genetic studies of rheumatoid arthritis using...

Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Medicine 1074

_____________________________ _____________________________

Genetic Studies of Rheumatoid Arthritis Using Animal Models.

BY

NIKLAS NORDQUIST

ACTA UNIVERSITATIS UPSALIENSISUPPSALA 2001

Dissertation for the Degree of Doctor of Philosophy (Faculty of Medicine) in MedicalGenetics presented at Uppsala University in 2001

ABSTRACT

Nordquist, N. 2001. Genetic studies of rheumatoid arthritis using animal models. ActaUniversitatis Upsaliensis. Comprehensive Summaries of Uppsala Dissertations fromthe Faculty of Medicine 1074. 48 pp. Uppsala. ISBN 91-554-5117-9.

Predisposition to autoimmune diseases such as, rheumatoid arthritis, diabetes, andmultiple sclerosis, is caused by the effect of multiple genes and a strong influencefrom the environment.

In this study, I have investigated genetic factors that confer susceptibility torheumatoid arthritis in a rat model. This work has led to the identification of severalchromosomal regions, containing uncharacterized genes that directly or indirectly areassociated to the arthritis development in these rats. We have observed that timing,gender, and genetic interactions are features that play a part in the effect that thesegenetic factors exert.

Unarguably, animal models for human disorders display differences to thehuman form of disease. An important fact is however that the same chromosomalregions are identified in both rodent and human studies, which suggests that there aregenetic factors that we have in common, which are involved directly or indirectly withan autoimmune response.

Focusing the interest on these similarities, and on the possibility to apply awide set of genetic tools, make animal models an invaluable, and probably necessary,instrument to dissect the genetic component of complex disorders. To fullycomprehend the genetic basis for a complex disorder like this, will requireunderstanding of how multiple genes interact with each other to cause disease.

We have been able to demonstrate that chronic arthritis, in a rat model forrheumatoid arthritis, is regulated by several genes and that these act during differenttemporal phases of the disease. These findings will hopefully contribute to ourunderstanding of the etiology and progression of rheumatoid arthritis.

Key words: Complex disease, autoimmune, rheumatoid arthritis, genetic, mapping,QTL, locus, linkage, animal model, rat.

Niklas Nordquist, Department of Genetics and Pathology, Rudbeck Laboratory, SE-751 85 Uppsala, Sweden

Niklas Nordquist 2001

ISSN 0282-7476ISBN 91-554-5117-9

Printed in Sweden by Lindbergs Grafiska HB, Uppsala 2001

I begynnelsen I

Det här avhandlingsarbetet begynte, liksom mycket annat, av en slump.Sommaren –93 feriearbetade jag som hantlangare åt min far vidrestaureringen av en kåk på Ekerölandet. Samtidigt sökte jag något somkunde vara mer relevant för mina pågående biologstudier. Ett samtal frånbygget till någon som hette Mats Sundvall, vid Institutionen förMedicinsk Genetik gav napp. Han kunde mycket väl tänka sig att ta in ensommarjobbare. Detta blev upptakten till forskarstudierna. Resten är nuhistoria. I efterhand anar jag att han var lika less på att köra geler som jagsenare skulle bli.

Niklas

II List of publications

List of publications

This thesis is based on the following papers, which are referred to in the text bytheir Roman numerals.

I. Carina Vingsbo-Lundberg*, Niklas Nordquist*, Peter Olofsson, MatsSundvall, Tore Saxne, Ulf Pettersson and Rikard Holmdahl. (1998).Genetic control of arthritis onset, severity and chronicity in a model forrheumatoid arthritis in rats, Nat Genet 20, 401-4.

II. Niklas Nordquist, Peter Olofsson, Carina Vingsbo-Lundberg, UlfPettersson and Rikard Holmdahl. (2000). Complex genetic control in arat model for rheumatoid arthritis, J Autoimmun 15, 425-432.

III. Peter Olofsson, Niklas Nordquist, Carina Vingsbo-Lundberg, AndersLarsson, Cecilia Falkenberg, Ulf Pettersson, Bo Åkerström and RikardHolmdahl. (2001). Genetic links between the acute phase response andarthritis development in rats, indicating a pathogenic role for the acutephase response, Arthritis Rheum In press.

IV. Shemin Lu*, Niklas Nordquist*, Jens Holmberg, Peter Olofsson, UlfPettersson and Rikard Holmdahl. Identification of two novel non-MHCloci which control pristane induced arthritis in rats - confirmation in acongenic strain

V. Niklas Nordquist, Hai-Tao Yang, Ludmila Elfineh, Carina Vingsbo-Lundberg, Kristin Bergsteinsdottir, Mats Sundvall, Rikard Holmdahl andUlf Pettersson. (1999). An improved genetic linkage map of the rat, RatGenome 5, 15-20.

* These authors have contributed equally to the work.

Reprints were made with permission from the publishers.

Table of contents III

Introduction................................................................. 1Genetic disorders......................................................... 1

Sibling relative risk.................................................. 1Twin studies........................................................ 1Simple vs. complex genetic disorders............................... 1Complex genetic disorders.......................................... 2

Conclusions “Genetic disorders” .........................................3

Benefits from using animal models...................................... 3Genetic tools in animal research.................................... 4

Congenic strains............................................... 6Recombinant inbred strains................................... 6Advanced intercross lines......................................6Transgenic strains..............................................7Knock-out, and conditional knock-out strains................ 7

The mouse or the rat? .............................................. 8Conclusions “Benefits from using animal models” ..................... 9

Autoimmunity............................................................ 10RA – Brief description and genetic components....................10Induced arthritis in animal models................................. 12Conclusions “Induced arthritis in animal models”................. 12

Quantitative trait loci analysis............................................13Genetic maps........................................................ 13QTL mapping........................................................ 13Phenotypes in QTL mapping........................................ 16Significance.......................................................... 16

Conclusions “Quantitative trait loci analysis” ........................... 17

IV Table of contents

Present investigation...................................................... 18Aim........................................................................18

Introduction.............................................................. 18Genetic predisposition to pristane-induced arthritis.............. 18

Results.....................................................................19Different genes control different phases of disease in

a model for rheumatoid arthritis in rats (Paper I)............19Complex genetic control in a rat model for rheumatoid

arthritis (Paper II)............................................. 20Genetic links between the acute phase response and arthritis

development in rats (Paper III)................................22Identification of two novel non-MHC loci controlling

pristane-induced arthritis (Paper IV)..........................23An improved genetic linkage map of the rat (Paper V).............24Summary of the papers............................................. 25

Discussion.................................................................25The animal model................................................... 25Overlapping QTLs suggests the presence of common

autoimmune susceptibility genes............................. 27Gender differences.................................................. 30Interactions..........................................................30Temporal shift of susceptibility loci.................................31

Future studies............................................................. 32

Concluding remarks...................................................... 33

Acknowledgements........................................................ 34

References.................................................................... 36

Abbreviations V

Abbreviations for animal models of autoimmunediseases

Aia Avridine induced arthritis

Bb Borrelia Burgdorferi induced arthritis

Cia Collagen induced arthritis

Eae Experimental allergic encephalomyelitis

Eau Experimental autoimmune uveitis

Idd Insulin dependent diabetes

Iddm Insulin dependent diabetes mellitus

Lmb Lupus in MRL x C57BL/6 mice

Lwb Lupus white and black

Oia Oil induced arthritis

Scwia Streptococcus cell wall induced arthritis

Sle Systemic lupus erythematosus

1 Genetic disorders

Introduction

Genetic disordersHuman genetic disorders occur as a consequence of abnormal functions in ourgenome. Which disorders then, are genetic disorders? In a broad sense, onecould say that our genes to some extent regulate all diseases. Many timesthough, it is obvious that a disease is caused by a pathogen from the outside. Itis for example a well-known fact that “the flu” is caused by the influenza virus.On the other hand, it is also well known that some people display mildsymptoms, while others get more severely affected. Differences in thesusceptibility to infectious agents can to some extent be ascribed to differentindividual genetic makeups (Dietrich, 2001).

In a more narrow sense, genetic disorders are commonly disorders thatare caused by malfunctioning genes or gene-products, or disorders where genescontribute substantially to the risk of becoming affected. So, how does onedetermine whether genes are of importance for a particular disease?

Sibling relative riskAn indication of a genetic component involved in predisposition to a disease isan increased sibling relative risk. This is a measure of the risk that, given oneaffected sibling in a sibling pair, the second sibling is also affected. It should benoted though, that this measure also includes the effect of shared environment.

Twin studiesMost disorders have a genetic component that explains parts of the observedphenotypic variability. In some cases, a predisposition for a certain disease canbe completely explained by genetic causes. Cystic fibrosis is a disease caused byone single defective gene, (Riordan et al., 1989). A carrier of the defective genewill, without exception, become affected. In other cases, the genetic componentmay explain only a small fraction of the total occurrence of a certain disorder.This can be exemplified with cervical cancer, where the genetic component hasbeen estimated to account for 27% of the variation in liability to disease(Magnusson et al., 2000). One can assess the magnitude of the role that geneticfactors play in a certain disorder by comparing concordance rates inmonozygotic and dizygotic twins. Since monozygotic twins are geneticallyidentical, and dizygotic twins share, on average, 50 % of their genes, thedifference in concordance between mono- and dizygotic twins can be ascribedto genetic causes.

Simple vs. complex disordersSome genetic disorders are inherited as simple Mendelian traits. These are themonogenic disorders, which are caused by a single gene. Cystic fibrosis,mentioned above, is an example of a simple genetic disorder, where “simple”

Genetic disorders 2

refers only to the predictability of the phenotypic outcome based on knowledgeof the genetic constitution. The causative gene in cystic fibrosis, CFTR, codes foran ion transporter on the surface of epithelial cells (Gregory et al., 1990).Mutations at certain positions in the CFTR gene make the gene productmalfunction. If an individual inherits one mutated copy of the gene from thefather, and one from the mother, that individual will be a CF genotype carrier,who will eventually develop cystic fibrosis. Another example of a simple geneticdisorder is hemophilia A, which is a blood disorder resulting in deficient bloodclotting. This disease is caused by mutations in a single gene that encodescoagulation factor VIII, which is necessary for normal blood clotting (Toole etal., 1984). This disease occurs predominantly in males (1 in 5-10,000 births),since the gene resides on the X-chromosome. Females, who have two X-chromosomes, will almost always have one functional copy of the gene.

Complex genetic disordersMost human genetic disorders are complex disorders, such as obesity, diabetes,certain types of cancer, and autoimmune disorders. These have been defined as“not exhibiting classic Mendelian inheritance caused by a single gene locus”(Lander and Schork, 1994). Complexities arise when the simple correspondencebetween genotype and phenotype breaks down. This complex behavior canoccur due to one or more of the following reasons.

Incomplete penetrance. This means that carrying a disease predisposinggenotype does not necessarily result in development of the disease. In otherwords, the disease predisposing gene may be necessary but not sufficient todevelop disease.

Genetic heterogeneity. The same disease can result from a defect in any one ofseveral genes. One can imagine that the end product in a biochemical pathwaywill be absent by blocking a reaction at any level in the reaction chain.

Phenocopy. Some individuals may develop a disease without carrying thepredisposing allele.

Polygenic inheritance. In general, complex traits, including genetic disorders aswell as normal physiological conditions are under the influence of many genes.Some genes may have a greater effect on the trait, while others have a smallereffect. The presence of several predisposing genes may be necessary to develop adisease. Genetic analysis of mice with insulin-dependent diabetes mellitus have,to date, revealed no less than 20 chromosomal regions harboring genes thataffect the susceptibility to develop the disease (Cornall et al., 1991; de Gouyonet al., 1993; Denny et al., 1997; Ghosh et al., 1993; Ikegami et al., 1995; McAleeret al., 1995; Morahan et al., 1994; Podolin et al., 1997; Prochazka et al., 1987;Rogner et al., 2001; Serreze et al., 1994; Todd et al., 1991).

3 Benefits from using animal models

Other transmission mechanisms. The mitochondria, which carry a smallcircular genome of its own, are inherited only through the maternal germ line.Leigh syndrome, which gives rise to pathological changes in the brain, is adisorder caused by mutations in the mitochondrial genome (Rahman et al.,1996).

Genomic trinucleotide repeats, which expand through the generations,have been shown to be associated with Huntington’s disease. These repeats arelocated in the Huntingtin gene and form a part of the transcript. An increasednumber of repeats causes earlier disease onset. The pathological mechanismthough, is still unknown (Andrew et al., 1993).

Genetic imprinting is another example of a type of genetic transmissionthat differs from traditional inheritance. After fertilization, some genes aresubjected to imprinting, which causes expression from either the paternal ormaternal allele to be silenced. The Prader-Willi syndrome is an example of adisorder that has been shown to be associated with imprinting. Loss ofexpression from paternally inherited genes in a region on chromosome 15, dueto erroneous imprinting, is one cause of the disease (Nicholls et al., 1998).

Conclusions “Genetic disorders”.Our genes are involved to a lesser or greater extent in regulation of mosthuman disorders. Assessing the importance of a genetic component for acertain disease can be done by looking at the degree of familial aggregation,and concordance rates in siblings and twins. Most human disorders areconsidered complex. The factors listed above gives rise to complex inheritancepatterns for many human disorders, and cause problems when performinggenetic analyses in human populations. Some of these problems arecircumvented by studying a restricted population, either with respect to a morenarrow disease definition, or a more genetically homogenous population, suchas ethnically or geographically isolated populations (e.g. Basques, Icelanders, orFinns).

Benefits from using animal modelsThe postulated genetic regulatory mechanisms of complex disorders, show thatthese disorders are not only difficult to study in humans, due to the presence offactors listed above, such as incomplete penetrance and heterogeneity, but alsoin animal models due to gene, and gene-environment interactions. Theidentification of these interactions is extremely difficult without the possibilityof an experimental approach where it is possible to study interacting agents ina presumably controlled background, such as in an animal model.

Even if humans would be the best species in which to identify genomicregions involved in genetic disorders, the problem still remains of how to provewhich genetic differences that are of importance. The confirmation that a

Benefits from using animal models 4

proposed alteration in the genome makes a difference to the course ofpathogenesis requires an experimental approach where the effect of thepredisposing alteration is either introduced onto an otherwise resistant geneticbackground, or replaced with a non-predisposing variant onto an otherwisesusceptible background.

The use of animal models has proven to be a powerful approach toidentify and understand genetic factors affecting the susceptibility anddevelopment of human disorders. The use of inbred strains of rats and micemakes it possible to avoid several of the inherent problems found in humangenetic analyses. First of all, genetic heterogeneity is avoided. Secondly, theenvironment can be controlled. Also, modeling a disease in experimentalanimals gives an obvious advantage in that the genetic and environmentalcomplexity can be manipulated.

There are animal models for a large number of human disorders.Examples of these are diabetes (Wicker et al., 1995), rheumatoid arthritis(Vingsbo et al., 1996), multiple sclerosis (Levine and Sowinski, 1973), obesity(Taylor and Phillips, 1996), hypertension (Kiprov, 1980), drug abuse (Berrettiniet al., 1994), cancer (Vogel et al., 1999), and behavioral disorders (Reeves et al.,1995).

The arguments most often raised against the use of animal models tostudy human disorders are that the diseases are not comparable, i.e. thatrodents and humans are too different to obtain biologically relevant resultsfrom rodent models of human disorders, and that most animal models requireinduction with something that triggers the development of disease. Eventhough the animal models display differences compared to the human form ofdisease, it is clear that there are also important parallels between them.

For example, an experimental model of rheumatoid arthritis, pristane-induced arthritis in rats (Vingsbo et al., 1996), fulfills the requirements neededfor being diagnosed as RA (Table 1). It has been observed that autoimmunedisorders often cluster in families, meaning that one disease is many timesaccompanied by other autoimmune diseases in these families. This wouldsuggest that many of these disorders share common factors that increase thesusceptibility (Becker et al., 1998; Jawaheer et al., 2001). The same has also beenobserved in some inbred rodent strains. Recent studies in animal models forautoimmune disorders have shown that there seem to be genetic factors thathave a function in the development of autoimmunity as a whole, rather than inspecific diseases (Bergsteinsdottir et al., 2000; Furuya et al., 2000; Jirholt et al.,1998; Kawahito et al., 1998). If these genetic factors could be identified inrodent models, it should be possible to find the corresponding factors, or theregulatory pathways where they exert their function, in humans.

Genetic tools in animal researchA lot of information on the genetic causes of a disorder can be obtained bycross breeding different strains of rats and mice. Typically, a disease susceptible


strain is crossed with a resistant strain. The offspring in the F1 generationbecome heterozygous for all autosomal loci throughout the genome, whichmeans that they obtain one gene from each parent at all positions on thechromosomes. An incidence of disease in the F1 generation indicates theinvolvement of dominant disease genes, meaning that one copy of a diseasegene is sufficient for the development of disease. The incidence in the F1generation also provides some information on the penetrance of the disease. Ifonly a portion of the progeny gets affected, it would indicate a reducedpenetrance of the disease genes, since all F1 progeny are genetically identical,except for the sex chromosomes. The next step in the breeding is to either crossF1 progeny to other F1 progeny to produce F2 intercross progeny, or to crossF1 progeny to one of the parental strains to produce F2 backcross progeny. Theillustration below outlines the different types of crosses discussed here (Figure1).

Figure 1. Breeding scheme to obtain BC2 backcross and F2 intercross progeny

The progeny in the F2 generation will have any combinations of genotypesthroughout the genome. Any given locus may display homozygosity for allelesfrom either of the parental strains. The progeny from the F2 generation can beused to search for genetic markers that co-segregate with disease, in order toidentify genomic regions that contain genes involved in disease development.This is further discussed in the chapter on quantitative trait locus analysisbelow.


Congenic strainIf one has already focused on an interesting genomic region, there are othertypes of genetic tools that can be applied to experimental animals to test thephenotypic effect exerted by genes in the region. One such tool is the breedingof congenic strains. Suppose that a region has been identified as beingassociated with a particular trait, and further, that this region originates from asusceptible strain of rats/mice. It is highly possible that additional regions, inthe genome of the susceptible strain regulate the trait under study. It wouldthen be of interest to isolate the effect that this particular region has on thetrait. By congenic breeding, one can isolate the genomic region on a differentgenetic background, by crossing the susceptible strain with a resistant strain,followed by crossing the progeny back to the resistant strain multiple times. Foreach generation of backcrossing, the genome of the progeny is enriched forgenes from the resistant strain. For each round of backcrossing, the interestingregion is typed with one or more genetic markers, to ensure that alleles fromthe susceptible strain are retained. After 12 generations of backcrossing it canbe assumed that the genome contains almost only alleles from the resistantstrain, except for the region that has been selected for, which is heterozygous atthis stage. An F1 intercross produces offspring that become homozygous for thetarget region. Since it is time consuming and expensive to breed animals forthis many generations, a more effective breeding technique is now commonlyused. This technique, named “marker assisted selection protocols” (MASPs),depends on the screening of progeny with genetic markers throughout thegenome, in order to select the offspring that show the highest rate ofbackground genome to be founders of the subsequent generation (Wakeland etal., 1997). This limits the time needed to obtain a clean congenic strain toapproximately 5 generations of breeding. In a recent study on a mouse modelfor SLE (systemic lupus erythematosus), Morel et.al. were able to show that apreviously identified susceptibility locus, Sle1, in fact contained three loci,which could independently cause the disease phenotype. The fine mapping ofthe Sle1 region was performed trough the analysis of congenic strains, whichmade it possible to reveal this cluster of functionally related loci (Morel et al.,2001).

Recombinant inbred strainsA similar, but more random, approach is to generate recombinant inbred (RI)strains. The genomes of these strains consist of random homozygous fragmentsfrom two different strains. A number of different RI strains can be a powerfultool to identify and isolate genomic regions affecting a trait, since a locus ofinterest can be narrowed down by comparing recombination events in them.

Advanced intercrossed lines F2 intercross progeny display a limited number of recombination events, whichresults in broad confidence intervals (CIs) for quantitative trait loci. These fairly


large chromosomal regions are not suitable for positional cloning or forcandidate gene identification. Advanced intercross lines (AIL) offer a possibilityto facilitate high-resolution mapping of these QTLs. Advanced intercross linesare produced by random intercrossing of progeny, starting in the F2generation. For each generation, the genome of the progeny is successivelyenriched in recombination events (Darvasi and Soller, 1995). In this way, largegenomic fragments, inherited from one of the parental strains, are split up tocontain smaller fragments from both founder strains. This could be comparedto the shuffling of cards. Imagine two decks of cards with different colors. Foreach turn of shuffling the chance will increase of finding two adjacent cardswith different colors. The AILs can be powerful in simultaneously limiting theconfidence intervals for several QTLs, which makes this method advantageousover congenic strains. Also, this method can make it possible to dissect QTLswith broad CIs into discrete regions that may contain more than one QTL.

Transgenic strainsIn mice and rats, it is possible to introduce single genes or genomic fragmentsby molecular genetic techniques, in order to produce transgenic animals. Theintroduced genetic material can originate from a different strain or even from adifferent species. This method involves the introduction of pure DNA bymicroinjection into the pronuclei of fertilized eggs. If the introduced DNA isstably incorporated into the genome it will be propagated to all cells in thedeveloping animal (Palmiter et al., 1982). In a study of gene function bytargeted deletions of genomic fragments in mice, Zhu et.al. were able toidentify a novel gene involved in the regulation of triglyceride production bythe use of the transgene technique (Zhu et al., 2000). A deletion of a 450kbgenomic region on mouse chromosome 11 caused the animals to suffer fromhypertriglyceridemia, liver and heart enlargement, growth retardation, andshortened life span. To identify the genes responsible, transgenecomplementation was used. Four mouse BACs (bacterial artificialchromosomes), covering the deleted region, and two human YACs (yeastartificial chromosomes) containing sequences orthologous to the deletedregion, were used to create transgenic mice. It turned out that one of thetransgenic animals, containing one of the human YACs, was indistinguishablefrom wildtype mice. Further analysis of the sequence of this human YAC,together with expression analysis, revealed that the absence of the gene OCTN2was responsible for the phenotype observed in mice that were homozygous forthe deletion.

Knock-out strains and conditional knock-out strainsWith mice, but not with rats, it is possible to create knockout animals, wherethe expression of a targeted gene or chromosomal region has been silenced.Knock-out mice are produced by growing embryonic stem cells in culture,creating targeted mutations in them and subsequently transplanting them into


pre-implantation mouse embryos to produce a chimera (te Riele et al., 1992;Thomas and Capecchi, 1987). This embryo will comprise wild-type cells as wellas lineages derived from the mutant cells. If the chimera has functional germcells that are derived from the mutated cells, offspring can be bred from thechimeras to produce heterozygous or homozygous mutant mice. This does notwork in other animals, since stem cells capable of producing germ linechimeras have so far not been reproducibly isolated except from certain strainsof mice.

Another, similar approach takes advantage of the P1 bacteriophagederived Cre recombinase, that catalyses recombination between specificrecognition sites, called loxP (Araki et al., 1995; Hamilton and Abremski, 1984).The introduction of loxP sites around a target sequence causes a recombinationbetween these sites, in the presence of Cre recombinase, which results inexcision or reversion of the target sequence. These methods can be verypowerful in studies of gene function. Two recent studies independentlyreported on the importance of a growth factor for normal limb buddevelopment, which was studied using the Cre-LoxP system (Lewandoski et al.,2000; Moon and Capecchi, 2000). Because the absence of this growth factorcauses early embryonic lethality, it had not been feasible to study the functionof this gene in vivo without a conditional system for silencing of the expression,offered by the Cre-loxP method.

The mouse or the rat?Historically, the mouse has been used extensively in genetic research. For thisreason, the pace at which genetic resources have been developed have beenhigher for the mouse than the rat. This includes genetic marker maps (Dietrichet al., 1996; Dietrich et al., 1994; Rhodes et al., 1998), a radiation hybrid map(Van Etten et al., 1999), YAC, BAC, and PAC genomic libraries (Chartier et al.,1992; Haldi et al., 1996; Osoegawa et al., 2000), and the whole mouse genomesequence (Celera Inc. 2001). However, during the past five years, high-throughput technologies for genotyping and sequencing have reduced the leadthat mouse genetic resources have had over the rat. Most resources that areavailable for the mouse are now also available for the rat, with the exception forthe large-scale sequencing of the rat genome, and the possibility to createknock-out rats, as mentioned above. On the other hand, the rat hastraditionally been used in research on mammalian physiology, mainly due toits comparatively larger size. As a result of this research, a wide variety of ratstrains have been selected for their expression of traits related to complexdiseases, such as diabetes, arthritis, multiple sclerosis, renal disease,hematological disorders, hypertension, autoimmune disorders, eye disorders,and behavioral disorders. With the development of genetic tools for the rat,these models have become an invaluable asset for the identification the geneticcomponents that regulate complex disorders. The recent advances in geneticand physical mapping of the mouse, rat, and man have created opportunities


for the construction of comparative maps (Nilsson et al., 2001; Remmers et al.,1999; Serikawa et al., 1999; Watanabe et al., 1999). These could make it feasibleto ‘jump’ between the species, if the desired methodology is not accessible forthe organism.

Conclusions “Benefits from using animal models”Unarguably, animal models for human disorders display differences comparedto the human form of the disorders. But, focusing the interest on theirsimilarities and on the possibility to apply a wide set of genetic tools on theexperimental animals, make animal models an invaluable, and probablynecessary, instrument to understand the genetic basis of complex disorders.

Autoimmunity 10

AutoimmunityA group of complex disorders that display genetic predisposition are theautoimmune diseases. Examples of autoimmune diseases are systemic lupuserythematosus (SLE), Hashimoto’s thyroiditis, multiple sclerosis, insulin-dependent diabetes mellitus and rheumatoid arthritis (RA). These arecharacterized by incorrect self-recognition, which leads to immune reactionsagainst either specific organs or non-organ specific targets. It has been reportedthat self-reactivity is a normal occurrence in the immune system, whichsuggests an essential role for this behavior, but also that the self-response iskept in control under normal conditions (McDevitt and Wakeland, 1998).

RA – Brief description and genetic componentsRheumatoid arthritis is an autoimmune disorder, which causes chronicinflammation in the small joints of the hands and feet. The criteria for beingdiagnosed with RA are shown in table 1.

The etiology and environmental factors that contribute to the risk ofdeveloping RA are poorly understood. It has been proposed fromepidemiological studies that coffee consumption (Heliovaara et al., 2000),smoking (Wolfe, 2000), oral contraceptives (Brennan et al., 1997), and highserum cholesterol levels (Heliovaara et al., 1996) are associated with anincreased risk of developing RA. An infectious etiology of rheumatoid arthritishas been postulated for a long time, but so far none has been identified.The prevalence, or population relative frequency, of RA is approximately 1% inCaucasian populations. Women are affected three to four times morefrequently than men. RA displays familial aggregation with a sibling relativerisk, λs = 8 (Wordsworth, 1995). This means that, given one affected sibling in asibling pair, the risk of a second sibling being affected is 8 times larger than forindividuals in the population on average. The concordance rate formonozygotic twins has been estimated to 15% (Silman et al., 1993). In a recentstudy of twins from Finland and the United Kingdom, the heratibility, or thevariation in liability to disease accounted for by genetic variation, wasestimated to 60% (MacGregor et al., 2000). Taken together, these data showthat genetic factors provide a substantial contribution to RA in the population.

Efforts made to identify genetic factors that explain the susceptibility toRA have to date come up with only modest results. A strong associationbetween the HLA region and RA has clearly been shown (Cornelis et al., 1998;Stastny, 1978).

11 Autoimmunity

Table 1. The American Rheumatism Association 1987 revised criteria for theclassification of rheumatoid arthritis (Arnett et al., 1988). Four of thesecriteria should be fulfilled to be diagnosed with RA. Abbreviations: PIP = proximal interphalangeal joint, MCP =metacarpophalangel joint, MTP = metatarsophalangeal joint.

It has been proposed that the association to MHC in autoimmunediseases is caused by the escape of self reactive T cells from thymic selection dueto poor self-binding properties of disease associated MHC class II molecules(Ridgway et al., 1999).

Two genome-scans of rheumatoid arthritis families have been published(Cornelis et al., 1998; Jawaheer et al., 2001). The sib-pair analysis performed byCornelis et.al. revealed HLA as a susceptibility locus. They could also identifyfourteen loci that did not reach the threshold for suggestive linkage. In a morerecent study on sib-pairs, Jawaheer et.al. could identify six susceptibility lociresiding on chromosomes 1, 4, 6 (HLA), 12, 16, and 17, respectively. None ofthese loci, however, except for the HLA locus, reached the threshold forsignificant linkage. The locus on chromosome 17 reached the level of asuggestive linkage.

Autoimmunity 12

Induced arthritis in animal modelsA number of rodent models of RA exist. These have the fact that the disease isinduced in common. There are a variety of induction methods, of which themost commonly used is heterologous or homologous type II collagen, acomponent of cartilage. The introduction of a joint-derived antigen triggers anautoimmune response against peripheral joints. Another method for inductionof arthritis in rodents is the use of an adjuvant, such as mycobacteriumtuberculosis in mineral oil (Pearson, 1956), avridine in mineral oil (Vingsbo etal., 1995), pristane (Vingsbo et al., 1996), or Freund’s mineral oil (Holmdahland Kvick, 1992). The adjuvants, which are considered to be non-antigenic, arebelieved to enhance a nonspecific immune response that somehow triggers anauto-reactive immune attack on the peripheral joints of the paws.

Conclusions “Induced arthritis in rats”Animal models of rheumatoid arthritis can provide important clues as to thecauses and regulatory mechanisms of the human form of disease. There areprobably species-specific differences between the human and rodent forms ofthe disease, but also important parallels that can be used to understand theetiology and development of the disease and to provide the basis forestablishing novel therapies.

13 Quantitative trait loci analysis

QTL analysisQuantitative trait loci (QTL) analysis is a method to identify chromosomalregions that affect the variability of a trait without prior knowledge of thegenetic causes of the observed trait variation.

Genetic mapsQTL analysis requires genetic maps, which are representations of geneticmarkers positioned within genomic regions. See figure 2 below. The distancesbetween the markers are determined by the frequency of recombinationbetween them. This represents a statistical distance measured in units ofcentimorgan (cM), i.e. the probability of a recombination event between themarkers, rather than a physical distance measured in number of nucleotides.There are different types of genetic markers. The most commonly used are theRFLP (restriction fragment length polymorphisms), SSLP (simple sequencelength polymorphisms), and SNP (single nucleotide polymorphisms) markers.They can be scored, either for size-differences between individuals, or for DNAsequence differences. This makes it possible to determine the marker genotypeof an individual.

Figure 2. A genetic map of rat chromosome 4 with 31 microsatellite markerspositioned along the chromosome.

Quantitative trait loci analysis 14

QTL mappingTo detect linkage between a marker and a putative QTL, one must have asegregating population, such as progeny from an experimental animal cross.

Figure 3. The illustration briefly outlines the principal of QTL analysis. Inpractice, computer programs designed for this purpose perform the statisticalcalculations used for testing the strength of the evidence for the presence of aQTL adjacent to a genetic marker.


Interval mapping is a commonly used method for detecting QTLs (Lander andBotstein, 1989). The principle behind interval mapping is to test a model forthe presence of a QTL between two mapped marker loci at many positions. Thestrength of a given model is tested using the maximum likelihood method,measured as the likelihood ratio (LR) or logarithm of the odds (LOD).Maximum likelihood involves searching for QTL parameters that give the bestestimation for the quantitative trait distributions that are observed over thepossible genotypic groups of a putative QTL. The models are evaluated bycomputing the likelihood of the observed distributions with and without fittinga QTL effect to them. The LOD score is the logarithm of the ratio between theselikelihood estimates.

Variations on the same theme include Composite Interval Mapping(CIM) (Zeng, 1994), and Multiple Interval Mapping (MIM) (Kao et al., 1999).The CIM method makes use of a combination of interval mapping and multipleregression. By using information from markers outside a defined interval, onecan avoid other QTLs affecting the test results within that interval. The MIMmethod allows for simultaneous scanning of multiple putative QTLs by thejoint analysis of multiple marker intervals.

The results obtained from QTL analysis provide information on thelocation of a putative QTL and the statistical strength of the presence of a QTLin that location, see figure 4.

Figure 4. Graphical illustration of results from a QTL scan of a genomicregion. The log likelihood for the presence of a QTL has been plotted againstthe genetic distance. The width of the shaded area represents the 95%confidence interval for the QTL, which comprises about 25 cM.

Quantitative trait loci analysis 16

The most commonly used experimental animal populations for QTL analysisare F2 intercross and BC2 backcross progeny. In these experiments, QTLmapping gives a very crude measurement of the QTL position, with a typicalconfidence interval of 10-30cM. This genetic distance translates into a physicaldistance of approximately 10-30 million basepairs, which could harborhundreds or even thousands of genes.

Phenotypes in QTL analysisSince QTL analysis is dependent on the measurement of a phenotype, it isimportant to consider the nature of these quantitative traits. For many of thecommonly used QTL software packages, such as Mapmaker/QTL (Lincoln et al.,1992) and QTL cartographer (Basten et al., 1994), the test statistics assume thetraits to be normally distributed. Deviations of the trait distribution fromnormality may result in false positive results or an erroneous estimation of thephenotypic effects of QTLs, or even failure to identify existing QTLs. This can, tosome extent, be overcome by variable transformation, whereby the distributionof trait values can be approximated to a normal distribution.

Many times, a scoring system such as disease severity is used for thequantification of a trait. The scoring system measures severity on an intervalscale. If the underlying trait variable is considered normally distributed, thescoring system poses no problem in the analysis, apart from reduced statisticalpower. If, on the other hand, it is not possible to determine the distribution ofthe underlying variable, one has to turn to other test methods, such as non-parametric analyses that do not have a prerequisite for a defined traitdistribution. Kruglyak & Lander (1995) have developed a QTL analysis methodthat can handle non-parametric traits. However, this method has only beenimplemented for sib-pair analysis at this point.

Multiple traits that show a strong correlation between them will, bynecessity, show association to the same QTL. This does not necessarily meanthat the underlying gene/genes in the QTL have a causative effect on all traits,or even for any of the traits studied. The variation observed for the correlatedtraits may be a secondary effect caused by the effect that the gene/genes haveon a primary causative trait. The observation that a phenotype under study isassociated to a certain QTL may therefore lead to erroneous conclusions aboutthe genetic regulation of that phenotype.

SignificanceWhat significance level should be chosen in order to be confident that a linkedlocus contains a true QTL? There is no definite answer to this question. Theconventional confidence level used in a genome scan is 95%. This means thatthe probability of a false positive result is 5%. When performing QTL analysisusing interval mapping, or variants thereof, these five percent must betranslated into the corresponding logarithm of the odds ratio (LOD). This valuewill vary, depending on the type of experimental cross that is being used.


Lander & Kruglyak (1995) have proposed a set of significance thresholds forvarious conditions that, for a long time, have been used as a standard forreporting linkage results. These proposed significance levels, however, assumesthe use of an infinitely dense map and traits that are normally distributed.

To establish a meaningful significance level in cases where the traitdistribution deviates from normality, one can perform permutation tests onthe data. Permutation testing is a method that results in an experiment-wisesignificance level. E.g. in an experimental animal cross where a genetic map hasbeen established, the trait values are shuffled randomly between the individualsto break any relation between genotype and phenotype. For each turn ofshuffling, a full QTL analysis is performed and the highest test statisticrecorded. This is subsequently repeated many times (102 – 104). The resultingdistribution of maximum test statistics shows what level of the test statisticthat yields, for example, a genome-wide nominal significance level of α=0.05. Ifthe genetic map contains gaps, the permutation test method can easily result inan overestimation of the test statistic needed for α=0.05. This is due to the factthat a gap in the genetic map results in a high probability of recombinationevents in the gap, which in turn increases the probability of fitting a QTL inthat region. If the genetic map suffers from large gaps, it could be advisable toinstead split these regions in to separate linkage groups before performingpermutation tests to obtain an experiment-wise significance level.

Multiple testing is another factor to consider in choosing an appropriatesignificance level. A common way to treat situations when multiple testing havebeen performed is to apply some kind of correction to the results in order toattain a nominal significance level of e.g. α=0.05. The Bon-Ferroni correction,which is commonly used states that an independent test is consideredsignificant if the probability of a given outcome is less than α/n, where ‘n’ isthe number of independent tests performed.

Conclusions “ Quantitative trait loci analysis”QTL analysis is a statistical method used to detect chromosomal regions thatshow association to a quantitative phenotype. The results from such analysescan provide an approximate location of a genomic region harboring agene/genes, which could be directly or indirectly responsible for the variabilityof the phenotype. In performing a QTL analysis one has to consider factors thatcan affect the results, such as, density of genetic markers, distribution of thephenotype, multiple tests performed, and level of significance.

Aim & Introduction 18

PRESENT INVESTIGATION

Aim

The aim of this study was to identify and characterize genetic components thataffect the susceptibility to rheumatoid arthritis in animal models.

Introduction

Pristane-induced arthritis in rats is an animal model for rheumatoid arthritis(RA) (Vingsbo et al., 1996). Pristane is a natural component in plants, as a partof the chlorophyll, and is therefore ingested and also absorbed by animals(Garrett et al., 1989). However, some strains of rats develop a disease thatmimics many aspects of RA following a subcutaneous injection with pristane.Rats develop the disease a few weeks after induction. The initial manifestationis a sudden inflammation in a peripheral joint whereafter the disease spreadsto other peripheral joints and usually develops along an active chronic diseasecourse. The inflammation is restricted to the peripheral joints and does seldomaffect other more centrally located joints or other tissues. As in RA, the jointsare usually affected symmetrically.

Genetic predisposition to pristane-induced arthritisThere is a genetic predisposition to the development of PIA. DA and Lew.1f ratsare highly susceptible, while E3 rats are almost resistant. Studies using MHC-congenic Lew rat strains have shown that the MHC locus contributes to thegenetic predisposition (Vingsbo et al., 1996). Interestingly, this effect was seenonly on the chronic development of disease, whereas no significant associationto the time of onset or the incidence was observed.

However, the influence of the MHC cannot explain the dramaticdifference in susceptibility to PIA seen between DA and E3, as the MHChaplotypes of these strains are equally permissive on a Lew background.Instead, there is a major non-MHC genetic influence that determines thedifference in PIA susceptibility observed between the DA and E3 strains. PIA isclearly a polygenic disease, as the penetrance of the disease is variable in a seriesof recombinant inbred strains made from DA and E3 (Vingsbo et al., 1996;Vingsbo-Lundberg et al., 1998).

The DA strain is highly susceptible to PIA and develops the disease withhigh incidence, pronounced joint erosions, high clinical severity and with achronic relapsing disease course. The Lew.1f strain is also highly susceptible, butdisplays less erosion of bone and cartilage during the acute and chronic phaseof disease compared to the DA strain. In contrast, the E3 strain is almostresistant. Among recombinant inbred strains from DA and E3, the DXEA strain

19 Results

develops mild disease with high incidence, the DXEB strain has more severe andchronic disease but with lower incidence whereas the DXEC strain is almostresistant (Vingsbo et al., 1996). This demonstrates that different genes controldifferent disease phenotypes. The disease development and progression in thestrains used is summarized in table 2:

Table 2. Development and progression of pristane induced arthritisin different rat strains.

Strain n Mean dayof onset

Mean maxclinical score

Frequency ofarthritis

Frequency ofchronic arthritis

DA 17 14 ± 2 11 100% 100%E3 23 38 ± 6 2 9% 0%Lew.1F 10 17 ± 4 12 100% 100%DXEA 20 31 ± 21 3 70% 10%DXEB 24 72 ± 34 5 33% 30%DXEC 20 32 ± 15 3 10% 0%

Results

Different genes control different phases of the disease in a model forrheumatoid arthritis in rats (Paper I)In this paper, we analyzed the progeny from a cross between the resistant E3strain and the susceptible DA strain. The aim of this study was to identifygenetic regions that could explain the difference in susceptibility to pristane-induced arthritis observed between these strains. The QTL analysis identifiedfive autosomal loci that affected the susceptibility to pristane-induced arthritis.

The development of pristane-induced arthritis can be viewed as havingthree phases. Firstly, there is the onset phase of disease, which is characterizedby an initial acute inflammation. This is followed by the second phase, whichexhibits a prolonged severe inflammation with bone and cartilage erosion.Eventually, the disease enters a chronic phase with continuing inflammationand progressive erosion/healing of bone and cartilage, which leads to improperbone formation, and thus malformation of the joints.

We found that the five identified loci were exclusively associated with acertain phase of the disease (table 3).

Results 20

Table 3. QTLs identified in the (E3xDA)f2 intercross

QTL Chr Trait Marker Inheritance LODVariationexplained

Regions associated with arthritis onset

Pia2 a 4 Day of onset D4Mgh14 E3 dominant 3.9 32 %Pia3 6 Day of onset D6Wox5 E3 dominant 4.5 25 %

Regions associated with arthritis severity and early joint erosion

Pia4 12 No. of affected paws day 28 D12Wox14 DA additive 8.4 22 %

Regions associated with arthritis chronicity

Pia5 b 4 Inflammation score day 120 D4Wox20 DA additive 4.5 24 %Pia6 14 No. of affected paws day 83 Csna DA recessive 4.9 19 %

a The effect of Pia2 was detected in female rats only (corresponding LOD score in male rats, 0.03).b The effect of Pia5 was detected in male rats only (corresponding LOD score in female rats, 0.2).

Two QTLs were associated with the onset of disease, Pia2 and Pia3,located on chromosome 4 and 6, respectively. Surprisingly, the diseasepromoting alleles at these loci originated from the resistant E3 strain.Alternatively, a protective effect was exerted by the corresponding DA alleles.The earlier onset that was caused by Pia2 was only observed in female rats.On chromosome 12 we identified a locus, Pia4 that showed a strongassociation to disease severity. Two QTLs, Pia5 and Pia6, were exclusivelyassociated with severity during the chronic phase of disease. Pia5, located onchromosome 4, was associated with histological scoring of inflammation at day120. This effect was only observed in male rats. Pia6, on chromosome 14 wasassociated with chronic severity. This association peaked at day 83 afterinduction with pristane.

These findings suggest that different sets of genes in the rat genomecontrol different events during disease development and further, that some lociexert their effect in a gender-specific manner.

Complex genetic control in a rat model for rheumatoid arthritis (Paper II)In this study, a cross between the susceptible strain DA and a resistantrecombinant inbred (RI) strain, DXEC, was analyzed. Despite carrying theseverity locus Pia4 on chromosome 12 and the chronicity locus Pia6 onchromosome 14, the DXEC strain was found to be almost resistant to pristane-induced arthritis. The aim of this experiment was to investigate whether newloci that confer susceptibility could be identified in a cross where twoimportant disease predisposing loci were shared in all progeny.

The RI strain DXEC originated from a cross between DA and E3.Consequently, its genome is a mixture of fragments from these two parental

21 Results

strains. An RI strain is defined as being homozygous for all loci, but these canoriginate from either of the founder strains. Crossing the DA strain and theDXEC strain results in enrichment for DA alleles, since both strains contributethese alleles.

In this cross, we could confirm the onset locus, Pia3, on chromosome 6.As was observed previously, this locus contains a gene/genes that causes earlierdisease onset if at least one allele is inherited from the E3 strain. A puzzlingfinding in this cross, was that Pia3 only showed an effect in males, whereas itwas equally permissive in males and females in the (E3xDA)F2 cross.

Two new loci were identified in this cross. One locus on chromosome 1,Pia8, was associated with disease severity. The allele that conferred susceptibilityoriginated from the resistant E3 strain and the effect was found to be restrictedto female rats. The second locus, Pia7, was located on chromosome 4 and it wasassociated with severity during the onset phase of disease. The underlyinggene/genes was found to have an effect in rats homozygous for the DA allele atthis locus.

We could also detect a suggestive linkage to the MHC region onchromosome 20. This locus was associated with severity at day 56. Thiscoincides with the time at which the disease usually progresses into the chronicphase. The justification for stating that the chronic phase starts at this time isthat animals affected at this point, in the majority of cases, remain affected.

There is a significant difference in severity (severity score day 35)between offspring from the two crosses (E3xDA)F2 and (DAxDXEC)f2. Thisdifference could be explained by the presence of Pia4 in all progeny from thelatter cross.

We did not find any evidence for an effect of the newly identified QTLsin the previous cross, (E3xDA)F2. There could be several reasons for this. Otherresearch groups have reported the identification of new QTLs after stratifyingtheir data with respect to a permissive genotype at a susceptibility locus(Remmers et al., 1996). One can imagine two loci that have an effect on acertain trait and further, that one locus shows a much stronger association withthe trait than the other. In this case, the variation caused by the second locuscould be “hidden”, or “masked” by the effect from the first, stronger, locus. If,on the other hand, one performs the analysis on a subset of data that containsonly animals with the permissive genotype at the first locus, the remaining traitvariability cannot be due to the first locus and is therefore caused by the secondlocus.

Another possible explanation as to why we did not detect these loci inthe previous cross, could be the presence of epistasis. Per definition, epistasis isthe influence of alleles at one locus on the phenotypic effects of alleles at otherloci (Bateson, 1909). It thus reflects the fact that the expression of geneticvariation is under the influence of other genes.

Thus, the reasons why we could not identify these new loci in the(E3xDA)F2 cross could be, either that a QTL with a large effect “masked” the

Results 22

smaller effect caused by the new QTLs or that the new QTLs are dependent oncertain alleles at another locus (i.e. epistasis). To test for these possibilities, were-analyzed the (E3xDA)F2 cross. This was performed by stratifying the datawith respect to the genotype at loci that are DA homozygous in the DXECstrain. In this way we tried to mimic the conditions of the (DAxDXEC)F2 cross.In the re-analysis of the (E3xDA)F2 cross we used, for a certain locus, only thoseanimals which had been scored as DA homozygous. This stratification was thenperformed for all loci where the alleles were derived from the DA strain, in theDXEC strain. The expected outcome, if there was a “masking”-effect, would bethat we could detect the new QTLs in the (E3xDA)F2 cross, given that we selectfor DA homozygosity at one of the QTLs identified earlier. Alternatively, ifepistasis was the cause, we would expect to identify the new QTLs, given that allanimals share a certain locus with alleles derived from the DA strain, but initself did not act as a QTL.

The analysis revealed that the loci Pia1 and Pia7 could be identified inthe (E3xDA)F2 cross and that the effect of these loci were dependent on DAhomozygosity at other loci on chromosome 14 and 18. These regulatory loci didnot, in themselves, have any detectable effect. However, they were aprerequisite for Pia1 and Pia7 in order to exert an effect on the trait variation.This would suggest the presence of epistasis. The number of animals left afterstratification was 17-20, which is a very low number for performing theanalysis. The multiple tests performed in the analysis should also be considered.Still, these QTLs show association to the same traits and with the sameinheritance pattern as in the (DAxDXEC)F2 cross.

In summary, these findings show that additional loci can be identified ifthe effects of other loci are neutralized. Also, results from this study suggestthat epistasis may have an important role in regulating the development ofpristane-induced arthritis.

Genetic links between the acute phase response and arthritis development inrats (Paper III) Previously, it had been observed that the levels of interleukin-6 (IL6) weredifferent between DA and E3 rats, with regards to pristane-induced arthritis.This study was performed in order to investigate the genetic regulation of IL-6and the related cytokines and their potential involvement in the regulation ofpristane-induced arthritis.

The acute phase response is marked by an instant increase of certainserum proteins upon infection and inflammation, which are the acute phaseproteins (Peri and Rinaldi, 1998). The measurement of the acute phase proteinsand the components associated with the cytokine system may provide valuableinformation about the initial processes in arthritis development. The levels ofα1-acid glycoprotein (AGP), α1-inhibitor3 (α-I3) and interleukin-6 (IL6), weredetermined in the F2 progeny from a cross between the rat strains DA and E3.

23 Results

These traits were used in QTL analysis as to identify loci explaining theirphenotypic variability.

The results revealed that QTLs, which explains a portion of the variabilityobserved from the acute phase proteins, were partly co-localized with regionsidentified previously to control arthritis development. The arthritis severitylocus, Pia4, on chromosome 12 (Vingsbo-Lundberg et al., 1998) was found tobe associated with both AGP (measured at day14) and IL6 (measured at day35). These associations could only be detected in male rats. Interleukin-6(measured at day 14) was found to be associated with a locus on chromosome14. This QTL overlaps with Pia6, which is associated with severity during thechronic phase of the disease. The effect of this QTL on levels of IL6 could onlybe detected in male rats, as well. Also, we identified a suggestive QTL, whichexplained parts of the observed variability for AGP (measured at day 14) thatcoincided with Ciaa3, a locus on the rat chromosome 1, which is reported to beassociated with anti-collagen antibody titers (Griffiths et al., 2000).

Thus, the question arises, are the observed correlations causative orsecondary effects? It has been reported that there is an association betweenacute phase proteins and arthritis development (Charles et al., 1999; O'Hara etal., 2000; Peri and Rinaldi, 1998). The results presented in this paper show thatthere is an association between certain acute phase proteins and differentdisease parameters of pristane-induced arthritis in rats. Furthermore, that thechromosomal regions contain genes, which could directly or indirectly regulatethe levels of these proteins. These regions coincide with regions identifiedpreviously that are involved in controlling the susceptibility to pristane-inducedarthritis. This does not necessarily mean that the same genes are operating onboth traits, even though it is a compelling thought. Regions that coincide couldargue for the presence of gene clusters, where the individual genes have incommon the fact that their products are parts of the samebiochemical/physiological pathways (1999; Morel et al., 2001). This couldexplain why so many of the autoimmune QTLs seem to be co-localized in thegenome.

Identification of two novel non-MHC loci which control pristane-inducedarthritis in rats (Paper IV)A cross between the pristane-induced resistant rat strain E3 and the susceptibleLew.1f strain was set up to confirm previous findings from the crosses betweenthe DA and E3 strains. Also, we wanted to map a susceptibility locus that co-segregates with the albino phenotype of Lew.1f, which is caused by alleles at theC-locus on chromosome 1. The development of pristane-induced arthritis inLew.1f rats differ somewhat from what is seen in DA rats, in that they show lesserosion of bone and cartilage during the acute and chronic phase of the disease.

The disease development in the (E3xLew.1f)F1 population displayed areduced incidence and severity, as compared to the Lew.1f strain. Interestingly,

Results 24

only female rats developed the disease in the F1 population, which furtherconfirms the observation that there is a strong gender bias in this disease.

In this cross we could identify two new QTLs on chromosomes 1 (Pia9)and 16 (Pia11), respectively. Also, we found a suggestive linkage tochromosome 14. The locus Pia4 identified previously, on chromosome 12, wasconfirmed in this cross with the same inheritance pattern as in the (E3xDA)F2cross.

In the same region as the C-locus, which is responsible for the albinophenotype of the Lew.1f strain, we identified the locus, Pia9, which wasassociated with arthritis severity during the acute phase of the disease. Thepredisposing allele originated from the Lew.1f strain. This was also confirmedin a congenic strain, where E3 alleles in the Pia9 region on a Lew.1fbackground, had a significant protective effect. Also, the Pia11 locus wasassociated with arthritis severity in the acute phase of the disease, but with thepredisposing allele derived from the resistant E3 strain.

A locus, Pia3, which was identified previously in the E3xDA cross, didnot show any effect in the E3xLew.1f cross. This locus was associated withdisease onset, where the E3 allele in a dominant mode caused an earlier diseaseonset. The fact that no effect of this locus could be detected in the present cross,would suggest that alleles from the susceptible strain DA, at the Pia3 locus,confer a protective effect, causing a later onset of disease.

Taken together, these findings suggest that different strains carry uniquepredisposing disease loci, as well as loci that are common between differentstrains.

An improved genetic map of the rat (Paper V)In order to locate integration points between three major genetic maps of therat genome (Bihoreau et al., 1997; Brown et al., 1998; Jacob et al., 1995), a high-resolution linkage map was constructed from various crosses between the ratstrains DA and E3. The genetic map is based upon genotype data from twoseparate projects. First, the rat arthritis project which consists of three crosses,(DAxE3)F2, (E3xDA)F2 and (DAxDXEC)F2, where DXEC is a recombinantinbred strain derived from a cross between the DA and E3 strains. Second, therat experimental autoimmune encephalomyelitis project, which consists of onecross, (E3xDA)F2. Together, these crosses comprise a total of 600 f2 progeny. Aset of microsatellite markers has been screened for size polymorphisms in theparental strains, of which approximately 320 have shown to be informative.

The size of the rat genome is approximately 2000 cM. Assuming onecrossover per 100 meioses, the information from 600 f2 progeny gives a mapresolution making it possible to order markers that are separated by a geneticdistance above 0.2 cM.

This map was also used to characterize four recombinant inbred (RI)strains, DXEA, DXEB, DXEC and DXER, all derived from the cross E3xDA. The

25 Discussion

strain DXER turned out to carry microsatellites of sizes that were not consistentwith either of the parental strains. Therefore, it is assumed that the DXER strainhas been contaminated with genes from another rat strain. For this reason, theDXER strain was excluded from further studies. The RI strains are of greatinterest when studying the genetics of autoimmunity, since the parental strainDA can develop induced autoimmune encephalomyelitis and induced arthritis.To get beyond the limit of resolution possible with QTL analysis the RI strainswill be a powerful tool in establishing the location of quantitative trait lociaffecting these disorders.

Summary of the papersUsing three different rat crosses, E3xDA, DAxDXEC and E3xLew.1F, we havebeen able to identify a number of genomic regions, which harbor unknowngenes that partially explain the variability of the susceptibility to pristane-induced arthritis. We have identified ten such regions, Pia1 thru 9 and 11. Eachof these regions has either an effect on disease onset, severity and joint erosionsor chronicity. These findings demonstrate that different genes control differenttemporal phases in this chronic self-perpetuative disease. Still, some of thepieces in this jigsaw puzzle are missing. This is demonstrated by the inability topredict the outcome based upon the genetic composition of the animal.

We have observed that the effects from some of these autosomal loci aregender specific. E.g., the locus Pia2 on the rat chromosome 6 is associated withthe onset of disease. There is a significant difference in disease onset betweenthe group of female rats, which carries the susceptible allele and those that donot. Among male rats, this locus does not show any effect. Thus indicating thatthe underlying gene is dependent upon a female context. Furthermore, thedisease-promoting allele is derived from the resistant strain E3. The finding thatthe resistant strain carries susceptibility alleles has also been observed in otherloci.

Furthermore, we have suggestive evidence for the presence of geneticepistasis. Two loci (Pia1 and Pia7) identified in the cross DAxDXEC were notobserved in the other crosses. However, when re-analyzing the E3xDA crosswith a model for a two-locus interaction, the loci were identified and weredependent on other loci in order to have any effect.

Discussion

The animal modelStrains of rodents, which have been used in studies of various characteristics,whether in the field of physiology, behavior, pathology or genetics, have incommon the fact that they have been subject to inbreeding for numerousgenerations, in order to isolate the desired traits. Undoubtedly, the use ofinbred strains is a powerful tool to study isolated variables in the absence of

Discussion 26

genetic variation and with control over the environmental factors. The geneticvariation among the available inbred rodent strains can be considered limiteddue to the restricted number of founders that have been used to create them(Beck et al., 2000).

In certain strains, prone to the development of autoimmune disease,there is a tendency to identify the same loci, regardless of what disease is beingstudied. The strains that were used in the study on genetic predisposition topristane-induced arthritis, were also used to study the genetic regulation of EAE(Bergsteinsdottir et al., 2000). In these studies, almost all loci associated to PIAco-localized with loci associated to EAE and to a large extent displayed an effecton the same traits. In fact, browsing through the literature on rat models forRA, shows that, out of 19 QTLs identified in five different crosses, that have incommon that they utilize the DA strain as the susceptible strain, only seven donot co-localize with any of the other QTLs (see table 4).

Table 4. QTLs identified in different crosses utilizing the rat strain DACrosses

DA x BN DA x ACI DA x F344 DA x LEW DA x E3

RNO 1 Cia2 Pia9 Pia8RNO 2 Cia7RNO 3 Cia11RNO 4 Cia13 Cia3, Aia3 Oia2 Pia2, Pia5, Pia7RNO 7 Cia4, Cia8RNO 10 Cia5 Cia5 Oia3RNO 12 Cia12 Pia4RNO 16 Pia11Co-localizing QTLs identified in the different crosses. RNO 1 : Pia9, Cia2RNO 4 : Pia5, Cia3, Aia3, and Pia7, Oia2, Cia13RNO 10 : Cia5, Oia3RNO 12 : Pia4, Cia12

The remaining twelve QTLs are distributed over five distinct chromosomalregions. These loci may harbor genes that have a common function in thedisease, which makes the DA strain predisposed to both collagen and adjuvantinduced arthritis. RA, for example, displays considerable clinical heterogeneityand may represent a set of distinct diseases; a syndrome that shares certaincharacteristics, rather than a single disease (Weyand et al., 1998). Each of thesesubtypes of RA may involve both specific and common pathogenic pathwaysthat lead to disease. From this perspective, the data support the idea that eachanimal model may represent only a limited part of the disease or a specificsubset of pathogenic events that can result in disease.

27 Discussion

Overlapping QTLs suggests the presence of common autoimmune susceptibilitygenesOver the past few years, the field of bioinformatics has evolved withtremendous speed. As a consequence of this, the scientific community has beenprovided with tools, available over the Internet, which has made retrieval ofcomplex information a simple task. Such undertakings were nearly impossible adecade ago. Some of these tools involve the databases that contain informationon homologies between species with regard to genes and genetic markers. Thishas made it possible to construct comparative genomic maps. One of the mostprominent sites for mouse genome informatics is maintained by the JacksonLaboratory (www.informatics.jax.org). This site, among many other things,contains comparative genome maps between several species, including mouse,rat, and man.

In QTL analysis, these comparative maps are very interesting, since theyallow for comparison of findings between the species. The illustrations below(figure 5) show the rat chromosomes where we have found loci that areassociated with pristane-induced arthritis. Also, a selection of QTLs for variousautoimmune diseases that are linked to these chromosomes are included. TheQTLs that co-localize with Pia 1-9 and 11 are tabulated in table 5.

Table 5. QTLs identified in other studies of autoimmune diseases, in rat, andmouse that co-localizes with Pia1-9 and 11.Locus Chr a QTLs of autoimmune diseasesPia1 20 Defining the MHC-locus*Pia2 4

rEae11(Bergsteinsdottir et al., 2000)Pia3 6

rEae9(Bergsteinsdottir et al., 2000), mTnfp(Libert et al., 1999)Pia4 12

rCia12(Remmers et al., 1996), rEae5(Bergsteinsdottir et al., 2000), rEau2(Sun

et al., 1999), mLbw3(Kono et al., 1994)Pia5 4

rAia3(Kawahito et al., 1998), rCia3(Griffiths et al., 2000), rEau1(Sun et al.,

1999), rScwia1(Wilder et al., 1999), mIdd20(Rogner et al., 2001),

mCia3(McIndoe et al., 1999) Pia6 14

rEae10(Bergsteinsdottir et al., 2000), mLmb2(Vidal et al., 1998), mBb3(Weis etal., 1999)

Pia7 4rCia13(Remmers et al., 1996), rCiaa4(Furuya et al., 2000), rOia2(Lorentzen et

al., 1998), mLbw4(Kono et al., 1994), mIdd19(Rogner et al., 2001),Pia8 1Pia9 1

rCia2(Griffiths et al., 2000), mCia7(Yang et al., 1999), mEae4(Butterfield et al.,1998)

Pia11 16

http://www.informatics.jax.org/

Discussion 28

Table 5. continued…

All QTLs have a prefix indicating in what species it was identified. r=rat, m=mouse.a Refers to chromosome in the rat.* All models for autoimmune disorders listed here show association to the MHC locus.

List of abbreviationsAia = Avridine induced arthritis Bb = Borrelia Burgdorferi induced arthritis Cia = Collagen induced arthritisCiaa = Collagen induced arthritis antibody Eae = Experimental allergic encephalomyelitis Eau = Experimental autoimmune uveitisIdd = Insulin dependent diabetesIddm = Insulin dependent diabetes mellitusLmb = Lupus in MRL x C57BL/6 miceLwb = Lupus white and blackOia = Oil induced arthritisScwia = Streptococcus cell wall induced arthritisSle = Systemic lupus erythematosusTnfp = Tumor necrosis factor protection

The use of different animal models for induced arthritis has resulted inthe identification of numerous loci associated with the disease. Some loci seemto be common between the models, while some loci are specific for a certainmodel. This is consistent with the observation that each model has its own setof phenotypic characteristics, while they at the same time displaying similaritiesto others.

There are numerous publications that claim that susceptibility loci forautoimmune diseases are clustered (Becker et al., 1998; Bergsteinsdottir et al.,2000; Cornelis et al., 1998; Furuya et al., 2000; Joe et al., 1999; Kawahito et al.,1998; Listwak et al., 1999; Martin et al., 1999; Merriman et al., 2001). Acompelling hypothesis would be that certain genes are of importance for theregulation of autoimmunity, rather than in regulating a specific disease. Itwould be of utmost interest to clone and characterize these genes, in order tounderstand the development of autoimmune disorders and to identify targetsfor novel therapeutic methods. On the other hand, one could say that these locicoincide due to a correlation with something that occurs after the actualoutbreak of autoimmunity, i.e. a secondary effect. However, an important factremains, and that is that many of these loci are identified in studies of bothrodents and humans, which suggests that there are common genetic factorsthat are involved, directly or indirectly, with an autoimmune response.

29 Discussion

Figure 5. Illustration of rat chromosomes where we have identified lociassociated with pristane-induced arthritis. The width of the shaded boxes showthe 95% CI for the PIA loci. The solid lines indicate loci that are associatedwith models for autoimmune disorders in rats. The dashed lines indicate mouseloci.

Discussion 30

Gender differencesThe susceptibility to autoimmune disorders is clearly different between femalesand males. From an epidemiological study of 24 autoimmune diseases in theUnited States, it was found that the risk of becoming affected was 2.7 timesgreater for women compared to men (Jacobson et al., 1997). Among SLEpatients, women are affected 9 times as often as men (Beeson, 1994). RA showsan overrepresentation of affected women, with a female to male ratio of 3:1(Beeson, 1994). The same phenomenon has also been observed in animalmodels of autoimmune disorders (Furuya et al., 2000; Gulko et al., 1998;Holmdahl et al., 1986; Joe et al., 2000).

Different explanations for this gender bias have been proposed,including loci residing on the sex chromosomes (Ebers et al., 1996; Jawaheer etal., 2001; Santiago et al., 1998), sex hormonal effects (Holmdahl et al., 1986;Jansson et al., 1994), imprinting (Pugliese et al., 1994), and non-random Xinactivation (Chitnis et al., 2000).In both the E3xDA, and E3xLew.1f crosses, it was observed that, in the F1population, female rats had a significantly earlier onset and a more severearthritis. In fact, only the female F1 progeny developed disease in the crossE3xLew.1f (4/5 female, and 0/4 male rats). Analysis of the cross E3xDA withgenetic markers on the X chromosome revealed no association with disease. Onthe other hand, we have identified three loci (Pia2, Pia5, and Pia8) that have aneffect in only one of the sexes. These findings suggest a genetic basis, exerted byautosomal genes for the gender bias.

The observation that specific loci exert an effect in only one of the sexeswill have implications for how the presence of predisposing genetic factorsshould be evaluated.

InteractionsIn these studies, we have found supporting evidence for the presence ofepistasis. In the cross E3xDA, we detected a suggestive epistatic effect. The use ofthe term ‘epistasis’ is not always the same. Here, I use the term ‘epistasis’ todescribe the interactive effect occurring between two loci, where the first locusis a QTL and the second locus determines whether or not the first locus willcause a phenotypic effect.

In our case, we observed that certain alleles at a locus were aprerequisite for another locus in order to have any phenotypic effect. It is acompelling thought to imagine this required locus as an ON/OFF switch forthe identified QTL. Previously, similar observations have been reported (Kuidaand Beier, 2000; Lark et al., 1995; Shimomura et al., 2001). These findings needto be confirmed, preferably in double congenic strains, in which to rule out thepossibility of random effects as the cause of detection.

31 Discussion

In summary, epistatic effects could play an important role incontributing to the genetic complexity of arthritis development andprogression. To fully comprehend the genetic basis for a complex trait like this,it will require an understanding of how multiple genes interact with oneanother to cause disease. The presence of gene interactions could also haveconsequences for how to evaluate the presence of predisposing genetic factorsin patients.

Temporal shift of susceptibility lociDifferent QTLs are identified at different temporal phases of the disease. In the PIA experiments presented here, disease development in the progenyfrom each cross have been followed and scored 1-2 times a week. The diseasescoring at each time point was subsequently used in QTL analysis. It turned outthat the effects of the severity associated loci were exerted in non-overlappingtemporal intervals (see figure 6).

Figure 6. The illustration shows the timing of different pristane-inducedarthritis associated loci. These loci were identified using clinical scoring atdifferent time points as quantitative traits.

This could be explained by the occurrence of critical events in the diseaseprogression that are driven by different sets of genes. One could argue thatthese effects are caused by the fact that individual rats develop disease atdifferent time points and that we have not used disease onset as a baseline forthe disease progression. I believe that this will not cause a problem in theanalysis, since the majority of the rats develop disease in a similar way, withrespect to time. The temporal shift has also been observed in a rat model forMS (Bergsteinsdottir et al., 2000).

The idea that these timing events could have a genetic basis couldcertainly be important in future experiments. E.g. in investigating differences ingene expression, it should be very interesting to specifically pay attention to

Future studies 32

changes in expression patterns in the temporal intervals, which could mark thetransition from an initial acute disease to severe erosive inflammation andeventually, the progression into a chronic phase.

Future studiesI can imagine two future approaches, both of course aimed at identifyingfactors that we can control and thereby, treat or prevent disease.

The first approach would directly focus on identifying and characterizinggenes in the genomic regions that show association to different autoimmunediseases. A strong argument in favor of this is that several autoimmunedisorders seem to be regulated by a common set of loci, which could containgenes that are of importance in the development of autoimmune disorders as awhole, rather than just for specific diseases. Today, the knowledge and thetechnology is available to commence with such a task. Many of the regions,which have been postulated to contain genes of importance in the developmentof autoimmune disorders, will need to be refined. This can be achieved by,congenic breeding or the use of advanced intercross lines. The available whole-genome sequence of man and mouse should make it possible to identify most,if not all, genes in the regions of interest. Also, gene expression profiling studiescould possibly identify candidate genes in these regions. Ultimately, the effectof the identified genes must be confirmed. A possible and powerful approachto this problem could be to utilize the transgene technique, in which eitherindividual genes or larger genomic fragments contained in artificialchromosomes, such as in BAC transgenic animals, could be tested.

We know that many genetic disorders are under the control of multiplegenes and environmental factors. We also know that the effects of these genescould be dependent on gender and on interactions with one another and theenvironment. As a consequence of this complex behavior, genes identified asbeing important in the pathogenesis will probably have different effectsdepending on the genetic and environmental conditions. Maybe it will turn outto be impossible to identify causative genetic factors without a betterunderstanding of the ‘genetic dynamics’.

An alternative future approach would therefore focus on investigatinghow the known disease-associated regions act together in well-definedexperiments utilizing the whole span of available techniques in experimentalanimals. I imagine this undertaking as a jigsaw puzzle, where some pieces maybe obtained from congenic strains containing different combinations of lociand other pieces obtained using transgene technique, with combinations ofpredisposing chromosomal regions. Ultimately, these experiments could answerquestions about how and which genes act together to provide conditions underwhich disease can occur.

33 Concluding remarks

Concluding remarksWhy do we carry genes that are involved in the development of severe diseases?Are genetic disorders caused by ‘disease genes’ or by unfortunate combinationsof naturally occurring polymorphisms?

Natural selection is the means by which organisms adapt to a changingenvironment by favoring individuals that carry gene variants that give them areproductive advantage. Why then, have we retained variants of genes thatconfer predisposition to severe diseases?

There are instances where the benefit of inheriting a potential diseasecausing gene variant seems to be greater than the risk of developing disease. Aclassical example, often brought forward in this discussion, is the case of Sicklecell anemia. This is an autosomal recessive disease, caused by a point mutationin the hemoglobin beta gene (HBB). The allele frequency of HBB is significantlyhigher in zones with a high incidence of malaria, since this gene has aprotective effect against this disease. This serves as an example of a gene thatgives the carrier a reproductive advantage, while it can simultaneously cause asevere blood disorder in its homozygous form.

Other disorders could have emerged quite recently as a result of newlifestyles that have been adopted along with global industrialization. ‘Normal’gene variants could thus suddenly start acting as disease associated genes whennew environmental factors are encountered.

It has also been proposed, that genes associated with complex disorders,which are neither sufficient nor necessary to cause disease, are subject to verylimited selective pressure (Pritchard, 2001).

Seemingly, there could be several alternative explanations to why wecarry genes that increase the risk of developing diseases. The examples above,illustrates the complexity of many disorders where genes can be both beneficialand deleterious, interact with the environment and with other genes.

In this study, I have investigated genetic factors that confer susceptibilityto rheumatoid arthritis in a rat model. This work has led to the identification ofseveral chromosomal regions, containing uncharacterized genes that directly orindirectly are associated with the arthritis development in these rats. We haveobserved that timing, gender and genetic interactions are features thatinfluence the effect that these genetic factors exert. My belief is that, intensiveinvestigations of how the underlying genes interact with each other and theenvironment will be crucial for understanding the etiology and progression ofthe disease.

Acknowledgements 34

Acknowledgements

Finally I would like to express my sincere gratitude to all the people who havehelped me in one way or another during my graduate studies.

Min handledare Ulf Pettersson för all hjälp och som har givit mig den härmöjligheten.

Min arbetskamrat Ludmila Elfineh som har slitit hårt med den laborativa delenav dessa projekt. Tack för all hjälp!

Hamid Darban, som också slitit varg med gelgjutning och genotypning. Tack förall hjälp! Och grattis till körkortet!

Mats Sundvall, som med sitt kunnande lade grunden för denna grupp och somgav mig chansen att prova på arbetet.

The former group-member Hai-Tao Yang, who introduced me to the practicalpart of this work, and has helped me a lot.

Jag vill särskilt tacka Elena Jazin med gruppen; Jorune Balciuniéne, EvaLindholm, Anja Castensson och Lina Emilsson som har agerat stödgrupp åtmig, Ludmilla och Hai-Tao.

Elisabeth Sandberg & Jeanette Backman. Tack för all hjälp med detadministrativa, inte minst för hjälp med hanteringen av fakturor, följesedlar,kopior av dessa, ansökningar, intyg... , som jag själv har vissa problem med.

Eddie Lundh & Stig Söderberg. Tack för all hjälp med det praktiska.

Rikard Holmdahl. Som jag anser ha varit min inofficiella handledare. Tack föratt du delar med dig av ditt brinnande intresse. Så brinnande att du till ochmed ringer hem till mina svärföräldrar på juldagen för att diskutera projekt.Det uppskattas!

Johan Jirholt. Tack för alla diskussioner och ditt sätt att lyckas marginalisera allaproblem. När skall ni inse att det är i Uppsala ni hör hemma?

I would also like to thank Carina Vingsbo-Lundberg, Peter Olofsson & SheminLu and all other members working at the section for inflammation research atLund University, for the collaboration. It’s too bad you don’t have your lab inUppsala.

35 Acknowledgements

Jag vill tacka Marta Alarcón med gruppen, Lucia Cavelier, Bo Johanneson,Cecilia Johansson, Ludmila Prokunina och Veronica Magnusson förgemensamma möten och diskussioner. Lucia, jag värderar högt de åsikter ochkommentarer som du bidragit med under produktionen av denna avhandling.Stort tack för detta!

Jag vill också rikta ett stort tack till de som hjälpt mig med datorer ochtillhörande problem: Per-Ivan Wyöni som visat mig hur man kan användascriptspråket Perl (’Pathologically Eclectic Rubbish Lister’). William Shannong &Viktor Persson. Toppenkul att jobba med er. Martti Tammi som introduceratlinux för mig och hjäpt mig med otaliga datorproblem.

Kim Andersson. Thank you very much for the linguistic help. There is a clearpattern in the corrections you have made. I have counted the number ofcommas in this thesis and out of 778 occurrences I had managed to correctlyplace less that 10% of them. So, in the future I guess I’ll just stop using them.

Slutligen vill jag passa på att tacka min familj och mina vänner för allt sånt sominte anknyter till studierna men som är så nödvändigt för att sträva vidare ochhålla humöret uppe.

Magnus Sundström, min studiekamrat och gode vän som jag spenderade enoförglömlig termin med i Edinburgh. Magnus Mohlin, Patrik Ohlsson, PatrikMagnusson och Micke Karlsson, studiekamrater och vänner som förgylltstudietiden i Uppsala.

Min barndomskamrat och gode vän Göran Eriksson.

Mina svärföräldrar Lena & Stefan Grönvall. ’08:an’ vill uttrycka sin innerligatacksamhet för er vänskap och er omsorg om barnbarnen.

Min kära Fru Jenny, våra barn Alma och Klas, Mamma Katarina och Pappa Lars,Mona och mina syskon Lisa & Gustav. Ni betyder allt för mig.

References 36

References

Andrew, S. E., Goldberg, Y. P., Kremer, B., Telenius, H., Theilmann, J., Adam, S.,Starr, E., Squitieri, F., Lin, B., Kalchman, M. A., and et al. (1993). Therelationship between trinucleotide (CAG) repeat length and clinical features ofHuntington's disease, Nat Genet 4, 398-403.

Araki, K., Araki, M., Miyazaki, J., and Vassalli, P. (1995). Site-specificrecombination of a transgene in fertilized eggs by transient expression of Crerecombinase, Proc Natl Acad Sci U S A 92, 160-4.

Arnett, F. C., Edworthy, S. M., Bloch, D. A., McShane, D. J., Fries, J. F., Cooper,N. S., Healey, L. A., Kaplan, S. R., Liang, M. H., Luthra, H. S., and et al. (1988).The American Rheumatism Association 1987 revised criteria for theclassification of rheumatoid arthritis, Arthritis Rheum 31, 315-24.

Basten, C. J., Weir, B. S., and Zeng, Z.-B. (1994). Zmap-a QTL cartographer. InProceedings of the 5th World Congress on Genetics Applied to LivestockProduction: Computing Strategies and Software, C. Smith, J. S. Gavora, B.Benkel, J. Chesnais, W. Fairfull, J. P. Gibson, B. W. Kennedy, and E. B. Burnside,eds. (Guelph, Ontario, Canada., the Organizing Committee, 5th WorldCongress on Genetics Applied to Livestock Production.), pp. 65-66.

Bateson, W. (1909). Mendel's Principles of Heredity (Cambridge/London/NewYork, Cambridge University Press).

Beck, J. A., Lloyd, S., Hafezparast, M., Lennon-Pierce, M., Eppig, J. T., Festing, M.F., and Fisher, E. M. (2000). Genealogies of mouse inbred strains, Nat Genet 24,23-5.

Becker, K. G., Simon, R. M., Bailey-Wilson, J. E., Freidlin, B., Biddison, W. E.,McFarland, H. F., and Trent, J. M. (1998). Clustering of non-majorhistocompatibility complex susceptibility candidate loci in human autoimmunediseases, Proc Natl Acad Sci U S A 95, 9979-84.

Beeson, P. B. (1994). Age and sex associations of 40 autoimmune diseases, Am JMed 96, 457-62.

Bergsteinsdottir, K., Yang, H. T., Pettersson, U., and Holmdahl, R. (2000).Evidence for common autoimmune disease genes controlling onset, severity,and chronicity based on experimental models for multiple sclerosis andrheumatoid arthritis, J Immunol 164, 1564-8.

37 References

Berrettini, W. H., Ferraro, T. N., Alexander, R. C., Buchberg, A. M., and Vogel, W.H. (1994). Quantitative trait loci mapping of three loci controlling morphinepreference using inbred mouse strains, Nat Genet 7, 54-8.

Bihoreau, M. T., Gauguier, D., Kato, N., Hyne, G., Lindpaintner, K., Rapp, J. P.,James, M. R., and Lathrop, G. M. (1997). A linkage map of the rat genomederived from three F2 crosses, Genome Res 7, 434-40.

Brennan, P., Bankhead, C., Silman, A., and Symmons, D. (1997). Oralcontraceptives and rheumatoid arthritis: results from a primary care-basedincident case-control study, Semin Arthritis Rheum 26, 817-23.

Brown, D. M., Matise, T. C., Koike, G., Simon, J. S., Winer, E. S., Zangen, S.,McLaughlin, M. G., Shiozawa, M., Atkinson, O. S., Hudson, J. R., Jr., et al.(1998). An integrated genetic linkage map of the laboratory rat, MammGenome 9, 521-30.

Butterfield, R. J., Sudweeks, J. D., Blankenhorn, E. P., Korngold, R., Marini, J. C.,Todd, J. A., Roper, R. J., and Teuscher, C. (1998). New genetic loci that controlsusceptibility and symptoms of experimental allergic encephalomyelitis ininbred mice, J Immunol 161, 1860-7.

Charles, P., Elliott, M. J., Davis, D., Potter, A., Kalden, J. R., Antoni, C., Breedveld,F. C., Smolen, J. S., Eberl, G., deWoody, K., et al. (1999). Regulation of cytokines,cytokine inhibitors, and acute-phase proteins following anti-TNF-alpha therapyin rheumatoid arthritis, J Immunol 163, 1521-8.

Chartier, F. L., Keer, J. T., Sutcliffe, M. J., Henriques, D. A., Mileham, P., andBrown, S. D. (1992). Construction of a mouse yeast artificial chromosomelibrary in a recombination-deficient strain of yeast, Nat Genet 1, 132-6.

Chitnis, S., Monteiro, J., Glass, D., Apatoff, B., Salmon, J., Concannon, P., andGregersen, P. K. (2000). The role of X-chromosome inactivation in femalepredisposition to autoimmunity, Arthritis Res 2, 399-406.

Chung, W. K., Zheng, M., Chua, M., Kershaw, E., Power-Kehoe, L., Tsuji, M., Wu-Peng, X. S., Williams, J., Chua, S. C., Jr., and Leibel, R. L. (1997). Geneticmodifiers of Leprfa associated with variability in insulin production andsusceptibility to NIDDM, Genomics 41, 332-44.

Cornall, R. J., Prins, J. B., Todd, J. A., Pressey, A., DeLarato, N. H., Wicker, L. S.,and Peterson, L. B. (1991). Type 1 diabetes in mice is linked to the interleukin-1receptor and Lsh/Ity/Bcg genes on chromosome 1, Nature 353, 262-5.

References 38

Cornelis, F., Faure, S., Martinez, M., Prud'homme, J. F., Fritz, P., Dib, C., Alves,H., Barrera, P., de Vries, N., Balsa, A., et al. (1998). New susceptibility locus forrheumatoid arthritis suggested by a genome- wide linkage study, Proc NatlAcad Sci U S A 95, 10746-50.

Darvasi, A., and Soller, M. (1995). Advanced intercross lines, an experimentalpopulation for fine genetic mapping, Genetics 141, 1199-1207.

de Gouyon, B., Melanitou, E., Richard, M. F., Requarth, M., Hahn, I. H., Guenet,J. L., Demenais, F., Julier, C., Lathrop, G. M., Boitard, C., and et al. (1993).Genetic analysis of diabetes and insulitis in an interspecific cross of thenonobese diabetic mouse with Mus spretus, Proc Natl Acad Sci U S A 90, 1877-81.

Denny, P., Lord, C. J., Hill, N. J., Goy, J. V., Levy, E. R., Podolin, P. L., Peterson, L.B., Wicker, L. S., Todd, J. A., and Lyons, P. A. (1997). Mapping of the IDDMlocus Idd3 to a 0.35-cM interval containing the interleukin-2 gene, Diabetes 46,695-700.

Dietrich, W. F. (2001). Using mouse genetics to understand infectious diseasepathogenesis, Genome Res 11, 325-31.

Dietrich, W. F., Miller, J., Steen, R., Merchant, M. A., Damron-Boles, D., Husain,Z., Dredge, R., Daly, M. J., Ingalls, K. A., O'Connor, T. J., and et al. (1996). Acomprehensive genetic map of the mouse genome, Nature 380, 149-52.

Dietrich, W. F., Miller, J. C., Steen, R. G., Merchant, M., Damron, D., Nahf, R.,Gross, A., Joyce, D. C., Wessel, M., Dredge, R. D., and et al. (1994). A geneticmap of the mouse with 4,006 simple sequence length polymorphisms, NatGenet 7, 220-45.

Ebers, G. C., Kukay, K., Bulman, D. E., Sadovnick, A. D., Rice, G., Anderson, C.,Armstrong, H., Cousin, K., Bell, R. B., Hader, W., et al. (1996). A full genomesearch in multiple sclerosis, Nat Genet 13, 472-6.

Furuya, T., Salstrom, J. L., McCall-Vining, S., Cannon, G. W., Joe, B., Remmers, E.F., Griffiths, M. M., and Wilder, R. L. (2000). Genetic dissection of a rat modelfor rheumatoid arthritis: significant gender influences on autosomal modifierloci, Hum Mol Genet 9, 2241-50.Garrett, L. R., Chung, J. G., Byers, P. E., and Cuchens, M. A. (1989). Dietaryeffects of pristane on rat lymphoid tissues, Agents Actions 28, 272-8.

Gauguier, D., Froguel, P., Parent, V., Bernard, C., Bihoreau, M. T., Portha, B.,James, M. R., Penicaud, L., Lathrop, M., and Ktorza, A. (1996). Chromosomal

39 References

mapping of genetic loci associated with non-insulin dependent diabetes in theGK rat, Nat Genet 12, 38-43.

Ghosh, S., Palmer, S. M., Rodrigues, N. R., Cordell, H. J., Hearne, C. M., Cornall,R. J., Prins, J. B., McShane, P., Lathrop, G. M., Peterson, L. B., and et al. (1993).Polygenic control of autoimmune diabetes in nonobese diabetic mice, NatGenet 4, 404-9.

Gregory, R. J., Cheng, S. H., Rich, D. P., Marshall, J., Paul, S., Hehir, K.,Ostedgaard, L., Klinger, K. W., Welsh, M. J., and Smith, A. E. (1990). Expressionand characterization of the cystic fibrosis transmembrane conductanceregulator, Nature 347, 382-6.

Griffiths, M. M., Wang, J., Joe, B., Dracheva, S., Kawahito, Y., Shepard, J. S.,Reese, V. R., McCall-Vining, S., Hashiramoto, A., Cannon, G. W., et al. (2000).Identification of four new quantitative trait loci regulating arthritis severity andone new quantitative trait locus regulating autoantibody production in ratswith collagen-induced arthritis, Arthritis Rheum 43, 1278-89.

Gulko, P. S., Kawahito, Y., Remmers, E. F., Reese, V. R., Wang, J., Dracheva, S. V.,Ge, L., Longman, R. E., Shepard, J. S., Cannon, G. W., et al. (1998). Identificationof a new non-major histocompatibility complex genetic locus on chromosome2 that controls disease severity in collagen- induced arthritis in rats, ArthritisRheum 41, 2122-31.

Haldi, M. L., Strickland, C., Lim, P., VanBerkel, V., Chen, X., Noya, D., Korenberg,J. R., Husain, Z., Miller, J., and Lander, E. S. (1996). A comprehensive large-insert yeast artificial chromosome library for physical mapping of the mousegenome, Mamm Genome 7, 767-9.

Hamilton, D. L., and Abremski, K. (1984). Site-specific recombination by thebacteriophage P1 lox-Cre system. Cre- mediated synapsis of two lox sites, J MolBiol 178, 481-6.

Heliovaara, M., Aho, K., Knekt, P., Impivaara, O., Reunanen, A., and Aromaa, A.(2000). Coffee consumption, rheumatoid factor, and the risk of rheumatoidarthritis, Ann Rheum Dis 59, 631-5.Heliovaara, M., Aho, K., Knekt, P., Reunanen, A., and Aromaa, A. (1996). Serumcholesterol and risk of rheumatoid arthritis in a cohort of 52 800 men andwomen, Br J Rheumatol 35, 255-7.

Holmdahl, R., Jansson, L., and Andersson, M. (1986). Female sex hormonessuppress development of collagen-induced arthritis in mice, Arthritis Rheum29, 1501-9.

References 40

Holmdahl, R., and Kvick, C. (1992). Vaccination and genetic experimentsdemonstrate that adjuvant-oil- induced arthritis and homologous type IIcollagen-induced arthritis in the same rat strain are different diseases, Clin ExpImmunol 88, 96-100.

Ikegami, H., Makino, S., Yamato, E., Kawaguchi, Y., Ueda, H., Sakamoto, T.,Takekawa, K., and Ogihara, T. (1995). Identification of a new susceptibility locusfor insulin-dependent diabetes mellitus by ancestral haplotype congenicmapping, J Clin Invest 96, 1936-42.

Jacob, H. J., Brown, D. M., Bunker, R. K., Daly, M. J., Dzau, V. J., Goodman, A.,Koike, G., Kren, V., Kurtz, T., Lernmark, A., and et al. (1995). A genetic linkagemap of the laboratory rat, Rattus norvegicus, Nat Genet 9, 63-9.

Jacob, H. J., Pettersson, A., Wilson, D., Mao, Y., Lernmark, A., and Lander, E. S.(1992). Genetic dissection of autoimmune type I diabetes in the BB rat, NatGenet 2, 56-60.

Jacobson, D. L., Gange, S. J., Rose, N. R., and Graham, N. M. (1997).Epidemiology and estimated population burden of selected autoimmunediseases in the United States, Clin Immunol Immunopathol 84, 223-43.

Jansson, L., Olsson, T., and Holmdahl, R. (1994). Estrogen induces a potentsuppression of experimental autoimmune encephalomyelitis and collagen-induced arthritis in mice, J Neuroimmunol 53, 203-7.

Jawaheer, D., Seldin, M. F., Amos, C. I., Chen, W. V., Shigeta, R., Monteiro, J.,Kern, M., Criswell, L. A., Albani, S., Nelson, J. L., et al. (2001). A genomewidescreen in multiplex rheumatoid arthritis families suggests genetic overlap withother autoimmune diseases, Am J Hum Genet 68, 927-36.

Jirholt, J., Cook, A., Emahazion, T., Sundvall, M., Jansson, L., Nordquist, N.,Pettersson, U., and Holmdahl, R. (1998). Genetic linkage analysis of collagen-induced arthritis in the mouse, Eur J Immunol 28, 3321-8.Joe, B., Griffiths, M. M., Remmers, E. F., and Wilder, R. L. (1999). Animalmodels of rheumatoid arthritis and related inflammation, Curr Rheumatol Rep1, 139-48.

Joe, B., Remmers, E. F., Dobbins, D. E., Salstrom, J. L., Furuya, T., Dracheva, S.,Gulko, P. S., Cannon, G. W., Griffiths, M. M., and Wilder, R. L. (2000). Geneticdissection of collagen-induced arthritis in Chromosome 10 quantitative traitlocus speed congenic rats: evidence for more than one regulatory locus and sexinfluences, Immunogenetics 51, 930-44.

41 References

Kanemoto, N., Hishigaki, H., Miyakita, A., Oga, K., Okuno, S., Tsuji, A., Takagi, T.,Takahashi, E., Nakamura, Y., and Watanabe, T. K. (1998). Genetic dissection of"OLETF", a rat model for non-insulin-dependent diabetes mellitus, MammGenome 9, 419-25.

Kao, C. H., Zeng, Z. B., and Teasdale, R. D. (1999). Multiple interval mapping forquantitative trait loci, Genetics 152, 1203-16.

Kawahito, Y., Cannon, G. W., Gulko, P. S., Remmers, E. F., Longman, R. E., Reese,V. R., Wang, J., Griffiths, M. M., and Wilder, R. L. (1998). Localization ofquantitative trait loci regulating adjuvant-induced arthritis in rats: evidence forgenetic factors common to multiple autoimmune diseases, J Immunol 161,4411-9.

Kiprov, D. (1980). Experimental models of hypertension, Cor Vasa 22, 116-28.

Kono, D. H., Burlingame, R. W., Owens, D. G., Kuramochi, A., Balderas, R. S.,Balomenos, D., and Theofilopoulos, A. N. (1994). Lupus susceptibility loci inNew Zealand mice, Proc Natl Acad Sci U S A 91, 10168-72.

Kruglyak, L., and Lander, E. S. (1995). A nonparametric approach for mappingquantitative trait loci, Genetics 139, 1421-8.

Kuida, S., and Beier, D. R. (2000). Genetic localization of interacting modifiersaffecting severity in a murine model of polycystic kidney disease, Genome Res10, 49-54.

Lander, E. S., and Botstein, D. (1989). Mapping mendelian factors underlyingquantitative traits using RFLP linkage maps, Genetics 121, 185-99.

Lander, E., and Kruglyak, L. (1995). Genetic dissection of complex traits:guidelines for interpreting and reporting linkage results, Nat Genet 11, 241-7.

Lander, E. S., and Schork, N. J. (1994). Genetic dissection of complex traits,Science 265, 2037-48.

Lark, K. G., Chase, K., Adler, F., Mansur, L. M., and Orf, J. H. (1995). Interactionsbetween quantitative trait loci in soybean in which trait variation at one locusis conditional upon a specific allele at another, Proc Natl Acad Sci U S A 92,4656-60.

References 42

Lewandoski, M., Sun, X., and Martin, G. R. (2000). Fgf8 signalling from the AERis essential for normal limb development, Nat Genet 26, 460-3.

Levine, S., and Sowinski, R. (1973). Experimental allergic encephalomyelitis ininbred and outbred mice, J Immunol 110, 139-43.

Libert, C., Wielockx, B., Hammond, G. L., Brouckaert, P., Fiers, W., and Elliott, R.W. (1999). Identification of a locus on distal mouse chromosome 12 thatcontrols resistance to tumor necrosis factor-induced lethal shock, Genomics 55,284-9.

Lincoln, S., Daly, M., and Lander, E. S. (1992). Mapping genes controllingquantitative traits with MAPMAKER/QTL 1.1, Whitehead Institute TechnicalReport.

Listwak, S., Barrientos, R. M., Koike, G., Ghosh, S., Gomez, M., Misiewicz, B.,and Sternberg, E. M. (1999). Identification of a novel inflammation-protectivelocus in the Fischer rat, Mamm Genome 10, 362-5.

Lorentzen, J. C., Glaser, A., Jacobsson, L., Galli, J., Fakhrai-rad, H., Klareskog, L.,and Luthman, H. (1998). Identification of rat susceptibility loci for adjuvant-oil-induced arthritis, Proc Natl Acad Sci U S A 95, 6383-7.

MacGregor, A. J., Snieder, H., Rigby, A. S., Koskenvuo, M., Kaprio, J., Aho, K.,and Silman, A. J. (2000). Characterizing the quantitative genetic contribution torheumatoid arthritis using data from twins, Arthritis Rheum 43, 30-7.

Magnusson, P. K., Lichtenstein, P., and Gyllensten, U. B. (2000). Heritability ofcervical tumours, Int J Cancer 88, 698-701.

Martin, A. M., Maxson, M. N., Leif, J., Mordes, J. P., Greiner, D. L., andBlankenhorn, E. P. (1999). Diabetes-prone and diabetes-resistant BB rats share acommon major diabetes susceptibility locus, iddm4: additional evidence for a"universal autoimmunity locus" on rat chromosome 4, Diabetes 48, 2138-44.

McAleer, M. A., Reifsnyder, P., Palmer, S. M., Prochazka, M., Love, J. M.,Copeman, J. B., Powell, E. E., Rodrigues, N. R., Prins, J. B., Serreze, D. V., and etal. (1995). Crosses of NOD mice with the related NON strain. A polygenicmodel for IDDM, Diabetes 44, 1186-95.

43 References

McDevitt, H. O., and Wakeland, E. K. (1998). Autoimmunity, Curr OpinImmunol 10, 647-8.

McIndoe, R. A., Bohlman, B., Chi, E., Schuster, E., Lindhardt, M., and Hood, L.(1999). Localization of non-Mhc collagen-induced arthritis susceptibility loci inDBA/1j mice, Proc Natl Acad Sci U S A 96, 2210-4.

Merriman, T. R., Cordell, H. J., Eaves, I. A., Danoy, P. A., Coraddu, F., Barber, R.,Cucca, F., Broadley, S., Sawcer, S., Compston, A., et al. (2001). Suggestiveevidence for association of human chromosome 18q12-q21 and its orthologueon rat and mouse chromosome 18 with several autoimmune diseases, Diabetes50, 184-94.

Moon, A. M., and Capecchi, M. R. (2000). Fgf8 is required for outgrowth andpatterning of the limbs, Nat Genet 26, 455-9.

Morahan, G., McClive, P., Huang, D., Little, P., and Baxter, A. (1994). Geneticand physiological association of diabetes susceptibility with raised Na+/H+exchange activity, Proc Natl Acad Sci U S A 91, 5898-902.

Morel, L., Blenman, K. R., Croker, B. P., and Wakeland, E. K. (2001). The majormurine systemic lupus erythematosus susceptibility locus, Sle1, is a cluster offunctionally related genes, Proc Natl Acad Sci U S A 98, 1787-92.

Morel, L., Tian, X. H., Croker, B. P., and Wakeland, E. K. (1999). Epistaticmodifiers of autoimmunity in a murine model of lupus nephritis, Immunity11, 131-9.

Nicholls, R. D., Saitoh, S., and Horsthemke, B. (1998). Imprinting in Prader-Willi and Angelman syndromes, Trends Genet 14, 194-200.

Nilsson, S., Helou, K., Walentinsson, A., Szpirer, C., Nerman, O., and Stahl, F.(2001). Rat-mouse and rat-human comparative maps based on gene homologyand high-resolution zoo-fish, Genomics 74, 287-98.

O'Hara, R., Murphy, E. P., Whitehead, A. S., FitzGerald, O., and Bresnihan, B.(2000). Acute-phase serum amyloid A production by rheumatoid arthritissynovial tissue, Arthritis Res 2, 142-4.

Osoegawa, K., Tateno, M., Woon, P. Y., Frengen, E., Mammoser, A. G., Catanese,J. J., Hayashizaki, Y., and de Jong, P. J. (2000). Bacterial artificial chromosomelibraries for mouse sequencing and functional analysis, Genome Res 10, 116-28.

References 44

Palmiter, R. D., Brinster, R. L., Hammer, R. E., Trumbauer, M. E., Rosenfeld, M.G., Birnberg, N. C., and Evans, R. M. (1982). Dramatic growth of mice thatdevelop from eggs microinjected with metallothionein-growth hormone fusiongenes, Nature 300, 611-5.

Pearson, C. M. (1956). Development of arthritis, periarthritis and periostitis inrats given adjuvants., Proc Soc Exp Biol Med 91, 95-101.

Peri, V., and Rinaldi, F. (1998). Acute phase and function in rheumatoidarthritis, J Rheumatol 25, 602-3.

Podolin, P. L., Denny, P., Lord, C. J., Hill, N. J., Todd, J. A., Peterson, L. B.,Wicker, L. S., and Lyons, P. A. (1997). Congenic mapping of the insulin-dependent diabetes (Idd) gene, Idd10, localizes two genes mediating the Idd10effect and eliminates the candidate Fcgr1, J Immunol 159, 1835-43.

Pritchard, J. K. (2001). Are rare variants responsible for susceptibility tocomplex diseases?, Am J Hum Genet 69, 124-37.

Prochazka, M., Leiter, E. H., Serreze, D. V., and Coleman, D. L. (1987). Threerecessive loci required for insulin-dependent diabetes in nonobese diabeticmice, Science 237, 286-9.

Pugliese, A., Awdeh, Z. L., Alper, C. A., Jackson, R. A., and Eisenbarth, G. S.(1994). The paternally inherited insulin gene B allele (1,428 FokI site) confersprotection from insulin-dependent diabetes in families, J Autoimmun 7, 687-94.

Rahman, S., Blok, R. B., Dahl, H. H., Danks, D. M., Kirby, D. M., Chow, C. W.,Christodoulou, J., and Thorburn, D. R. (1996). Leigh syndrome: clinical featuresand biochemical and DNA abnormalities, Ann Neurol 39, 343-51.

Reeves, R. H., Irving, N. G., Moran, T. H., Wohn, A., Kitt, C., Sisodia, S. S.,Schmidt, C., Bronson, R. T., and Davisson, M. T. (1995). A mouse model forDown syndrome exhibits learning and behaviour deficits, Nat Genet 11, 177-84.

Remmers, E. F., Griffiths, M. M., Longman, R. E., Gulko, P. S., Kawahito, Y.,Chen, S., Chang, L., Shepard, J., Ge, L., Dracheva, S., et al. (1999). An integratedrat genetic map: analysis of linkage conservation with the mouse and humanmaps, Transplant Proc 31, 1549-54.

Remmers, E. F., Longman, R. E., Du, Y., O'Hare, A., Cannon, G. W., Griffiths, M.M., and Wilder, R. L. (1996). A genome scan localizes five non-MHC locicontrolling collagen-induced arthritis in rats, Nat Genet 14, 82-5.

45 References

Rhodes, M., Straw, R., Fernando, S., Evans, A., Lacey, T., Dearlove, A.,Greystrong, J., Walker, J., Watson, P., Weston, P., et al. (1998). A high-resolution microsatellite map of the mouse genome, Genome Res 8, 531-42.

Ridgway, W. M., Fasso, M., and Fathman, C. G. (1999). A new look at MHC andautoimmune disease, Science 284, 749,751.

Riordan, J. R., Rommens, J. M., Kerem, B., Alon, N., Rozmahel, R., Grzelczak, Z.,Zielenski, J., Lok, S., Plavsic, N., Chou, J. L., and et al. (1989). Identification ofthe cystic fibrosis gene: cloning and characterization of complementary DNA,Science 245, 1066-73.

Rogner, U. C., Boitard, C., Morin, J., Melanitou, E., and Avner, P. (2001). Threeloci on mouse chromosome 6 influence onset and final incidence of type idiabetes in nod.c3h congenic strains, Genomics 74, 163-71.

Santiago, M. L., Mary, C., Parzy, D., Jacquet, C., Montagutelli, X., Parkhouse, R.M., Lemoine, R., Izui, S., and Reininger, L. (1998). Linkage of a majorquantitative trait locus to Yaa gene-induced lupus- like nephritis in (NZW xC57BL/6)F1 mice, Eur J Immunol 28, 4257-67.

Serikawa, T., Kuramoto, T., Yokoi, N., Andoh, Y., Cui, Z., Kondo, Y., Komeda, K.,and Kitada, K. (1999). Rat genetic map and a rat/mouse/human comparativegene map, Transplant Proc 31, 1537-40.

Serreze, D. V., Prochazka, M., Reifsnyder, P. C., Bridgett, M. M., and Leiter, E. H.(1994). Use of recombinant congenic and congenic strains of NOD mice toidentify a new insulin-dependent diabetes resistance gene, J Exp Med 180,1553-8.

Shimomura, K., Low-Zeddies, S. S., King, D. P., Steeves, T. D., Whiteley, A.,Kushla, J., Zemenides, P. D., Lin, A., Vitaterna, M. H., Churchill, G. A., andTakahashi, J. S. (2001). Genome-wide epistatic interaction analysis revealscomplex genetic determinants of circadian behavior in mice, Genome Res 11,959-80.

Silman, A. J., MacGregor, A. J., Thomson, W., Holligan, S., Carthy, D., Farhan, A.,and Ollier, W. E. (1993). Twin concordance rates for rheumatoid arthritis:results from a nationwide study, Br J Rheumatol 32, 903-7.

Stastny, P. (1978). Association of the B-cell alloantigen DRw4 with rheumatoidarthritis, N Engl J Med 298, 869-71.

References 46

Sun, S. H., Silver, P. B., Caspi, R. R., Du, Y., Chan, C. C., Wilder, R. L., andRemmers, E. F. (1999). Identification of genomic regions controllingexperimental autoimmune uveoretinitis in rats, Int Immunol 11, 529-34.

Taylor, B. A., and Phillips, S. J. (1996). Detection of obesity QTLs on mousechromosomes 1 and 7 by selective DNA pooling, Genomics 34, 389-98.

te Riele, H., Maandag, E. R., and Berns, A. (1992). Highly efficient gene targetingin embryonic stem cells through homologous recombination with isogenicDNA constructs, Proc Natl Acad Sci U S A 89, 5128-32.

The MHC sequencing consortium (1999). Complete sequence and gene map ofa human major histocompatibility complex. The MHC sequencing consortium,Nature 401, 921-3.

Thomas, K. R., and Capecchi, M. R. (1987). Site-directed mutagenesis by genetargeting in mouse embryo-derived stem cells, Cell 51, 503-12.

Todd, J. A., Aitman, T. J., Cornall, R. J., Ghosh, S., Hall, J. R., Hearne, C. M.,Knight, A. M., Love, J. M., McAleer, M. A., Prins, J. B., and et al. (1991). Geneticanalysis of autoimmune type 1 diabetes mellitus in mice, Nature 351, 542-7.

Toole, J. J., Knopf, J. L., Wozney, J. M., Sultzman, L. A., Buecker, J. L., Pittman, D.D., Kaufman, R. J., Brown, E., Shoemaker, C., Orr, E. C., and et al. (1984).Molecular cloning of a cDNA encoding human antihaemophilic factor, Nature312, 342-7.

Wakeland, E., Morel, L., Achey, K., Yui, M., and Longmate, J. (1997). Speedcongenics: a classic technique in the fast lane (relatively speaking), ImmunolToday 18, 472-7.

Van Etten, W. J., Steen, R. G., Nguyen, H., Castle, A. B., Slonim, D. K., Ge, B.,Nusbaum, C., Schuler, G. D., Lander, E. S., and Hudson, T. J. (1999). Radiationhybrid map of the mouse genome, Nat Genet 22, 384-7.

Watanabe, T. K., Bihoreau, M. T., McCarthy, L. C., Kiguwa, S. L., Hishigaki, H.,Tsuji, A., Browne, J., Yamasaki, Y., Mizoguchi-Miyakita, A., Oga, K., et al. (1999).A radiation hybrid map of the rat genome containing 5,255 markers, Nat Genet22, 27-36.

Weis, J. J., McCracken, B. A., Ma, Y., Fairbairn, D., Roper, R. J., Morrison, T. B.,Weis, J. H., Zachary, J. F., Doerge, R. W., and Teuscher, C. (1999). Identificationof quantitative trait loci governing arthritis severity and humoral responses inthe murine model of Lyme disease, J Immunol 162, 948-56.

47 References

Weyand, C. M., Klimiuk, P. A., and Goronzy, J. J. (1998). Heterogeneity ofrheumatoid arthritis: from phenotypes to genotypes, Springer SeminImmunopathol 20, 5-22.

Wicker, L. S., Todd, J. A., and Peterson, L. B. (1995). Genetic control ofautoimmune diabetes in the NOD mouse, Annu Rev Immunol 13, 179-200.

Vidal, S., Kono, D. H., and Theofilopoulos, A. N. (1998). Loci predisposing toautoimmunity in MRL-Fas lpr and C57BL/6-Faslpr mice, J Clin Invest 101, 696-702.

Wilder, R. L., Griffiths, M. M., Remmers, E. F., Cannon, G. W., Caspi, R. R.,Kawahito, Y., Gulko, P. S., Longman, R. E., Dracheva, S. V., Du, Y., et al. (1999).Localization in rats of genetic loci regulating susceptibility to experimentalerosive arthritis and related autoimmune diseases, Transplant Proc 31, 1585-8.

Vingsbo, C., Jonsson, R., and Holmdahl, R. (1995). Avridine-induced arthritis inrats; a T cell-dependent chronic disease influenced both by MHC genes and bynon-MHC genes, Clin Exp Immunol 99, 359-63.

Vingsbo, C., Sahlstrand, P., Brun, J. G., Jonsson, R., Saxne, T., and Holmdahl, R.(1996). Pristane-induced arthritis in rats: a new model for rheumatoid arthritiswith a chronic disease course influenced by both major histocompatibilitycomplex and non-major histocompatibility complex genes, Am J Pathol 149,1675-83.

Vingsbo-Lundberg, C., Nordquist, N., Olofsson, P., Sundvall, M., Saxne, T.,Pettersson, U., and Holmdahl, R. (1998). Genetic control of arthritis onset,severity and chronicity in a model for rheumatoid arthritis in rats, Nat Genet20, 401-4.

Vogel, K. S., Klesse, L. J., Velasco-Miguel, S., Meyers, K., Rushing, E. J., andParada, L. F. (1999). Mouse tumor model for neurofibromatosis type 1, Science286, 2176-9.

Wolfe, F. (2000). The effect of smoking on clinical, laboratory, and radiographicstatus in rheumatoid arthritis, J Rheumatol 27, 630-7.

Wordsworth, P. (1995). Genes and arthritis, Br Med Bull 51, 249-66.

References 48

Yang, H. T., Jirholt, J., Svensson, L., Sundvall, M., Jansson, L., Pettersson, U., andHolmdahl, R. (1999). Identification of genes controlling collagen-inducedarthritis in mice: striking homology with susceptibility loci previously identifiedin the rat, J Immunol 163, 2916-21.

Zeng, Z. B. (1994). Precision mapping of quantitative trait loci, Genetics 136,1457-68.

Zhu, Y., Jong, M. C., Frazer, K. A., Gong, E., Krauss, R. M., Cheng, J. F., Boffelli,D., and Rubin, E. M. (2000). Genomic interval engineering of mice identifies anovel modulator of triglyceride production, Proc Natl Acad Sci U S A 97, 1137-42.

genetic studies of rheumatoid arthritis using...

Documents