RUNX2 belongs to the RUNT domain family of transcription factors of which three have been identified in humans (RUNX1, RUNX2 and RUNX3). RUNX proteins are vital for metazoan development and participate in the regulation of cellular differentiation and cell cycle progression (Coffman, 2003). RUNX2 is required for proper bone formation by driving the differentiation of osteoblasts from mesenchymal progenitors during development (Ducy et al, 1997; Komori et al, 1997; Otto et al, 1997). RUNX2 is also vital for chondrocyte maturation by promoting the differentiation of chondrocytes to the hypertrophic phenotype (Enomoto et al, 2000). The consequences of completely disrupting the RUNX2 locus in mice provided compelling and conclusive evidence for the biological importance of RUNX2 where knockout mice died shortly after birth with a complete lack of bone formation (Komori et al, 1997; Otto et al, 1997). A further indication of the requisite role of RUNX2 in skeletal development was the discovery that RUNX2 haploinsufficiency in humans and mice caused the skeletal syndrome Cleidocranial Dysplasia (CCD) (Mundlos et al, 1997; Lee et al, 1997). A unique feature of RUNX2 is the consecutive polyglutamine and polyalanine tracts (Q/A domain). Mutations causing CCD have been observed in the Q/A domain of RUNX2 (Mundlos et al, 1997). The Q/A domain is an essential part of RUNX2 and participates in transactivation function (Thirunavukkarasu et al, 1998). Previous genotyping studies conducted in our laboratory identified several rare RUNX2 Q/A variants in addition to a frequently occurring 18 base pair deletion of the polyalanine tract termed the 11Ala allele. Analysis of serum parameters in 78 Osteoarthritis patients revealed the 11Ala allele was associated with significantly decreased osteocalcin. Furthermore, analysis of 11Ala allele frequencies within a Geelong Osteoporosis Study (GOS) fracture cohort and an appropriate age matched control group revealed the 11Ala allele was significantly overrepresented in fracture cases indicating an association with increased fracture risk. To further investigate the 11Ala allele and rare Q/A variants, 747 DNA samples from the Southeast Queensland bone study were genotyped using PCR and PAGE. The experiment served two purposes: 1) to detect additional rare Q/A variants to enrich the population of already identified mutants and 2) have an independent assessment of the effect of the 11Ala allele on fracture to either support or refute our previous observation which indicated the 11Ala allele was associated with an increased risk of fracture in the GOS. From the 747 samples genotyped, 665 were WT, 76 were heterozygous for the 11Ala allele, 5 were homozygous for the 11Ala allele and 1 was heterozygous for a rare 21 bp deletion of the polyglutamine tract. Chi-square analysis of RUNX2 genotype distributions within fracture and non-fracture groups in the Southeast Queensland bone study revealed that individuals that carried at least one copy of the 11Ala allele were enriched in the fracture group (p = 0.16, OR = 1.712). The OR of 1.712 was of similar magnitude to the OR observed in the GOS case-control investigation (OR = 1.9) providing support for the original study. Monte-Carlo simulations were used to combine the results from the GOS and the Southeast Queensland bone study. The simulations were conducted with 10000 iterations and demonstrated that the maximum probability of obtaining both study results by chance was less than 5 times in two hundred (p < 0.025) suggesting that the 11Ala allele of RUNX2 was associated with an increased fracture risk. The second element of the research involved the analysis of rare RUNX2 Q/A variants identified from multiple epidemiological studies of bone. Q/A repeat variants were derived from four populations: the GOS, an Aberdeen cohort, CAIFOS and a Sydney twin study. Collectively, a total of 20 rare glutamine and one alanine variants were identified from 4361 subjects. All RUNX2 Q/A variants were heterozygous for a mutant allele and a wild type allele. Analysis of incident fracture during a five year follow up period in the CAIFOS revealed that Q-variants (n = 8) were significantly more likely to have fractured compared to non-carriers (p = 0.026, OR 4.932 95% CI 1.2 to 20.1). Bone density data as measured by quantitative ultrasound was available for CAIFOS. Analysis of BUA and SOS Z-scores revealed that Q-repeat variants had significantly lower BUA (p = 0.031, mean Z-score of -0.79) and a trend for lower SOS (p = 0.190, mean Z-score of -0.69). BMD data was available for all four populations. To normalize the data across the four studies, FN BMD data was converted into Z-scores and the effect of the Q/A variants on BMD was analysed using a one sample approach. The analysis revealed Q/A variants had significantly lower FN BMD (p = 0.0003) presenting with a 0.65 SD decrease. Quantitative transactivation analysis was conducted on RUNX2 proteins harbouring rare glutamine mutations and the 11Ala allele. RUNX2 proteins containing a glutamine deletion (16Q), a glutamine insertion (30Q) and the 11Ala allele were overexpressed in NIH3T3 and HEK293 cells and their ability to transactivate a known target promoter was assessed. The 16Q and 30Q had significantly decreased reporter activity compared to WT in NIH3T3 cells (p = 0.002 and 0.016, for 16Q and 30Q, respectively). In contrast 11Ala RUNX2 did not show significantly different promoter activation potential (p = 0.54). Similar results were obtained in HEK293 cells where both the 16Q and 30Q RUNX2 displayed decreased reporter activity (p=0.007 and 0.066 for 16Q and 30Q respectively) whereas the 11Ala allele had no material effect on RUNX2 function (p = 0.20). The RUNX2 gene target reporter assay provided evidence to suggest that variation within the glutamine tract of RUNX2 was capable of altering the ability of RUNX2 to activate a known target promoter. In contrast, the 11Ala allele showed no variation in RUNX2 activity. The third feature of the research served the purpose of identifying potential RUNX2 gene targets with particular emphasis on discovering genes cooperatively regulated by RUNX2 and the powerful bone promoting agent BMP2. The experiment was conducted by creating stably transfected NIH3T3 cells lines overexpressing RUNX2 or BMP2 or both RUNX2 and BMP2. Microarray analysis revealed very few genes were differentially regulated between standard NIH3T3 cells and cells overexpressing RUNX2. The results were confirmed via RT-PCR analysis which demonstrated that the known RUNX2 gene targets Osteocalcin and Matrix Metalloproteinase-13 were modestly induced 2.5 fold (p = 0.00017) and 2.1 fold (p = 0.002) respectively in addition to identifying only two genes (IGF-II and SCYA11) that were differentially regulated greater than 10 fold. IGF-II and SYCA11 were significantly down-regulated 27.6 fold (p = 1.95 x 10-6) and 10.1 fold (p = 0.0002) respectively. The results provided support for the notion that RUNX2 on its own was not sufficient for optimal gene expression and required the presence of additional factors. To discover genes cooperatively regulated by RUNX2 and BMP2, microarray gene expression analysis was performed on standard NIH3T3 cells and NIH3T3 cells stably transfected with both RUNX2 and BMP2. Comparison of the gene expression profiles revealed the presence of a large number of differentially regulated genes. Four genes EHOX, CCL9, CSF2 and OSF-1 were chosen to be further characterized via RT-PCR. Sequential RT-PCR analysis on cDNA derived from control cells and cells stably transfected with either RUNX2, BMP2 or both RUNX2/BMP2 revealed that EHOX and CSF2 were cooperatively induced by RUNX2 and BMP2 whereas CCL9 and OSF-1 were suppressed by BMP2. The overexpression of both RUNX2 and BMP2 in NIH3T3 fibroblasts provided a powerful model upon which to discover potential RUNX2 gene targets and also identify genes synergistically regulated by BMP2 and RUNX2. The fourth element of the research investigated the role of RUNX2 in the ascorbic acid mediated induction of MMP-13 mRNA. The study was carried out using NIH3T3 cell lines stably transfected with BMP2, RUNX2 and both BMP2 and RUNX2. The cell lines were grown to confluence and subsequently cultured for a further 12 days in standard media or in media supplemented with AA. RT-PCR analysis was used to assess MMP-13 mRNA expression. The RT-PCR results demonstrated that AA was not sufficient for inducing MMP-13 mRNA in NIH3T3 cells. In contrast RUNX2 significantly induced MMP-13 levels 85 fold in the absence of AA (p = 0.0055) and upregulated MMP-13 mRNA levels 254 fold in the presence of AA (p = 0.0017). The results demonstrated that RUNX2 was essential for the AA mediated induction of MMP-13 mRNA in NIH3T3 cells. The effect of BMP2 on MMP-13 expression was also investigated. BMP2 induced MMP-13 mRNA transcripts a modest 3.8 fold in the presence of AA (p = 0.0027). When both RUNX2 and BMP2 were overexpressed in the presence of AA, MMP-13 mRNA levels were induced a massive 4026 fold (p = 8.7 x 10-4) compared to control cells. The investigation revealed that RUNX2 was an essential factor for the AA mediated induction of MMP-13 and that RUNX2 and BMP2 functionally cooperated to regulate MMP-13 mRNA levels.

Bioinformatics is a fast-developing field, which makes use of computational methods to analyse and structure biological data. An important branch of bioinformatics is structure and function prediction of proteins, which is often based on finding relationships to already characterized proteins. It is known that two proteins with very similar sequences also share the same 3D structure. However, there are many proteins with similar structures that have no clear sequence similarity, which make it difficult to find these relationships. In this thesis, two methods for annotating protein domains are presented, one aiming at assigning the correct domain family or families to a protein sequence, and the other aiming at fold recognition. Both methods use hidden Markov models (HMMs) to find related proteins, and they both exploit the fact that structure is more conserved than sequence, but in two different ways. Most of the research presented in the thesis focuses on the structure-anchored HMMs, saHMMs. For each domain family, an saHMM is constructed from a multiple structure alignment of carefully selected representative domains, the saHMM-members. These saHMM-members are collected in the so called "midnight ASTRAL set", and are chosen so that all saHMM-members within the same family have mutual sequence identities below a threshold of about 20%. In order to construct the midnight ASTRAL set and the saHMMs, a pipe-line of software tools are developed. The saHMMs are shown to be able to detect the correct family relationships at very high accuracy, and perform better than the standard tool Pfam in assigning the correct domain families to new domain sequences. We also introduce the FI-score, which is used to measure the performance of the saHMMs, in order to select the optimal model for each domain family. The saHMMs are made available for searching through the FISH server, and can be used for assigning family relationships to protein sequences. The other approach presented in the thesis is secondary structure HMMs (ssHMMs). These HMMs are designed to use both the sequence and the predicted secondary structure of a query protein when scoring it against the model. A rigorous benchmark is used, which shows that HMMs made from multiple sequences result in better fold recognition than those based on single sequences. Adding secondary structure information to the HMMs improves the ability of fold recognition further, both when using true and predicted secondary structures for the query sequence. / Bioinformatik är ett område där datavetenskapliga och statistiska metoder används för att analysera och strukturera biologiska data. Ett viktigt område inom bioinformatiken försöker förutsäga vilken tredimensionell struktur och funktion ett protein har, utifrån dess aminosyrasekvens och/eller likheter med andra, redan karaktäriserade, proteiner. Det är känt att två proteiner med likande aminosyrasekvenser också har liknande tredimensionella strukturer. Att två proteiner har liknande strukturer behöver dock inte betyda att deras sekvenser är lika, vilket kan göra det svårt att hitta strukturella likheter utifrån ett proteins aminosyrasekvens. Den här avhandlingen beskriver två metoder för att hitta likheter mellan proteiner, den ena med fokus på att bestämma vilken familj av proteindomäner, med känd 3D-struktur, en given sekvens tillhör, medan den andra försöker förutsäga ett proteins veckning, d.v.s. ge en grov bild av proteinets struktur. Båda metoderna använder s.k. dolda Markov modeller (hidden Markov models, HMMer), en statistisk metod som bland annat kan användas för att beskriva proteinfamiljer. Med hjälp en HMM kan man förutsäga om en viss proteinsekvens tillhör den familj modellen representerar. Båda metoderna använder också strukturinformation för att öka modellernas förmåga att känna igen besläktade sekvenser, men på olika sätt. Det mesta av arbetet i avhandlingen handlar om strukturellt förankrade HMMer (structure-anchored HMMs, saHMMer). För att bygga saHMMerna används strukturbaserade sekvensöverlagringar, vilka genereras utifrån hur proteindomänerna kan läggas på varandra i rymden, snarare än utifrån vilka aminosyror som ingår i deras sekvenser. I varje proteinfamilj används bara ett särskilt, representativt urval av domäner. Dessa är valda så att då sekvenserna jämförs parvis, finns det inget par inom familjen med högre sekvensidentitet än ca 20%. Detta urval görs för att få så stor spridning som möjligt på sekvenserna inom familjen. En programvaruserie har utvecklats för att välja ut representanter för varje familj och sedan bygga saHMMer baserade på dessa. Det visar sig att saHMMerna kan hitta rätt familj till en hög andel av de testade sekvenserna, med nästan inga fel. De är också bättre än den ofta använda metoden Pfam på att hitta rätt familj till helt nya proteinsekvenser. saHMMerna finns tillgängliga genom FISH-servern, vilken alla kan använda via Internet för att hitta vilken familj ett intressant protein kan tillhöra. Den andra metoden som presenteras i avhandlingen är sekundärstruktur-HMMer, ssHMMer, vilka är byggda från vanliga multipla sekvensöverlagringar, men också från information om vilka sekundärstrukturer proteinsekvenserna i familjen har. När en proteinsekvens jämförs med ssHMMen används en förutsägelse om sekundärstrukturen, och den beräknade sannolikheten att sekvensen tillhör familjen kommer att baseras både på sekvensen av aminosyror och på sekundärstrukturen. Vid en jämförelse visar det sig att HMMer baserade på flera sekvenser är bättre än sådana baserade på endast en sekvens, när det gäller att hitta rätt veckning för en proteinsekvens. HMMerna blir ännu bättre om man också tar hänsyn till sekundärstrukturen, både då den riktiga sekundärstrukturen används och då man använder en teoretiskt förutsagd. / Jeanette Hargbo.