1. Compiling polymorphic miRNA-target interactions: the Patrocles database. Samuel Hiard 1, Xavier Tordoir 2, Wouter Coppieters 2, Carole Charlier 2 and Michel Georges 2 1 Bioinformatics and Modeling, GIGA & Department of Electrical Engineering and Computer Science – University of Liège, Sart-Tilman B28, Liège, Belgium 2 Unit of Animal Genomics, Department of Animal Production, Faculty of Veterinary Medicine & CBIG, University of Liège (B43), 20 Boulevard de Colonster, 4000-Liège, Belgium. Introduction miRNA-mediated gene silencing emerges as a key regulator of cellular differentiation and homeostasis to which metazoans devote a considerable amount of sequence space. This sequence space is bound to suffer its toll of mutations of which some will be selectively neutral while others will be advantageous or more often at least slightly deleterious. DNA sequence polymorphisms (DSP) occurring within this sequence space certainly contribute to phenotypic variation including disease susceptibility and agronomically important traits. An important question is how important their contribution actually is. DSP may affect miRNA-mediated gene regulation by perturbing core components of the silencing machinery, by affecting the structure or expression level of miRNAs, or by altering target sites (Table 1). DSP in core components of the silencing machinery may affect its overall efficacy. Mutations that drastically perturb RNA silencing will obviously be rare given their predictable highly deleterious consequences. Yet, DSP with subtle effects on gene function may occur. As distinct targets may be more or less sensitive to variations in miRNA concentration or silencing efficiency, such DSP may affect some pathways more than others. Specific miRNA-target interactions may be influenced by mutations affecting either the miRNA or its target. On the miRNA side of the equation: (i) the sequence of the mature miRNA may be altered, thereby either stabilizing or destabilizing its interaction with targets, (ii) mutations in the pri- or pre-miRNA may affect stability or processing efficiency, (iii) mutations acting in cis or trans on the pri-miRNA promoter may influence transcription rate, and (iv) Copy Number Variants (CNV) may affect the number of copies of the miRNA or the integrity of the pri-miRNA host. On the target side of the equation: (i) mutations may affect functional target sites thereby destabilizing or stabilizing the interaction with the miRNA, (ii) mutations may create illegitimate miRNA target sites (either in the 3’UTR or maybe even in other segments of the transcript) which will be particularly relevant if occurring in antitargets, (iii) mutations causing polymorphic alternative polyadenylation may affect a gene’s content in target sites. Categories of DNA sequence polymorphisms (DSP) affecting miRNA-mediated gene regulation Table 1 Screen shot Compiling candidate pSNPS Quantifying miRNA putative target co-expression Why? For a pSNP to be affect function, miRNA and putative target need to have overlapping expression domains. To assist in the identification of relevant pSNPs, we therefore have devised a way to quantify the degree of co-expression for miRNA-gene pairs How? Gene Expression : SymAtlas (http://symatlas.gnf.org/SymAtlas/) miRNA Expression : - Fahr et al. 2005 - Compute observed frequency - Compute expected frequency ( ) - Kolmogorov-Smirnov test P-Value : Determined by 1000 random permutation of genes + KS - 80% of miRNAs hosted by genes - Deduce expression from corresponding gene expression - Experimental data CoExpression : First try : - CoExpression of known antitarget gene and miRNA is quite low - Why? This function doesn’t differenciate moderate coexpression across all tissues and extremely high coexpression in one tissue Patrocles finder Why? Patrocles is built using the public information provided by Ensembl. But the laboratories that work on SNPs often discover new ones. So, there must be a tool that allows these labs to obtain the information about stabilized, destabilized or illegitimate target sites How? End users must provide one or two sequences for, respectively, (i) the analysis of the presence of octamers or (ii) the comparison of the two sequences regarding to the content in octamers. They also have to possibility to provide an alignment of each sequence if they care about conservation. Screen shots Acknowledgements PAI P5/25 from the Belgian SSTC (n° R.SSTC.0135), EU “Callimir” STREP project. C.C. is chercheur qualifié from the FNRS. Abstract Using positional cloning, we have recently identified the mutation responsible for muscular phenotype of the Texel sheep. It is located in the 3’UTR of the GDF8 gene - a known developmental repressor of muscle growth - and creates an illegitimate target site for miRNA expressed in the same tissue. This causes miRNA-mediated translation inhibition of mutant GDF8 transcripts which leads to muscle hypertrophy. We followed up on this finding by searching for common polymorphisms and mutations that affect either (i) RNAi silencing machinery components, (ii) miRNA precursors or (iii) target sites. These might likewise alter miRNA-target interaction and could be responsible for substantial differences in gene expression level. They have been compiled in a public database (“Patrocles”: www.patrocles.org), where they are classified in (i) DNA sequence polymorphisms (DSP) affecting the silencing machinery, (ii) DSP affecting miRNA structure or expression and (iii) DSP affecting miRNA target sites. DSP from the last category were organized in four classes: destroying a target site conserved between mammals (DC), destroying a non-conserved target site (DNC), creating a non-conserved target site (CNC), or shifting a target site (S). To aid in the identification of the most relevant DSP (such as those were a target site is created in an antitarget gene), we have quantified the level of coexpression for all miRNA-gene pairs.www.patrocles.org Analysis of the numbers of Patrocles-DSP as well as their allelic frequency distribution indicates that a substantial proportion of them undergo purifying selection. The signature of selection was most pronounced for the DC class but was significant for the DNC and CNC class as well, suggesting that a significant proportion of non-conserved targets is truly functional. The Patrocles database allowed for the selection of DSP that are likely to affect gene function and possibly disease susceptibility. The effect of these DSP is being studied both in vitro and in vivo. In conclusion, Patrocles-DSP could be widespread and underlie an appreciable amount of phenotypic variation, including common disease susceptibility. Evidence for purifying selection against pSNPs of conserved and non-conserved target sites Why? What is the evidence that any of the candidate pSNPs listed above truly affect gene function and hence phenotype? Indirect evidence that a significant proportion of them are functional can be obtained from population genetics. Indeed, pSNPs without appreciable effect on gene function will evolve neutrally, subject only to the vagaries of random genetic drift while pSNPs affecting gene function may undergo positive, negative or balancing selection via their effect on phenotype. Selection may leave distinct signatures on the level of inter-species divergence, intra-species variability, allelic distribution and linkage disequilibrium How? - Generation of 100 random sets of SNPs - Processed through pipeline Results:  Less pSNPs in real data  Differences between X and L  pSNPs that destroy conserved target site are highly underrepresented (expected)  pSNPs that either destroy or create non- conserved target site are also underrepresented (  functional even if not conserved across mammals) Mutations in miRNAs For specific miRNAs a)mutations in the mature miRNA (table 2) 6 SNP in the miR seed (yellow) 11 SNP in the mature miR (white) b)mutations in the pre-miRNA may affect stability or processing efficacy, 71 SNP in the premiR: eg.: * * * c)mutations acting in cis or trans on the pri-miRNA promoter (or host gene) may influence transcription rate: For the 474 human miRNAs in Rfam (oct 2006): - 186 host genes for 229 miR (48.3%) - 245 miR without host gene We identified miRNA host genes characterized by inherited variation in expression levels, reasoning that this might affect the cellular concentration of passenger miRNAs. We compiled host genes influenced by both trans- and cis-acting “expression QTL” (eQTL) identified either by linkage analysis or by association studies and host genes having shown allelic imbalance in heterozygous individuals (review by Pastinen et al., 2006; Spielman et al, 2006). At least eight host genes were found amongst the differentially regulated genes reported in these studies. An additional one is showing allelic imbalance. d)Copy Number Variants (CNV) may affect the number of copies of the miRNA or the integrity of the pri-miRNA host: A first CNV map of the human genome has been recently constructed (Redon et al., 2006). We found 43 miRNAs residing in regions involved in CNV, 19 without known host gene and 24 in a host gene which were completely (18) or partially (6) included in a CNV. Table 2: DSP in mature miRNA Globally a)DSP in components of the RNA silencing machinery may affect its overall efficacy. We followed 19 genes involved in miR biology for coding SNP, CNV, eQTL and allelic imbalance: CNV encompass Drosha and DGCR8 genes and 6 genes present non synomymous mutations (table 3) Table 3: non synonymous SNP in components of miR pathway miR mediated translational inhibition of the Texel MSTN allele The g+6723G-A natural polymorphism causes translational inhibition of the Texel MSTN allele by creating an illegitimate target site for two miRNA expressed in the same tissue, this leads to muscle hypertrophy. Schematic representation of the MSTN gene and sequence context of the polymorphic miRNA-MSTN interaction (left). Muscle hypertrophy in Texel compared to wild-type Romanov sheep (right). Texel Reduction of >3X Reduction of ~1.5X Allelic imbalance of MSTN at the mRNA level Texel allele (A) < WT allele (G) in heterozygous animals Reduced circulating MSTN protein in Texel (T1) vs WT (W1) genomic cDNA 12 Kd MSTN Nature Genetics, 2006 Romanov In Human ConservedNot conserved Created X : 0 L : 0 B : 0X : 5282 L : 7967 B : 858 Destroyed X : 913 L : 639 B : 225X : 4524 L : 7365 B : 708 Polymorphic X : 0 L : 0 B : 0X : 391 L : 691 B : 85 Shifted X : 202 L : 361 B : 16 ConservedNot conserved Created X : 0 L : 0 B : 0X : 3661 L : 4325 B : 592 Destroyed X : 424 L : 269 B : 73X : 3363 L : 4157 B : 529 Polymorphic X : 0 L : 0 B : 0X : 1000 L : 1313 B : 197 Shifted X : 14 L : 21 B : 11 In Mouse X = Xie et al. 2005 : Predicted putative miRNA target sites by identification of octamer motifs in 3’UTRs characterized by unusually high motif conservation scores (i.e. proportion of conserved amongst all occurrences). L = Lewis et al. 2005 : Reverse complement of (A + 2  8) of mature miRNA (MiRBase) B = Both miRNA derived expression CoExpression distribution of known antitargets CoExpression Score Nb of Known antitargets TargetmiRNASilencing machinery DSP altering miRNA recognition sites in the target Altering existing target sites. Stabilizing or destabilizing the interaction with the miRNA Creating illegitimate target sites DSP altering the target’s 3’UTR e.g. polymorphic polyadenylation DSP altering the sequence of the miRNA. Stabilizing or destabilizing the interaction with the target (pSNP) DSP altering the concentration of the miRNA Copy Number Variants emcompassing the pri-miRNA DSP altering the transcription rate of the pri-miRNA. Cis or trans-acting DSP affecting the processing efficiency of the pri- or pre-miRNA DSP altering the amino-acid sequence of silencing components DSP altering the concentration of silencing components Copy Number Variants encompassing silencing components