Identification of differentially expressed genes in fibroblasts derived from patients with Dupuytren's Contracture

Dupuytren's contracture (DC) is the most common inherited disease of connective tissue in humans [1] and an autosomal dominant form of the disease was recently mapped to the long arm of chromosome 16 [2]. The disease is characterized by the appearance of small nodules of hyperproliferative cells within the palmar fascia that, over time, give rise to large bands of contracted, collagen-rich fibrotic tissue (diseased cords), a hallmark of the disease [3, 4]. If left untreated, this disease may impose severe limitations on hand function. It is a familial disorder that is highly prevalent in individuals of Northern European extraction [5, 6] and is observed less frequently among other ethnicities [7]. The manifestations of Dupuytren's are usually noticeable between the ages of 40 to 60 and with a higher incidence in men than in women [8].

Although the pathogenesis of DC disease has not been fully explained it is clear that genetics plays an important role; traumatic factors may also be important and may explain the male gender bias of the disease. In addition, a number of metabolic conditions that negatively affect wound healing processes in general have been statistically associated with DC including: diabetes mellitus (8%), alcoholism (10%), smoking, and HIV infection [9–12]. Finally, there is a puzzling connection with epilepsy (2%).

The mainstay of treatment is surgery, but no specific surgical approach has proved to be consistently more effective than others at curing this condition as the trauma associated with surgery itself can lead to recurrence. Possible alternatives to surgery include injection of steroids, γ-interferon [13, 14], use of creams based on vitamin E, dimethylsulphoxides, drugs inducing hypo-uricaemia, ultrasonic therapy [15] and clostridial collagenase injection [16]. However, these medical treatments appear to be either temporary alternatives to surgical intervention with only limited success at best, or are still under clinical assessment.

Previous studies have identified dysregulation among multiple structural proteins in DC, including: type I and type III collagens; the extracellular matrix (ECM) proteins fibronectin, tenascin C, and laminin; as well as matrix metalloproteinases in the diseased fascial cords [17, 18]. Previous studies have also shown that a number of signaling molecules such as transforming growth factor β (TGFβ), epidermal growth factor (EGF) and platelet derived growth factor (PDGF) are differentially regulated in DC [19, 20]. No specific causative gene has yet been identified and identification of susceptibility loci may help to unravel the pathogenesis of this common disease. Our laboratory has investigated the pattern of inheritance of DC in a large Swedish family and has shown that DC is inherited as an autosomal dominant disorder with incomplete penetrance by the end of the fifth decade [2]. A genome-wide scan at a resolution of ~8 cM for all autosomes established linkage to a single 6 cM region between markers D16S419 and D16S3032 on chromosome 16q. In contrast, Bayat et al., (2005) [21] identified a heteroplasmic mitochondrial mutation associated with DC. These investigators have also identified associations between polymorphisms in the human TGFβRI and Zf9 genes with DC [22, 23] raising the possibility that DC may be the common result of multiple disparate genetic lesions.

Previously investigators have conducted differential gene expression profiling of tissues collected from Dupuytren's patients (using normal palmar fascia as controls [24–26]). In the present study we compare the gene expression patterns in fibroblasts derived from DC and carpal tunnel patients, as fibroblasts are believed to play an important role in the genesis of DC and fibroproliferation is one of the hallmarks of the disorder [27, 28]. Since DC has been hypothesized to result from an impaired wound healing response, cellular elements such as myofibroblasts (primarily derived from fibroblasts), which are implicated both in the contraction of wound granulation tissue and the scar-like disease cords that define DC, are attractive targets for study. Fibroblasts/myofibroblasts actively control both extracellular matrix remodeling and scar deposition; processes which are central to wound healing and may be dysregulated in DC. In this context, the present study was designed as an attempt to identify key genes whose expression is significantly altered in the principal cell type of interest during the development of Dupuytren's contracture. We have followed the recommendations of the Microarray Quality Control Consortium (MAQC) [29], which recommends the use of multiple microarray platforms for the quantitative characterization of gene expression. Accordingly, we used both the GE Code Link™ and Illumina™ systems for cross validation.

The present study was designed to identify differences in the gene expression patterns between primary Dupuytren's contracture-derived fibroblasts and primary carpal tunnel-derived control fibroblasts. This was achieved employing two different microarray platforms, the GE CodeLink™ and Illumina™, using RNA extracted from primary cultures of carpal tunnel- and DC-derived fibroblasts within passages 2–5 of tissue culture (P2–P5). Furthermore, several high interest candidate genes identified by analysis of the microarray data were further validated using quantitative real time RT-PCR, and the online Ingenuity Knowledge Based Analysis system was used to identify signaling pathways that could serve as potential therapeutic targets to limit the recurrence of DC disease.

Gene Products Validated Using Real time RT-PCR

Real-time RT-PCR was used to confirm the expression levels of three candidate genes: proteoglycan 4 (PRG4), fibulin-1 (FBLN-1) transcript variant D and type XV collagen alpha 1 chain (Figure 1, 2 &3). These genes were all identified as downregulated in the data sets of both microarray based platforms (although SAM analysis picked them out only in one). Real-time RT-PCR results showed that PRG4 mRNA levels in DC fibroblasts was less than 10% of that in control fibroblasts. Similarly the FBLN-1 and collagen genes showed expression levels of less than 40% of those observed in the controls. Thus, in all three cases, the quantitative RT-PCR results confirmed the microarray findings.

The microarray analyses of DC fibroblasts versus controls reported here identified far more genes as downregulated in the disease samples than upregulated. This is consistent with previous microarray findings from DC tissues wherein the authors similarly report substantially more downregulated genes than upregulated genes [31], although a countervailing report identified more genes as upregulated than downregulated on tissues derived from DC [24]. The present study using fibroblast cultures did not identify some of the previously reported genes that might be involved in DC, eg. alpha-1 integrins, MafB [25] or various MMPs/TIMPS [26]. This may be due to differences between tissues (where multiple cell types may be present, possibly interacting with each other) and cultured cells expressing a solitary phenotype. Our results across two microarray platforms are internally consistent and indicate that fibroblast cell populations in DC generally preferentially downregulate selected gene expression relative to control cells.

Dupuytren's disease presents two fibrotic structures that are clearly identifiable: the nodule and the cord [32, 4]. The nodule is a relatively vascular tissue containing a dense population of fibroblasts, with a high proportion being myofibroblasts. In contrast, the cord is a collagen-rich structure that is relatively avascular and acellular, and with a lesser (but still significant) abundance of myofibroblasts. The slow progression of the disease, combined with a lack of awareness as to eventual severity and that surgical resection is rarely performed in the absence of pronounced finger contracture, limits the availability of nodule tissue for experimental study.

In the present study, we identified by microarray analyses candidate genes from cord-derived fibroblasts that might serve as targets for intervention to limit the recurrence of palmar fascia fibrosis. One of the hallmarks of DC is abnormal proliferation of fibroblasts [33, 34] with excessive production of collagen (predominantly type III collagen) along with elevated levels of other ECM proteins, mainly fibronectin, tenascin and laminin [17, 35, 36]. Higher expression of laminin and tenascin has been reported in nodular tissues [35] and an elevated level of fibronectin expression has been found in both nodular tissues and in cord derived fibroblasts [36]. Since alterations in ECM proteins appear to play an important role in DC, we looked for differences in the gene expression patterns of several ECM proteins between normal and diseased derived palmar fascia fibroblasts. Intriguingly, we did identify changes in the expression of other ECM proteins including, types XIV and XV collagens, PRG4 and FBLN1 transcript variant D which have not been previously associated with DC. The three fibroblast-derived ECM genes were initially identified through SAM analysis of the Illumina™ data, but were not identified by SAM using the CodeLink™ data. However, direct inspection of the CodeLink raw data did show a statistically significant downregulation of these genes that was confirmed by real-time RT-PCR analysis (Table 4 & Figure 1, 2 &3).

Milani et al., (1994) [37] reported that type XV collagen (undulin) is constitutively expressed in normal hepatic liver, but that its levels are elevated in fibrotic liver, suggesting its participation in the rearrangement of connective tissue during fibrotic disease. Since DC is a fibroproliferative disease, high levels of undulin might be expected. Our analysis however, indicates that type XV collagen mRNA levels are relatively lower in Dupuytren's-derived fibroblasts, suggesting that overexpression of this gene product is not universal in fibrotic disorders but may vary with tissue type and location. The downregulation of type XV collagen in DC-derived fibroblasts warrants further study as it has been shown previously that lack of type XV collagen results in skeletal myopathy and cardiovascular defects in mice, demonstrating that type XV collagen can prove a significant contributor in the pathogenesis of musculoskeletal disorders [38].

We also identified two other ECM proteins, PRG4 and FBLN1, as being downregulated in DC-derived fibroblasts. PRG4, also commonly referred to as megakaryocyte stimulating factor, articular superficial zone protein and lubricin, is a multifaceted cytoprotective glycoprotein that independently or additively contributes to boundary lubrication in synovial joints and at articular cartilage-cartilage interfaces [39, 40]. Recently Alazami et al., (2006) [41] have corroborated the earlier findings of Marcelino et al., (1999) [42] demonstrating that camptodactyly-arthropathy-coxavara-pericarditis (CACP) syndrome (where one or more fingers are curved inwards on the palm (flexed) and cannot be straightened) is an autosomal recessive disorder caused by mutations in the PRG4 gene. Moreover, PRG4 deficiency has also been implicated in the pathogenesis of osteoarthritis [43], thus making this an attractive candidate gene for further study. The fibulin-1 gene, which encodes both an extracellular matrix protein and a secreted plasma glycoprotein, has been found to be disrupted in a chromosomal translocation (12;22) which results in haploinsufficiency for the FBLN1-D variant in synpolydactyly, a congenital condition characterized by a union of fingers or toes and/or increase in the number of toes or fingers. Biochemical analyses of synpolydactyly fibroblasts showed significantly reduced levels of FBLN1-D polypeptide incorporated in the ECM [44]. In summary, PRG4, fibulin-1 and type XV collagen are all potential candidates for being involved in the pathogenesis of DC and the biological role of these ECM proteins in this disease warrants further study. Interestingly, Ingenuity Knowledge Based Analysis identified fibulin-1 as a focus gene; it is one of the many genes included in Network 1 and appears to have interaction with other ECM proteins, namely laminin 4, versican (chondritin sulfate proteoglycan 2) along with the transcription factor Sp1 (SP1) and connective tissue growth factor (CTGF).

In addition to the genes discussed above, we identified several others that produced concordant results on the two microarray platforms. These included three metabolic genes involved in alcohol metabolism: aldehyde dehydrogenase 2 family (ALDH2), aldo-keto reductase family, member C1 (AKR1C1), aldehyde dehydrogenase 3 family, member A1 (ALDH3A1), all of which were downregulated in the DC palmar fascia fibroblasts relative to controls and observed previously by Pan et al. (2003) [31]. ALDH2 was identified as a focus gene and listed as a target in various signaling cascades (Table 6) through Ingenuity Knowledge Based Analysis. The analysis also suggests that TNF might increase the expression of ALDH2. The data provides further evidence that these metabolic genes have the potential to play a role in the progression of DC. In our studies, ALDH2 and AKR1C1 were identified by both platforms, whereas ALDH3A1 was not identified by the CodeLink™ platform. However, a close examination of the data points for ALDH3A1 in the CodeLink™ data set does reveal a statistically significant reduction (p < 0.005) with greater than a 3-fold change between the two study groups (Table 4), albeit not enough to satisfy the extremely stringent parameters of SAM analysis. The downregulation of these three genes involved in alcohol metabolism is congruent with the observation that alcohol consumption is linked with Dupuytren's disease.

There is also some evidence to show that a relationship exists between the pathogenesis of Dupuytren's contracture induced by various oxidative stress molecules, including superoxide radical (O_2-), hydrogen peroxide (H₂O₂) and the hydroxyl radical (OH-) [45, 46]. It is hypothesized that the progressive restriction of capillaries with age, smoking and other environmental factors leads to a condition of localized hypoxia resulting in increased levels of xanthine oxidase and the subsequent production of free radicals. In the present study, we found mitochondrial superoxide dismutase 2, cysteine dioxygenase type 1, and selenoprotein P, plasma, 1 (SEPP1) to be downregulated in DC-derived fibroblasts. Low levels of these enzymes may render DC-derived fibroblasts deficient in their ability to detoxify superoxide radicals or to oxidize cysteine residues as an antioxidant defense mechanism [45, 47]. Clusterin, a secreted mammalian chaperone associated with stress-associated cell survival (anti-apoptotic gene) [48] is also downregulated in DC, contrasting with its overexpression in several human cancers such as prostate, breast, and squamous cell carcinoma [49, 50].

Chitinase-3-like 2 (CHI3L2), a secreted glycoprotein component of the ECM which has been linked to early detection and prognosis of ovarian cancer, was identified in this study. CHI3L2 appears to be downregulated in DC as opposed to increased expression seen in ovarian cancer patients [51]. Angiopoietin-like factor (CDT6), a homologue to the angiopoietin gene family, is a glycoprotein that is involved in vascular morphogenesis and maintenance through binding to the vascular endothelial Tie2 receptor [52]. CDT6 is also found to be downregulated in DC-derived fibroblasts. The significance of altered CHI3L2 and CDT6 expression to the progression of Dupuytren's contracture is as yet unclear.

The finding that serum/glucocorticoid regulated kinase (SGK) message is downregulated in DC fibroblasts is contrary to previous studies that document a high level of expression of SGK transcript in fibrosing disorders such as Crohn's disease, fibrosing pancreatitis, diabetic nephropathy, lung fibrosis and liver scirrhosis [53]. F-box protein 2 (FBXO2), which plays a crucial factor in the ubiquitin-mediated degradation of cellular regulatory proteins [54], is also noted to be downregulated in DC-derived fibroblasts. Hydroxysteroid (11-beta) dehydrogenase transcript variant 1 catalyzes the interconversion of biologically inactive glucocorticoid (cortisone) to active glucocorticoid (cortisol); it has been shown to play a predominant role in obesity, type 2 diabetes [55] and hypertension but its role in fibrotic disease is yet to be determined.

Of interest is the observation that 29 of the dysregulated genes identified in the DC fibroblasts are of unknown function, suggesting that entirely new pathways may be operational here that have not been previously characterized. An understanding of these genes and their protein products may provide new insights into DC and other disorders potentially associated with DC. In summary, our study reveals a variety of candidate genes, some of known function, some of unknown function, that together with additional genetic studies will point the way to a better understanding of this most common of human connective tissue disorders.

The present study using fibroblasts derived from DC patients has identified a variety of candidate genes that may be involved in the pathophysiology of palmar fibrosis which may ultimately serve as attractive molecular targets for alternative therapies. Future studies will further correlate the changes in the mRNA levels to the protein expression both at the cellular and tissue level which will provide further clues to intervene and limit the progression/recurrence of DC.