• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of ajhgLink to Publisher's site
Am J Hum Genet. May 2006; 78(5): 793–803.
Published online Mar 22, 2006. doi:  10.1086/503849
PMCID: PMC1474041

Cis- and Trans-Acting Gene Regulation Is Associated with Osteoarthritis

Abstract

Osteoarthritis (OA) is a complex disease of the skeleton and is associated with aging. Both environmental and genetic factors contribute to its pathogenesis. We set out to identify novel genes associated with OA, concentrating on regulatory polymorphisms allowing for differential expression. Our strategy to identify differentially expressed genes included an initial transcriptome analysis of the peripheral blood mononuclear cells of six patients with OA and six age-matched healthy controls. These were screened for allelic expression imbalances and potentially regulatory single-nucleotide polymorphisms (SNPs) in the 5′ regions of the genes. To establish disease association, disparate promoter SNP distributions correlating with the differential expression were tested on larger cohorts. Our approach yielded 26 candidate genes differentially expressed between patients and controls. Whereas BLP2 and CIAS1 seem to be trans-regulated, as the absence of allelic expression imbalances suggests, the presence of allelic imbalances confirms cis-regulatory mechanisms for RHOB and TXNDC3. Interestingly, on/off–switching suggests additional trans-regulation for TXNDC3. Moreover, we demonstrate for RHOB and TXNDC3 statistically significant associations between 5′ SNPs and the disease that hint at regulatory functions. Investigating the respective genes functionally will not only shed light on the disease association but will also add to the understanding of the pathogenic processes involved in OA and may point out novel therapeutic approaches.

Osteoarthritis (OA [MIM 165720]) is the most common musculoskeletal disease associated with aging.1 Clinical problems include pain and joint stiffness, which often lead to significant disability and joint replacement. Although OA has previously been considered a mere consequence of age-related cartilage degeneration, it is now accepted that OA is the result of active disease processes. Imbalances between the synthesis and degradation of the extracellular matrix and between the proliferation and apoptosis of the articular chondrocytes are the most widely discussed disease processes.25

OA is a complex disease with multiple environmental and genetic factors contributing to its pathogenesis.69 To elucidate the genetics, whole-genome linkage analyses and candidate-gene approaches have been performed.9 Whereas the former may provide large QTLs without revealing the disease-associated genes themselves, the latter can yield only small aspects of the genetics.1015

The description of differential gene expression as a driving force in evolution led to an increased appreciation of regulatory gene polymorphisms.16 Although the differential expression of genes allows for a flexible response to a changing environment and, thus, may be beneficial for the population as a whole, adverse effects for the individual may occur. Indeed, regulatory gene polymorphisms are likely candidates for an association with complex diseases: moderate alterations in gene expression can implement subtle differences over a long period of time and may render the organism susceptible to cumulative pathogenic impacts.17 In OA, regulatory gene polymorphisms within the IL1 locus have been described elsewhere to be associated with the disease.18,19

We set out to identify novel regulatory gene polymorphisms associated with OA. To that end, we performed transcriptome analyses of peripheral blood mononuclear cells (PBMC) from OA patients and controls, and we identified genes that were differentially expressed. We based our study on the assumption that regulatory SNPs leading to differential expression in PBMC may do so in articular chondrocytes as well. Since we were primarily interested in regulatory polymorphisms acting in cis, we screened our candidate genes for allelic expression imbalances and promoter SNPs associated with the disease. Indeed, we found two genes exerting allelic expression imbalances, which confirmed the presence of cis-regulatory polymorphisms. In addition, we found, in the same two genes, 5′ SNPs that show a statistically significant association with OA. We, therefore, consider our approach suitable to identify disease-associated differential gene expression resulting from regulatory gene polymorphisms.

Subjects and Methods

Study Population

For microarray analyses, six patients with knee OA awaiting joint replacement surgery and six age-matched healthy controls were analyzed. Whereas the controls were without any previously recorded joint complaints, the patients’ diagnosis of OA was based on clinical evaluations, excluding metabolic causes. All six patients with OA were recruited from the Charité University Hospital in Berlin and were selected for comparable and low synovitis scores of the joints replaced. Determining the synovitis score included screening the tissues for hyperplasia/enlargement of the synovial lining cell layer, activation of resident cells/synovial stroma, and inflammatory infiltrations.20

For the OA association study, SNP analysis was performed on 171 patients undergoing joint replacement surgery and 182 healthy controls. Patients with OA as well as healthy controls were of European white ethnicity. All patients fulfilled the criteria of the American College of Rheumatology regarding OA and were recruited from different hospitals within the Berlin area. The mean age of the patients was 69.4 ± 10.9 years, with a male-to-female ratio of 1:2.9 and a hip-to-knee ratio of 1:2.1. The mean age of the controls was 39.6 ± 9.8 years, with a male-to-female ratio of 1:1.9. The study was approved by the local ethics committee, and informed consent was obtained from all participants.

Microarray Analyses

Fifty milliliters of peripheral blood was obtained, and erythrocytes were lysed using ammoniumchloride/Tris at 37°C. Leukocytes were then stained on ice with monoclonal antibodies and Microbeads (Miltenyi Biotec). The CD15+ cells were depleted using automated separation with autoMACS (Miltenyi Biotec). From the CD15 population, CD14+ monocytes, CD19+ B cells, CD4+ T helper cells, and CD8+ cytotoxic T cells were positive selected by fluorescence-activated cell sorting, with use of the DiVantage (BD Biosciences) or the MoFlo (DakoCytomation). The purities of the isolated cell fractions were >98%. All samples were treated equally. Cells were lysed directly after sorting, with the use of RLT lysis buffer (Qiagen) containing β-mercaptoethanol. Total RNA was isolated using the RNeasy Mini Kit (Qiagen) and was treated with DNAse (RNase free DNase Set [Qiagen]), to remove genomic DNA. Double-stranded cDNA was synthesized using SuperScript II RNase H Reverse Transcriptase and the SuperScript II Kit (Invitrogen) and was converted to Biotin-labeled cRNA via in vitro transcription, with use of the BioArray HighYield Transcription Labeling Kit (Enzo). The labeled cRNA was then fragmented and hybridized to Affymetrix HG-U133A chips. All preparation procedures and quality controls were performed in accordance with the Affymetrix standard protocol.

Analyses of Allelic Imbalances

Genomic DNA was extracted from peripheral blood with use of the Invisorb Blood Giga Kit (Invitek). Identification of a SNP to be used as allelic marker within the transcribed region was performed via PCR amplification and sequencing of the UTRs. Information about 5′ and 3′ UTRs were derived from public databases. UTRs were sequenced for their full lengths. Primers were designed using the Primer3 software (Whitehead Institute for Biomedical Research). The resulting 18–24mer primers were suitable for amplifying and sequencing genomic DNA fragments of ~700 bp (for detailed primer sequence information, see tables 1 and and2).2). In cases where 3′ UTRs were longer than 700 bp (PADI2 and VNN1 [MIM 603570]), multiple primer pairs were used. The PCR was performed using the High Fidelity PCR Master (Roche), and the conditions were 30 cycles of 94°C for 1 min, 55°C–68°C for 1 min, and 72°C for 1 min. Before sequencing, the PCR products were purified using the Invisorb PCRapid Kit (Invitek), to remove primers and PCR reagents.

Table 1
Primers for Amplification and Sequencing of the 5′ UTRs
Table 2
Primers for Amplification and Sequencing of the 3′ UTRs

Pyrosequencing technology (Biotage AB) was used to confirm differential gene expression at the allelic level. The cDNA was synthesized using SuperScript II RNase H Reverse Transcriptase (Invitrogen) and a mixture of random hexamere and anchored oligo-dT primers. The RNA templates used were the same as those used for microarray analysis. cDNA fragments between 165 and 253 bp in length were PCR amplified using the primers RHOB (MIM 165370) 5′ Bio-TGTGTGTCTGTTCGACTCCC-3′ (forward), RHOB 5′-GCAGTACCCGGGGTCTATGT-3′ (reverse), BLP2 5′ Bio-CAGGCCCACAGTATTTCTTA-3′ (forward), BLP2 5′-GTGGGAGTGTTTCTGTTGTT-3′ (reverse), CIAS1 (MIM 606416) 5′ Bio-ACAGCTCTGTGATCCTTCCG-3′ (forward), CIAS1 5′-TACATGAGGTCACCAAGAGG-3′ (reverse), TXNDC3 (MIM 607421) 5′ Bio-ATGGAACTTCGCCGAGGTCA-3′ (forward), and TXNDC3 5′-CGGTGTAGAACAGGGGTCAG-3′ (reverse), and the PCR conditions were 45 cycles of 95°C for 15 s, 50°C–60°C for 30 s, and 72°C for 15 s. For each gene and each individual, four PCR samples were run in parallel. The PCR products were then purified via the Invisorb PCRapid Kit (Invitek). The sequencing primers—RHOB 5′-CCGCTGGTGAAGG-3′, BLP2 5′-AGTTTTGCAGTGTAATACAT-3′, CIAS1 5′-TGAGGCGCTGTGAT-3′, and TXNDC3 5′-CCTCTAAGCTAGACTG-3′—were defined using the SNP Primer Design software version 1.0.1 (Biotage AB). Before pyrosequencing analysis, the biotinylated DNA strand was prepared using streptavidin-coated paramagnetic beads (Dynal), in accordance with the manufacturer’s instructions. Measurements were performed with the PSQ96 system (Biotage AB) and, in the case of the CIAS1 gene, with the PSQ96HS system as well. Each sample was run in quadruplicates (see above).

Genotyping

SNPs in the promoter regions and in the 3′ UTRs of the genes were identified via PCR amplification and sequencing of genomic DNA fragments. Promoter fragments were located 5′ of the transcription start site and were at least 500 bp long. Information on the 3′ UTR was derived from public databases, and 3′ UTRs were sequenced in full length. For detailed primer information, see tables 2 and and3.3. To establish SNP associations, 171 patients and 182 controls were analyzed for the presence of the promoter SNPs, excepting VNN1, via an allele-specific PCR assay. Two different forward primers were designed with the most 3′ base representing either the SNP or the wild-type allele. Each forward primer was combined with a different reverse primer, so that the PCR products differed in size. For detailed primer information, see table 4. PCR was performed using the PCR Master Mix (Promega), with 45 cycles of 94°C for 1 min, 62°C–65°C for 1 min, and 72°C for 1 min. The PCR products were analyzed on 2% agarose gels (GTQ, Roth).

Table 3
Primers for Amplification and Sequencing of the Promoter Regions
Table 4
Primers for Analyzing SNP Associations via Allele-Specific PCR

Because the presence of the SNP at position −137 of the VNN1 gene generates a recognition sequence for the endonuclease MnlI, this SNP was genotyped via PCR and restriction analysis. First, a 301-bp fragment was amplified using the primers 5′-GCAGCCCTAGCGAACTAATG-3′ and 5′-CAGCTGCAGTGAAAGTGTCC-3′. PCR was performed using the PCR Master Mix (Promega), with 30 cycles of 94°C for 1 min, 60°C for 1 min, and 72°C for 1 min. PCR products were then purified using the Invisorb PCRapid Kit (Invitek), before 4 μl of the purified PCR product was digested using 10 units of MnlI (MBI Fermentas) at 37°C. After that, the DNA fragments were analyzed on 2.5% agarose gels (GTQ, Roth).

Gene Expression Analyses in Articular Chondrocytes

Total RNA was isolated from the articular cartilage of patients with OA, via the guanidinium-isothiocyanate method. Directly after joint replacement surgery, cartilage pieces were obtained from the joint surfaces and were frozen in liquid nitrogen. Then, the tissue was minced in liquid nitrogen using a mortar and pestle, was resuspended in guanidinium-isothiocyanate solution, and was homogenized using a syringe and needle. After centrifugation, the supernatant was added to a cesium chloride solution and was ultracentrifugated for 15–20 h at 20,000 rpm. The RNA pellet was then resuspended in diethyl pyrocarbonate–H2O, followed by the removal of remaining proteins via a phenol/chloroform extraction. Finally, the RNA was precipitated by the addition of yeast tRNA, sodium acetate, and ethanol. cDNA synthesis was performed as described above. Cartilage-specific gene expression was analyzed via PCR amplification of cDNA, with use of the exon-overlapping primers CIRBP (MIM 602649) 5′-CCGCTTGCGTCAGGGACCTG-3′ (forward), CIRBP 5′-CAGCCATTGGAAGGACGATCT-3′ (reverse), BLK (MIM 191305) 5′-GAAGTTGCTCGTTGTCGCT-3′ (forward), BLK 5′-TCATCCAGGTGTTCATC-3′ (reverse), DKFZP434O1427 5′-CGAGGCCAGAGAGAAAAGAC-3′ (forward), DKFZP434O1427 5′-CCAGCTACTCCCTTCTTTGG-3′ (reverse), FIZZ3/RETN (MIM 605565) 5′-TGCAGGATGAAAGCTCTCTG-3′ (forward), FIZZ3/RETN 5′-TCTCCAGGTTTATTTCCAGCTC-3′ (reverse), VNN1 5′-ACGTGGCAATTTTGCTTTTC-3′ (forward), VNN1 5′-TCACCAAGGTAACAGCAGGA-3′ (reverse), TXNDC3 (forward) 5′-GATGAGGCAGCTGAAGAACA-3′, and TXNDC3 (reverse) 5′-CACTAAAGAGAAAAACAGGTTCACA-3′. The PCR conditions were 45 cycles of 94°C for 1 min, 50°C–55°C for 1 min, and 72°C for 1 min. PCR amplification of β-actin–specific cDNA was used as a positive control, with the primers 5′-CGGGAAATCGTGCGTGACAT-3′ (forward) and 5′-GAACTTTGGGGGATGCTCGC-3′ (reverse) and 20 cycles of 95°C for 1 min, 65°C for 2 min, and 72°C for 1.5 min. PCR products were visualized via agarose gel electrophoresis.

Bioinformatic and Statistical Analyses

Differences in mRNA expression levels between patients with OA and healthy controls were assessed via Affymetrix technology. For bioinformatics analyses, we combined two different software tools, the Affymetrix Data Mining Tool (DMT) (v. 3.01) and Genes@work (v. October 2002). With use of the DMT, comparisons were made between the signal log ratios of individual chips, followed by two-sided Mann-Whitney tests, to identify P values <.05. The only genes considered for further analysis were those that showed a statistically significant differential expression for 50% of the comparisons made for CD4+, CD8+, and CD14+ cells, and for 100% of the comparisons made for CD19+ B cells. For the Genes@work analysis, we preselected “present” genes, performed pattern analyses based on the signal values, and chose the pattern containing the highest number of genes.

We assessed SNP associations by performing the χ2 test, including Yate’s continuity correction, based on the number of alleles. P values were subsequently corrected in accordance with Bonferroni. Testing for Hardy-Weinberg equilibrium (HWE) was performed using the software Arlequin v2.0.

Results

Transcriptome Analysis of PBMC Reveals 26 Genes Differentially Expressed in Patients with OA

To identify genes differentially expressed in patients with OA and in healthy controls, we performed transcriptome analyses on PBMC from six patients and six age-matched controls. In detail, we isolated the CD4+ and the CD8+ T cells, the CD19+ B cells, and the CD14+ monocytes, to >98% purity each. From these 48 cell populations, first-strand cRNA was synthesized. For the controls, two individual cRNAs of each cell type and cell type–specific pools were hybridized to HG-U133A gene chips. For the patients, our goal was to hybridize as many individual cRNAs as possible. However, since some individual yields of mRNA were <5 μg, samples had to be pooled. In total, 26 HG-U133A gene chips were probed, as outlined in table 5.

Table 5
Summary of Microarrays Probed

For bioinformatic analyses, we combined two different software programs, to generate lists of genes either up- or down-regulated in patients with OA. We used the DMT to calculate statistically significant expression differences between patients and controls. The second software program, Genes@work, was used to search for similarities in gene expression among patients or controls and to produce lists of genes for which the expression is similar among the patients but different from the controls. Since both methods yield long lists of differentially expressed genes, we formed the intersections to lend more weight to genes identified via two different methods. Figure 1 illustrates our results. In total, the DMT revealed 210 differentially expressed genes for the CD4, 484 for the CD8, 376 for the CD14, and 90 for the CD19+ cells. In contrast, larger lists resulted from the Genes@work analysis, yielding 555 differentially expressed genes for the CD4, 2,329 for the CD8, 2,208 for the CD14, and 3,978 for the CD19+ cells. The intersections contain 86, 322, 181, and 86 genes, respectively.

Figure  1
Combination of two software tools, which reduces the number of differentially expressed candidate genes and lends more weight to genes identified via two different methods. For each cell type, lists of differentially expressed genes were compiled using ...

It was our goal to identify among these differentially expressed genes those candidates that carry cis-regulatory gene polymorphisms. Assuming that cis-regulation will be manifested in any cell type expressing the gene of interest, we disregarded all the genes that, even though present in all four cell types (see table 6), did not show differential expression in all cell types. However, we did allow for cell type–specific differential gene expression, provided that these genes were expressed exclusively in that particular cell type. To avoid mere inflammation markers, we excluded those genes that also showed a differential expression in PBMC from patients suffering from chronic inflammatory joint diseases (A. Gruetzkau, personal communication). We thus reduced the list of candidate genes to the 26 listed in table 7. The signal values for these 26 genes obtained from all gene chips are listed in table 6. The mean fold changes were 1.37 for the CD19+ B cells, 1.48 for the CD4+ T helper cells, 1.53 for the CD8+ cytotoxic T cells, and 1.72 for the CD14+ monocytes. The corresponding SDs were 0.73, 0.76, 1.06, and 0.57, respectively.

Table 6
Signal Values and Detection Calls Obtained from All Gene Chips[Note]
Table 7
Differentially Expressed Genes in OA

Figure 2 outlines the experimental workflow pursued to identify cis-regulation and disease association among the 26 candidate genes. First, the genomic DNA of the six patients and six controls were screened for allelic differences in the 5′ and 3′ UTRs of all candidate genes. Any SNP detected was considered for pyrosequencing, which would reveal allelic imbalances and proof of cis-regulation. Moreover, SNPs in the 5′ regions were directly analyzed for disease association, provided that they showed a disparate distribution among the six patients and six controls that correlated with the differential expression seen on the gene chips. Disease association then prompted the analysis of gene expression in articular chondrocytes, to strengthen a functional relevance for the pathogenesis of OA. Genes carrying a disease-associated SNP and expressed in chondrocytes will be considered for further functional analysis. This analysis could be overexpression or knockdown in chondrocytic cells, for the investigation of the genes' impact on apoptosis and cytokine response. Genes showing neither disparate promoter SNP distribution nor disease association underwent no further analysis.

Figure  2
Workflow for the identification of cis-regulated and disease-associated genes among the 26 candidates. Our screening focuses on allelic expression imbalances, disease-associated promoter polymorphisms, and expression in articular chondrocytes.

Analyses of Large Cohorts Reveal Disease Associations for 5′ SNPs in the RHOB and TXNDC3 Genes

To identify new polymorphisms or to confirm previously published ones in the 5′ regions of our candidate genes, we sequenced the genomic DNA from the six patients and the six controls. In detail, we sequenced up to 700 bp upstream of the respective transcriptional start sites, and we also sequenced the 5′ UTRs if polymorphisms had been indicated. For seven genes, no polymorphisms were found. All the other 5′ SNPs were tested, patient by patient, for a correlation with the respective expression signals (see table 6). Figure 3 summarizes the most promising results. Whereas, for RHOB, BLK, CIRBP, and VNN1, an increased expression in the patients correlates with an increased frequency of the minor promoter allele, the situation is reversed for DKFZP434O1427. Here, the decreased expression in the six patients is paralleled by a decreased frequency of the minor promoter allele. For FIZZ3/RETN, an increased expression in the patients correlates with a decreased frequency of the minor allele. TXNDC3 represents yet another situation, since the gene is switched off in the controls, as shown by mean signal values of 79.6 in monocytes. However, in the patients, elevated signal values directly correlate with the presence of the T allele at position 428. Signal values are 96.5 and 99 for the CC, 178.6 for the CT, and 208.5 for the TT genotype (see table 6).

Figure  3
Correlation between the presence of promoter SNPs and expression signals, which hints at disease-associated regulatory polymorphisms. Expression signals from CD14+ monocytes (RHOB, CIRBP, DKFZP434O1427, FIZZ3/RETN, and VNN1) and CD19+ B cells (BLK) were ...

To screen a large cohort of patients with OA for an association with any of these 5′ SNPs, we established high-throughput PCR assays for the A/G SNP at position −165 of RHOB, for the C/T SNP at position −1218 of BLK, for the G/A SNP at position −342 of CIRBP, for the G/T SNP at position −81 of DKFZP434O1427, for the C/G SNP at position −180 of FIZZ3/RETN, and for the C/T SNP at position 428 of TXNDC3. For the A/C SNP at position −137 of VNN1, a combined PCR/restriction assay was established. A total of 171 patients and 182 controls were analyzed, and table 8 summarizes the results. Allele distributions for all controls, except for DKFZP434O1427, and for all patients, except for RHOB and VNN1, conform to HWE. Whereas no disease association was found for BLK, CIRBP, DKFZP434O1427, FIZZ3/RETN, and VNN1, statistically significant associations with OA resulted for the −165 promoter SNP of RHOB and for the 428 SNP of TXNDC3, with corrected P values of .0007 each.

Table 8
5′ SNPs in the RHOB and TXNDC3 Genes Show Association with OA

Gene Expression in Chondrocytes Supports a Role of Candidate Genes in the Pathogenesis of OA

Because our initial transcriptome analysis was performed on PBMC, formal proof of gene expression in chondrocytes would lend weight to the role of the candidate genes in the pathogenesis of OA. To that extent, we extracted mRNA from the cartilage of patients with OA undergoing joint replacement surgery, transcribed it into cDNA, and assayed for the presence of BLK-, CIRBP-, DKFZP434O1427-, FIZZ3/RETN-, VNN1-, and TXNDC3-specific transcripts. RHOB has been shown elsewhere to be expressed in articular chondrocytes.21 Figure 4 summarizes the data. Whereas CIRBP, DKFZP434O1427, and TXNDC3 are expressed abundantly, BLK, FIZZ3/RETN, and VNN1 are expressed at lower levels, and sequence-specific PCR products could be visualized only in samples with high concentrations of cDNA. We thus conclude that our differentially expressed candidate genes RHOB, BLK, CIRBP, DKFZP434O1427, FIZZ3/RETN, VNN1, and TXNDC3 are expressed in articular chondrocytes.

Figure  4
Gene expression in chondrocytes supporting a role for candidate genes in the pathogenesis of OA. RNA isolated from the cartilage of patients with OA undergoing joint replacement surgery was transcribed into cDNA and was assayed for the presence of specific ...

Allelic Expression Imbalances Confirm Cis-Regulation for RHOB and TXNDC3

To provide direct proof for cis-regulation, we screened for expression imbalances. We used the pyrosequencing method, to quantify the amount of transcript arising from each allele. The requirements for this method—a suitable SNP allowing for the discrimination of both alleles on the cDNA level and individuals heterozygous for this particular SNP—were met by 16 of our candidate genes. However, because of homopolymeric stretches close to the SNP, sequences leading to template loops during the assay, absence of sufficient amounts of mRNA, and very low frequency of that particular SNP allowed for the successful analysis of RHOB, BLP2, CIAS1, and TXNDC3 only. Table 9 lists the dbSNP accession numbers of the SNPs used, their positions with respect to the transcriptional start site, the heterozygous donors, and the relative expressions arising from the sets of two alleles. Whereas no allelic imbalances could be observed for BLP2 and CIAS1, differential expression on the allele level and, thus, cis-regulation is confirmed for both RHOB and TXNDC3.

Table 9
RHOB and TXNDC3 Show Allelic Imbalances

Absence of Allelic Imbalances and On/Off–Switching of Genes Suggests Trans-Regulation

The absence of allelic imbalances for BLP2 and CIAS1 suggest trans-regulatory elements controlling the gene expression (table 4). Alternative evidence for transregulation is provided for FBXL8 (MIM 609077), MGAM (MIM 154360), and TXNDC3. These genes are switched off in the healthy donors, as indicated by “absent” and “marginal” expression signals on the gene chips. At the same time, they are switched on in the patients (see table 6). Interestingly, TXNDC3 gene regulation occurs both in cis and in trans.

Discussion

Ever since differential gene expression has been implicated in driving evolutionary processes,16 regulatory gene polymorphisms have gained increasing attention. Although the differential expression of genes allows for a flexible response to a changing environment and may thus be beneficial for the population as a whole, adverse effects for the individual may occur.22 Indeed, regulatory gene polymorphisms within the IL1 locus have been described elsewhere to be associated with OA.18,19

Our approach to identify novel OA-associated regulatory gene polymorphisms is based on an initial transcriptome analysis used to define genes that are differentially expressed between patients and controls. We chose PBMC for our transcriptome analysis for two reasons: (1) Patient and control material is easily accessible and can be obtained under identical conditions, and (2) even though gene expression may be strictly cell type–specific, numerous genes are expressed in a spatially less-exquisite fashion. Indeed, data from our own lab and the Affymetrix technical notes confirm that 40%–60% of all genes represented on the HG-U133A gene are transcribed in any tissue or cell type tested.23 The overlap of genes simultaneously expressed in various cell types is all the larger the closer the cells are related. Indeed, both immune cells and chondrocytes are of mesodermal origin, the development of the articular cartilage/bone system and of the adaptive immune system is phylogenetically tightly linked, and both systems use the same cytokines and growth factors as messenger molecules.2 Our approach is, therefore, likely to identify genes associated with the systemic character of OA but is bound to miss those expressed in a cartilage-specific manner. Indeed, IL-1β was upregulated in some patients’ monocytes; however, the difference between patients and controls was not significant enough to pass our stringent bioinformatics analyses.

We identified 26 genes that are differentially expressed between patients with OA and controls. Whereas RHOB and S100B (MIM 176990) have been directly linked to OA before,21,24 GH1 (MIM 139250) has been shown to be involved in chondrocyte differentiation in the growth plate and to be associated with short stature (SS [MIM 604271]).25 Other genes—among them RHOB, BLP2, CIAS1, DUSP1 (MIM 600714), DYSF (MIM 603009), and S100B—have been implicated in apoptosis and regulation of cell cycle and repair.21,2630 A link to the OA characteristic of imbalance of cartilage degeneration and regeneration is, therefore, plausible. Other feasible candidates are transcription and translation factors like DKFZP434O1427, FBXL8, EIF4A1 (MIM 602641), and SUI1,3134 even though their specific targets need to be identified before the link to OA can directly be explained. Another candidate, VNN1, is a glycosyl-phosphatidylinositol–anchored cell surface molecule, which has, so far, been implicated in the homing of T-cell progenitors to the thymus and in sexual development in the mouse.35,36 Interestingly, in mouse testis, VNN1 is regulated by Sox-9, a transcription factor involved in chondrocytic differentiation during development and still expressed in articular chondrocytes.37,38 An additional interesting candidate is TXNDC3, which shows cis-regulation and carries a 5′ SNP with a statistically significant association with OA. Likewise, the thioredoxin protein NM23-H8 encoded by TXNDC3 has been described elsewhere to be expressed in testis.39,40 Here, we add the expression in monocytes and chondrocytes; however, the exact role of TXNDC3 for the pathogenesis of OA needs to be further investigated.

To identify regulatory gene polymorphisms responsible for the differential expression, we pursued two different routes. The first one focused directly on polymorphisms in the 5′ regions of the genes, which were screened (1) for a disparate distribution among patients and controls and (2) for a correlation between the SNP haplotype and the signal value. Even though the analysis of six patients and six controls uses too few subjects for any firm conclusions to be drawn, we demanded an obvious correlation between the presence of a regulatory gene polymorphism and the differential expression documented in signal values. This correlation is graphically documented for six genes and their SNPs in figure 3. These six SNPs and the 5′ SNP of TXNDC3 were subsequently analyzed for disease associations, with the use of larger cohorts. Interestingly, except for the DKFZP434O1427 gene, the tendencies for the rare alleles found in the six patients and six controls initially analyzed were again reflected in the larger cohorts. The 5′ SNP in the TXNDC3 gene is located in the second exon, which is noncoding, and the direct correlation of the T allele with an elevated expression hints either at a tight linkage disequilibrium with a regulatory SNP or at some regulatory function. The latter is supported by the observation that TXNDC3 transcripts seem to be alternatively spliced in monocytes and chondrocytes, with the chondrocyte transcripts lacking exon 2 (data not shown). In total, we identified two OA-associated polymorphisms in the genes encoding RHOB and TXNDC3. And, whereas no direct statistical significance could be observed for the other five genes tested, the genotype distributions for VNN1 and DKFZP434O1427 and the resulting differences in the Hardy-Weinberg disequilibria between patients and controls suggest that the analysis of larger cohorts may yield more disease associations. The second route pursued the identification of OA-associated regulatory gene polymorphisms and focused on differential expression detectable on the allele level. To that end, we performed pyrosequencing analyses and used SNPs in the transcribed regions of the genes, to differentiate both alleles. We did not, however, assume that a SNP in the 3′ UTR might be the regulatory polymorphism itself, nor do we know about any linkage disequilibrium with a 5′ regulatory SNP. Any negative result could consequently be a false negative one. Even though pyrosequencing is a very powerful method, there are numerous technical restrictions, and, among our 26 candidates, pyrosequencing analysis could be performed for 4 genes only. Two of them, RHOB and TXNDC3, showed allelic imbalances and, thus, direct proof of a cis-regulated gene expression. In contrast, BLP2 and CIAS1 showed a balanced expression hinting at trans-regulation. It should be mentioned, though, that balanced expression per se does not exclude the possibility of cis-regulation and differences in allelic expression in individuals not tested here. Supporting trans-regulation for BLP2 and CIAS1 are the lack of suggestive 5′ polymorphisms in both genes, yet it was beyond the scope of this article to scan more distantly for SNPs. Also beyond scope was the follow up on trans-regulated gene expression.

In summary, we identified 26 genes that are differentially expressed in OA patients and that may contribute to the pathogenesis of the disease. We confirmed allelic imbalances and, hence, a cis-regulated gene expression for TXNDC3 and RHOB. For both, 5′ SNPs that show a statistically significant association with OA have been identified. Whether these SNPs are true regulatory gene polymorphisms needs to be confirmed with molecular methods. The elucidation of the specific function of the various genes in the context of OA will help the understanding of the pathogenesis of the disease and will hint at novel therapeutical approaches.

Acknowledgments

This work was supported by grants from the Bundesministerium für Bildung und Forschung (BerlInflame) and the Deutsche Forschungs-Gemeinschaft (Mu844/7-1) (to B.M.). The authors are grateful to Robert Goertsches, for help with the HWE, and to Peter Nürnberg, for helpful discussion.

Web Resources

Accession numbers and URLs for data presented herein are as follows:

Online Mendelian Inheritance in Man (OMIM), http://www.ncbi.nlm.nih.gov/Omim (for OA, VNN1, RHOB, CIAS1, TXNDC3, CIRBP, BLK, FIZZ3/RETN, FBLX8, MGAM, S100B, GH1, SS, DUSP1, DYSF, and EIF4A1)

References

1. Buckwalter JA, Saltzman C, Brown T, Schurman DJ (2004) The impact of osteoarthritis: implications for research. Clin Orthop 427:S6–S15 [PubMed]
2. Muller B (2002) Cytokine imbalance in non-immunological chronic disease. Cytokine 18:334–339 [PubMed] [Cross Ref]10.1006/cyto.2002.0882
3. Aigner T, Rose J, Martin J, Buckwalter J (2004) Aging theories of primary osteoarthritis: from epidemiology to molecular biology. Rejuvenation Res 7:134–145 [PubMed] [Cross Ref]10.1089/1549168041552964
4. Lorenzo P, Bayliss MT, Heinegard D (2004) Altered patterns and synthesis of extracellular matrix macromolecules in early osteoarthritis. Matrix Biol 23:381–391 [PubMed] [Cross Ref]10.1016/j.matbio.2004.07.007
5. Sharif M, Whitehouse A, Sharman P, Perry M, Adams M (2004) Increased apoptosis in human osteoarthritic cartilage corresponds to reduced cell density and expression of caspase 3. Arthritis Rheum 50:507–515 [PubMed] [Cross Ref]10.1002/art.20020
6. Tanzi RE (1999) A genetic dichotomy model for the inheritance of Alzheimer’s disease and common age-related disorders. J Clin Invest 104:1175–1179 [PMC free article] [PubMed]
7. Felson DT (2004) An update on the pathogenesis and epidemiology of osteoarthritis. Radiol Clin North Am 42:1–9, v [PubMed] [Cross Ref]10.1016/S0033-8389(03)00161-1
8. Spector TD, MacGregor AJ (2004) Risk factors for osteoarthritis: genetics. Osteoarthritis Cartilage 12:S39–S44 [PubMed] [Cross Ref]10.1016/j.joca.2003.09.005
9. Peach CA, Carr AJ, Loughlin J (2005) Recent advances in the genetic investigation of osteoarthritis. Trends Mol Med 11:186–191 [PubMed] [Cross Ref]10.1016/j.molmed.2005.02.005
10. Knowlton RG, Katzenstein PL, Moskowitz RW, Weaver EJ, Malemud CJ, Pathria MN, Jimenez SA, Prockop DJ (1990) Genetic linkage of a polymorphism in the type II procollagen gene (COL2A1) to primary osteoarthritis associated with mild chondrodysplasia. N Engl J Med 322:526–530 [PubMed]
11. Horton WE Jr, Lethbridge-Cejku M, Hochberg MC, Balakir R, Precht P, Plato CC, Tobin JD, Meek L, Doege K (1998) An association between an aggrecan polymorphic allele and bilateral hand osteoarthritis in elderly white men: data from the Baltimore Longitudinal Study of Aging (BLSA). Osteoarthritis Cartilage 6:245–251 [PubMed] [Cross Ref]10.1053/joca.1998.0117
12. Leppavuori J, Kujala U, Kinnunen J, Kaprio J, Nissila M, Heliovaara M, Klinger N, Partanen J, Terwilliger JD, Peltonen L (1999) Genome scan for predisposing loci for distal interphalangeal joint osteoarthritis: evidence for a locus on 2q. Am J Hum Genet 65:1060–1067 [PMC free article] [PubMed]
13. Stefansson SE, Jonsson H, Ingvarsson T, Manolescu I, Jonsson HH, Olafsdottir G, Palsdottir E, Stefansdottir G, Sveinbjornsdottir G, Frigge ML, Kong A, Gulcher JR, Stefansson K (2003) Genomewide scan for hand osteoarthritis: a novel mutation in matrilin-3. Am J Hum Genet 72:1448–1459 [PMC free article] [PubMed]
14. Forster T, Chapman K, Marcelline L, Mustafa Z, Southam L, Loughlin J (2004) Finer linkage mapping of primary osteoarthritis susceptibility loci on chromosomes 4 and 16 in families with affected women. Arthritis Rheum 50:98–102 [PubMed] [Cross Ref]10.1002/art.11427
15. Hunter DJ, Demissie S, Cupples LA, Aliabadi P, Felson DT (2004) A genome scan for joint-specific hand osteoarthritis susceptibility: the Framingham Study. Arthritis Rheum 50:2489–2496 [PubMed] [Cross Ref]10.1002/art.20445
16. Enard W, Khaitovich P, Klose J, Zollner S, Heissig F, Giavalisco P, Nieselt-Struwe K, Muchmore E, Varki A, Ravid R, Doxiadis GM, Bontrop RE, Paabo S (2002) Intra- and interspecific variation in primate gene expression patterns. Science 296:340–343 [PubMed] [Cross Ref]10.1126/science.1068996
17. Baumgart M, Moos V, Schuhbauer D, Muller B (1998) Differential expression of major histocompatibility complex class II genes on murine macrophages associated with T cell cytokine profile and protective/suppressive effects. Proc Natl Acad Sci USA 95:6936–6940 [PMC free article] [PubMed] [Cross Ref]10.1073/pnas.95.12.6936
18. Moos V, Rudwaleit M, Herzog V, Hohlig K, Sieper J, Muller B (2000) Association of genotypes affecting the expression of interleukin-1beta or interleukin-1 receptor antagonist with osteoarthritis. Arthritis Rheum 43:2417–2422 [PubMed] [Cross Ref]10.1002/1529-0131(200011)43:11<2417::AID-ANR7>3.0.CO;2-R
19. Meulenbelt I, Seymour AB, Nieuwland M, Huizinga TW, van Duijn CM, Slagboom PE (2004) Association of the interleukin-1 gene cluster with radiographic signs of osteoarthritis of the hip. Arthritis Rheum 50:1179–1186 [PubMed] [Cross Ref]10.1002/art.20121
20. Krenn V, Morawietz L, Haupl T, Neidel J, Petersen I, Konig A (2002) Grading of chronic synovitis—a histopathological grading system for molecular and diagnostic pathology. Pathol Res Pract 198:317–325 [PubMed] [Cross Ref]10.1078/0344-0338-5710261
21. Gebhard PM, Soder S, Bau B, Aigner T (2004) Down-regulation of the GTPase RhoB might be involved in the pre-apoptotic phenotype of osteoarthritic chondrocytes. Front Biosci 9:827–833 [PubMed]
22. Mitchison NA, Muller B, Segal RM (2000) Natural variation in immune responsiveness, with special reference to immunodeficiency and promoter polymorphism in class II MHC genes. Hum Immunol 61:177–181 [PubMed] [Cross Ref]10.1016/S0198-8859(99)00141-X
23. Affymetrix (2003) Affymetrix technical note: design and performance of the GeneChip Human Genome U133 Plus 2.0 and Human Genome 133A 2.0 Arrays. (http://www.affymetrix.com/support/technical/technotes/hgu133_p2_technote.pdf) (accessed March 21, 2006)
24. Mohr W, Kuhn C, Pelster B, Wessinghage D (1985) S-100 protein in normal, osteoarthrotic, and arthritic cartilage. Rheumatol Int 5:273–277 [PubMed] [Cross Ref]10.1007/BF00541355
25. Millar DS, Lewis MD, Horan M, Newsway V, Easter TE, Gregory JW, Fryklund L, Norin M, Crowne EC, Davies SJ, Edwards P, Kirk J, Waldron K, Smith PJ, Phillips JA 3rd, Scanlon MF, Krawczak M, Cooper DN, Procter AM (2003) Novel mutations of the growth hormone 1 (GH1) gene disclosed by modulation of the clinical selection criteria for individuals with short stature. Hum Mutat 21:424–440 [PubMed] [Cross Ref]10.1002/humu.10168
26. Kajkowski EM, Lo CF, Ning X, Walker S, Sofia HJ, Wang W, Edris W, Chanda P, Wagner E, Vile S, Ryan K, McHendry-Rinde B, Smith SC, Wood A, Rhodes KJ, Kennedy JD, Bard J, Jacobsen JS, Ozenberger BA (2001) Beta-amyloid peptide-induced apoptosis regulated by a novel protein containing a g protein activation module. J Biol Chem 276:18748–18756 [PubMed] [Cross Ref]10.1074/jbc.M011161200
27. Rosengren S, Hoffman HM, Bugbee W, Boyle DL (2005) Expression and regulation of cryopyrin and related proteins in rheumatoid arthritis synovium. Ann Rheum Dis 64:708–714 [PMC free article] [PubMed] [Cross Ref]10.1136/ard.2004.025577
28. Unoki M, Nakamura Y (2001) Growth-suppressive effects of BPOZ and EGR2, two genes involved in the PTEN signaling pathway. Oncogene 20:4457–4465 [PubMed] [Cross Ref]10.1038/sj.onc.1204608
29. Anderson LV, Davison K, Moss JA, Young C, Cullen MJ, Walsh J, Johnson MA, Bashir R, Britton S, Keers S, Argov Z, Mahjneh I, Fougerousse F, Beckmann JS, Bushby KM (1999) Dysferlin is a plasma membrane protein and is expressed early in human development. Hum Mol Genet 8:855–861 [PubMed] [Cross Ref]10.1093/hmg/8.5.855
30. Ikura M, Osawa M, Ames JB (2002) The role of calcium-binding proteins in the control of transcription: structure to function. Bioessays 24:625–636 [PubMed] [Cross Ref]10.1002/bies.10105
31. von Rotz RC, Kins S, Hipfel R, von der Kammer H, Nitsch RM (2005) The novel cytosolic RING finger protein dactylidin is up-regulated in brains of patients with Alzheimer’s disease. Eur J Neurosci 21:1289–1298 [PubMed] [Cross Ref]10.1111/j.1460-9568.2005.03977.x
32. Ota T, Suzuki Y, Nishikawa T, Otsuki T, Sugiyama T, Irie R, Wakamatsu A, et al (2004) Complete sequencing and characterization of 21,243 full-length human cDNAs. Nat Genet 36:40–45 [PubMed] [Cross Ref]10.1038/ng1285
33. Chan CC, Dostie J, Diem MD, Feng W, Mann M, Rappsilber J, Dreyfuss G (2004) eIF4A3 is a novel component of the exon junction complex. Rna 10:200–209 [PMC free article] [PubMed] [Cross Ref]10.1261/rna.5230104
34. Sheikh MS, Fernandez-Salas E, Yu M, Hussain A, Dinman JD, Peltz SW, Huang Y, Fornace AJ Jr (1999) Cloning and characterization of a human genotoxic and endoplasmic reticulum stress-inducible cDNA that encodes translation initiation factor 1(eIF1(A121/SUI1)). J Biol Chem 274:16487–16493 [PubMed] [Cross Ref]10.1074/jbc.274.23.16487
35. Bernard P, Tang P, Liu S, Dewing P, Harley VR, Vilain E (2003) Dimerization of SOX9 is required for chondrogenesis, but not for sex determination. Hum Mol Genet 12:1755–1765 [PubMed] [Cross Ref]10.1093/hmg/ddg182
36. Aurrand-Lions M, Galland F, Bazin H, Zakharyev VM, Imhof BA, Naquet P (1996) Vanin-1, a novel GPI-linked perivascular molecule involved in thymus homing. Immunity 5:391–405 [PubMed] [Cross Ref]10.1016/S1074-7613(00)80496-3
37. de Crombrugghe B, Lefebvre V, Behringer RR, Bi W, Murakami S, Huang W (2000) Transcriptional mechanisms of chondrocyte differentiation. Matrix Biol 19:389–394 [PubMed] [Cross Ref]10.1016/S0945-053X(00)00094-9
38. Wilson MJ, Jeyasuria P, Parker KL, Koopman P (2005) The transcription factors steroidogenic factor-1 and SOX9 regulate expression of Vanin-1 during mouse testis development. J Biol Chem 280:5917–5923 [PubMed] [Cross Ref]10.1074/jbc.M412806200
39. Sadek CM, Damdimopoulos AE, Pelto-Huikko M, Gustafsson JA, Spyrou G, Miranda-Vizuete A (2001) Sptrx-2, a fusion protein composed of one thioredoxin and three tandemly repeated NDP-kinase domains is expressed in human testis germ cells. Genes Cells 6:1077–1090 [PubMed] [Cross Ref]10.1046/j.1365-2443.2001.00484.x
40. Matsuo Y, Akiyama N, Nakamura H, Yodoi J, Noda M, Kizaka-Kondoh S (2001) Identification of a novel thioredoxin-related transmembrane protein. J Biol Chem 276:10032–10038 [PubMed] [Cross Ref]10.1074/jbc.M011037200

Articles from American Journal of Human Genetics are provided here courtesy of American Society of Human Genetics
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...