Gene expression profiling of arthritis using a QTL chip reveals a complex gene regulation of the Cia5 region in mice


One of the major quantitative trait loci (QTLs) associated with arthritis in crosses between B10.RIII and RIIIS/J mice is the Cia5 on chromosome 3. Early in the congenic mapping process it was clear that the locus was complex, consisting of several subloci with small effects. Therefore, we developed two novel strategies to dissect a QTL: the partial advanced intercross (PAI) strategy, with which we recently found the Cia5 region to consist of three loci, Cia5, Cia21 and Cia22, and now we introduce the QTL-chip strategy, where we have combined congenic mapping with a QTL-restricted expression profiling using a novel microarray design. The expression of QTL genes was compared between parental and congenic mice in lymph node, spleen and paw samples in five biological replicates and in dye-swapped experiments at three time points during the induction phase of arthritis. The QTL chip approach revealed 4 genes located in Cia21, differently expressed in lymph nodes, and 14 genes in Cia22, located within two clusters. One cluster contains six genes, differently expressed in spleen, and the second cluster contains eight genes, differently expressed in paws. We conclude the QTL-chip strategy to be valuable in the selection of candidate genes to be prioritized for further investigation.


Rheumatoid arthritis (RA) and other complex diseases or syndromes are typically the result of combinations of multiple contributing genes, environmental factors, gene–gene interactions and/or gene–environment interactions. Despite large efforts, the disease etiology remains largely unknown. The collagen-induced arthritis (CIA) model in mice is used to study the pathology of the disease.1 Over 90% of the mouse and human genomes can be partitioned into corresponding regions of conserved synteny, and key pathways are most probably also conserved, opening the possibility to study the biology in mice and translate the results to the human situation.2, 3, 4

Over 30 quantitative trait loci (QTL) for CIA in mice have been identified (listed at However, the transition from QTL to QTG (quantitative trait gene) is not straightforward. One reason for the experienced difficulties is that several of the identified disease loci have turned out to be clusters of genes with small individual effects. In general, it is difficult to narrow down the congenic fragments to less than a few hundred genes by breeding. Therefore, in order to identify QTGs, we need alternative strategies to reduce the number of candidate genes. We have recently developed the partial advanced intercross (PAI) strategy to increase penetrance through genetic interactions, and at the same time accumulate valuable recombinations.5 In addition, in order to facilitate the genetic mapping, our group and others have studied subphenotypes with higher penetrance than the originally mapped more complex phenotype, such as the production of auto-antibodies and CD4/CD8 T-cell ratio.6, 7 Another approach to reduce the number of candidate genes is to investigate the expression of the genes in the QTL.8, 9 In this way, polymorphisms in regulatory or promoter regions resulting in differences in mRNA expression will be detected. This kind of experiment is well suited in a distinct system like a congenic strain, where the noise of genetic differences between strains is reduced.

The combination of expression and mapping studies has proven successful in previous studies with complex traits. Several candidate QTGs for hypertension were identified by integrating global expression profiling with linkage analysis.10 Moreover, the Ifi202 gene was identified in expression studies of a congenic strain for the Nba2 locus in the NZB/NZW model of SLE.11 Also, the importance of the complement component C5 in a model for asthma12 and the CD36 gene for insulin resistance was demonstrated by the same technique.13 In these experiments the global gene expression was studied and subsequently the QTL gene identified. However, the use of a genome-wide approach as a method to identify QTGs is not straightforward. Firstly, on a genome wide chip it is difficult to assure the presence of all the genes in the QTL region and, secondly, the risk of false- positives and -negatives is increased with the number of genes investigated.

To avoid some of these problems when examining the Cia5 locus on mouse chromosome 3, we have used a different microarray approach where only genes located in the QTL region were studied. The Cia5 locus was identified in a cross between the arthritis resistant RIIIS/J strain and the susceptible B10.RIII strain.14 The Cia5 congenic strain with an RIIIS/J fragment inserted onto the B10.RIII susceptible background has a reduced severity of CIA compared to the parental strain.5 In order to select candidate genes for further investigations, we have studied the QTL gene expression profile during the priming phase of arthritis. We initially investigated the expression differences across the genome, which gave us an indication that the Cia5 region contains several differently expressed genes. However, problems with gene annotations and multiple testing on the genome wide chip led us to construct a custom made microarray, containing only genes in the arthritis-linked Cia5 region. The expression QTL genes was studied in three tissues and compared between the congenic and the parental strain, before and after immunization. This is the first time a congenic strain is extensively profiled exclusively for the expression of candidate genes within the QTL.


Expression profiling of 11 000 genes reveals differently expressed genes within the Cia5 region

In order to get an indication of the immunological process in the arthritis-resistant Cia5 congenic strain we compared the gene expression in the congenic mice with the expression in the parental B10. RIII mice. In this pilot study, we used the single dye Affymetrix system to measure the expression of 11 000 genes across the genome. Since it is unknown how the protection from CIA development is mediated in the Cia5 congenic strain, we chose to profile three whole tissues: lymph nodes, spleen and paws. We were primarily interested in finding inciting or etiological factors, and therefore chose to investigate samples taken early in the disease process, before the visible onset of arthritis. Within a few days after immunization with collagen type II (CII), the immune system is primed and after 1 or 2 weeks, signs of immune activation can be detected in the joints.1 However, the macroscopic signs of arthritis in the B10.RIII strain have in our hands never been detected before day 17 after immunization.5 Interestingly, several genes located in the congenic fragment were differently expressed and of these Chi3l3 showed the most pronounced expression difference between the two strains (FC –23.5 at day 12 in spleen). Some of the genes were differently expressed in more than one of the tissues (Chi3l3) or at more than one time point (Gstm5, Sprr2I) supporting the interpretation of a true difference. However, the experimental setup with pooled samples did not suffice for any statistical analysis on the biological variance. In addition, when comparing the genes present on the genome wide chip with genes currently annotated to the Cia5 region, it became clear that not all genes in the region were present on the chip. Consequently, the results were used as an indication of a complex gene regulation within Cia5.

Profiling the QTL gene expression between the Cia5 congenic and the parental strain using a QTL chip

To be able to explore the QTL gene expression differences in more detail and with higher statistical power, we prepared a custom made QTL chip containing a more accurate selection of QTL genes compared to the genome wide chip (covering the region seen in Figure 1). The QTL chip is a two-dye system where the sample is hybridized together with a tissue-specific reference sample. We focused on a 15 Mb region, surrounding the Cia5 peak marker,14 containing 163 genes. As for the genome wide chip we prepared samples from lymph nodes, spleen and paws from Cia5 congenic and parental mice. However, in this experiment five biological replicates in dye-swapped experiments were investigated at three time points during the induction phase of arthritis (Figure 2). Of the genes represented on the chip, roughly 50% were expressed in spleen, 70% in lymph nodes and 60% in paws (data not shown) and, in total, 26 genes had a significant expression difference (P-value of 0.05) between the congenic and the parental strain (Table 1). The genes that exhibited the largest differences in fold change between the congenic and the parental strain at several time points were Chi3l3 and Vav3. The Chi3l3 gene was downregulated in the congenic mice at all time points in spleen (Figure 3a) and also in lymph nodes at day 6. The Vav3 gene was, in contrast to Chi3l3, upregulated at all time points in spleen as well as in lymph nodes at day 0 and in paws at day 6 and 12. Vav3 is the only gene with an expression difference seen in all three tissues (Table 1). The Gpr89 gene was also upregulated at all time points in spleen and at day 0 in lymph nodes (Figure 3b).

Figure 1

A map of the congenic fragment and the region covered by the QTL chip. The black area on the chromosome represents the Cia5 congenic region. The QTL chip region is shown to the left. The Mb positions are according to the NCBI assembly 33.

Figure 2

The experimental setup for the QTL gene expression experiment. Paw, spleen and lymph node samples were taken at day 0 (unimmunized mice) and at days 6 and 12 after immunization with CII in IFA. Five biological replicates were taken for each strain at the selected time points. Each sample was stained with both Cy5 and Alexa and then hybridized with a tissue-specific reference sample.

Table 1 Differently expressed genes in the Cia5 congenic compared to the parental strain B10.RIII in the QTL-chip experiment
Figure 3

Expression differences between the Cia5 congenic and B10.RIII parental mice in the QTL chip experiment. The results are given as fold change compared to the tissue-specific reference samples. Sp=spleen. LN=lymph nodes. (a) Expression of Chi3l3 is downregulated in spleen in Cia5 congenic mice (independent of day). (b) Expression of Gpr89 is upregulated in spleen in Cia5 congenic mice (independent of day). (c) Expression of Ptpn22 is downregulated in lymph nodes in the Cia5 congenic mice (day 0).

As Chi3l3, the Ptpn22 gene was downregulated in the congenic strain but only at one time point, day 0 in lymph nodes (Figure 3c). The CD53 gene had a small but highly significant difference in expression between the Cia5 congenic and the parental B10.RIII strain at day 0 in lymph nodes and at day 12 in spleen. The expression difference seen in spleen is due to an upregulation of CD53 in the parental strain as a result of the immunization (Figure 4a). The reverse expression pattern was seen for Gstm1, which is upregulated in the parental strain at day 0 in spleen and lymph nodes. However, after immunization, the difference in Gstm1 expression in lymph nodes was reduced due to a downregulation in the parental strain (Figure 4b). The results are summarized in Table 1, where the expression differences in the congenic Cia5 strain compared to the parental B10.RIII strain are indicated as Δfold change, which is a measurement of expression differences, not expression levels.

Figure 4

Different expression patterns as a result of the immunization. The results are given as fold change compared to the tissue-specific reference samples. (a) The expression of CD53 in spleen is upregulated in the parental strain but not in the congenic strain due to the immunization as seen at day 12. (b) The expression of Gstm1 in lymph nodes is upregulated in the parental strain but not in the congenic strain at day 0 but decreases after immunization.

Validation of the expression differences found with the QTL chip

Of the genes located in the Cia5 region that were found to be differently expressed in the genome-wide expression analysis, six were also included on the QTL chip. Three of them, Chi3l3, Gstm5 and Sprr2D, were confirmed, whereas Sprr2I, Sprr2F and 2810052M02Rik were not. The Vav3 gene that was highly differently expressed between the two strains in the QTL chip experiment was not present on the Affymetrix chip.

Four of the most interesting expression differences, found for Chi3l3, Vav3, Gstm1 and Ptpn22 were tested in 3–6 new biological samples from each strain using real-time RT-PCR. Ptpn22 was confirmed to be downregulated in the congenic strain compared to the parental in lymph nodes (fold change 2.4 in B10.RIII compared with congenic mice, P=0.04), and Chi3l3 was confirmed in spleen. The expression of Chi3l3 was nearly absent or undetectable in the congenic mice in contrast to the parental mice, which all had a high expression of the gene (Figure 5). For Gstm1 the same tendency as seen on the QTL chip was detected (fold change 2.3 in B10.RIII compared with congenic mice, P=0.08). We were not able to confirm the very low levels of Vav3 expression in the parental mice. Instead, both strains showed comparable expression with real-time RT-PCR (data not shown).

Figure 5

Validation of the Chi3l3 expression differences using real-time RT-PCR. Chi3l3 expression in Cia5 congenic mice was nearly absent in contrast to the high expression seen in the parental B10.RIII mice.

Location of the differently expressed genes

The differently expressed genes, Gstm1, Gstm2, Gstm5, Gstm7, Sort1, Sars1, Gnai and Dda3, are clustered in a 0.4 Mb region between 108.4 and 108.8 Mb (Figure 6). Notably, all but Sars1 and Gnai are differently expressed in the paw, the arthritis target organ. Moreover, all but Gstm7 are downregulated in the congenic mice. Chi3l3, CD53, Atp5f, 1110003H18Rik, 2010012G17Rik and 2610212E17Rik are located in a 1.2 Mb region between 106.4 and 107.6 Mb and are differently expressed in spleen. The genes Ptpn22, CD2 and Olfm13, which are downregulated in lymph nodes, are located between 101.1 and 103.8 Mb. Comparing the location of these genes with the recently published results from the PAI5 reveals that the two clusters between 106.4 and 107.6, and 108.4 and 108.8 Mb are located in the Cia22 region and that Ptpn22, CD2, Tspan-2 and Olfm13 are located in the Cia21 region (Figure 6).

Figure 6

A map of the 19 differently expressed genes between the B10.RIII parental and the Cia5 congenic strain located in the Cia21 and Cia22 loci. The genes that were upregulated in the congenic mice compared to the parental mice are written to the left. The genes that were downregulated in the congenic mice compared to the parental mice are written to the right. The Mb positions are according to the NCBI assembly 33.


Positional cloning of QTLs with small effects is a difficult task and alternative strategies in order to reduce the number of candidate genes are needed. By investigating the expression of genes within the Cia5 QTL region, we have found 26 genes that are differently expressed during the induction phase of arthritis in the Cia5 congenic compared to the parental strain. The expression patterns detected include a constitutive expression difference between the two strains at all time points as seen for Gpr8 and Chi3l3, and a decrease or increase in expression difference after immunization as seen for Gstm1 and CD53. Several other genes were differently expressed at one specific time point, for example, Ptpn22, Gstm7 and Tspan-2. Four of the differently expressed genes are located in the recently described Cia21 region and 14 in the Cia22 region identified in the same study.5 The differently expressed genes in Cia22 are located in two clusters, containing six and eight genes, respectively. The expression differences for one gene with a highly significant P-value (P=0.005), Chi3l3, and one with a P-value of 0.03, Ptpn22, were tested and confirmed with real-time RT-PCR.

The advantage of using a QTL chip, compared to a genome-wide approach, is the ability to focus on a specific region of interest. We used both genome-wide and QTL-restricted expression analysis in the current study. The genome wide expression analysis revealed several differently expressed genes; however, not all genes in the region were present on the genome-wide chip. Moreover, gene annotations are constantly changing, which is difficult to follow on a global level but possible in a specific region of interest. With the custom made QTL chip we could carefully select which genes to include insuring that all genes annotated to the QTL chip region were present on the chip. In combination with the high number of replicates, this enabled us to detect several new candidate genes for arthritis.

The experimental setup for the QTL chip allowed us to identify differently expressed genes with high probability and also to detect low expression differences. In total, 26 genes had a P-value of 0.05 and given the number of genes on the chip, approximately eight could be false positives. The fact that the expression difference for Ptpn22, which had a P-value of 0.0267, was confirmed indicates that a P-value 0.05 as a threshold is reasonable. We have avoided using a Δfold change value as threshold since the biological relevance for such a threshold is lacking. A small expression difference could have equally important biological effects as a large difference. The CD53 gene has a low Δfold change of −0.84 at day 12 in spleen (P-value 0.0108) but with nonoverlapping values, indicating a true difference in expression.

The differently expressed genes identified in this study are strong candidates not only because of their location but also because of their known immunological functions. Both the Gstm1 and Gstm2 were found to be expressed at a higher level in the parental than in the Cia5 congenic strain. This difference was observed at day 0, however; after immunization the difference in expression decreased due to a downregulation in the parental strain (Figure 4b). The members of the glutathione s-transferase (GST) superfamily are important in the protection of cells from reactive oxygen species due to their ability to utilize substrates of a wide range of products of oxidative stress.15 Based on their thought role in protecting cells from oxidative stress, several GSTs have been suggested as candidate genes for cancer,16 arthritis,17 hypertension18 and asthma.19 However, the lower levels of Gstm1 and Gstm2 in the congenic strain might play a protective role during the induction phase of CIA, since reduced oxidative burst response has been shown to promote activation of arthritogenic T cells, both in pristane- induced arthritis (PIA) in rats20 and in CIA in mice.21

Tspan-2, which is one of the candidates for Cia21,5 is also a possible candidate for the insulin-dependent diabetes loci 10, Idd10, mapped to the same region.22 The gene was upregulated in spleen in the Cia5 congenic compared to the parental strain 12 days after immunization. Tspan-2 encodes a protein participating in the membrane dynamics required for proper T-cell signaling.23 Another T-cell-associated gene, Ptpn22, which was downregulated in paw samples from unimmunized Cia5 congenic mice, is also located within the Cia21 region. Ptpn22 encodes the protein LYP, which is an inhibitor of T-cell activation.24, 25 Different allelic variants of Ptpn22 have recently been associated with several autoimmune disorders in humans including RA and Type I diabetes26, 27, 28 The upregulation of Tspan-2 and the downregulation of Ptpn22 could indicate a higher activation state of T cells in the congenic strain. This activation might be required for self-reactive T cells to undergo apoptosis or to become T regulatory cells; however, further experiments are needed in order to answer this question. Nevertheless, further support of the congenic T cells being activated is the expression differences of the Vav3 gene, one of the candidate genes for Cia22. The expression of Vav3 in the parental strain is nearly absent on the QTL chip whereas a clear expression is seen in the real-time RT-PCR experiment. This could indicate a polymorphism in the coding region where the probe on the QTL chip is binding. The Vav family of Rho guanine nucleotide exchange factors is thought to orchestrate signaling events downstream of lymphocyte antigen receptors and is essential in the formation of an adaptive immune response.29 Vav3 has been reported to be a positive regulator of TCR-induced transcription30 and important in positive and negative thymocyte selection.31 Vav3 is thus a possible candidate not only for Cia22, but also for the Trmq6 locus that controls CD4/CD8 T-cell ratio linked to the same region.6

The most markedly downregulated gene in the congenic strain was the Chi3l3, where the expression in the congenic mice is either not detectable or very low compared with the parental mice. Chi3l3 is an eosinophil chemotactic factor mainly expressed in neutrophils and macrophages.32, 33 The protein is upregulated by Th2 cytokines such as IL-4 by a STAT6-dependent mechanism.32 Th2 responses and extensive eosinophilic infiltration in arthritic joints have previously been demonstrated in CIA34 and a low amount of Chi3l3 could thus play a protective role.

It is possible that the protective effect observed in the congenic strain represents the sum of effects from all the differently expressed genes. Whether all the genes are polymorphic, or whether one polymorphic gene or transcription factor (not necessarily differently expressed) affects the expression of all of these genes, needs further investigation. A possibility is that the differently expressed genes are members of the same pathway and, consequently, interdependent. The fact that the differently expressed genes are clustered in the genome may indicate a common regulator. To answer this question we need to do a high-resolution mapping in combination with QTL gene expression studies and sequencing of the candidate genes.

The selection of tissues is central to the study of gene expression. For a number of diseases, like arthritis, where many cell types and organs are involved in the disease progression, the choice of tissue or cell type is not obvious. Therefore, we chose to array samples from three different tissues. Lymph nodes and spleen were chosen because they are known to be involved in the priming phase of arthritis and paws because they are the target tissue of the autoimmune attack. Since we have studied complex tissues, any expression difference may reflect either an increased number of a certain cell type or the increased expression of a gene in a constituent cell population. This will be further investigated by isolation of cell populations within spleen, lymph nodes and paws followed by gene expression analysis.

Although we have found several differently expressed genes in the Cia5 fragment, this alone is not proof of them causing the CIA phenotype. We cannot exclude that the expression differences between the congenic and parental strain are affecting a different trait other than arthritis. However, in such a defined system as a congenic strain, this is unlikely. Moreover, even if the genes in the congenic fragment responsible for the lower severity of CIA might not be differently expressed, the knowledge of what genes are regulated by the congenic fragment will still help reveal the pathways involved.

In summary, we have found 26 genes in the Cia5 fragment that are differently expressed between the Cia5 congenic and the parental B10.RIII strain. These genes are likely to be involved in protecting the Cia5 congenic mice from developing CIA and will be prioritized in the selection of candidate genes that will be further investigated. In QTL mapping, there is no golden standard strategy; but the QTL chip is definitely a valuable tool in the search for genes underlying complex traits.

Materials and methods

Animal experiments

Mice were bred and kept in the animal house of the Medical Inflammation Research Unit, Lund University. The mice were kept in climate-controlled environment with 12–h light/dark cycles, fed with standard rodent chow and water ad libitum (as defined in Jan Klein, Tübingen, Germany, originally provided the B10.RIII mice and RIIIS/J mice were obtained from Jackson Laboratories (Bar Harbor, ME, USA). The production of Cia5-region congenic mice (denoted by R3) is described elsewhere.5 Parental B10.RIII and Cia5 congenic 12-week-old mice were immunized with 100 μg bovine CII emulsified in IFA (incomplete Freund's adjuvant)(Difco, Detroit) by intradermal (i.d.) immunization at the base of the tail. Paw, spleen and lymph node samples were taken from unimmunized mice (day 0) and at day 12 after immunization for the Affymetrix MG_U74A chip and at days 0, 6 and 12 after immunization for the QTL chip (Figure 2). Five biological replicates were taken for each strain at the selected time points.

Sample preparation and labeling

Paws, spleens and lymph nodes (in that order) were dissected out and immediately either snap frozen in liquid nitrogen (paws) or transferred into tubes with 1 ml of RNAlater™ (Ambion, Austin, TX, USA) (lymph nodes and spleens). Care was taken to ensure that the time between time of death and harvest of each of the tissues was as rapid and consistent as possible. Total RNA was extracted from homogenized samples using RNeasy extraction kit (Qiagen, Valencia, CA, USA) following the manufacturer's instructions. RNA concentrations were determined photometrically (260, 280 nm) and the quality was checked on a 1.5% agarose gel. For the Affymetrix experiment, total RNA was synthesized to cRNA and labeled according to the GeneChip® expression analysis (technical manual provided by Affymetrix). For the QTL chip, each RNA sample was transcribed into cDNA and labeled with either Alexa (Molecular Probes, Leiden, Netherlands) or Cy5 (PerkinElmer Life Sciences, Boston, MA, USA) according to instructions provided by Qiagen.


For the Affymetrix chip, RNA samples from the same strain, tissue and time point were pooled and hybridized, making the total number of hybridizations twelve (2 days × 3 tissues × 2 strains).

For the QTL chip, 2 μg of RNA from each sample was hybridized with a tissue-specific reference sample, made by pooling 2 μg of RNA from all samples from that specific tissue. Hybridizations were performed according to instructions provided by Qiagen. The total number of hybridizations in the QTL chip experiment was 180 (2 strains × 3 tissues × 3 time points × 5 replicates × 2 stainings). This experimental design is according to the ‘modified dye-swap’ as delineated by Churchill35 (Figure 2).

Design of the QTL chip

The selection of genes was based on both the publicly available mouse assembly based on the NCBI build 30 ( and on the assembly from the Celera Discovery System (subscription restricted). Lists of all genes in the Cia5 region were downloaded from both sites and visually compared. Selection criteria were set so that only genes present on both annotations were selected. The selection of 163 genes was based on the location of the Cia5 peak marker from the original intercross experiment.14 The selection of genes was unbiased.The function of the genes was not taken into account. However, a few genes located in the Cia5 fragment but not within the selected QTL region were added to the list because they had previously been reported as possible candidates22 or because they were differently expressed in the genome wide profiling experiment. The complete gene list is presented in Supplementary Table 1.

We utilized custom-made SensiChips, which are spotted with 70-mer oligonucleotide probes, provided by Zeptosens. Each probe is spotted in duplicate on the chip. The probes for the 163 genes were designed by the bioinformatics department at Qiagen and they were thoroughly tested for mismatching and cross-hybridization. A few genes were excluded from the chip because of the risk of cross-hybridization, for example,Chi3l4, which has a high-sequence similarity with Chi3l3. However, the probe for Chi3l3 is specific and does not bind Chi3l4.


RNA was extracted from congenic and parental mice as described above. Expression was investigated by RT-PCR on iCycler IQ™ (BIO-RAD Laboratories, CA, USA). The iQ™ SYBR® Green Supermix was used for the reactions and performed with 0.3 μM of the forward and reverse primers with a TA of 60°C. The primers used were Vav3-forward IndexTerm5′-TCC TCC TTC AGG AGC TGG TA-3, Vav3-reverse IndexTerm5′-CCA AGT CCT TCA TTG CAT CC-3′, Ptpn22-forward IndexTerm5′-CAA TCC ACC AAG TAC AAG GCC GAC-3′, Ptpn22-reverse IndexTerm5′-CTA CCA GGC TGT GAT CAT AGG GCA –3′, Chi3l3-forward IndexTerm5′-AAC CGT CAG ATA TTC AAT CAG TCA, Chi3l3-reverse IndexTerm5′-AGC TTT ACG CAT TTC CTT CAC-3′, Gstm1-forward IndexTerm5′-CTT GAA GAC CAT CCC TGA GAA A-3, Gstm1-reverse IndexTerm5′- TGA GTG CCC GTG TAG C-3, HPRT-forward IndexTerm5′-TGG ATA CAG GCC AGA CTT TGT TG-3′, HPRT-reverse IndexTerm5′-GCA GAT GGC CAC AGG ACT A-3′, B-actin-forward IndexTerm5′-ACA CCC GCC ACC AGT TCG C-3′ and B-actin-reverse IndexTerm5′-ATG GGG TAC TTC AGG GTC AGG ATA-3′. Normalization was performed to either HPRT (Vav3 and Chi3l3) or B-actin (Ptpn22) with comparable primer efficiencies. Fold changes for the expression of Vav3, Gstm1 and Ptpn22 in the Cia5 congenic strain compared to the B10.RIII parental mice were derived using the comparative 2−ΔΔCT method.

Statistical analysis

For the genome wide expression profiling with Affymetrix, the fold change values were calculated with the Microarray Suit Software version 4.0 and a fold change >3 was considered interesting. For the QTL chip analysis, the images were analyzed with the SensiChip View software version 2.1 to extract the fluorescence intensities and to get an impression of the quality of the chips. This yielded ratios for the signal intensities of the red channel vs the green channel. In order to combine the two measurements of the dye swap, the mean of the log2 ratios of the signal ratios was calculated. This log2 value represents the difference in RNA expression between an individual mouse and the tissue-specific reference sample and is referred to as the fold change. To test for significant expression differences, a Wilcoxon rank test was performed as implemented in R.36 To be able to make a direct comparison between the congenic and the parental strain, the difference of the average fold changes between the two strains was compared and is referred to as Δfold change, which still is a logarithmic value of the base 2. Genes were considered differently expressed when the expression difference had a P-value equal to or lower than 0.05.


  1. 1

    Holmdahl R, Bockermann R, Backlund J, Yamada H . The molecular pathogenesis of collagen-induced arthritis in mice—a model for rheumatoid arthritis. Ageing Res Rev 2002; 1: 135–147.

    CAS  Article  Google Scholar 

  2. 2

    The Mouse Genome Consortium. Initial sequencing and comparative analysis of the mouse genome. Nature 2002; 420: 520–562.

  3. 3

    Vyse TJ, Todd JA . Genetic analysis of autoimmune disease. Cell 1996; 85: 311–318.

    CAS  Article  Google Scholar 

  4. 4

    Serrano-Fernandez P, Ibrahim SM, Zettl UK et al. Intergenomic consensus in multifactorial inheritance loci: the case of multiple sclerosis. Genes Immun 2004; 5: 615–620.

    CAS  Article  Google Scholar 

  5. 5

    Johannesson M, Karlsson J, Wernhoff P et al. Identification of epistasis through a partial advanced intercross reveals three arthritis loci within the Cia5 QTL in mice. Genes Immun 2005; 6: 175–185.

    CAS  Article  Google Scholar 

  6. 6

    Karlsson J, Johannesson M, Lindvall T et al. Genetic interactions in Eae2 control collagen induced arthritis and the CD4+/CD8+ T cell ratio. J Immunol 2005; 174: 533–541.

    CAS  Article  Google Scholar 

  7. 7

    Morel L, Croker BP, Blenman KR et al. Genetic reconstitution of systemic lupus erythematosus immunopathology with polycongenic murine strains. Proc Natl Acad Sci USA 2000; 97: 6670–6675.

    CAS  Article  Google Scholar 

  8. 8

    Hitzemann R, Malmanger B, Reed C et al. A strategy for the integration of QTL, gene expression, and sequence analyses. Mamm Genome 2003; 14: 733–747.

    CAS  Article  Google Scholar 

  9. 9

    Fischer G, Ibrahim SM, Brockmann GA et al. Expressionview: visualization of quantitative trait loci and gene-expression data in Ensembl. Genome Biol 2003; 4: R77.

    Article  Google Scholar 

  10. 10

    Hubner N, Wallace CA, Zimdahl H et al. Integrated transcriptional profiling and linkage analysis for identification of genes underlying disease. Nat Genet 2005; 37: 243–253.

    CAS  Article  Google Scholar 

  11. 11

    Rozzo SJ, Allard JD, Choubey D et al. Evidence for an interferon-inducible gene, Ifi202, in the susceptibility to systemic lupus. Immunity 2001; 15: 435–443.

    CAS  Article  Google Scholar 

  12. 12

    Karp CL, Grupe A, Schadt E et al. Identification of complement factor 5 as a susceptibility locus for experimental allergic asthma. Nat Immunol 2000; 1: 221–226.

    CAS  Article  Google Scholar 

  13. 13

    Aitman T, Glazier AM, Wallace CA et al. Identification of Cd36 (Fat) as an insulin-resistance gene causing defective fatty acid and glucose metabolism in hypersensive rats. Nat Genet 1999; 21: 76–83.

    CAS  Article  Google Scholar 

  14. 14

    Jirholt J, Cook A, Emahazion T et al. Genetic linkage analysis of collagen-induced arthritis in the mouse. Eur J Immunol 1998; 28: 3321–3328.

    CAS  Article  Google Scholar 

  15. 15

    Hayes JD, Strange RC . Glutathione S-transferase polymorphisms and their biological consequences. Pharmacology 2000; 61: 154–166.

    CAS  Article  Google Scholar 

  16. 16

    Yuille M, Condie A, Hudson C et al. Relationship between glutathione S-transferase M1, T1, and P1 polymorphisms and chronic lymphocytic leukemia. Blood 2002; 99: 4216–4218.

    CAS  Article  Google Scholar 

  17. 17

    Mattey DL, Hutchinson D, Dawes PT et al. Smoking and disease severiy in rheumatoid arthritis. Association with polymorphism at the Glutathione S-Transferase M1 locus. Arthritis Rheum 2002; 46: 640–646.

    CAS  Article  Google Scholar 

  18. 18

    McBride MW, Brosnan MJ, Mathers J et al. Reduction of Gstm1 expression in the stroke-prone spontaneously hypertension rat contributes to increased oxidative stress. Hypertension 2005; 45: 786–792.

    CAS  Article  Google Scholar 

  19. 19

    Brasch-Andersen C, Christiansen L, Tan Q et al. Possible gene dosage effect of glutathione-s-transferases on atopic asthma: using real-time PCR for quantification of GSTM1 and GSTT1 gene copy numbers. Hum Mutat 2004; 24: 208–214.

    CAS  Article  Google Scholar 

  20. 20

    Olofsson P, Holmberg J, Tordsson J et al. Positional identification of Ncf1 as a gene that regulates arthritis severity in rats. Nat Genet 2003; 33: 25–32.

    CAS  Article  Google Scholar 

  21. 21

    Hultqvist M, Olofsson P, Holmberg J et al. Enhanced autoimmunity, arthritis, and encephalomyelitis in mice with a reduced oxidative burst due to a mutation in the Ncf1 gene. Proc Natl Acad Sci USA 2004; 101: 12646–12651.

    CAS  Article  Google Scholar 

  22. 22

    Penha-Goncalves C, Moule C, Smink LJ et al. Identification of a structurally distinct CD101 molecule encoded in the 950-kb Idd10 region of NOD mice. Diabetes 2003; 52: 1551–1556.

    CAS  Article  Google Scholar 

  23. 23

    Delaguillaumie A, Harriague J, Kohanna S et al. Tetraspanin CD82 controls the association of cholesterol-dependent microdomains with the actin cytoskeleton in T lymphocytes: relevance to co-stimulation. J Cell Sci 2004; 117: 5269–5282.

    CAS  Article  Google Scholar 

  24. 24

    Cohen S, Dadi H, Shaoul E, Sharfe N, Roifman CM . Cloning and characterization of a lymphoid-specific, inducible human protein tyrosine phosphatase, Lyp. Blood 1999; 93: 2013–2024.

    CAS  PubMed  Google Scholar 

  25. 25

    Cloutier JF, Veillette A . Cooperative inhibition of T-cell antigen receptor signaling by a complex between a kinase and a phosphatase. J Exp Med 1999; 189: 111–121.

    CAS  Article  Google Scholar 

  26. 26

    Smyth D, Cooper JD, Collins JE et al. Replication of an association between the lymphoid tyrosine phosphatase locus (LYP/PTPN22) with Type 1 diabetes, and evidence for its role as a general autoimmunity Locus. Diabetes 2004; 53: 3020–3023.

    CAS  Article  Google Scholar 

  27. 27

    Bottini N, Musumeci L, Alonso A et al. A functional variant of lymphoid tyrosine phosphatase is associated with type I diabetes. Nat Genet 2004; 36: 337–338.

    CAS  Article  Google Scholar 

  28. 28

    Begovich AB, Carlton VE, Honigberg LA et al. A missense single-nucleotide polymorphism in a gene encoding a protein tyrosine phosphatase (PTPN22) is associated with rheumatoid arthritis. Am J Hum Genet 2004; 75: 330–337.

    CAS  Article  Google Scholar 

  29. 29

    Fujikawa K, Miletic AV, Alt FW et al. Vav1/2/3-null mice define an essential role for Vav family proteins in lymphocyte development and activation but a differential requirement in MAPK signaling in T and B cells. J Exp Med 2003; 198: 1595–1608.

    CAS  Article  Google Scholar 

  30. 30

    Zakaria S, Gomez TS, Savoy DN et al. Differential regulation of TCR-mediated gene transcription by Vav family members. J Exp Med 2004; 199: 429–434.

    CAS  Article  Google Scholar 

  31. 31

    Costello PS, Walters AE, Mee PJ et al. The Rho-family GTP exchange factor Vav is a critical transducer of T cell receptor signals to the calcium, ERK, and NF-kappaB pathways. Proc Natl Acad Sci USA 1999; 96: 3035–3040.

    CAS  Article  Google Scholar 

  32. 32

    Welch JS, Escoubet-Lozach L, Sykes DB et al. TH2 cytokines and allergic challenge induce Ym1 expression in macrophages by a STAT6-dependent mechanism. J Biol Chem 2002; 277: 42821–42829.

    CAS  Article  Google Scholar 

  33. 33

    Harbord M, Novelli M, Canas B et al. Ym1 is a neutrophil granule protein that crystallizes in p47phox-deficient mice. J Biol Chem 2002; 277: 5468–5475.

    CAS  Article  Google Scholar 

  34. 34

    Chen Y, Rosloniec E, Goral MI, Boothby M, Chen J . Redirection of T cell effector function in vivo and enhanced collagen-induced arthritis mediated by an IL-2R beta/IL-4R alpha chimeric cytokine receptor transgene. J Immunol 2001; 166: 4163–4169.

    CAS  Article  Google Scholar 

  35. 35

    Churchill GA . Fundamentals of experimental design for cDNA microarrays. Nat Genet 2002; 32: 490–495.

    CAS  Article  Google Scholar 

  36. 36

    Gentleman RC, Carey VJ, Bates DM et al. Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 2004; 5: R80.

    Article  Google Scholar 

Download references


We thank Isabell Bohlin and Carlos Palestro for help with the animal care, Ildiko Toth and Cecilia Andersson for technical assistance with the RNA preparations and the real-time RT-PCR, respectively, Valérie Lemée and Andreas Abel (Zeptosens) for help with the SensiChip preparations and Vincent Benoit (Zeptosens) for help with the image analysis of the SensiChips. This work was supported by grants from the Swedish Medical Research Council, the Swedish Foundation for Strategic Research, the Swedish Association against Rheumatism, the Crafoord, Lundberg, the Kock and Österlund Foundations and EU FP5 (QLG1-CT-2001-01407 ‘EUROME’).

Author information



Corresponding author

Correspondence to M Johannesson.

Additional information

Supplementary Information accompanies the paper on Genes and Immunity website (

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Johannesson, M., Olsson, L., Lindqvist, AK. et al. Gene expression profiling of arthritis using a QTL chip reveals a complex gene regulation of the Cia5 region in mice. Genes Immun 6, 575–583 (2005).

Download citation


  • quantitative trait locus
  • complex trait mapping
  • microarray
  • collagen-induced arthritis

Further reading