Introduction
Genetic and functional approaches are the two main ones used in identification of genes
that, when mutated, cause disease. In genetic approaches to the study of hereditary disease,
genetic mapping methods are employed to identify the position within the human genome of
a genetic factor influencing development of a disorder. Then, the gene or genes within that
region are screened for evidence of involvement in the disorder. In contrast, in functional
approaches to study of genetic disease, ‘candidate genes’ that may be mutated in individuals
with the disease under study are identified based on other information. This information can
include the biology of the disorder and the known or predicted function of the protein encoded
by the gene, as well as other knowledge about the biology of the gene or protein, such as
the times during development and tissues in which the gene or protein is expressed and the
subcellular or extracellular location of the protein.
In the past, genetic approaches to disease gene identification were often considered, with
some validity, to be less direct and more laborious than functional approaches. However, with
recent improvements in laboratory technology, analytic approaches, and available data (the
latter including the nearly complete sequence of the human genome), genetic approaches can
now be applied with steadily increasing efficiency to study of an ever-expanding set of diseases.
In recent years, the application of genetic approaches has contributed greatly to our understanding
of hereditary disorders, and this productive use of genetic approaches is likely to
continue increasing.
Human genetics has changed dramatically during the past two decades, as the following
two examples illustrate. In 1983, the initial localization of the gene mutated in Huntington
disease (HD) to human chromosome 4p was one of the earliest successes in use of genetic
mapping to genetically localize a previously unknown disease gene and involved a combination
of very hard work and substantial luck. Selected members of two large pedigrees (including
one with more than 3000 identified members) were genotyped, along with other unaffected
relatives. Luck came into play in that one of the first 12 genetic markers (in this case,
restriction fragment length polymorphisms analyzed using Southern hybridization techniques)
tested was linked to HD.1 A decade and the work of many dozens of researchers affiliated with a number of collaborating and competing laboratories was required for the subsequent identification of the gene mutated in HD (i.e., the HD ‘disease gene’) by a collaborative group of 58
authors from six institutions.2
In 1994, a gene mutated in some cases of benign recurrent intrahepatic cholestasis (BRIC)
was mapped to human chromosome 18q through study of four patients and six unaffected
relatives.3 The following year, the progressive familial intrahepatic cholestasis type 1 (PFIC1) locus was mapped to the same region through study of two patients and their parents.4 It then took less than three years, and the work of many fewer individuals than for HD, to identify the gene mutated in these two disorders.5
Twenty years after the initial mapping of the HD gene, identification of disease genes for
disorders with straightforward inheritance patterns (like HD) borders on the routine. Often,
the time lapsing between initiation of a mapping project and identification of a disease gene is
on the order of a few years; typically, such projects now require the concentrated efforts of only
a few individuals to succeed.
One of the key advantages of genetic approaches over many functional ones is that genetic
approaches depend very little upon knowledge about the underlying disease biology. Genetic
approaches permit discovery of a disorder's genetic etiology, even if the identity of the disease
gene could not have been predicted, given current understanding of cell, molecular, and disease
biology. Therefore, genetic approaches allow a comprehensive screening of the possible genetic
factors leading to development of disease and facilitate unexpected, unpredictable discoveries.
This impressive power of genetic approaches has been one factor driving their increased use.
The identification of the BRIC/PFIC1 gene, ATP8B1 (also called FIC1, for familial intrahepatic cholestasis 1), illustrates this point. Prior to identification of ATP8B1, it was hypothesized that the BRIC/PFIC1 gene most likely encoded a canalicular bile acid transporter that
was a member of the ATP-binding cassette (ABC) protein family.6 The reasonableness of this hypothesis was borne out when the gene mutated in a different form of PFIC was identified; the PFIC2 gene is indeed an ABC protein that transports bile acids across the canalicular membrane. In contrast, ATP8B1 is not an ABC transporter but belongs to a different family of
membrane transporters, the P-type ATPases.7 ATP8B1 is expressed in a wide variety of tissues, suggesting that it plays a role important in many tissues, rather than being specifically a canalicular bile acid transporter.5 Within the liver, the protein is found in the canalicular membrane
of hepatocytes, and in cholangiocytes.8,9 ATP8B1 may function in transport of
aminophospholipids between membrane leaflets.8 Initial evaluation of a mouse model suggests that ATP8B1 may participate in regulation of intestinal bile salt absorption.10 Further studies to determine the function of ATP8B1 are currently under way and promise to shed new light on the biology of bile acid transport.
Genetic and functional approaches to the study of hereditary disease complement each other
and can be applied together. Such an approach facilitated rapid identification of the PFIC2
gene; initially, the PFIC2 locus was genetically mapped to chromosome 2q24.11 An ABC transporter gene lay in the region and was indeed the PFIC2 gene, ABCB11 (also called BSEP, for bile salt export protein).12 If no such excellent candidate gene lay in the region, it likely would
have taken longer to identify the disease gene.
In recent years, genetic approaches have been applied to the study of cholestatic disorders,
with exciting results. In this chapter, the basic concepts in human genetics that are helpful for understanding of these and future advances will be reviewed, using as illustrations examples from hereditary cholestatic disease.
Our DNA: The Basics
The nucleus of each human somatic cell carries 46 chromosomes: two copies of each of the
22 autosomal chromosomes, and either two copies of the X chromosome (in females), or one
copy each of the X and Y chromosomes (in males). Each cell also contains numerous mitochondria,
which have their own ˜16 kb (kilobase pair) genome. A single copy of the human
nuclear genome (one copy each of chromosomes 1-22, X and Y) is ˜3.2 Gigabase pairs in
length and has been estimated to contain approximately 30,000–35,000 protein-coding genes;
approximately 1.5% of our DNA is protein-coding sequence, hidden within a large excess of
non-coding DNA.13,14 This non-coding DNA contains introns, as well as genes that produce
non-coding RNAs, including transfer RNA, ribosomal RNA, small nuclear RNA, small nucleolar
RNA, microRNAs, small interfering mRNAs, and small temporal RNAs.15,16 Non-coding
DNA also contains numerous other functionally important sequences, including those necessary
for maintenance of chromosome ends, proper segregation of chromosomes during cell
division, and initiation and regulation of gene expression. Much of our genome has no currently
known function; this sequence includes pseudogenes, and the roughly 48% of our DNA
that consists of interspersed repetitive elements. This latter group includes many types of
transposon-derived repeats, as well as simple sequence repeats.13
Review of a few terms may aid understanding of the discussion that follows. The term locus refers to the chromosomal location of a gene or DNA marker. A specific DNA sequence is
polymorphic if it varies between individuals, and the different sequence variants are alleles. A haplotype is a set of specific co-inherited alleles; typically, these alleles are co-inherited because they are present at neighboring genetic markers. An individual carrying the same allele at both
copies of a locus is homozygous for that allele, while someone carrying two different alleles is
heterozygous. Genetic markers are specific DNA sequences that are polymorphic and employed in mapping of disease genes. When a single disorder can be caused by mutation in different genes, locus heterogeneity is present; low-gamma-glutamyl-transpeptidase (GGT) PFIC exhibits locus heterogeneity, since mutations in either ATP8B1, ABCB11, or one or more as-yet-unidentified genes can cause the disease.17 Allelic heterogeneity exists if different
mutations in a single disease gene can cause the same disorder; many genetic disorders manifest
allelic heterogeneity, complicating mutation screens.
Steps in Genetic Mapping
Genetic mapping studies typically involve a series of steps: identification of the likely mode
of inheritance of the disease; application of experimental and analytical methods to initially
map the disease gene and then refine its location; identification of candidate genes; screening
of candidate genes for mutation; and determination of the functional consequences of the mutation(s).
Mode of Inheritance
Table 1
Genetics of cholestatic disorders and disorders of bilirubin metabolism and transport: genes identified or mapped
| ATP8B1 disease: PFIC1, BRIC, Greenland Familial Cholestasis (GFC) | 18q21 | ATP8B1 (FIC1)5,18 | Genetic | AR (BRIC: incomplete penetrance) |
| ABCB11 disease: PFIC2 (BRIC?) | 2q24 | ABCB11 (BSEP)12,19,20 | Genetic and functional | AR |
| ABCB4 disease: PFIC3, cholestasis of pregnancy (ICP) | 7q21 | ABCB4 (MDR3)2125 | Functional | PFIC3: AR ICP: AD, sex-limited, incomplete penetrance |
| Alagille syndrome (AGS) | 20p12 | JAG12628 | Genetic | AD |
| Dubin-Johnson syndrome | 10q24 | ABCC2 (cMOAT)29 | Functional | AR, incomplete penetrance |
| Crigler-Najjar syndrome | 2q37 | UGT1A130,31 | Functional | AR |
| Neonatal giant cell hepatitis | 8q21.3 | CYP7B132 | Functional | AR |
| Wilson Disease | 13q14 | ATP7B33,34 | Genetic and functional | AR |
| Cystic fibrosis | 7q31 | ABCC7 (CFTR)3537 | Genetic | AR |
| Lymphedema- cholestasis syndrome (LCS) | 15q38 | unknown | Genetic | AR |
| North American Indian Childhood Cirrhosis (NAIC) | 16q2239 | CIRHIN83 | Genetic | AR |
| ICP | 2p1340 | Unknown | Genetic (association study) | Unknown (sex-limited) |
| Cystic fibrosis modifier locus for meconium ileus (CFM1) | 19q1341 | Unknown | Genetic | ? |
Table 2
Mendelian modes of inheritance: typical features
| Autosomal recessive (AR) | No (unless extensive consanguinity or very common) | 25% | No (unless mother is a carrier) | No (unless father is a carrier) | In equal proportions, with similar severity | Consanguinity seen more often than with other modes of inheritance |
| Autosomal dominant (AD) | Yes | 50% | Yes | Yes | In equal proportions, with similar severity | Fewer than 50% of siblings may be affected if parents are unaffected, and one is germline mosaic for a new mutation |
| X-linked recessive | Yes, but generations be ‘skipped’, due to unaffected status of carrier females | Transmission from unaffected mother: 50% of sons, and no daughters, affected. Transmission from affected father: no offspring are affected. | No | Yes | Females rarely affected, since they have 2X chromosomes; heterozygous females may occasionally manifest disease due to skewed X-inactivation | |
| X-linked dominant | Yes | Maternal transmission: 50% of sons and daughters are affected. Paternal transmission: all daughters, but no sons, are affected. | No | Yes | Females at least twice as likely to be affected as males (some such disorders are lethal prenatally in males, so even fewer affected males are seen) | With small amounts of data, it can be difficult to distinguish X-linked dominant from AD inheritance. |
| Y-linked | Yes | 50% | Yes | Not applicable | Only males are affected | Seen very rarely, as there are few genes on the Y chromosome |
| Mitochondrial | Yes | Maternal transmission: ˜all children affected. Affected fathers do not transmit the disorder to their children | No | Yes | Yes, in equal proportions | As mitochondrial DNA is maternally inherited, affected men do not transmit such disorders. |
Understanding the mode of inheritance of a disease can focus genetic studies, permits selection
of the most appropriate form of genetic analysis, and is important when counseling affected
families. Additionally, many forms of genetic analysis require specification of the mode of inheritance of a disorder. The modes reported for cholestatic diseases to date are mostly autosomal recessive and autosomal dominant (
Table 1), so these are discussed in some detail
below. Basic features of several other modes are indicated in
Table 2.
For an autosomal recessive (AR) disorder, disease develops only if both copies of the ‘disease gene’ in an individual possess deleterious mutations. Generally, AR disorders are seen in
one or more siblings in a family, but not in parents or children of patients. Heterozygous
carriers are usually clinically normal but may exhibit biochemical abnormalities. The ratio of
males to females affected is typically ˜1:1, and on average, one-fourth of the children of two
heterozygous carrier parents will be affected. Consanguinity is seen more often in AR disease
than in disorders with other modes of inheritance and has proven helpful in mapping of
PFIC1, AR BRIC, PFIC2, LCS, and North American Indian childhood cirrhosis (NAIC).3,4,11,12,38,39,42 In consanguineous families, affected children are typically homozygous for a
disease mutation. Patients with an AR disease who carry two different disease mutations
are compound heterozygotes.
In a simple autosomal dominant (AD) disorder, one defective copy of the disease gene is sufficient to produce disease. In a typical pedigree for an AD disease, individuals in multiple
generations will be affected, no generations will be skipped, males and females are equally
likely to be affected, and if affected, to transmit the disorder. In a family in which one parent
has an AD disorder, 50% of the children, on average, will also have the disorder. Homozygotes,
when they occur, are generally more severely affected than heterozygotes. A form of BRIC
demonstrating AD inheritance has recently been described.43
As mitochondrial DNA is maternally transmitted, affected fathers do not transmit disorders with a mitochondrial mode of inheritance to their children. Of note with regard to mitochondrial
inheritance is that many genetic diseases in which mitochondrial defects occur do not demonstrate mitochondrial inheritance. This is because many proteins essential for normal function of mitochondria are encoded by the nuclear genome. Navaho neurohepatopathy may be an example of such a disorder; it is AR, but appears to involve depletion of mitochondrial DNA.44
Chromosomal abnormalities are large enough to be visible using cytogenetic techniques and include changes in chromosome number (polyploidy, aneuploidy) or structure (i.e., translocations,
deletions, inversions, duplications). Some such abnormalities can be genetically transmitted,
although review of their inheritance patterns is beyond the scope of this chapter. Before
embarking on extensive genetic mapping studies of a disorder, it is sensible to have karyotyping
performed on some patients to ensure that they do not have a chromosomal abnormality. For
example, early identification of a large deletion in a patient with Alagille syndrome (AGS)
narrowed the focus of mapping studies for this disorder to a region on chromosome 20.45
Other features can influence the manifestations and apparent inheritance pattern of a genetic
disorder. Disorders that present at different rates or with different symptoms in males
versus females can be due to mutation in autosomal genes. Such disorders are sex-influenced
or sex-limited; intrahepatic cholestasis of pregnancy (ICP) is a sex-limited disorder. Another
important feature of a genetic disease is its penetrance—i.e., the proportion of people with a
disease-causing genotype who actually develop the disease. Penetrance can be complete (i.e.,
all patients with a disease-causing genotype develop the disease) or incomplete, and also may
be age-dependent; BRIC due to mutation in ATP8B1 exhibits incomplete, age-dependent
penetrance.46
In a disorder with variable expressivity, patients differ in the severity and/or constellation of disease manifestations they suffer. Sometimes variable expressivity is due to mutations with
effects of differing severity on protein function, but in other cases, substantial variability is seen
even between patients possessing the same disease mutation. Both BRIC and Alagille syndrome
have substantially variable expressivity.28,46
Several other features should be kept in mind when considering the pattern of inheritance
of a genetic disorder:
-
A new mutation may have occurred. For example, a new mutation for a dominant disease may occur in a parent, so that a child is affected, although the parent was not. If the possibility of new mutation is not considered, the disorder in that family might be thought recessive. In Alagille syndrome, as in many AD disorders that reduce reproductive success, a majority of the mutations are ‘de novo’.28 Such a new mutation can occur in a single germ
cell in the parent, or the parent can be mosaic for it— i.e., a proportion of the parent's cells
carry the mutation. In a dominant disorder, parental germline mosaicism for a new disease
mutation means that multiple children in the family can suffer from an AD disorder, although
neither parent is affected.
-
Some disorders show anticipation, in which the age of onset decreases, and/or disease severity increases (on average), with each successive generation. Anticipation is seen most
typically in disorders caused by expansion of a trinucleotide repeat, such as HD, as the
repeat can further expand in successive generations.
-
Imprinting is said to occur when inheritance of the same mutation has a different effect on the child, depending upon whether it was inherited from the mother or father.
Sometimes, a disorder's mode of inheritance can be evaluated statistically using a formal
segregation analysis. In other cases, especially for rare disorders, and given today's typically
smaller families, too few patients and family members are available to permit statistically definitive
segregation analysis; nevertheless, it is often possible to identify the most likely mode of
inheritance. Knowledge of the mode of inheritance of a disorder can help greatly in identifying
the genetic etiology of the disease; for example, if a disorder demonstrates X-linked inheritance,
a genetic screen need only be performed for the X-chromosome, rather than for the entire
nuclear genome.
Complex disorders are those which do not exhibit simple Mendelian inheritance patterns but are multifactorial (i.e., influenced by multiple genetic and environmental factors). Such
disorders may be oligogenic (influenced by a small number of genetic loci) or polygenic (influenced by many loci). Given an adequate study sample, the extent to which a trait is inherited
can be estimated. For example, in one study, sisters and mothers of women with ICP were
found to have ˜12 times greater risk of developing ICP than were women in the general popu-
lation;47 however, both genetic and environmental factors could contribute to this increased relative risk. A number of susceptibility loci may exist for a disorder; a susceptibility locus is one at which mutation increases the risk of developing the disease but does not lead inevitably
to disease. Modifier loci may influence the phenotype of a disorder; in cystic fibrosis (CF), a
modifier locus for meconium ileus has been mapped.41
Localizing a Disease Gene: Genetic Markers
A number of experimental and analytic approaches can be used to map a disease gene.
Experimentally, these approaches rely upon genotyping of genetic markers, although the number
and type of markers vary. The genetic markers most frequently used in mapping studies to
date are polymorphic simple sequence repeats (SSRs). These are widely distributed, short,
tandemly repeated sequences. The most commonly employed SSRs include dinucleotide and
tetranucleotide repeats. The number of copies of the repeat unit varies between alleles, so alleles
differ in length. The inheritance pattern can be assessed by amplifying the repeat from genomic
DNA using the polymerase chain reaction (PCR), with unique primers flanking the repeat,
and then electrophoresing the PCR products to separate the alleles by size. Over 8,000 polymorphic
SSRs have been positioned on a comprehensive genetic map, for an average density of
1 SSR every ˜400 kb;48 a typical genome screen might involve typing of 200–800 of these markers. A major advantage of SSRs for genetic mapping is that they are highly polymorphic and consequently, very informative in genetic analyses. SSRs have been successfully employed
in genetic mapping of PFIC1, AR BRIC, PFIC2, LCS, and NAIC, amongst other disorders.3,4,11,38,39
The use of single nucleotide polymorphisms (SNPs; genomic sites at which a single base varies between alleles) has been increasing. An advantage of SNPs is that they are extremely
common; 2 chromosomes differ from each other at ˜1 bp in 1,300.13,14 Also, SNPs have a lower mutation rate than do SSRs. Numerous SNP genotyping technologies are at various
stages of development; these include methods employing electrophoresis, oligonucleotide microarrays, mass spectrometry, fluorescent microtiter plate reading, or flow cytometry, among others. Some of these methods are amenable to extremely high-throughput genotyping, compared to what can be achieved with SSRs.49 A disadvantage of SNPs relative to SSRs is that SNPs are less polymorphic, since most SNPs have only 2 alleles.
Genotyping data can be interpreted using various statistical or empiric forms of analysis;
what is most appropriate in a specific situation depends upon characteristics of the disorder,
the type and size of sample available, and characteristics of the population(s) from which the
sample is derived. Genetic mapping data can be evaluated using linkage analysis and/or population
genetic mapping.
Localizing a Disease Gene Part 1: Linkage Analysis
Figure 1
.
Evidence for or against linkage. This figure shows two pedigrees of families in which an autosomal recessive (AR) disorder occurs. Affected individuals are indicated by black symbols, and unaffected individuals, by unfilled symbols. Data from two genetic markers (A and B) flanking a region of interest are
shown. Numbers 1–4 represent different marker alleles. The rectangles of different patterns indicate the
different copies of the region present in each family. Data for the pedigree on the left are consistent with
linkage to the interval between markers A and B, as both affected children have inherited the same paternal
and maternal copies of the region. Data for the pedigree on the right are inconsistent with linkage to this
interval; while both children have inherited the same paternal copy of the region, they have inherited
different copies of the region from their mother.
Linkage analysis is a powerful, family-based approach to disease mapping. It makes use of the fact that specific copies of the genomic region containing the disease gene are co-inherited
with the disease within a family; this reflects lack of recombination between the disease mutation
and neighboring genetic markers, due to their close proximity. Within a family, individuals
who share a disease will typically share alleles at markers near the disease gene (). The particular alleles co-inherited with the disease often differ between families, reflecting allelic
heterogeneity or ancestral genetic recombination events. Results of linkage analysis are reported
as LOD scores representing the relative likelihood that a disease locus and a genetic marker are
genetically linked (with a recombination fraction theta), rather than that they are genetically
unlinked. A LOD score of at least +3.3 is typically considered evidence of linkage from a
genome-wide screen. A LOD score of −2 or below excludes disease linkage to a region. Linkage
analysis permitted mapping of the CF gene.
5052
Standard parametric linkage analysis requires specification of a genetic model, including the mode of inheritance, penetrance, and frequency of the disease in the population, and often
also necessitates estimation of marker allele frequencies in the population. For disorders with a
poorly understood genetic model, other mapping approaches are more appropriate. Additionally,
locus heterogeneity within the collection of families studied complicates standard linkage
analysis and, if not properly taken into account, can lead to incorrect interpretation of results.
Finally, extremely complex pedigrees present computational difficulties for linkage analysis.
Homozygosity mapping is a particularly powerful form of parametric linkage analysis applicable to recessive conditions in consanguinous families. In such a family, patient(s) are
likely to be homozygous by descent for a single disease mutation, and for alleles at nearby
genetic markers, i.e., both the mutation and marker alleles were inherited from a single ancestor
shared by their mother and father. Homozygosity mapping identifies segments of homozygous
DNA in patients. Such an approach was employed to map the loci for Wilson disease (WD) and PFIC2.11,53,54
When the genetic model of a disorder cannot be determined, modified forms of linkage
analysis may be used. When penetrance of a disorder is unknown, an affecteds-only parametric
linkage analysis can be performed, in which the phenotype of unaffected individuals is considered unknown. This approach results in some loss of statistical power. If the mode of inheritance of a disorder is also unknown, a nonparametric linkage approach, such as affected sib pair analysis, can be performed. Regions that are shared by affected siblings or other relatives more often than expected by chance are identified; such approaches are useful in mapping susceptibility loci for complex traits.
Localizing a Disease Gene Part 2: Population Genetic Mapping
Figure 2
.
Population genetic mapping. This figure illustrates the principles behind this mapping approach, as applied to an AR disease. A common ancestor, or founder, has introduced a disease mutation into the
population. This mutation is represented by the black box on the rectangle representing the chromosomal
region in which the mutation occurred. The shaded areas flanking the mutation represent the version of the
surrounding chromosomal sequence that was present on the chromosome when it was introduced by the
founder, i.e., the ancestral disease-associated haplotype. As the mutation is passed down through the
generations, recombination events occur, shrinking the size of this haplotype. The white areas represent
those areas that no longer share the ancestral haplotype, due to recombination. Many generations after
introduction of the disease mutation into the population, those affected individuals in the population who
have inherited the mutation introduced by the common ancestor still share the ancestral haplotype surrounding
the disease mutation.
Population genetic mapping can also be employed to map disease genes. In this approach, the genome is screened in patients from a single population to identify chromosomal segments
that the patients share identical by descent (IBD). The concept behind this approach is that
members of a population are relatives, even if their exact relationships are unknown. Distantly
related patients may share the same disease mutation, and the same version of the chromosomal
sequence surrounding the mutation, inherited from a common ancestor, especially if the
disease is rare (). Due to genetic recombination, the greater the number of generations that have passed since the introduction of the mutation into the population, the smaller this
shared region will be. The presence of a shared haplotype, or set of specific co-inherited alleles,
at several consecutive genetic markers indicates IBD sharing of such an ancestral region. Such
sharing reflects the presence of
linkage disequilibrium (LD, i.e., a non-random association)
between a disease and particular genetic marker alleles.
Population genetic mapping approaches can be applied in circumstances where successful
linkage analysis is difficult or impossible. In population genetic mapping, mode of inheritance
and penetrance of the disorder need not be specified, and allelic and/or locus heterogeneity may
be present without resulting in erroneous interpretation of results (although it does reduce
statistical power). Also, precise family relationships need not be known, and it is not necessary
to identify large pedigrees with multiple patients; DNA need only be obtained from single
patients (and if possible, their parents). Additionally, population genetic approaches often permit
further refinement of a disease gene's localization than is possible with linkage analysis.
A disadvantage of population genetic studies is that, if the patients are very distantly related, evidence of LD may only be detectable over a very small genomic region, necessitating screening
of a large sample with densely spaced genetic markers. This potential problem can be
minimized through careful choice of study populations. Recently founded genetically isolated
populations are ideal for population genetic mapping, as the size of the genomic region(s) in
LD with the disease is likely to be relatively large, and more easily detectable. If such a population
was founded by a relatively small number of individuals, the risk of extensive allelic and/or
locus heterogeneity, which reduce power to detect a disease locus, is decreased.
Population genetic mapping was successfully employed in the mapping of BRIC, PFIC1,
and LCS.3,4,38 A modification of this approach was used to map the NAIC locus; a genome screen was performed on three pools of DNA, one each from patients, unaffected siblings, and parents. Markers with enrichment of an allele in the patient pool, as compared to the other two pools, were identified. Those regions were then further characterized to identify the NAIC
region.39
Typically, association studies are used to evaluate candidate genes (or regions) in less consanguineous populations. DNA from a set of patients is collected and data from a polymorphism
of interest is generated and statistically evaluated to determine whether one of the alleles
is overrepresented in patients as compared to controls. Large patient samples are usually needed
for such studies, and patients and controls must be carefully matched. False-positive results can
be caused by factors such as undetected population stratification in the sample used, as well as
effects of natural selection. Some association tests, such as the transmission disequilibrium test
and the more recently developed genomic control and structured association tests, account for
population stratification.55 In Finnish women, an association study using markers spanning chromosome 2 identified a region that may contain a susceptibility locus for ICP;40 another study of this same population suggested that a polymorphism in the gene encoding the angiotensin-converting enzyme (not located on chromosome 2) may be associated with ICP.56 Due to the high rate of false-positive results obtained in association studies, confirmation of these results will be important.
Localizing a Disease Gene Part 3: Fine-Mapping
Often the initial mapping of a disease locus to a region is too imprecise to permit immediate
identification of the actual disease gene, particularly if the region contains many genes, and
none of them are especially promising from a functional standpoint. The location of a disease
gene is genetically further refined through study of additional patients, and/or use of additional
genetic markers. Often a combination of analytic approaches and populations is used for
study of a single disease, as one approach or population may be most useful for initial mapping
of a disease locus, and another for the refinement of localization.
To initially localize ATP8B1, for example, haplotypes shared by distantly related patients were identified. 3,4 The position of the locus was further refined using linkage analysis in additional families and finally, detailed haplotype analysis of a larger set of PFIC1 and BRIC patients
from multiple populations.42,57 This analysis permitted identification of shared disease
haplotypes, and one disease-associated deletion, and localized the disease gene to a ˜1 cM
(centimorgan) interval, greatly facilitating identification of ATP8B1.
Candidate Gene Identification
Once a disease gene has been mapped, genes within the candidate region must be identified.
In the past, this required much laborious and clever laboratory work, but it has become
vastly easier. With the availability of the human genome sequence, a substantially complete
inventory of the genes in a given region can be obtained without performing any experiments.
At a public or private website devoted to the human genome (e.g., http://www.ensembl.org/
Homo_sapiens/, http://www.genome.ucsc.edu/goldenPAth/help/hgTracksHelp.html, http://
www.ncbi.nlm.nih.gov/, or http://www.ncbi.nlm.nih.gov/genome/guide/human/ ), the genomic
region of interest can be selected, and a list of known or predicted genes in the region displayed.
Sequence from the region can be used to scan databases of transcribed sequences. To find genes
that have not yet been identified through laboratory experiment, sequence from the region can
be evaluated using computer programs designed to identify genes within genomic sequence.
Nevertheless, difficulties remain. There are still gaps and errors in the genome sequence. Also,
the process of computationally predicting the presence and structure of genes, given DNA
sequence, is imperfect. Particularly in the case of genes that do not demonstrate substantial
sequence homology to known genes, it is possible that a gene might be present, but not predicted
to exist.
Usually, one or more promising candidate genes are identified in the region. When multiple
genes are present, those screened first for mutation are usually those that data suggest are the
most promising candidates, given the known biology of the disease. A gene that encodes a
functionally uncharacterized protein can be evaluated as a candidate using information such as
whether it has homology to proteins that have known functions consistent with the disease
phenotype, and/or whether it is expressed in the tissue(s) most affected by the disease. Genetic
and functional approaches to disease gene identification often meet at this point.
Genetic Polymorphisms and Mutations: Types and Identification
Table 3
| Single base-pair change | Missense (altered amino acid) |
| Nonsense (stop codon) |
| Splicing (prevention of normal splicing, induction of abnormal splicing) |
| Regulatory (change transcription levels or pattern, or mRNA stability> |
| Deletion | Frameshift (change of reading frame) |
| In-frame (removal of amino acids) |
| Splicing |
| Regulatory |
| Gene deletion |
| Microdeletion/contiguous gene syndrome (loss or disruption of multiple genes) |
| Insertion (including duplication and repeat expansion) | Frameshift In-frame (addition of amino acids) Regulatory Splicing |
| Inversion | Disruption of normal gene structure |
| Chromosomal abnormalities | Changes in chromosome number (polyploidy, trisomy, etc.) or structure (translocations, deletions, inversions, duplications) |
The categories of mutation that can occur in our DNA are summarized in
Table 3. For most disorders, the most common category of disease mutation identified is the single base pair substitution, in which a single base pair of DNA is replaced with another base pair. Deletions,
inversions and insertions (including duplications and repeat expansions) also occur.
Deleterious mutations are changes in the DNA sequence that lead to development of, or increased susceptibility to, disease. Neutral polymorphisms or normal variants are DNA
sequence changes that have no apparent functional significance. When it is unclear whether a sequence change has any functional consequence, it can conservatively be referred to as a variant.
A number of methods for identifying disease mutations are available. The types and proportion
of total mutations detected depends on the method used. Most commonly used methods involve as an initial step PCR amplification of the coding sequence of the candidate gene; to screen for regulatory mutations, the gene's promoter should also be identified and screened. Genes can be screened for mutation using RNA or genomic DNA. Screening of genomic DNA
usually requires more effort, as each exon of the gene is amplified by PCR, then analyzed. Where RNA from tissue in which the candidate gene is expressed is available, the transcript can be amplified from it. Unfortunately, patient RNA from relevant tissues is often not readily available. For example, ATP8B1 has 27 coding exons, and screening of the genomic DNA encoding it requires generation and evaluation of 24 separate PCR products (in three cases,
small introns enable two exons to be included in a single PCR product). The coding portion of the ATP8B1 transcript is 3.8 kb in length, so when patient RNA from a tissue in which ATP8B1 is expressed is available, the coding portion of the gene can be amplified in substantially fewer PCR reactions.
Once PCR products have been generated, they can be analyzed using one of several methods;
the ‘gold standard’ method is DNA sequencing. However, generation and analysis of DNA sequence can be comparatively expensive and time-consuming, so other methods are sometimes used for screening, especially when a large number of samples is involved. Also, heterozygous mutations may occasionally go unnoticed, due to base-calling errors; sequencing both strands of DNA can minimize this problem.
Numerous other approaches to identification of mutations through analysis of PCR products
have been developed. One commonly used approach is single strand conformation polymorphism
analysis (SSCP); in this approach PCR products are denatured and then electrophoresed
through a non-denaturing gel. Mutations may alter the conformation, and thus the
mobility, of the single-stranded PCR product; however, some mutations are typically not detected.
Three other approaches, denaturing gradient gel electrophoresis (DGGE), mismatch
cleavage, and denaturing high performance liquid chromatography (DHPLC), use various techniques
to detect differences between homoduplex and heteroduplex DNA molecules. Heteroduplexes
are double-stranded DNA molecules in which the sequence differs between the two strands. They are formed by denaturation and reannealing of a PCR product; as a patient may be homozygous for a mutation, the mutation detection rate for these approaches is highest if the patient PCR product is mixed with a PCR product from a control sample prior to denaturation,
to ensure that heteroduplexes as well as homoduplexes will form. DHPLC in particular lends itself to automation, and reportedly detects over 95% of mutations; however, the necessary equipment is expensive.58 In general, these techniques identify samples that possess sequence changes, and then the region containing the sequence change is sequenced from that sample to precisely characterize the mutation. The presence of a frequent neutral polymorphism may necessitate a lot of extra sequencing. In the future, mutation screening may increasingly be performed using oligonucleotide microarrays; these techniques are still being perfected.
Although PCR-based methods of mutation detection are efficient and require only small
quantities of patient DNA or RNA, some types of mutation are difficult or impossible to
detect using them. Therefore, other methods still play a role. Use of karyotyping and forms of
fluorescence in situ hybridization (FISH) allow detection of changes in chromosome number,
as well as large deletions or other rearrangements. Preparation of standard genomic Southern
blots, followed by hybridization with probes from the candidate region, and/or pulsed field gel
electrophoresis (PFGE) of genomic DNA digested with restriction enzymes that yield large
fragments, followed by Southern blotting and hybridization, can permit detection of deletions
and rearrangements of intermediate size. Although comparatively large amounts of DNA and
labor are required for such studies, they can be worthwhile, particularly in patients in whom
mutations remain unidentified after use of other screening methods.
Regardless of the method used for mutation detection, findings in patients should be compared
with those in a control sample (ideally ethnically matched, although this is not always
possible), to help distinguish disease-causing mutations from neutral polymorphisms. If a particular
disease mutation is found to occur frequently in patients, a specific assay for its efficient
detection can be developed.
Functional Consequences of Mutation
There are a variety of ways in which a disease mutation in a gene can ultimately affect the
function of the encoded protein; for example a mutation may affect transcription, mRNA
stability, translation, or protein stability, localization, or function. The functional consequences
of a mutation can often be predicted based upon the sequence change induced; nevertheless,
functional studies are extremely valuable in confirming, refining, or changing these predictions.
-
Transcription: a mutation may occur in a promoter or enhancer element of a gene, and prevent transcription of the gene, or alter the levels, timing, and/or tissue distribution of expression. An example of such a mutation was recently described in the Wilson's Disease
(WD) gene, ATP7B; a 15-bp deletion in the promoter region of the gene was present on
60.5% of Sardinian WD chromosomes, and reduced transcriptional activity by 75% in
expression assays.59 Mutations can also change splicing patterns of a gene.
-
mRNA stability and translation: Another way in which a mutation may effect mRNA levels is by decreasing the stability of the transcript. Mutation in a polyadenylation site may
prevent polyadenylation of the transcript, and lead to decreased stability of the mRNA, and/or inhibition of its translation. The presence of a sequence change that leads to premature termination of translation (such as a nonsense mutation, or an insertion, deletion, or splicing mutation that leads to a frameshift) can also lead to mRNA degradation through nonsense-mediated mRNA decay.60 Rarely, the initiation codon may be mutated.
-
Protein stability or localization: a mutation may have no effect on transcript levels, but alter the stability or localization of the protein. A mutation causing abnormal folding of a protein may result in its ubiquitination and subsequent degradation. Coding sequence mutations
may alter localization signals, or prevent post-translational modification, necessary
for delivery of a protein to its correct location. Several examples of mutations leading to
liver disease through such mechanisms are known. For instance, the common WD mutation,
H1069Q, appears to cause a protein-folding defect resulting in degradation of the
protein.61 Similarly, the common ABCC7 mutation in CF patients, deltaF508, causes ABCC7 to be incompletely glycosylated, mislocalized, and degraded.62,63 Two JAG1 mutations found in AGS lead to abnormal glycosylation and mislocalization of JAG1.64
-
Protein function: normal levels of a protein may be produced, and the protein may be delivered to its proper location, but a change in the amino acid sequence of the protein may prevent it from functioning properly; its function may be partially or completely destroyed.
For example, two ABCC7 mutations found in CF patients have been shown not to prevent
normal protein localization, but to disrupt protein function.62,65 Occasionally, mutations may instead cause a ‘gain of function’, in that the protein has higher activity than normal, or functions in ways or circumstances different from those of the normal protein. Dominant
negative mutations are loss-of-function mutations in which, in heterozygous individuals,
the mutated protein interferes with the function of the normal versions of the protein; generally, this effect occurs with multimeric proteins.
Genetics of Cholestatic Disorders
Application of genetic approaches has greatly increased our understanding of the molecular
biology of cholestasis. Through use of genetic and functional approaches, the genes mutated in
a number of disorders have been identified; others have been genetically mapped (
Table 1; a helpful source of more information and references on these disorders is the Online Mendelian Inheritance in Man database at http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=OMIM).
There remains much knowledge to be gained through further genetic study of cholestatic disorders.
For example, a recent report indicates that at least one additional low-gamma-GT PFIC
gene remains to be identified; in some families segregating low-GGT PFIC, the disease does
not map to either
ATP8B1 or to
ABCB11.
17 Mutation of
ABCB11 in a BRIC patient has
recently been reported, and a third BRIC locus waits to be identified, as a family in which
BRIC demonstrates an autosomal dominant mode of inheritance unlinked to either
ATP8B1
or to
ABCB11 has been described.
20,43 It is clear that many cholestatic disorders exhibit extensive allelic and/or locus heterogeneity, and this is both a challenge with respect to identifying the genetic etiologies of these disorders, as well as a rich source of information about the liver and the enterohepatic circulation in health and disease. Given the rapid pace and high efficiency
with which genetic research can now move, we will soon be learning much that is new
about the molecular basis of cholestasis.
This new knowledge will eventually extend beyond increased understanding of rare cholestatic
diseases; common genetic polymorphisms associated with increased or decreased susceptibility
to relatively common, less obviously ‘genetic’ disorders will likely be identified, as will factors
influencing disease severity, rate of progession, age of onset, and other similar variables. Work
in this area is already beginning, with the mapping of a modifier locus for meconium ileus in
CF and of potential susceptibility loci for ICP in the Finnish population.40,41,56 It is likely that some of the sequence variants eventually found to influence more common hepatic disorders
will occur in genes that encode proteins involved in hepatic transport and metabolism; these
genes may include those initially functionally linked to cholestasis through their identification
as disease genes for rare disorders.
Association studies of candidate genes have already revealed genetic factors that potentially
influence development of a number of hepatic disorders. In primary biliary cirrhosis (PBC),
alleles have been identified that appear associated with increased or decreased disease susceptibility
or rate of disease progression; these polymorphisms include ones in the major histocompatibility
locus (MHC), as well as one in the cytotoxic T lymphocyte-associated antigen-4 (CTLA-4) gene, one in the interleukin 1 gene, another in the vitamin D receptor, and a 4-SNP haplotype in the mannose-binding lectin (MBL) gene. 6671 Variation in susceptibility to primary sclerosing cholangitis (PSC), alcoholic liver disease, and chronic viral hepatitis may also be associated with alleles of polymorphisms in the MHC. 7277 PSC susceptibility also appears associated with a functional polymorphism in the stromelysin (MMP-3) gene;78 other genetic
associations to development of severe alcoholic liver disease have also been identified, including
an association with a common amino acid change in manganese superoxide dismutase, one
with a variant in the promoter of the CD14 endotoxin receptor, and a third with variants in the
interleukin-1beta gene.7981 In chronic hepatitis C, progression of liver disease may be associ-
ated with particular variants in the transforming growth factor beta 1 and angiotensin II genes.82 In evaluating these types of study, it is important to remember that false-positive evidence for association is sometimes obtained, and replication is necessary to confirm results. Also, due to
the existence of linkage disequilibrium (LD) between closely linked genetic markers, evidence
for association between a disease and a polymorphic allele does not necessarily indicate that
that allele itself is the functional cause of the association; another nearby genetic variant in LD
with the polymorphism studied may be the functional cause of the association, i.e., it may
affect production or function of a protein, and thus influence disease development. Understandably,
many genetic association studies of common liver diseases have focused on evaluation
of genes encoding proteins involved in function of the immune system. It will be interesting
to see whether associations between sequence variants in hepatic transporter genes and
particular disease features, such as ability to maintain adequate hepatic function despite liver
damage, may also be found.
Free Full text in PMC