CHUK protein, human

PWMs for 16 transcription factors have previously been collated [17 (link)] from various in vitro SELEX and DNaseI footprinting experiments that used purified transcription factor protein and naked DNA [70 (link)] (BDTNP, unpublished data). For convenience these are provided in Additional file 13. These PWMs were used to identify all DNA sequences that match them genome-wide at P-values <0.04 using Fimo [107 (link)]. For each factor, these recognition site occurrences were then divided into two groups depending on whether the matches were located within 5% FDR DNaseI accessible regions or whether they were in inaccessible chromatin. The recognition sites were then further broken down into cohorts in R based on P-values as follows:
For each cohort, the maximum ChIP-chip signal from the relevant factor within 250 bp of each sequence match was determined using input DNA normalized ChIP-chip scores calculated as Array hybridization signal for factor immunoprecipitation/Array hybridization signal for input crosslinked DNA (see Figure 2 in [14 (link)]) except that natural numbers, not log2, were used here. The 95% confidence interval about the median of these scores was calculated using R's box plot function (Figure 4; Additional file 8).
In addition, ten permutations of each original PWM were generated by shuffling the order of positions in the weight matrices for each permutation. If any permutation that matched any other of the randomly generated permutations for that factor or the normal PWM of one of the other 15 factors (P < 0.05 defined using Tomtom [108 (link)]) it was discarded and a new permutation was generated. The set of sequence matches to these scrambled PWMs were then identified throughout the genome, separated into those in open or closed chromatin and binned into groups based on affinity in the same manner as for the genuine motifs. The maximum ChIP-chip scores within 250 bp of each scrambled recognition site occurrence was determined and the median of this peak score was determined over the entire set of ten scrambled PWMs for each factor and the 95% confidence limits calculated as for the matches to the genuine PWMs (Figure 4; Additional file 8).
To correlate accessibility with ChIP-chip scores (Figure 5a), peaks in accessibility at stage 5 were annotated with maximum input DNA normalized ChIP-chip scores within 75 bp of each peak for the 16 factors with well-characterized in vitro binding specificities (Figure 4; Additional file 8). The peaks were ranked by accessibility and the correlation between level of accessibility and ChIP-chip score was calculated using R's Pearson correlation function. The DNaseI peaks were then ranked, separated into cohorts of 200 similarly accessible peaks and the median peak in ChIP-chip signal for each cohort was determined and plotted using R's heat map function scaling rows to account for inherent differences in ChIP-chip signal between factors. A similar process was used to correlate accessibility with the presence of recognition sites for each of the 16 factors (Figure 5b). The same PWMs for the factors derived from in vitro DNA binding data, described above, were employed to identify all sequence matches to these matrices within 75 bp of peaks of accessibility with P < 0.003 using Fimo [107 (link)] (that is, matches that fell into at least the -3 cohort from Figure 4). The correlation between the level of accessibility and the number of PWM matches was calculated using R's Pearson correlation function. For each factor, the peaks in accessibility were ranked and divided into cohorts of 200 and the sum of all recognition sites was added over each cohort and plotted in R using the heat map function, while scaling rows to one another in order to account for differences in information content between PWMs.

The AIA rats were sacrificed on day 23 post-adjuvant induction. The CIA mice were sacrificed on day 33 post-primary immunization. The animals were dissected to obtain synovitis of joint tissues. The total RNA of tissues was extracted in TRNzol Universal reagent by vigorous mechanical shaking. The cDNA synthesis was performed using PrimeScript RT reagent kit and subjected to qPCR reaction using TB Green Premix Ex Taq (Tli RNaseH Plus) kit on ABI 7500 systems (Applied Biosystems, Foster City, CA). The relative mRNA expression was normalized by 2^-ΔΔCt method, and the primer sequences for Tnf, Il1b, Il6, Tab2, Tab3, Ikka, and Actb are listed in Supplementary Table 1.

The cell culture and miR-23b transfection were conducted as described in the section Cell apoptosis analysis. The total RNA of treated cells was extracted by TRNzol Universal reagent, and 2 μg RNA was utilized for the cDNA synthesis using PrimeScript RT reagent kit. Subsequently, 1 μg cDNA was applied for qPCR reaction by TB Green Premix Ex Taq (Tli RNaseH Plus) kit on ABI 7500 systems (Applied Biosystems, Foster City, CA). The relative expression levels of Tnf, Il1b, Il6, Tab2, Tab3, and Ikka were measured by 2^-ΔΔCt method using Actb as the reference gene. The sequences of primers used are listed in Supplementary Table 1.