence variants. WaveMaker software provides instrument and analysis control to precisely optimize assay conditions. The system offers a powerful alternative to gel electrophoresis with rapid protocols and high throughput capabilities, so that it has become the most popular method for mutation scanning both in research and diagnostic. The main disadvantage is the high cost of system and reagents.

- The single-strand conformation polymorphism (SSCP) or analysis (SSCA) (Orita et al, 1989) has been the most widely used method for scanning genes, due to its technical simplicity. It relies upon the principle that secondary structure contributes to the mobility of single-stranded DNA in electrophoretic gels under non-denaturing conditions. Single-stranded DNAs are generated by denaturation of the PCR products and separated on a non-denaturing polyacrylamide gel. A fragment with a single base modification generally forms a different conformer and migrates differently when compared with wild-type DNA. However it has a major limitation due to unpredictable electrophoresis behaviour of the mutant strands, so different sets of conditions of electrophoresis (eg gel temperature and/or presence of glycerol) have to be conducted to achieve maximum sensitivity. Furthermore, the sensitivity decreases as the size of the fragment being analyzed increases. Multiple refinements and improvements of SSCA have been described, including methylcellulose as molecular sieving agent and post-PCR fluorescent labelling dye. Reduction in migration times, better resolution and reproducibility may be reached using capillary electrophoresis and fluorescence detection (Larsen et al., 2001, Doi K et al, 2004). Variations of the method include dideoxyfingerprinting (ddF) (Liu et al, 1996) and bi-directional dideoxyfingerprinting (bi-ddF), in which dideoxy chain terminated products are analysed by SSCP.

Several techniques exploit the fact that mismatched bases are sensitive to binding or cleavage by enzymes or chemicals. After PCR amplification, the wild type and variant products are subjected to denaturation/renaturation to create heteroduplex molecules. After incubation with resolvases or chemicals, the products are resolved electrophoretically side by side to score for the presence of mismatch-cleaved DNAs. Cleavage methods mostly work on heteroduplex molecules, cleaving the helix at the distorded region caused by the mismatch. This means that homozygous or hemizygous mutations would not be detected unless the sample is hybridized with a reference sample to form the mismatched heteroduplex. These methods are able to scan larger fragments than conformation-based techniques, and the size of the cleaved product roughly indicates the localization of the mutation, however they require considerable post-PCR manipulation.

- Ribonuclease mismatch cleavage. The single-strand specificity of RNase has been utilized to digest RNA:RNA or RNA:DNA heteroduplexes. The technique has been adapted to commercially available nonisotopic RNase cleavage assays (NIRCA) (Goldrick MM, 1996) with higher mutation rates than in the original versions. However the major drawback of these methods is that they require in vitro synthesis of RNA.

- The Chemical cleavage method (CCM) (Cotton et al, 1988) employed hydroxylamine and osmium tetroxide to modify DNA at the site of base mismatch. The modified DNA is then cleaved with piperidine and the reaction products are separated by PAGE. CCM can detect mutations in DNA fragments up to 1.4 Kb in length with almost 100% efficiency. CCM has evolved over the years from radiolabelled to non-toxic fluorescently labelled multiplex methods. However CCM is not well suited to routine lab uses because of the number of analytic steps.

- Enzyme mismatch cleavage (EMC) by resolvases (Solaro et al, 1993) and other endonucleases including single strand specific nucleases (Till et al, 2004) holds promise for the future, since standardization and automation using capillary electrophoresis and microdevices should be achieved with relative ease. However, more experience is needed with these methods, particularly for the analysis of heterozygous states.

- The Base excision sequence scanning (BESS-T and BESS-G) (Hawkins GA 1997) also known as glycosylase mediated polymorphism detection assay (GMPD) does not require the formation of heteroduplexes: it detects mutations by cleavage of PCR products at modified nucleotides (for instance thymidine replaced by uridine), generating defined series of fragments similarly to a sequencing ladder. The usage of uracylglycosylase may be interesting because in about 80% of all base substitutions, a thymidine is involved at the sense or the antisense DNA strand.

- Mismatch binding proteins (MutS and homologs) have been used either in direct or in exonuclease protection assays. Given the normal role of bacterial mismatch-repair proteins to replace mis-incorporated bases in newly synthesized DNA, extensive efforts have been made to adapt these proteins for mutations detection assays. These proteins bind to the site of mismatch and do not cleave the DNA.

The protein truncation test (PTT) (Roest et al, 1993) selectively detects translation-terminating mutations, which are revealed on the protein level by SDS-PAGE. PCR is performed with RNA or DNA using forward primers containing a T7 promoter sequence and an eukaryotic translation initiation sequence. Subsequently, PCR products are used as templates in coupled transcription-translation reactions. The size of labeled translation products is analyzed by gel electrophoresis : stop codons generated by frameshift or nonsense mutations lead to a reduced size of translated proteins. The assay is best suited for diseases where translating termination mutations predominate, as it avoids the confusion with silent polymorphisms. The PTT assay is usually performed on cDNA copies of the mRNA of interest, where segments gathering together many exons can be scanned in only one reaction. For instance, RT-PCR-PTT has been sucessfully applied to the detection of point mutation in the large DMD (Duchenne muscular dystrophy) gene. In some cases, it can be done directly from genomic DNA to monitor large exons of several kbs. The translated products may be visually inspected, or separated by chromatography or capillary electrophoresis, allowing an objective measurement of truncated proteins. A recent modification using N-terminally tagged test peptides and measuring peptide masses by MALDI-TOF should allow to detect non only truncations but also amino acid substitutions (Garwin et al., 2000). Recently a high-throughput solid-phase protein truncation test (HTS-PTT) has been described where capture and detection of translation products are accomplished in a single well using a standard 96-well microtiter plate enzyme-linked immunosorbent assay (ELISA) format and chemiluminescence readout (Gite et al, 2003).

Direct DNA sequencing is, in theory, the most accurate technique for point mutation detection. However, even the present state-of-the-art capillary Sanger-DNA sequencing with laser-induced multi-colour fluorescence detection faces difficulties in accurately sequencing GC-rich regions owing to the compression of the DNA fragments. In cases where the mutation affects only one of the two alleles of the gene, the peak heights corresponding to wild type and mutant alleles are not always present at the predicted ratio of 1:1. Therefore, most genetic testing laboratories sequence both strands of DNA to achieve maximum sensitivities and to provide independent confirmation of any putative mutation identified. Direct sequencing of PCR products is often used as a final step of any mutation scanning procedure, both confirming and identifying the sequence alteration.

-The next future of sequencing could be mass spectrometry (MS). Because DNA sequences are determined by accurately measuring the mass of the DNA fragments, DNA must be stringently purified and be free from salts and other contaminants. In the SPC sequencing (solid-phase-capturable biotinylated dideoxynucleotide terminators) (Ju J, 2002), DNA sequencing fragments that are terminated by biotin-ddNTPs at the 3’end in a cycle-sequencing reaction are bound to a streptavidin-coated solid phase. All excess primers, salts and falsely terminated DNA fragments are washed away to provide pure DNA samples for MALDI-TOF (matrix-assisted laser desorption/ionization time-of-flight) MS detection. Segments up to 100 bp can be read in a very short time. Moreover, in case of frameshift mutation in the heterozygote, the sequences of both alleles can be simultaneously read and the mutation site can be unambiguously identified (Ruparel et al, 2004).

Mutation detection in clinical diagnostic labs is applied predominantly to patient's genomic DNA. While this approach is expected to detect the majority of mutations causing single gene disorders, an increased number of mutations are coming to light that will not be detected using standard mutation screening methods on genomic DNA. In order to offer comprehensive mutation screening in genes with large introns such as the dystrophin gene, an RNA based approach is likely to be required. If a gene with many exons has to be scanned for unknown mutations, testing by RT-PCR offers several advantages: i) the coding sequence can be scanned through a much smaller number of PCR reactions than genomic DNA, ii) the procedure can detect aberrant splicing caused by sequence changes in introns or in exons, iii) some large deletions or duplications that would remain undetectable with DNA testing can be detected by aberrant sizes of RT-PCR products. However, RNA is more difficult to obtain and handle due to rapid degradation of mRNA, low expression in available tissues including illegitimate or ectopic expression from blood white cells. A potential limitation is unstable mRNA generated by some mutations (truncative mutations can result in nonsense-mediated mRNA decay) so that the RT-PCR product from an heterozygous individual may show only the normal allele. It is possible to confirm the expression of the two alleles through heterozygosity for a polymorphism.

A potential advantage of mRNA analysis is to avoid the co-amplification of a gene with pseudogenes (in the genomic DNA approach, it is necessary to designe primers specific to the gene and not to its pseudogenes).

Sequencing RT-PCR products without previous scanning may be, for some genes, the method of choice to detect and identify the mutation in a single step. However much time can be wasted investigating artefacts, especially if the sequencing template is not of the highest quality. Whatever the method, it is necessary to confirm RNA alterations in genomic DNA of the patients. The group of Estivill reported the first description of an alternative splicing induced by cold shock conditions in a human pre-mRNA. The accuracy of the splicing mechanism was perturbed after keeping samples for a short period of time at room temperature, resulting in the insertion of a 31-bp cryptic exon in NF1 mRNA (Ars et al, 2000). Although rare, this phenomenon must be taken into account when using RNA technology as it can lead to misdiagnosis of a disease-causing mutation.

Many mutations are difficult or impossible to detect with the techniques described above. For example, if the genomic region examined is deleted from the mutant allele, the PCR product that is obtained from genomic DNA will be exclusively derived from the wild-type allele, leading to the false conclusion that this region is normal. The type of mutations that are refractory to detection include deletions encompassing one or more exons, insertions, translocations or a variety of alterations that affect the expression of the gene at the RNA or protein level. Intronic mutations that affect splicing can be revealed through the analysis of RNA. Polymorphisms, when present within an mRNA transcript, can be used to assess relative levels of expression of the two alleles of the gene. Quantitative hybridization, quantitative PCR or Southern blotting can theoretically detect deletions of one or a few exons. Gross rearrangements are easily detected at the cytogenetic level by using FISH-based methods (deletions or duplications in the order of the megabase), array-CGH-based methods (deletions or duplications in the order of 200-500 kb) or by using Southern blotting (deletions or duplications in the order of the kilobase). The Southern blot method is on the way of being abandoned by most diagnostic laboratories because it requires big amounts of DNA of good quality, uses radioactive probes and takes several days. Consequently, as mid-size deletions or duplications are rarely screened, they are usually underestimated. The publication of the draft sequence of the human genome has facilitated the development of novel techniques for assessment of copy number at multiple loci (reviewed by Sellner and Taylor, 2004)

Quantitative Real-time PCR is now the method of choice for absolute and relative quantification of DNA and RNA template molecules in a variety of applications including allele imbalances, detection of deletions in genes and gene dosage. Real-time PCR is also becoming increasingly important in the diagnosis of tumors, such as for the detection and monitoring of minimal residual diseases, the identification of micrometastases in colorectal cancer, neuroblastoma, and prostate cancer. It has been used to quantify amplifications of oncogenes, or deletions of tumor suppressor genes in tumor samples. Also, the results of DNA chip experiments are validated by real-time PCR quantifications. However, these technologies pose many challenges to the clinical diagnostic setting where the interpretation of data generated can be affected by many factors including the cutoffs to distinguish positive from negative samples, the assay controls and quantification calibrators used and the ability to accurately distinguish specific from non-specific products. There is a need for greater standardization of the methods used and inter-laboratory validation of the technologies. Also the number of alleles that can be analyzed at the same time is limited.

Large heterozygous deletions or duplications in genes can easily be detected by using various formats of fluorescent multiplex PCR assays designed to amplify simultaneously multiple target sequences under quantitative conditions (Yau et al, 1996; Robinson et al, 2000, Saugier-Veber et al, 2001); the sensitivity of the method has recently been improved in QMPSF (quantitative multiplex PCR of short fluorescent fragment) (Casilli et al. 2002).

Two new methods rely on semi-quantitative analysis of PCR products using specifically bound probes that are amplified with universal primers, which allows multiplexing of numerous targets.

MAPH (Multiplex Amplifiable Probe hybridization) relies on the ability of multiple short DNA probes (100-300 bp) to be recovered quantitatively by use of a common primer pair after hybridization to genomic DNA immobilized on a membrane (Armour et al, 2000). The probes are designed to be sufficiently different in sizes to be resolved by electrophoresis. Each probe is flanked with a pair of common synthetic primer binding site. After washing to remove unbound probe, the remaining specifically bound probe fragments (that are present in amount proportional to their target copy number) are stripped from the membrane and are further amplified by use of the common primer pair. PCR products are then separated by electrophoresis and the relative quantitation of band intensities or peak heights (depending on the method of detection) will indicate deletions or duplications. The method has been applied to the detection of constitutional genomic deletions in genes (White et al, 2002) or in multiple subtelomeric regions of chromosomes (Hollox et al, 2002). Recently it was demonstrated that MAPH can also be a reproducible high-thoughput method suitable for the assessment of acquired genomic imbalances of multiple loci in tumor DNA samples with heterogeneous cell populations (Reid et al, 2003).

In MPLA (Multiplex Ligation-dependent Probe Amplification) (Shouten et al, 2002), each target sequence in genomic DNA is hybridized in solution to two adjacent complementary probes. The probes are subsequently joined by a ligation reaction and copy sequences are amplified in a multiplex PCR reaction using fluorescent-labeled universal primers attached to the probes. Only probes hybridized to the target sequence will be ligated and subsequently amplified in the PCR reaction. One of the two adjacent probes has a variable length random fragment to allow size-based electrophoretic separation of multiplex PCR products. The long probes are M13-derived whereas the short probes are synthetic. The peak area of each amplified product will reflect the relative copy number of the target sequence. Up to 40-50 target sequences can be analyzed simultaneously (Hogervorst et al, 2003). Commercially available kits have been designed for several tumor suppressor genes as well as subtelomeric regions of all chromosomes. Recently the use of MLPA for rapid high throughput prenatal detection of common aneuploidies has been assessed (Slater et al, 2003).

Several microArray-based CGH (comparative genomic hybridization) assays have been recently tested for detecting gross duplications and deletions, especially in tumour cells, using DNA purified from BAC clones spaced along a chromosome segment (Cai et al, 2002) and there are now commercially available arrays (Hui et al, 2002).

Converting diploid to haploid mutants. In the approach called « Conversion », patient cells are fused with a specially designed rodent cell line, creating hybrids that stably retain a subset of human chromosomes (Yan H 2000). About 25% of the derived hybrids contain a single copy of any human chromosome of interest. The diploid nature of the human genes has been converted to a haploid state, in which mutations are easier to detect. For example, deletion of a single exon will be easily detected from genomic analysis of hybrids, as no PCR product will be observed from the hybrid containing the mutant allele. Mutations affecting the expression of the gene can be revealed through analysis of the RT-PCR products of the haploid hybrids, whereas analyses of diploid genome are uninformative or difficult to interpret. However, due to time and expense associated with the generation of hybrids, this approach cannot be used in diagnostic labs.

Diseases such as congenital disorders and hereditary diseases are now been recognized as major health burdens. The birth incidence of infants with congenital disorders is generally estimated to be about 30-65/1000, and an estimated six out of ten people will, by the age of 60, develop a disease that is at least partially genetically determined. Due to recent progress in molecular genetics, it is now evident that inherited predisposition is important in a number of common diseases of later life, such as atherosclerosis, coronary heart disease, hypertension, diabetes mellitus, some rheumatic, oncological and mental illnesses, and cancer, which develop into severe handicaps in predisposed people. It is probable that genetic approaches to the prevention of common diseases will emerge as one of the dominant strategies for the improvement of health. Moreover, pharmacogenomics is an emerging field examining the genetic basis for individual variations in response to therapeutics (drug efficacy and toxicity). Mutation and SNP detection is therefore becoming more and more important, and much effort will be invested in the development of cost-effective sequencing methods and high-throughput scanning or screening techniques.

Molecular diagnosis is the detection of pathogenic mutations in DNA and RNA samples to aid in detection, diagnosis, subclassification, prognosis and monitoring response to therapy (Trent et al, 2003). Use of molecular genetic testing includes several categories of testing within genetic services that can be divided into those services that target whole population with a view to identifying those at increased risk, and those that focus on the needs of families which are affected by a genetic disorder (Donnai D, 2002). There are several categories of genetic tests :

-Diagnosis. DNA test to confirm clinical diagnosis - ex, positive Friedrieich's ataxia testing is 100% sensitive and 100% specific.

-Carrier testing or prenatal testing. Detect Heterozygotes among individuals at risk for reproductive planning purposes. In the US, in a 30-year period, 51,000 carriers for Tay-Sachs disease in Ashkenazi populations have been identified, resulting in the detection of 1,400 two-carrier couples (Khoury MJ, 2003).

-Prenatal diagnosis. For couples known to be at high risk of transmitting disease alleles, PND permits informed decision making about continuing the pregnancy - ex, severe haemoglobinopathies.

-Pre-symptomatic (or predictive) DNA testing. Allows genetic disorder to be detected in advance of clinical presentation and includes at least three different indications: i) Predictive testing without treatment – for example, Huntington disease screening identifies people who will develop the disease with the aim to enable personal decision making; ii) Predictive testing with treatment – ex, FAP testing identifies family members requiring early colectomy. Once the parental FAP mutation is known, the offspring can have the test to determine if he has inherited the normal parental gene (he is no longer at risk) or the mutated one (he will develop FAP at some future date); iii) Predispositional testing – ex, BRCA1 testing provides the probability of developping breast cancer and related cancers.

-Community or population screening. An example is newborn screening - ex: cystic fibrosis: testing provides improved nutrition and pulmonary management of affected children. Infants are first tested with the use of an immunoreactive trypsinogen assay; if the result is positive, the test is followed up with a DNA test (31 common mutations) of the original specimen of dried blood spot.

Prescription of the genetic test is usually ordered by a physician with genetic background and not offered directly to the consumer, in order to explain, in collaboration with the laboratory, the benefits, limitations and possible adverse consequence of the test (Amos and Patnaik, 2002). Pre-analytical information is essential for assessing the appropriateness of testing for a given patient, and, depending on the disorder, may be critical for interpretation of test results. For example, interpretation of CF mutation analysis requires knowledge of the indication for testing, family history and ethnic background of the patient because the mutation detection rates are less than 97% for almost all ethnic groups. The interpretation of a heterozygous genotype (detection of only one mutation) is fundamentally different for a diagnostic versus a carrier test. Moreover, for carrier tests, the prior and revised carrier risks are based on family history, as well as the population prevalence of carriers in the patient’s ethnic group. Carrier risk revision for a patient with negative mutations analysis and a positive family history is enhanced by knowledge of the familial mutation (Amos and Patnaik, 2002).

As rapid technological advances produce systems that are capable of increased sensitivity, throughput and quantitative potential, new sets of validation issues need to be addressed. These include the reproducibility of the techniques, both intra and inter-laboratory, in terms of sensitivity, accuracy, precision and the interpretation of data where arbitrary selection of cutoff points to separate positive and negative results are often used. The issue of quality assurance is also highlighted in considering that predictive and susceptibility genetic testing is often performed on asymptomatic persons, and interpretation of results may or may not be supported by other findings, such as family history. Furthermore, molecular genetic tests are often performed only once in the life of an individual and may never be reported or confirmed. There is need also for better understanding of appropriate genetic testing practices among physicians. In Europe, QA standards for human molecular genetic testing are continually updated through guidelines and recommendations of the EMQN network. There is a major need in positive control samples containing well-defined mutations associated with disease. Very few laboratories have sufficient resources to develop themselves immortalized cell repositories representing most common human mutations (Williams et al, 2003).

Because large heterozygous deletions involving one or several exons most probably will be missed by routine PCR-based techniques, diagnostic DNA analyses should include a comprehensive examination of the whole relevant gene in the patient and confirmation of carrier status in both parents. This should allow one to distinguish "homozygosity" for a point mutation from a loss of heterozygosity due to a large deletion in one allele.

The improved understanding of the molecular aetiology of genetic disorders through the identification of thousands of causative mutations is now altering our perception of disease transmission. The classical model of individual alleles segregating into families according to Mendel's law, that allowed the discovery of many single-gene disorders, does not explain a growing number of disorders where it appears that phenotypic effects can result from the combined action of alleles in many genes. Even for a disease like cystic fibrosis, considered as a pure Mendelian recessive disease, mutational CFTR genotypes cannot predict the phenotype in each patient, which probably depends on a discrete number of alleles at different loci, some of them already identified. In reality, there is a continuum between Mendelian and complex traits (Badano et al, 2002), with "oligogenic disorders" emerging in the middle of this spectrum. Retinitis pigmentosa (RP) was the first example of proven 'digenism" : Kajiwara et al, 1994) showed that heterozygous mutations in both the ROM1 gene (retinal outer segment membrane protein1) and the RDS gene (peripherin/retinal degeneration) were required in some seemingly dominant RP pedigrees to cause disease. The digenic RDS mutation prevents the formation of functional RDS-RDS homocomplexes, with the null ROM1 mutation further decreasing the amount of functional complexes available, leading to photoreceptor degeneration.

The methods described above will detect nucleic acids alterations but do not necessarily define their biological significance. Some sequence changes, such as those causing frameshift or chain-terminations, are obviously functionally deleterious to the protein, but missense mutations leading to amino acid substitutions or base change in introns and untranslated regions are often problematic. Additional information including family and population studies may help to determine the significance of a sequence alteration. Extensive databases for genes and diseases can be extremely useful for interpreting the significance of missense mutations and other sequence variants. Equally important are functional studies of the mutant proteins after in vitro expression, which make invaluable the contribution of basic research to genetic testing.

In conclusion, much of molecular biology in diagnostic is still both tedious and time-consuming, involving endless pipetting, mixing, incubating, pouring gels, deciphering electropherograms, extracting nucleic acids, and so forth. Despite major improvements, scanning genes for unknown mutations in a service laboratory still remains a considerable problem.

Abeysinghe SS, Stenson PD, Krawczak M, Cooper DN. Gross Rearrangement Breakpoint Database (GRaBD). Hum Mutat. 2004, 23:219-21.

Screening for mutations has become a cornerstone in the study of inherited and complex diseases. The dominant method for determining sequence variations is DNA sequencing, also known as the gold standard to determine sequence variations. However, DNA sequencing is expensive, both in regard to consumables and labor cost. In many laboratories high throughput mutation screening therefore calls for a simpler screening procedure than DNA sequencing. As a consequence many mutation screening methods have been developed over the past 20 years. Most of these methods have been designed for mutation discovery, but some are mostly used for identification of known mutations.

The most commonly used methods to screen for mutations include methods for detection of mobility shifts in a separation matrix in form of a gel, such as a polyacrylamide gel, a liquid polymer, or a reusable reverse-phase HPLC column. The mobility shift methods can be subdivided into two fundamentally different types of mutation detection methods. One group of methods relies on the ability to fractionate heteroduplex DNA formed between mutant and normal DNA from homoduplexes (Denaturing gradient gel electrophoresis (DGGE), constant gradient capillary electrophoresis (CGCE), temperature gradient capillary electrophoresis (TGCE) or denaturing high performance liquid chromatography (DHPLC)). The other group of methods relies on the ability to fractionate mutant single stranded DNA (ssDNA) from normal ssDNA (single strand conformation polymorphism (SSCP), Dideoxy Fingerprinting (ddF), endonuclease cleavage SSCP), based on the different folding properties of the ssDNA.

When using these methods for discovery, all of these mobility shift methods are usually performed in combination with a final determination of the DNA sequence. In contrast, when performing screening for known mutation, the mobility shift methods may in some cases stand alone.

Heteroduplex analysis (HA) is a simple mutation detection method performed in polyacrylamide gel systems . The analysis is performed by melting the double stranded DNA and simply allowing the strands to reanneal. In case of a heterozygote the complementary strands of the two different alleles also anneal to form a heteroduplex. The heteroduplex has a different conformation from the homoduplex. Thus, the heteroduplex may be separated from the homoduplex in a non-denaturing gel or liquid polymer . It has been used in rapid detection of known mutations or in combination with single strand conformation polymorphism . Recently, microchip-based HD detection has been developed . However, no commercially available instrument has so far been developed to perform this analysis.

Denaturing Gradient Gel Electrophoresis (DGGE) was first developed in 1983 . The principle of DGGE is based on the fact that the melting behavior of a DNA fragment is sequence dependent and that dsDNA melts in discrete domains rather than in a zipper-like fashion . Furthermore, a partially denatured DNA molecule travels much slower in a gel than a DNA molecule in its native conformation. The melting profile of a given DNA fragment may be predicted using dedicated software, such as MELT87 or SQHTX .

Two homoduplexes, only differing by a single base, or two heteroduplexes melt differently in a denaturing gradient and may be distinguished due to differences in migration rate in a polyacrylamide gel with an increasing concentration of denaturant or denaturing conditions.

The critical step is the determination of the melting profile of a given DNA region and choosing the adequate primers. If the DNA fragment of interest has a very low melting point it is necessary to make an artificial high melting domain. This is accomplished by introducing a so-called GC-clamp at the 5’-end of one of the PCR primers. This makes the method applicable to almost any DNA fragment of interest and gives a very high level of sensitivity .

Temporal temperature gradient gel electrophoresis (TTGE) and constant gradient gel electrophoresis are both methods based on the theoretical considerations behind DGGE, using either temperature gradient as the increasing denaturing condition or keeping the denaturing conditions constant at the condition predetermined to give the maximal separation between two homoduplex molecules .

High throughput methods derived methods include temperature gradient capillary electrophoresis (TGCE) and Constant Denaturing Capillary Electrophoresis (CDCE) . In TGCE a temperature gradient is applied during electrophoresis in stead of the gradient of denaturant in the case of DGGE. This temperature gradient may be controlled from within the capillary via Joule heat generated from a voltage ramp during the run or using commercial instruments . In CDCE the concentration and the type of denaturant is kept constant, and therefore this requires a preliminary optimization assay at varying denaturant compositions and concentrations or as is most often the case at various temperatures by repeated injections of sample at increasing temperatures.

The temperature gradient methods (TTGE and TGCE) are especially useful for mutation discovery, whereas the constant gradient methods (CGGE and CGCE) primarily should be used for mutation detection of known mutations.

Both CGCE and TGCE have large potentials for fast screening of mutations in DNA, and both methods should in theory be highly sensitive, since they are based on the same principle as DGGE, which has sensitivity above 95% . A comparative study showed that CGCE identified all mutations in the study , but TGCE remains to be validated.

Denaturing HPLC (DHPLC) is a method that is based on the well-known HPLC technique, used in protein purification for many years, and heteroduplex analysis (see above). By running a DNA sample over a column packed with alkylated poly (styrene-divinylbenzen) or alkylated silica particles with affinity for DNA, it is possible to separate heteroduplex complexes from homoduplexes at pre-calculated melting temperatures (Tm). The Tm, and consequently, the assay temperature can be precalculated by using Tm-computer programs such as DHPLCMelt (, http://insertion.standford.edu/melt.html). The homo- and heteroduplex DNAs are eluted with triethyl ammonium acetate as the mobile phase against a gradient of acetonitrile . As the assay can be performed at a pre-calculated temperature, it should not be necessary to perform more than one assay per sample; however, in reality many laboratories assay at two or three temperatures in order to attain maximal sensitivity . The method has recently been excellently reviewed by Xiao and Oefner . With the development of dedicated DHLPC instruments and validation of the method in a number of laboratories this method has become one of the most widely applied methods for mutation screening. DHPLC instruments have been designed both for unlabeled fragment analysis and for fluorescence analysis.

Also capillary columns for DHPLC analysis have been developed, which will increase the throughput radically . These systems can be directly combined with a mass spectrometry analysis for a final identification of the mutation .

Single strand conformation polymorphism (SSCP) analysis developed by Hayashi and coworkers has been extensively used for screening unknown point mutations for many years. The theoretical background for SSCP is (1) the conformation of a given ssDNA-fragment is sequence specific and (2) ssDNA molecules with different conformations have different mobility properties in a non-denaturing polymer.

The method involves a first denaturation step of a PCR product, followed by a subsequent rapid cooling on ice. This rapid cooling allows for formation of ssDNA rather than reannealing to dsDNA fragments and the ssDNA folds into sequence dependent tertiary structures due to intramolecular base-pairing. The folded DNA fragments may subsequently be separated by non-denaturing polyacrylamide gel electrophoresis or in a non-denaturing liquid polymer by capillary electrophoresis, and the mutant DNA fragments are distinguished from the wild-type fragments by showing a different migration pattern.

Single-stranded DNA secondary structures are strongly dependant on the physical environment, such as temperature, polymer, buffer and/or additives such as glycerol. Therefore, these parameters are varied by performing several runs using different conditions, most often by varying the temperature Thus, in order to achieve a high sensitivity of SSCP the analysis is normally performed using at least two different electrophoresis conditions, which of course increases the workload per sample.

The conversion into using CE-instruments for DNA sequencing has given a great number of laboratories the possibility to perform CE-SSCP, where automation and multicolor fluorescence-detection has dramatically improved the throughput and sensitivity of SSCP and made it a very attractive alternative to the conventional gel-based SSCP methods . In automated CE-SSCP it is possible to program the instrument to analyze each sample at several temperatures without intervention. Recently, it was shown that CE-SSCP performed at a single temperature using 5% dimethyl polyacrylamide polymer had a +90% sensitivity .

Fluorescence labeling is most often attained by using fluorescent-labeled primers, allowing for distinction between strands and for multiplexing of several PCR fragments, which may increase throughput . However, fluorescent-labeled primers are expensive and in order to reduce assay price, post–PCR labeling of DNA for SSCP has been performed using specific primers with either a GTT extension on the forward primer or an ATT extension on the reverse primer. Following PCR amplification fragments are terminally labeled with a strand specific fluorescent dNTP using the Klenow enzyme, and unincorporated dNTPs are degraded by alkaline phosphatase in a one-tube setup . The Klenow enzyme fills in the protruding ends of the PCR fragments and degrades unused amplification primers, which otherwise may hybridize to the ssDNA and result in anomalous peaks that often complicates data analysis when using end-labeled primers . This method markedly reduces the cost of the consumables, but also increases the workload.

Prototypes of chip-based capillary electrophoresis instruments have been developed and shown to markedly reduce assay time and still maintaining the high sensitivity in SSCP .

Dideoxy fingerprinting (ddF) is a hybrid method of Sanger DNA sequencing and SSCP analysis . The analysis is performed in two steps. Initially, a sequencing reaction is performed using only one dideoxy nucleotide. Subsequently, the chain termination products are analyzed by SSCP, using a single electrophoresis condition. ddF gives a sequence dependent pattern (or "fingerprint") caused by the specific migration of each folded termination product in a gel or a liquid polymer as discussed above. The use of dideoxy termination should abolish the need for analysis at more than one temperature in a gel-based system or a capillary electrophoresis system . The method has been used for detection of rifampicin resistance mutations in M. tuberculosis or in screening for disease associated mutations . However, CE-ddF cannot be considered a simple method due to the need for optimization for each fragment of interest in order to achieve sufficient robustness for routine applications.

Another approach that has been applied to increase the sensitivity of SSCP is to perform a combined PCR, endonuclease cleavage and SSCP analysis .

Generally, it has yet to be validated whether combination methods, such as ddF, endonuclease cleavage/SSCP or the like, increase the sensitivity compared to the parent method. Even though they may increase the sensitivity somewhat, these combination methods increase the workload and complexity of the mutation detection methods, which in turn may result in a higher error rate and therefore a lower sensitivity.

SSCP and HA has been combined in a number of studies and have been shown to increase the sensitivity markedly in gel-based systems or in capillary electrophoresis based systems , . As outlined above, this combination system is based on two fundamentally different properties contrary to the above combination methods, which is the reason why this method may be a more sensitive method than either SSCP or HA. Contrary to the above mentioned combination methods, SSCP-HA is not necessarily more complex to perform and the workload is at least in one study about equal to conventional SSCP analysis .

A number of factors are to be considered when choosing a mutation screening method. First of all, the method should be simple in respect to sample handling and post-assay data analysis in order to minimize labor cost and assay errors. Here, the various mobility shift methods vary to some degree, but all in all most are considered relatively simple to perform. However, the combination methods often require additional steps, which may increase sensitivity but also increase assay time and risk of errors.

Secondly, sensitivity is always an issue. Over the years many studies have focused on sensitivity. Generally, studies of assay sensitivity can always be reproached due to bias of the validation setup. However, most mobility shift methods are c