Molecular Vision 2004; 10:758-772 <http://www.molvis.org/molvis/v10/a91/>
Received 12 August 2004 | Accepted 7 October 2004 | Published 7 October 2004		 Download
Reprint

Analysis of transcriptional regulation of the small leucine rich proteoglycans

Elena S. Tasheva,1 Bernward Klocke,2 Gary W. Conrad1
 
 

1Kansas State University, Division of Biology, Manhattan, KS; 2Genomatix Software GmbH, München, Germany

Correspondence to: Elena S. Tasheva, Division of Biology, Ackert Hall, Kansas State University, Manhattan, KS, 66506-4901; Phone: (785) 532-6553; FAX: (785) 532-6653; email: est@ksu.edu

Abstract

Purpose: Small leucine rich proteoglycans (SLRPs) constitute a family of secreted proteoglycans that are important for collagen fibrillogenesis, cellular growth, differentiation, and migration. Ten of the 13 known members of the SLRP gene family are arranged in tandem clusters on human chromosomes 1, 9, and 12. Their syntenic equivalents are on mouse chromosomes 1, 13, and 10, and rat chromosomes 13, 17, and 7. The purpose of this study was to determine whether there is evidence for control elements, which could regulate the expression of these clusters coordinately.

Methods: Promoters were identified using a comparative genomics approach and Genomatix software tools. For each gene a set of human, mouse, and rat orthologous promoters was extracted from genomic sequences. Transcription factor (TF) binding site analysis combined with a literature search was performed using MatInspector and Genomatix' BiblioSphere. Inspection for the presence of interspecies conserved scaffold/matrix attachment regions (S/MARs) was performed using ElDorado annotation lists. DNAseI hypersensitivity assay, chromatin immunoprecipitation (ChIP), and transient transfection experiments were used to validate the results from bioinformatics analysis.

Results: Transcription factor binding site analysis combined with a literature search revealed co-citations between several SLRPs and TFs Runx2 and IRF1, indicating that these TFs have potential roles in transcriptional regulation of the SLRP family members. We therefore inspected all of the SLRP promoter sets for matches to IRF factors and Runx factors. Positionally conserved binding sites for the Runt domain TFs were detected in the proximal promoters of chondroadherin (CHAD) and osteomodulin (OMD) genes. Two significant models (two or more transcription factor binding sites arranged in a defined order and orientation within a defined distance range) were derived from these initial promoter sets, the HOX-Runx (homeodomain-Runt domain), and the ETS-FKHD-STAT (erythroblast transformation specific-forkhead-signal transducers and activators of transcription) models. These models were used to scan the genomic sequences of all 13 SLRP genes. The HOX-Runx model was found within the proximal promoter, exon 1, or intron 1 sequences of 11 of the 13 SLRP genes. The ETS-FKHD-STAT model was found in only 5 of these genes. Transient transfections of MG-63 cells and bovine corneal keratocytes with Runx2 isoforms confirmed the relevance of these TFs to expression of several SLRP genes. Distribution of the HOX-Runx and ETS-FKHD-STAT models within 200 kb of genomic sequence on human chromosome 9 and 500 kb sequence on chromosome 12 also were analyzed. Two regions with 3 HOX-Runx matches within a 1,000 bp window were identified on human chromosome 9; one located between OMD and osteoglycin (OGN)/mimecan genes, and the second located upstream of the putative extracellular matrix protein 2 (ECM2) promoter. The intergenic region between OMD and mimecan was shown to coincide with different patterns of DNAse I hypersensitivity sites in MG-63 and U937 cells. ChiP analysis revealed that this region binds Runx2 in U937 cells (mimecan transcript note detectable), but binds Pitx3 in MG-63 cells (expressing high level of mimecan), thereby demonstrating its functional association with mimecan expression. Upon comparing the predictions of S/MARs on the relevant chromosomal context of human chromosomes 9 and 12 and their rodent equivalents, no convincing evidence was found that the tandemly arranged genes build a chromosomal loop.

Conclusions: Twelve of 13 known SLRP genes have at least one HOX-Runx module match in their promoter, exon 1, intron 1, or intergenic region. Although these genes are located in different clusters on different chromosomes, the common HOX-Runx module could be the basis for co-regulated expression.

Introduction

The process of transcription is the key element in gene expression and, as such, an attractive control point for regulation of gene expression in cell and tissue specific manners. It is not surprising that considerable research has been conducted on elucidating the mechanisms by which genes are regulated [1-7]. Current views of transcriptional regulation incorporate the histone code hypothesis which proposes that different combinations of histone modifications function as recognition signals for transcription factors and that promoters bear the histone code: H3 hyperacetylation and methylation of lysine 4 [1]. Various models demonstrate how the two types of complexes, the nucleosome remodeling complexes of the SWI/SNF (switch/sucrose non-fermentable) type, which use the energy of ATP-hydrolysis to alter histone-DNA contacts, and the enzymatic complexes that modify histones by acetylation, methylation, phosphorylation, and ubiquitinylation participate in chromatin remodeling [1-7]. Models also explain how regulatory motifs act at a distance and involve looping to bring regulatory elements in contact with distant promoters. Regulatory motifs are seen as binding sites for proteins that induce chemical modifications and structural alterations propagating down the fiber [5]. In addition to studying regulation of transcription through a variety of biological and biochemical approaches, recently there has been much interest in the possibility of using bioinformatics approaches to identify gene regulatory elements [8-10]. Large scale cross-species DNA sequence comparisons reveal regions of highly conserved non-coding sequences located upstream of transcription initiation sites (gene promoters), or in introns and intergenic regions (enhancers, silencers, scaffold/matrix attachment region (S/MARs), and locus control regions) [11-15]. Functional analyses of these conserved regions show that they represent cis-regulatory elements that control expression of nearby genes. Data indicate that these regulatory regions have modular structure and that regulatory effects of a control region depend on the specific combination of elements, as well as the order and orientation in which they occur [8,10,16]. The ability of a given sequence-specific transcription factor to interact with both co-activators and co-repressors and/or to recruit histone-modifying proteins is thought to provide a simple means for generating on-off switches in a tissue specific manner during the cell-cycle, and in development [1-4,7]. In addition to the order and orientation of units of transcription regulatory regions, the abundance of transcriptional cofactors in a certain cell type also influences the ability of site specific factors to regulate gene expression [17]. Thus, a combination of bioinformatics and wet lab experiments has emerged as a successful method for detecting cis-regulatory elements in many genomic loci, including those for homeodomain containing transcription factors (HOX), immunoglobulin, β-globin, and IL4/IL13/IL5-cytokine gene clusters [18-21].

The small lecine rich proteoglycans (SLRPs) are a well-known family of secreted proteoglycans present in many connective tissues. Crucial roles of these macromolecules in matrix assembly and modulation of cellular growth have been demonstrated extensively in the literature [22-25]. In the eye, SLRPs have been shown to be equally important for development and maintenance of corneal transparency and for providing the structural link between the neural retina and retinal pigment epithelium [26,27]. Alterations in the balance of corneal SLPGs result in increased hydration and loss of corneal transparency. In vitro data demonstrate that these molecules regulate axon guidance and synapse formation during the development of nervous tissue and the vertebrate retina. Individual roles of several SLRPs have been demonstrated by production of gene knock-out animals. All single or double SLRP-null mice generated so far displayed abnormal collagen fibrillogenesis and developed a variety of diseases such as osteoporosis, osteoarthritis, muscular dystrophy, Ehlers-Danlos syndrome, and corneal pathology, e.g. diseases that appear to result from collagen defects [28-33]. Phenotypic changes in different SLRP-null mice are mild, indicating that these proteins can compensate for one another, as evidenced by an increased amount of lumican in fibromodulin-null mice [31]. Although the molecular bases for these compensatory mechanisms presently are unknown, it is likely that this might occur at the level of transcription. The notion of regulated expression of the SLRPs at the level of transcription is supported by their genomic organization. Thus, 10 of the 13 members of the SLRP gene family are organized in clusters on three chromosomes: opticin (OPTC), proline arginine rich end leucine rich repeat protein (PRELP), and fibromodulin (FMOD) on human chromosome 1q3 (their syntenic equivalents on mouse chromosome 1); asporin (ASPN), osteoadherin/osteomodulin (OMD), and OGN/mimecan on human chromosome 9q2 (mouse chromosome 13); dermatan sulfate proteoglycan 3/epiphycan (DSPG3), keratocan (KERA), lumican (LUM), and decorin (DCN), on human chromosome 12q (mouse chromosome 10, Figure 1). Such clustered chromosomal arrangements resemble the organization of other genomic loci that have been shown to be precisely regulated in time and space to ensure normal development. For example, mammals have 39 HOX genes that are clustered in four genomic loci, and their spatial and temporal transcriptional activation correspond to the gene order along the cluster [12,20]. Similarly, the human β-globin locus consists of five erythroid specific genes that are expressed sequentially during development [34]. As with other gene clusters, elucidation of the mechanisms of transcriptional regulation of the SLRP gene clusters may provide information about important control points for regulating the expression of these genes in particular cell types or in response to specific signals. In addition, increased understanding of transcriptional regulation of these genes will enable development of therapies directed against the right molecular targets for the treatment of different eye pathologies such as those caused by surgical procedures, systemic diseases, and tumors.

In this report, we describe the results of our analysis of the SLRP gene clusters on human chromosomes 1, 9, and 12 that were obtained using a combination of bioinformatics, a review of relevant publications, and wet lab experiments. Because co-regulated promoters often utilize the same framework of TF binding sites, our findings suggest that the HOX-Runx models may be the basis for co-regulated expression of the SLRP genes.

Methods

Bioinformatics resources and genome databases

For these studies, tools from the Genomatix software package (Genomatix Suite release 3.0) were used. These tools are shown in Table 1. More detailed information is available at Genomatix.

DNAse I hypersensitivity PCR assay

Previously published protocols were used with modifications [35-37]. Briefly, MG-63 (human osteosarcoma cell line) and U-937 (human promyelocytic cell line) cells were harvested by centrifugation. After two washes with PBS, cells were resuspended in 0.5 ml of permeabilizing buffer (15 mM Tris [pH 7.5], 60 mM KCl, 15 mM NaCl, 5 mM MgCl2, 0.5 mM EGTA, 300 mM sucrose, and 0.5 mM 2-mercaptoethanol) supplemented with 0, 500, 2,000, or 10,000 U/ml of DNase I (Roche Applied Sciences, Indianapolis, IN). An equal volume of permeabilizing buffer supplemented with 0.1% lysolecithin (Sigma-Aldrich, St Louis, MO) was added, and the reaction was incubated at room temperature for 4 min. Genomic DNA was isolated using standard procedures. PCR was performed in 50 μl reactions using 100 ng of genomic DNA, 100 ng of each primer, 0.5 mM dNTPs and 1 unit of Taq polymerase (Promega Corp., Madison, WI). The cycle of denaturing at 94 °C for 30 s, annealing at 60 °C for 30 s, and extension at 72 °C for 2 min was repeated 30 times. The PCR products were resolved by agarose gel electrophoresis and visualized with ethidium bromide staining. Primers synthesized by Integrated DNA Technologies Inc. (Coralville, IA) were used in this study, and listed in Table 2.

Chromatin immunoprecipitation

ChIP analysis was carried out essentially as described, with minor modifications [38]. Briefly, MG-63 and U937 cells were collected, re-suspended in media containing 2.5% formaldehyde and incubated at room temperature for 15 min to crosslink protein DNA complexes. Cells were washed twice with ice cold TBS buffer and the pelleted nuclei were resuspended in 1 ml of ChIP lysis buffer (150 mM NaCl, 25 mM Tris, pH 7.5, 1% Trixon X-100, 0.1% SDS, 0.5% deoxycholate), and the protease inhibitor cocktail (Complete Mini, Roche Diagnostics, Mannheim, Germany). The samples were sonicated ten times for 20 s each, with 1 min of cooling on ice in between sonications, and then pre-cleared by incubation with protein A Sepharose beads for 2 h at 4 °C. The various primary antibodies were then added and samples were incubated overnight at 4 °C. Immunocomplexes were precipitated for 3 h with protein A-Sepharose beads and the precipitates were washed once with 1 ml of RIPA buffer (50 mM Tris, pH 8, 150 mM NaCl, 0.1% SDS, 0.5% deoxycholate, 1% Nonidet P-40, and 1 mM EDTA), once with 1 ml of High Salt buffer (50 mM Tris, pH 8, 500 mM NaCl, 0.1% SDS, 0.5% deoxycholate, 1% Nonidet P-40, and 1 mM EDTA), once with 1 ml of LiCl buffer (50 mM Tris, pH 8, 1 mM EDTA, 250 mM LiCl, 1% Nonidet P-40, and 0.5% deoxycholate), and twice with 1 ml of TE (10 mM Tris [pH 8] and 1 mM EDTA). All washes were for 5 min, rotating, at 4 °C. The samples were treated with 200 μg/ml proteinase K for 3 h at 55 °C. Formaldehyde crosslinks were reversed by overnight incubation at 65 °C. The DNA was isolated by phenol-chloroform extraction and ethanol precipitation. The primers for PCR amplifications are listed in the DNAse hypersensitivity assay. The following antibodies were used; anti-PEBP2αA (sc-12488), anti-Pitx3 (sc-19307), anti-upstream stimulatory factor-1 (USF-1, sc-229), anti-c-Myc (sc-42X); anti-IRF-2 (sc-498). All antibodies were obtained from Santa Cruz Biotechnology, Inc., Santa Cruz, CA.

Plasmids, transient transfection of mammalian cells, and semi-quantitative RT-PCR

Mammalian expression plasmids for Runx2 (MRIPV and MASNSL isoforms) and control plasmid (pCMV5) were gifts from Dr. Jennifer J. Westendorf (University of Minnesota, Minneapolis, MN) [39]. The first Runx2 isoform, also known as PEBP2aA1, type I and p56, is a 513 amino acid protein that initiates in exon 2 at the sequence MRIPV [40]. The second isoform, also known as til-1, type II or p57, initiates in exon 1 at the sequence MASNSL and is 15 amino acids longer than the first isoform [41]. Primary bovine corneal keratocytes and MG-63 cells were transiently transfected using FuGENE 6 transfection reagent (Roche Applied Sciences, Indianapolis, IN) according to the standard protocol (9 μl reagent per 3 μg DNA). Total RNA was isolated using Totally RNA, Total RNA Isolation Kit (Ambion Inc., Austin, TX). RNA (2 μg) was reverse-transcribed using the anchor primer oligonucleotide (dT)18, and Superscript II Reverse Transcriptase (Life Technologies, Inc., Gaithersburg, MD). The single stranded cDNA products (2 μl) were used as templates in PCR amplification reactions as described [42]. Gene specific primers used for human and bovine SLRPs are listed in Table 3.

Results

Promoter sets and identification of candidate transcription factors and frameworks

Promoter sequences for the 13 proteoglycan genes (Figure 1) were extracted from genome sequences for human (NCBI build 34), mouse (MGSCv3 R3), and rat (NCBI build 2) using Genomatix' Gene2Promoter software. These orthologous promoter sets are listed in Table 4. The presence of more than one putative transcription start site (TSS) in some genes reflects the fact that several transcripts are mapped to the same locus. The rodent BGN promoters were omitted because there were stretches of Ns within the sequence, indicating sequencing or assembly ambiguities.

The average length of these promoters was adjusted to 600 bp, 500 bp upstream of the most 5' mapped TSS and 100 bp downstream of the most 3' mapped TSS in the set (Table 4). These promoter sets were used to search with MatInspector [43]. The search produced a large and complex output and the results showed that orthologous promoters from human, mouse, and rat are similarly organized. To obtain matches that were more likely to identify functional sites, we combined the TF binding site analysis with a literature analysis using Genomatix' BiblioSphere. BiblioSphere allows analyzing gene/gene, and gene/transcription factor relations from their co-citation in PubMed abstracts. The group of proteoglycan genes was used as input to BiblioSphere. This approach allowed us to reveal relations between the cluster of proteoglycan genes and other genes/transcription factors that were co-cited with all or some of the genes from the input cluster. The results of this type of analysis, termed Cluster Centered BiblioSphere, are shown in Figure 2. The co-citation analysis from Figure 3 shows that OMD, FMOD, BGN, and DCN are co-cited with Runx2, whereas DCN, LUM, and mimecan are co-cited with IRF1 (interferon regulatory factor 1). We therefore inspected the outputs from MatInspector searches on all of the orthologous promoter sets for matches to IRF factors and Runx factors. The most striking similarities were found for the Runx matches in the orthologous CHAD and OMD promoters. In these six promoters Runx matches are found close to the putative TSS. To find potential regulatory models containing either Runx or IRF, the set of CHAD and OMD promoters was subjected to a search for common frameworks of TF binding sites with FrameWorker. Of note, framework is a description of two or more transcription factor binding sites (elements) arranged in a defined order and orientation within a defined distance range. The model is a computational description of a framework for the purpose of computer-assisted detection of frameworks in DNA sequences. FrameWorker determined 10 models of 2 elements and 3 models of 3 elements. Resulting models were evaluated as follows. The variation of the distance between TF binding sites within the model matches should be small. The initial parameters allowed models with a distance of 10-100 bp (a maximal variation of 90 bp). Models with a variation exceeding 30 bp were rejected. Further, the strand orientation of TF binding sites had to be conserved and models with a conserved position relative to the TSS in all promoters were higher ranked than others. A further ranking is done according to the specificity score (p value) FrameWorker calculates. The number of matches within 5,000 random promoters is determined and subsequently the probability to obtain an equal or greater number of matching sequences in a randomly drawn sample of the same size is calculated. The scores are useful for ranking but their absolute value is less significant since it depends much on the size of the sequence set, which usually is small. The HOX-Runx model fulfils these criteria; the minimal distance of absolutely strand conserved elements is 50-52 bp, the positions of model matches are conserved near the TSS in all six promoters, and the FrameWorker specificity score is third ranked among the two element models (p value of 3.82x10-5). Furthermore, the possible importance of Runx is indicated by co-citation with OMD. From the 3 element models the erythroblast transformation specific forkhead signal transducers and activators of transcription (ETSF-FKHD-STAT) was further investigated, since it was the only one with small distance variation and strand conservation for the first two elements, and fairly conserved positioning around -300 bp relative to the TSS. Scanning the human genome with these two models showed 0.31 occurrences per 10,000 bp for the HOX-Runx model and 0.08 occurences per 10,000 bp for the ETSF-FKHD-STAT. Both values are within the quality standard of the experimentally verified modules in Genomatix' Promoter Module Library. The models as generated by FrameWorker are shown in Figure 3.

These models were used to scan the remaining proteoglycan promoters. The HOX-Runx model was found in three other human promoters, DSPG3, KERA, and NYX (Table 5). Since the expression of mimecan is governed by elements within exon 1 [44,45], the models were used to search for matches in the first exon and first intron of the input genes. A number of model matches were found in intron 1 or exon 1 sequences of ASPN, BGN, CHAD, DCN, DSPG3, KERA, LUM, OMD, and PRELP (Table 5). Notably, among these are BGN, DCN, and FMOD, which display no matches in their proximal promoter sequences, but have a co-citation support for Runx2 by BiblioSphere. In contrast to Runx, no models could be derived from the promoter sequences of those SLRP genes co-cited with IRF.

Taken together, the results above show that 11 of the 13 human SLRP genes have at least one HOX-Runx match in their promoter, exon 1, or intron 1 sequences. Although these genes are located in different clusters on different chromosomes, the common HOX-Runx model could be the basis for a co-regulated expression.

Chromosomal context

Control of an entire cluster of genes at the chromosomal level would require them to reside within one chromosomal loop structure that is accessible to the transcription machinery. Usually such a loop is about 200 kb or less. S/MARs often define the borders of chromosomal loops [46,47]. Thus, one prerequisite of an analysis of clusters is that their genes reside on the same sequence contig in relative proximity to each other. Because the SLRP genes on human chromosome 1 were found on different sequence contigs, this cluster was not analyzed at the chromosomal level. On chromosome 9 there is a cluster of ASPN, OMD, and OGN. Upstream to these three genes another gene, extracellular matrix protein 2 (ECM2), coding for an extracellular matrix protein is found (Figure 1 and Figure 4). All these genes reside within approximately 180 kb making them suitable for common control mechanisms at the chromosomal level. The same arrangement of genes is found on chromosome 13 of mouse and chromosome 17 of rat. The genes for human DCN, LUM, KERA, and DSPG3 are found within a region of about 280 kb on chromosome 12. Inspection of S/MAR annotations in ElDorado did not give hints for the presence of chromosomal loops, for the human gene clusters, or for their synthenic rodent equivalents.

Next, the sequences of the chromosomal regions were extracted and scanned for the presence of matches to the models previously defined within the proximal promoter regions. In many cases, TFs that are functional in the proximal promoter are also involved in enhancer/repressor function. The clustered co-occurrence of putatative TF binding sites has been successfully exploited to determine enhancer elements in Drosophila [48,49]. Therefore, accumulations of model matches in intergenic regions may indicate similar regulatory elements. However, our models imply the additional constraints of a defined site orientation and distance range between sites. They are expected to occur less frequently than the accumulation of the respective single sites. We therefore allowed an enlarged window size of 1,000 bp in contrast to the 500 bp and 700 bp sliding windows used in [48] and [49], respectively. Figure 4 shows that there are two occurrences of three matches to the HOX-Runx model within a 1,000 bp window on human chromosome 9. The first is found in the intergenic region between OGN/mimecan and OMD, the second is upstream of the putative promoter of ECM2. Given the overall low frequency of occurrence of this model within the human genome sequence, the vicinity of three model matches could indicate a regulatory module for ECM2 and either a downstream regulatory module for OMD or an upstream regulatory module for OGN/mimecan. The region on chromosome 12 was analyzed in the same way. However, model matches were much more evenly distributed and no occurrence of three or more matches within a 1,000 bp window were observed.

DNAse I hypersensitivity site formation on human chromosome 9

The relevance of the intergenic region with HOX-Runx models to mimecan expression was analyzed by DNAse I hypersensitivity (DH) PCR and ChIP assays. MG-63 cells (shown to express high levels of mimecan) and U937 cells (mimecan transcripts are not detectable) were used in these studies [44]. The cells were subjected to in vivo DNAse I treatment via membrane permeabilization, and the isolated genomic DNA was used for PCR amplifications. Primer sets that amplify separately the three HOX-Runx elements were used (Figure 5A, ig1, ig2, and ig3). To ensure the validity of the assay, we also analyzed the human mimecan promoter (Figure 5A, e1). It was shown previously that USF1 is a transcriptional activator of bovine and human mimecan promoters [45], therefore a DH site should be detected in this region. In our DH assay, an increasing DNAse I concentration correlates with a decreasing PCR product, as the template (if accessible) is degraded by DNAse I treatment. As shown in Figure 5B, the ig1 region was sensitive in MG-63 cells after treatment with as little as 2,000 U/ml of DNAse I. Interestingly, in U937 cells, ig1 and ig3 regions also were found to be sensitive to DNAse I treatment, although a higher concentration of DNAse I (10,000 U/ml) was need for their detection. The ig2 region was not sensitive to DNAse I in both cell types. In agreement with our expectations, the mimecan promoter region was DNAse I sensitive in MG-63 cells, but not in U-937 cells (Figure 5B, e1).

ChIP analysis was used to determine transcription factors that occupy the three Hox-Runx models in the intergenic region (Figure 5C). Pitx3 was present at the ig1 region in MG-63 cells but not in U937 cells, whereas Runx2 was present at the ig2 region in U-937 cells but not in MG-63 cells. These results suggest that Pitx3 may act as a positive regulator of mimecan transcription in MG-63 cells, whereas Runx2 has an opposite effect in U937 cells. Consistent with our previous reports, USF1 was present at the e1 region in MG-63 cells, but not in U937 cells. The presence of IRF2 at e1 in MG-63 cells (Figure 5C, e1) also is consistent with our previous data that demonstrate the involvement of IRF2 in transcriptional regulation of human mimecan [45,50]. The presence of Pitx3 at e1 in U937 cells (Figure 5C, e1) was surprising. Mutually exclusive binding of Pitx3 or USF1 seems a plausible explanation for these results. Of note, Pitx and Runx sites within the first exon of human mimecan are not a part of the HOX-Runx model because the distance between these two sites is different than the distance described for the model.

Taken together, these data indicate that the intergenic region between OGN/mimecan and OMD is associated with the regulated transcription of the human mimecan gene.

Runx2 transcription factors affect expression of several SLRP genes

To test the effect of Runx2 transcription factors on expression of SLRPs we performed transient transfection experiments using MG-63 cells and primary bovine corneal keratocytes. These cells were chosen because both express mimecan, thereby allowing a comparison between bone specific and cornea specific transcriptional regulation of this gene by Runx2 factors. In addition, MG-63 cells also express decorin and biglycan, whereas corneal keratocytes express lumican and low levels of chondroadherin, and currently are the only cell type known that expresses keratocan. We limited our analysis to testing only the two isoforms of Runx2 TFs for the following reasons. First, Runx2 was the factor shown to bind the intergenic region in U937 cells, suggesting a repressor function on mimecan expression (Figure 5C). Second, the results from BiblioSphere analysis show co-citation of this transcription factor with DCN, BGN, FMOD, and OMD in other cell types. Third, considering the fact that there are about 40 HOX genes in vertebrates and most of them give rise to at least two isoforms, testing all of the HOX and Runx transcription factors that potentially could bind to Hox-Runx modules described above will require detailed analyses that are beyond the aim of this study.

The results from transient transfections of MG-63 cells are shown in Figure 6. Overexpression of Runx2 (p56 isoform) led to decreased levels of mimecan and BGN mRNAs, whereas overexpression of Runx2 (p57 isoform) led to further decrease in mimecan mRNA and increase in biglycan mRNA. Overexpression of both isoforms had no effect on the level of decorin mRNA. Similar experiments using bovine corneal keratocytes are shown in Figure 7. Overexpression of Runx2 isoforms led to a slight decrease in mimecan mRNA and an increase in KERA and CHAD mRNAs. The level of LUM mRNA remained unchanged.

Taken together these results show that Runx2 TFs affect the expression of 5 of the 7 SLRPs analyzed in this study. Unchanged expression of DEC and LUM indicates that Runx2 is not a transcriptional regulator of decorin in MG-63 cells and lumican in bovine corneal keratocytes. Whether another member of the Runx family of TF could change the expression of DEC and LUM in these cell types remain to be determined.

Discussion

We have used a bioinformatics approach to compare promoters of the SLRPs between human, mouse, and rat. Compared to approaches used by others, such as those used to search for cis-regulatory modules involved in pattern formation in the Drosophila genome [49], the main difference in our approach is that first it generates a defined model from promoter sequences and then it tries to find multiple occurrences in related intergenic regions. This approach is more stringent than the searches for site clustering in intergenic regions [49]. Furthermore, it implies that the very same TF modules that are active in regulation at the level of the proximal promoter also are involved in regulation at the level of intergenic elements, including enhancers and locus control regions (LCRs). Such involvement has been demonstrated for the hematopoietic lineage specific transcription factor GATA-1 [51]. By searching for a defined model instead of searching for the occurrences of single TF binding sites we identified common regulatory frameworks in promoters of the SLRP genes. One such framework, HOX-Runx, was detected within the proximal promoter, exon 1, or intron 1 sequences of 11 of the 13 human proteoglycan genes. In addition, three HOX-Runx frameworks were found in the intergenic region between OGN/mimecan and OMD and shown by DH and ChIP assays to be associated with expression of mimecan, thereby increasing the list of SLRPs genes that contain this model to 12 of 13. The only remaining SLRP that could not be shown to contain the framework is opticin. However, because the gene cluster on human chromosome 1 was not scanned for the model, as the clusters on chromosomes 9 and 12 were, the presence of an intergenic HOX-Runx framework similar to those on chromosome 9, that might affect the expression of opticin, cannot be excluded at this time.

The homeobox gene family of transcription factors was first identified in Drosophila, encoding proteins that play fundamental roles in coordinating development and morphogenesis [52]. Analysis of their structure shows that they contain a highly conserved DNA binding domain 60 residues long, termed the homeodomain or homeobox. Based on differences in the amino acid at position nine of the DNA recognition helix of these proteins, they are subdivided into large subgroups. The homeodomain proteins related to Drosophila bicoid have a lysine at this position and bind to the sequence TCTAATCCC, whereas HD-proteins related to Drosophila antennapedia and fushi-tarazu have glutamine at this position and bind to sequence TCAATTAA [53,54]. The DNA binding specificity of these proteins is further affected by interactions between different homeodomain proteins [55]. In addition, various Hox genes are known to produce alternative transcripts encoding different isoforms whose physiological relevance during development is not yet understood [56]. There are only eight HOX proteins in Drosophila but about 40 in vertebrates. Pitx genes, also referred to as the RIEG/PITX homeobox gene family, are of particular interest as potential regulators of the SLRP genes because of their important role in eye, tooth, pituitary, and umbilical region development [57-60]. Pitx3 is associated with anterior segment mesenchymal dysgenesis (ASMD), congenital cataracts, and development of dopaminergic neurons in the substantia nigra [61-64]. The finding that Pitx3 binds the intergenic region on human chomosome 9 in MG-63 cells, but not in U937 cells, indicates that this TF is associated with regulated expression of mimecan.

The first member of the Runx family of transcription factors, runt, was also identified in Drosophila as a regulatory gene, which functions in establishing segmentation patterns during embryogenesis, and also in sex determination and neurogenesis [65]. A second Drosophila Runx gene, lozenge, is a multifunctional transcription factor that is required for cell patterning in the eye and for hematopoiesis [66]. There are three Runx genes in mammals: Runx1 is required for the formation of hematopoietic stem cells and is a frequently mutated gene in human leukemia [67,68]. Runx2 is required for osteogenesis and is associated with cleidocranial dysplasia [69,70], and Runx3 controls neurogenesis in dorsal root ganglia and cell proliferation in the gastric epithelium, and is frequently deleted or silenced in human gastric cancer [71,72]. Runx proteins activate or repress gene expression by binding to a TGPuGGTPu DNA sequence [73]. It has been hypothesized that Runx factors are actually organizers that facilitate the assembly of transcriptional regulatory complexes on gene regulatory elements [74]. Indeed, Runx factors have been shown to interact with many transcription factors and also with histone deacetylase 6 [39,75]. Our finding that Runx2 binds the intergenic region on human chomosome 9 in U937 cells is consistent with these reports (Figure 5C). Runx2 could attract histone deacetylase(s) to this region, thereby repressing transcription of mimecan in this cell type. It is of interest to note that most of the biological activities of Runx proteins are similar to those of TGF-β superfamily cytokines [76].

Both the homeodomain and runt domain containing transcription factors seem to be reasonable candidates for participating in transcriptional regulation of the SLRP family members for several reasons: First, TGF-β signaling has been shown to affect the expression of both homeodomain containing and runt domain containing transcription factors [77,78]. Second, TGF-β growth factors and cytokines also have been shown to be principal regulators of SLRP expression and matrix remodeling during development, inflammation and diseases [79-82]. Thus, it seems likely that HOX and Runx TF are mediators of TGF-β effects on SLRP expession. Supportive to this notion are several reports demonstrating that both homeodomain containing and runt domain containing TFs also regulate the expression of other ECM molecules, including collagen type I and type V, and procollagen lysyl hydrolase, i.e., similarly to TGF-β and SLRPs, HOX and Runx TF also are involved in matrix remodeling [83-87]. Third, the results from bioinformatics analysis, transient transfection, ChIP, and DH assays presented here support the above hypothesis. Forth, four SLRP members already have been co-cited with Runx2 (Figure 3). A hypothetical model to explain how HOX and Runt families of transcription factors may mediate the effects of TGF-β on expression of the SLRP genes is shown in Figure 8. The model is based on results presented in this study and also is supported by data from several publications [79-88]. It is consistent with current views on interactions between certain members of the SLRPs and cytokines of the TGF-β superfamily. It is applicable to different cell types and suggests that depending on the cell type (as well as developmental stage or the type of tissue injury) different combinations of Hox and Runx TF may regulate the expression of SLRPs thereby maintaining/remodeling the ECM accordingly. Finally, the model is easily testable, and will serve as a focus for our future studies.

Acknowledgements

This work was supported by NIH Grant EY13395 to GWC and EST.

References

1. Strahl BD, Allis CD. The language of covalent histone modifications. Nature 2000; 403:41-5.

2. Vignali M, Hassan AH, Neely KE, Workman JL. ATP-dependent chromatin-remodeling complexes. Mol Cell Biol 2000; 20:1899-910.

3. Wolffe AP. Transcriptional regulation in the context of chromatin structure. Essays Biochem 2001; 37:45-57.

4. Hassan AH, Neely KE, Vignali M, Reese JC, Workman JL. Promoter targeting of chromatin-modifying complexes. Front Biosci 2001; 6:D1054-64.

5. Cook PR. Nongenic transcription, gene regulation and action at a distance. J Cell Sci 2003; 116:4483-91.

6. Smale ST, Kadonaga JT. The RNA polymerase II core promoter. Annu Rev Biochem 2003; 72:449-79.

7. Papadakis ED, Nicklin SA, Baker AH, White SJ. Promoters and control elements: designing expression cassettes for gene therapy. Curr Gene Ther 2004; 4:89-113.

8. Werner T, Fessele S, Maier H, Nelson PJ. Computer modeling of promoter organization as a tool to study transcriptional coregulation. FASEB J 2003; 17:1228-37.

9. Lenhard B, Sandelin A, Mendoza L, Engstrom P, Jareborg N, Wasserman WW. Identification of conserved regulatory elements by comparative genome analysis. J Biol 2003; 2:13.

10. Wasserman WW, Sandelin A. Applied bioinformatics for the identification of regulatory elements. Nat Rev Genet 2004; 5:276-87.

11. Frisch M, Frech K, Klingenhoff A, Cartharius K, Liebich I, Werner T. In silico prediction of scaffold/matrix attachment regions in large genomic sequences. Genome Res 2002; 12:349-54.

12. Kmita M, Tarchini B, Duboule D, Herault Y. Evolutionary conserved sequences are required for the insulation of the vertebrate Hoxd complex in neural cells. Development 2002; 129:5521-8.

13. Frazer KA, Tao H, Osoegawa K, de Jong PJ, Chen X, Doherty MF, Cox DR. Noncoding sequences conserved in a limited number of mammals in the SIM2 interval are frequently functional. Genome Res 2004; 14:367-72.

14. Wang Z, Fan H, Yang HH, Hu Y, Buetow KH, Lee MP. Comparative sequence analysis of imprinted genes between human and mouse to reveal imprinting signatures. Genomics 2004; 83:395-401.

15. Dickmeis T, Plessy C, Rastegar S, Aanstad P, Herwig R, Chalmel F, Fischer N, Strahle U. Expression profiling and comparative genomics identify a conserved regulatory region controlling midline expression in the zebrafish embryo. Genome Res 2004; 14:228-38.

16. Duret L, Bucher P. Searching for regulatory elements in human noncoding sequences. Curr Opin Struct Biol 1997; 7:399-406.

17. Fry CJ, Farnham PJ. Context-dependent transcriptional regulation. J Biol Chem 1999; 274:29583-6.

18. Cocea L, Dahan A, Ferradini L, Reynaud CA, Weill JC. Negative regulation of Ig gene rearrangement by a 150-bp transcriptional silencer. Eur J Immunol 1998; 28:2809-16.

19. Li Q, Peterson KR, Fang X, Stamatoyannopoulos G. Locus control regions. Blood 2002; 100:3077-86.

20. Manzanares M, Wada H, Itasaki N, Trainor PA, Krumlauf R, Holland PW. Conservation and elaboration of Hox gene regulation during evolution of the vertebrate head. Nature 2000; 408:854-7.

21. Loots GG, Locksley RM, Blankespoor CM, Wang ZE, Miller W, Rubin EM, Frazer KA. Identification of a coordinate regulator of interleukins 4, 13, and 5 by cross-species sequence comparisons. Science 2000; 288:136-40.

22. Hocking AM, Shinomura T, McQuillan DJ. Leucine-rich repeat glycoproteins of the extracellular matrix. Matrix Biol 1998; 17:1-19.

23. Iozzo RV. The biology of the small leucine-rich proteoglycans. Functional network of interactive proteins. J Biol Chem 1999; 274:18843-6.

24. Kresse H, Schonherr E. Proteoglycans of the extracellular matrix and growth control. J Cell Physiol 2001; 189:266-74.

25. Ameye L, Young MF. Mice deficient in small leucine-rich proteoglycans: novel in vivo models for osteoporosis, osteoarthritis, Ehlers-Danlos syndrome, muscular dystrophy, and corneal diseases. Glycobiology 2002; 12:107R-16R.

26. Plaas AH, West LA, Thonar EJ, Karcioglu ZA, Smith CJ, Klintworth GK, Hascall VC. Altered fine structures of corneal and skeletal keratan sulfate and chondroitin/dermatan sulfate in macular corneal dystrophy. J Biol Chem 2001; 276:39788-96.

27. Tanihara H, Inatani M, Koga T, Yano T, Kimura A. Proteoglycans in the eye. Cornea 2002; 21:S62-9.

28. Danielson KG, Baribault H, Holmes DF, Graham H, Kadler KE, Iozzo RV. Targeted disruption of decorin leads to abnormal collagen fibril morphology and skin fragility. J Cell Biol 1997; 136:729-43.

29. Chakravarti S, Magnuson T, Lass JH, Jepsen KJ, LaMantia C, Carroll H. Lumican regulates collagen fibril assembly: skin fragility and corneal opacity in the absence of lumican. J Cell Biol 1998; 141:1277-86.

30. Xu T, Bianco P, Fisher LW, Longenecker G, Smith E, Goldstein S, Bonadio J, Boskey A, Heegaard AM, Sommer B, Satomura K, Dominguez P, Zhao C, Kulkarni AB, Robey PG, Young MF. Targeted disruption of the biglycan gene leads to an osteoporosis-like phenotype in mice. Nat Genet 1998; 20:78-82.

31. Svensson L, Aszodi A, Reinholt FP, Fassler R, Heinegard D, Oldberg A. Fibromodulin-null mice have abnormal collagen fibrils, tissue organization, and altered lumican deposition in tendon. J Biol Chem 1999; 274:9636-47.

32. Tasheva ES, Koester A, Paulsen AQ, Garrett AS, Boyle DL, Davidson HJ, Song M, Fox N, Conrad GW. Mimecan/osteoglycin-deficient mice have collagen fibril abnormalities. Mol Vis 2002; 8:407-15 <http://www.molvis.org/molvis/v8/a48/>.

33. Liu CY, Birk DE, Hassell JR, Kane B, Kao WW. Keratocan-deficient mice display alterations in corneal structure. J Biol Chem 2003; 278:21672-7.

34. Stamatoyannopoulos G, Grosveld F. Hemoglobin switching. In: Stamatoyannopoulos G, Majerus PW, Perlmutter RM, Varmus H, editors. The molecular basis of blood Diseases. 3rd ed. Philadelphia: WB Saunders; 2001. p.135-182.

35. Ymer S, Jans DA. In vivo chromatin structure of the murine interleukin-5 gene region: a new intact cell system. Biotechniques 1996; 20:834-8,840.

36. Yoo J, Herman LE, Li C, Krantz SB, Tuan D. Dynamic changes in the locus control region of erythroid progenitor cells demonstrated by polymerase chain reaction. Blood 1996; 87:2558-67.

37. Zabel MD, Byrne BL, Weis JJ, Weis JH. Cell-specific expression of the murine CD21 gene depends on accessibility of promoter and intronic elements. J Immunol 2000; 165:4437-45.

38. Im H, Grass JA, Johnson KD, Boyer ME, Wu J, Bresnick EH. Measurement of protein-DNA interactions in vivo by chromatin immunoprecipitation. Methods Mol Biol 2004; 284:129-46.

39. Kahler RA, Westendorf JJ. Lymphoid enhancer factor-1 and beta-catenin inhibit Runx2-dependent transcriptional activation of the osteocalcin promoter. J Biol Chem 2003; 278:11937-44.

40. Ogawa E, Maruyama M, Kagoshima H, Inuzuka M, Lu J, Satake M, Shigesada K, Ito Y. PEBP2/PEA2 represents a family of transcription factors homologous to the products of the Drosophila runt gene and the human AML1 gene. Proc Natl Acad Sci U S A 1993; 90:6859-63.

41. Xiao ZS, Thomas R, Hinson TK, Quarles LD. Genomic structure and isoform expression of the mouse, rat and human Cbfa1/Osf2 transcription factor. Gene 1998; 214:187-97.

42. Tasheva ES, Ke A, Deng Y, Jun C, Takemoto LJ, Koester A, Conrad GW. Differentially expressed genes in the lens of mimecan-null mice. Mol Vis 2004; 10:403-16 <http://www.molvis.org/molvis/v10/a51/>.

43. Quandt K, Frech K, Karas H, Wingender E, Werner T. MatInd and MatInspector: new fast and versatile tools for detection of consensus matches in nucleotide sequence data. Nucleic Acids Res 1995; 23:4878-84.

44. Tasheva ES, Maki CG, Conrad AH, Conrad GW. Transcriptional activation of bovine mimecan by p53 through an intronic DNA-binding site. Biochim Biophys Acta 2001; 1517:333-8.

45. Tasheva ES. Analysis of the promoter region of human mimecan gene. Biochim Biophys Acta 2002; 1575:123-9.

46. Dijkwel PA, Hamlin JL. Matrix attachment regions are positioned near replication initiation sites, genes, and an interamplicon junction in the amplified dihydrofolate reductase domain of Chinese hamster ovary cells. Mol Cell Biol 1988; 8:5398-409.

47. Loc PV, Stratling WH. The matrix attachment regions of the chicken lysozyme gene co-map with the boundaries of the chromatin domain. EMBO J 1988; 7:655-64.

48. Halfon MS, Grad Y, Church GM, Michelson AM. Computation-based discovery of related transcriptional regulatory modules and motifs using an experimentally validated combinatorial model. Genome Res 2002; 12:1019-28.

49. Berman BP, Nibu Y, Pfeiffer BD, Tomancak P, Celniker SE, Levine M, Rubin GM, Eisen MB. Exploiting transcription factor binding site clustering to identify cis-regulatory modules involved in pattern formation in the Drosophila genome. Proc Natl Acad Sci U S A 2002; 99:757-62.

50. Tasheva ES, Conrad GW. Interferon-gamma regulation of the human mimecan promoter. Mol Vis 2003; 9:277-87 <http://www.molvis.org/molvis/v9/a39/>.

51. Horak CE, Mahajan MC, Luscombe NM, Gerstein M, Weissman SM, Snyder M. GATA-1 binding sites mapped in the beta-globin locus by using mammalian chIp-chip analysis. Proc Natl Acad Sci U S A 2002; 99:2924-9.

52. Lewis EB. A gene complex controlling segmentation in Drosophila. Nature 1978; 276:565-70.

53. Harrison SC, Aggarwal AK. DNA recognition by proteins with the helix-turn-helix motif. Annu Rev Biochem 1990; 59:933-69.

54. Treisman J, Gonczy P, Vashishtha M, Harris E, Desplan C. A single amino acid can determine the DNA binding specificity of homeodomain proteins. Cell 1989; 59:553-62.

55. Gehring WJ, Qian YQ, Billeter M, Furukubo-Tokunaga K, Schier AF, Resendez-Perez D, Affolter M, Otting G, Wuthrich K. Homeodomain-DNA recognition. Cell 1994; 78:211-23.

56. Dintilhac A, Bihan R, Guerrier D, Deschamps S, Pellerin I. A conserved non-homeodomain Hoxa9 isoform interacting with CBP is co-expressed with the 'typical' Hoxa9 protein during embryogenesis. Gene Expr Patterns 2004; 4:215-22.

57. Andersen B, Rosenfeld MG. Pit-1 determines cell types during development of the anterior pituitary gland. A model for transcriptional regulation of cell phenotypes in mammalian organogenesis. J Biol Chem 1994; 269:29335-8.

58. Szeto DP, Rodriguez-Esteban C, Ryan AK, O'Connell SM, Liu F, Kioussi C, Gleiberman AS, Izpisua-Belmonte JC, Rosenfeld MG. Role of the Bicoid-related homeodomain factor Pitx1 in specifying hindlimb morphogenesis and pituitary development. Genes Dev 1999; 13:484-94.

59. Semina EV, Reiter R, Leysens NJ, Alward WL, Small KW, Datson NA, Siegel-Bartelt J, Bierke-Nelson D, Bitoun P, Zabel BU, Carey JC, Murray JC. Cloning and characterization of a novel bicoid-related homeobox transcription factor gene, RIEG, involved in Rieger syndrome. Nat Genet 1996; 14:392-9.

60. Ryan AK, Blumberg B, Rodriguez-Esteban C, Yonei-Tamura S, Tamura K, Tsukui T, de la Pena J, Sabbagh W, Greenwald J, Choe S, Norris DP, Robertson EJ, Evans RM, Rosenfeld MG, Izpisua Belmonte JC. Pitx2 determines left-right asymmetry of internal organs in vertebrates. Nature 1998; 394:545-51.

61. Semina EV, Ferrell RE, Mintz-Hittner HA, Bitoun P, Alward WL, Reiter RS, Funkhauser C, Daack-Hirsch S, Murray JC. A novel homeobox gene PITX3 is mutated in families with autosomal-dominant cataracts and ASMD. Nat Genet 1998; 19:167-70.

62. Semina EV, Murray JC, Reiter R, Hrstka RF, Graw J. Deletion in the promoter region and altered expression of Pitx3 homeobox gene in aphakia mice. Hum Mol Genet 2000; 9:1575-85.

63. Rieger DK, Reichenberger E, McLean W, Sidow A, Olsen BR. A double-deletion mutation in the Pitx3 gene causes arrested lens development in aphakia mice. Genomics 2001; 72:61-72.

64. Hwang DY, Ardayfio P, Kang UJ, Semina EV, Kim KS. Selective loss of dopaminergic neurons in the substantia nigra of Pitx3-deficient aphakia mice. Brain Res Mol Brain Res 2003; 114:123-31.

65. Coffman JA. Runx transcription factors and the developmental balance between cell proliferation and differentiation. Cell Biol Int 2003; 27:315-24.

66. Yan H, Canon J, Banerjee U. A transcriptional chain linking eye specification to terminal determination of cone cells in the Drosophila eye. Dev Biol 2003; 263:323-9.

67. North T, Gu TL, Stacy T, Wang Q, Howard L, Binder M, Marin-Padilla M, Speck NA. Cbfa2 is required for the formation of intra-aortic hematopoietic clusters. Development 1999; 126:2563-75.

68. Look AT. Oncogenic transcription factors in the human acute leukemias. Science 1997; 278:1059-64.

69. Ducy P, Zhang R, Geoffroy V, Ridall AL, Karsenty G. Osf2/Cbfa1: a transcriptional activator of osteoblast differentiation. Cell 1997; 89:747-54.

70. Lee B, Thirunavukkarasu K, Zhou L, Pastore L, Baldini A, Hecht J, Geoffroy V, Ducy P, Karsenty G. Missense mutations abolishing DNA binding of the osteoblast-specific transcription factor OSF2/CBFA1 in cleidocranial dysplasia. Nat Genet 1997; 16:307-10.

71. Inoue K, Ozaki S, Ito K, Iseda T, Kawaguchi S, Ogawa M, Bae SC, Yamashita N, Itohara S, Kudo N, Ito Y. Runx3 is essential for the target-specific axon pathfinding of trkc-expressing dorsal root ganglion neurons. Blood Cells Mol Dis 2003; 30:157-60.

72. Li QL, Ito K, Sakakura C, Fukamachi H, Inoue K, Chi XZ, Lee KY, Nomura S, Lee CW, Han SB, Kim HM, Kim WJ, Yamamoto H, Yamashita N, Yano T, Ikeda T, Itohara S, Inazawa J, Abe T, Hagiwara A, Yamagishi H, Ooe A, Kaneda A, Sugimura T, Ushijima T, Bae SC, Ito Y. Causal relationship between the loss of RUNX3 expression and gastric cancer. Cell 2002; 109:113-24.

73. Meyers S, Downing JR, Hiebert SW. Identification of AML-1 and the (8;21) translocation protein (AML-1/ETO) as sequence-specific DNA-binding proteins: the runt homology domain is required for DNA binding and protein-protein interactions. Mol Cell Biol 1993; 13:6336-45.

74. Wang SW, Speck NA. Purification of core-binding factor, a protein that binds the conserved core site in murine leukemia virus enhancers. Mol Cell Biol 1992; 12:89-102.

75. Westendorf JJ, Zaidi SK, Cascino JE, Kahler R, van Wijnen AJ, Lian JB, Yoshida M, Stein GS, Li X. Runx2 (Cbfa1, AML-3) interacts with histone deacetylase 6 and represses the p21(CIP1/WAF1) promoter. Mol Cell Biol 2002; 22:7982-92.

76. Miyazono K, Maeda S, Imamura T. Coordinate regulation of cell growth and differentiation by TGF-beta superfamily and Runx proteins. Oncogene 2004; 23:4232-7.

77. Grienenberger A, Merabet S, Manak J, Iltis I, Fabre A, Berenger H, Scott MP, Pradel J, Graba Y. Tgfbeta signaling acts on a Hox response element to confer specificity and diversity to Hox protein function. Development 2003; 130:5445-55.

78. Lee KS, Hong SH, Bae SC. Both the Smad and p38 MAPK pathways play a crucial role in Runx2 expression following induction by transforming growth factor-beta and bone morphogenetic protein. Oncogene 2002; 21:7156-63.

79. Kinsella MG, Bressler SL, Wight TN. The regulated synthesis of versican, decorin, and biglycan: extracellular matrix proteoglycans that influence cellular phenotype. Crit Rev Eukaryot Gene Expr 2004; 14:203-34.

80. Long CJ, Roth MR, Tasheva ES, Funderburgh M, Smit R, Conrad GW, Funderburgh JL. Fibroblast growth factor-2 promotes keratan sulfate proteoglycan expression by keratocytes in vitro. J Biol Chem 2000; 275:13918-23.

81. Demoor-Fossard M, Galera P, Santra M, Iozzo RV, Pujol JP, Redini F. A composite element binding the vitamin D receptor and the retinoic X receptor alpha mediates the transforming growth factor-beta inhibition of decorin gene expression in articular chondrocytes. J Biol Chem 2001; 276:36983-92.

82. Tiede K, Stoter K, Petrik C, Chen WB, Ungefroren H, Kruse ML, Stoll M, Unger T, Fischer JW. Angiotensin II AT(1)-receptor induces biglycan in neonatal cardiac fibroblasts via autocrine release of TGFbeta in vitro. Cardiovasc Res 2003; 60:538-46.

83. Penkov D, Tanaka S, Di Rocco G, Berthelsen J, Blasi F, Ramirez F. Cooperative interactions between PBX, PREP, and HOX proteins modulate the activity of the alpha 2(V) collagen (COL5A2) promoter. J Biol Chem 2000; 275:16681-9.

84. Kern B, Shen J, Starbuck M, Karsenty G. Cbfa1 contributes to the osteoblast-specific expression of type I collagen genes. J Biol Chem 2001; 276:7101-7.

85. Hjalt TA, Amendt BA, Murray JC. PITX2 regulates procollagen lysyl hydroxylase (PLOD) gene expression: implications for the pathology of Rieger syndrome. J Cell Biol 2001; 152:545-52.

86. Chen XD, Fisher LW, Robey PG, Young MF. The small leucine-rich proteoglycan biglycan modulates BMP-4-induced osteoblast differentiation. FASEB J 2004; 18:948-58.

87. Ghannam G, Takeda A, Camarata T, Moore MA, Viale A, Yaseen NR. The oncogene Nup98-HOXA9 induces gene transcription in myeloid cells. J Biol Chem 2004; 279:866-75.

88. Ge G, Seo NS, Liang X, Hopkins DR, Hook M, Greenspan DS. Bone Morphogenetic Protein-1/Tolloid-related Metalloproteinases Process Osteoglycin and Enhance Its Ability to Regulate Collagen Fibrillogenesis. J Biol Chem 2004; 279:41626-33.

89. Balint E, Lapointe D, Drissi H, van der Meijden C, Young DW, van Wijnen AJ, Stein JL, Stein GS, Lian JB. Phenotype discovery by gene expression profiling: mapping of biological processes linked to BMP-2-mediated osteoblast differentiation. J Cell Biochem 2003; 89:401-26.

90. Lee TH, Klampfer L, Shows TB, Vilcek J. Transcriptional regulation of TSG6, a tumor necrosis factor- and interleukin-1-inducible primary response gene coding for a secreted hyaluronan-binding protein. J Biol Chem 1993; 268:6154-60.

91. Silverstein DM, Travis BR, Thornhill BA, Schurr JS, Kolls JK, Leung JC, Chevalier RL. Altered expression of immune modulator and structural genes in neonatal unilateral ureteral obstruction. Kidney Int 2003; 64:25-35.

92. Frech K, Danescu-Mayer J, Werner T. A novel method to develop highly specific models for regulatory units detects a new LTR in GenBank which contains a functional promoter. J Mol Biol 1997; 270:674-87.

93. Frech K, Quandt K, Werner T. Muscle actin genes: a first step towards computational classification of tissue specific promoters. In Silico Biol 1998; 1:29-38.

Tasheva, Mol Vis 2004; 10:758-772 <http://www.molvis.org/molvis/v10/a91/>
©2004 Molecular Vision <http://www.molvis.org/molvis/>
ISSN 1090-0535