Search

Additional file 10: Figure S7. Cellular back mapping identifies co-localizing protein populations at extranuclear DNA segments. A. images shown in Fig. 4, all channels. B. Analysis of KDM6A TPR and NPM1: Cellular back mapping identifies a colocalizing population (orange) and free KDM6A TPR (green) and NPM1 (red) at cytoplasmic DNA segments. C. scatterplot analysis and cellular back mapping of KDM6A variants and DAPI as indicator of KDM6A-DNA complexes. In KDM6A WT scatterplot we identified four populations: 1 = cytoplasmic KDM6A (without DAPI) and 4 = DAPI (without KDM6A), as well as, populations #2 and #3, which resemble different ratios of KDM6A-DNA complexes. Population 2 is located at the lamina and in nucleolar puncta, and population 3 is located within the nucleoplasm. Substitution variant T726K shows minor differences in populations #2 and #4, as the lamina in #2 is less clearly visible, but strongly enriched in population #4. Contour scatterplot of JmjC or TPR with DAPI identified only three populations, with population #1 being absent. Back mapping clearly indicates compartment-wise localization of different DAPI-KDM6A JmjC or TPR compositions.

Additional file 5: Figure S2. Activity assay and data analysis. In this section, we describe our data analysis workflow. A. delta (grey shaded area) in fluorescence emission spectrum of eGFP-KDM6A WT lysate before incubation and after incubation with GFP-trap dynabeads. For data analysis, the amount of pulled fluorescent protein was determined by subtracting the eGFP fluorescence emission (470 nm excitation) after incubation with GFP-trap dynabeads from eGFP fluorescence emission before incubation with eGFP-trap dynabeads. Fluorescence intensities in the range of 500–530 nm were integrated subsequently (later referred to as integrated fluorescence). B. Maximum-normalized spectra obtained from crude cell lysate of eGFP, eGFP-KDM6A WT or eGFP-KDM6A Q1133A samples. All fluorescence emission spectra of the substitution variants had a similar profile between 500 and 530 nm. This was used to compare the amount of protein for each variant. Q1133A variant is shown exemplarily for all other substitution variants. C. Fluorescence emission spectra of WT and ΔJmjC. The high signal of ΔJmjC towards the blue end of the spectrum is due to scatter, which could indicate protein instability under the given (low salt) buffer conditions. D. 1–238 ng demethylated product (H3K27me2 peptide, MW = 2945.5 g/mol) was obtained from at least nine data points from three independent experiments, each in triplicates and fitted with a four parameters logistic regression (4-PL) according to Eq. 1: y = d + a − d 1 + x c b $$ \mathrm{y}=\mathrm{d}+\frac{\mathrm{a}-\mathrm{d}}{1+{\left(\frac{\mathrm{x}}{\mathrm{c}}\right)}^{\mathrm{b}}} $$ (1) with y = assay readout absorption, x = amount of product in A., amount of KDM6A species, either by fluorescence in B. or amount of protein in D., a = background signal, d = maximum signal, b = slope factor, c = c50, x-value at half maximum y.The following fit results were obtained: a = 0.29, b = 1.27, d = 1.73 and c = 62.4 ng. E. The amount of protein was plotted against absorption (shown for KDM6A WT) and fitted accordingly with 4-PL, with c being the only open parameter. c50 for WT was determined with 518,000 +/− 56,000 AU. To correct for background and maximum signal fluctuation between experiments, they were normalized to the values obtained from the product curve fit (a = 0.29, d = 1.73). F. A defined amount of recombinant eGFP (27 kDa) was used to convert the x-axis from fluorescent signal to the amount of protein. The linear dependency was used to calculate the amount of eGFP-KDM6A (181 kDa) per fluorescent unit. The molar integrated fluorescence (500–530 nm) from emission spectra for recombinant eGFP (MW = 26.9 kDa) was determined as 2.34E+ 18 AU/mol from the relation 1 ng eGFP = 86,900 [AU]. Therefore, 1 ng eGFP-KDM6A WT would yield an integrated fluorescence signal of 1.29E+ 03 AU. G. Data from E. and F. were used to transform the x-Axis with the fluorescent signal into the amount of KDM6A protein,. Approximated from D., the respective c50 always corresponded to 62.4 product. This correlation was used to calculate the specific activity for each variant. Given the fixed time of 240 min per assay, the specific activity under these conditions was calculated. The c50 for all available substitution variants was obtained in the same manner, fixing all parameters except c, using it as a relative measure of activity. At least four independent triplicate measurements were obtained for each substitution variant. H. absorption for truncated variants, three repeats with triplicates each. Since the truncated variants showed an unusual spectrum with a high scatter fraction at lower wavelengths (see C.), a quantification was impossible. Therefore, we qualitatively confirmed, that all truncated variants with a JmjC domain are catalytically active (green boxes) and those without JmjC domain are catalytically dead (red boxes). The distribution of the signal over the background is an evidence for activity in all truncated variants with a JmjC domain. ΔJmjC and TPR do not exceed assay background levels (zero), while WT, ΔIDR, ΔTPR and JmjC are all above background. I. Activity of T726K is dependent on post-expression time. WT activity does not change significantly between 24 h (black) to 48 h (grey), whereas in T726K activity is slightly lower than WT after 24 h (blue) and strongly reduced after 48 h (cyan) when comparing c50-values.

Additional file 4: Figure S1. Bioinformatic analysis of KDM6A tetratricopeptide repeats. A. Alignment of annotated KDM6A TPRs using the TPRprediction and alignment tool published in [37]. All annotated KDM6A TPRs with the given amino acid range (see label) were compared for their TPR conservation to the consensus sequence as published in [38] with the most conserved amino acids W4, L7, G8, Y11, A20, Y24, A27 and P30. Identity and p-values are shown on the right site. The TPR motif could not be identified for TPR3 (aa 170–199). Color coding according to CLUSTAL W alignment. B. Secondary structure prediction of TPR3 using the Quick2D on the MPI Bioinformatics toolkit server [39, 40] indicated a high probability for aa 14–26 to form an α-helix, but aa 1–10 tend to form either an α-helix or a β-strand. Of note, the canonical aa 34 TPR repeating unit forms a helix-turn-helix motif [38]. In general, the N-terminal TPRs 1–4 show less strict conservation towards the canonical TPR sequence compared to the C-terminal TPRs 6–8.

Additional file 3: Table S3. P-Values and scoring results from Figs. 5 and 6. A. P-Values from Fig. 5. B. Scoring results used in Fig. 5. C. P-Values from Fig. 6.

Additional file 9: Figure S6. Pathway enrichment analysis of MS data from KDM6A-tagGFP2 cell lines. A. Data from quadruplicates of three independent experiments of two different KDM6A-tagGFP2 cell lines using the reactome database. MS data sets were applied to the reactome analysis database. The resulting output shows significantly enriched pathways. The five most interesting pathways are depicted: cell cycle, chromatin organization, DNA repair, RNA metabolism including rRNA processing and protein metabolism including ribosome biogenesis and post-translational modification of histones. B. Top ten proteins identified by MS in KDM6A-tagGFP2 stable cell lines VM-CUB1 and RT-112. Among the top ten proteins in KDM6A pull down and subsequent MS, Nucleophosmin (NPM1), Nucleolin, ribosomal subunits 40S and 60S, as well as histone variants, are detectable. Mutational KDM6A and KMT2C/D status and experiments with RT-112 and VM-CUB1 have been recently published [18].

Additional file 8: Figure S5. Western Blot control experiments and raw blots. A. eGFP-KDM6A TPR whole cell lysate and the corresponding IP fraction was separated on a Western Blot and stained with eGFP (CST) or RBBP5 (CST) antibodies. RBBP5 precipitated with eGFP-KDM6A TPR. B. Dynabeads washing control on same membrane, washed with 3x RIPA like buffer (RLB). While KDM6A WT pulls a clearly visible RBBP5 band and a low WDR5 band, both are undetectable in eGFP-transfected cells alone (negative control). C. Raw blots from Fig. 3. Lysate and Co-IP for selected KDM6A variants and eGFP. IP and corresponding lysate lanes were also on the same blot. All experimental conditions for the blots presented were kept constant. D. Raw blots from Fig. 1B/D. Protein lanes shown in main figures are in red boxes. The PageRuler Plus Prestained protein ladder (Thermo Fisher) was used for orientation.

Additional file 6: Figure S3. Control experiments for antibody staining specificity and endogenous KDM6A protein levels. SW-1710 and T-24 cells were transfected with KDM6A WT or ΔIDR variants and stained with two different antibodies (sc-514,859, Santa Cruz biotechnology and CST-33510, Cell Signaling) raised against epitopes within the KDM6A central region (IDR) and 2nd antibody labeled with AbStar Red (Abberior). Clearly, both antibodies (in red channel) detect transfected KDM6A WT protein (very strong signals in green channel), but not the ΔIDR variant. The untransfected control cells show endogenous KDM6A in both cell lines stained with both KDM6A antibodies.

Additional file 7: Figure S4. Localization study of KDM6A variants in normal and invasive urothelial (cancer) cells. A. KDM6A substitution and deletion variants are localized in the cytoplasm and nucleus in T-24 cells. Substitutions cause mild effects, whereas deletion variants cause stronger cellular responses such as DNA release and mitotic errors. B. Representative KDM6A deletion variants are localized in the cytoplasm and nucleus in HBLAK. Two different phenotypes have been observed: a milder phenotype (majority of cells) and a phenotype (minority of cells) which reproducibly causes stronger cellular responses such as DNA release and mitotic errors. DAPI counterstaining of the nucleus and the respective overlay of both channels (DAPI in blue and GFP in green) is displayed. Images are taken in 2048 × 2048 resolution using a widefield microscope Olympus FX81 with 60x oil objective with post-processing using HuygensPro20.10. Effects of KDM6A variants in SW-1710 cells is shown in Fig. 2.

As people increasingly rely on online media and recommender systems to consume information, engage in debates and form their political opinions, the design goals of online media and news recommenders have wide implications for the political and social processes that take place online and offline. Current recommender systems have been observed to promote personalization and more effective forms of informing, but also to narrow the user’s exposure to diverse content. Concerns about echo-chambers and filter bubbles highlight the importance of design metrics that can successfully strike a balance between accurate recommendations that respond to individual information needs and preferences, while at the same time addressing concerns about missing out important information, context and the broader cultural and political diversity in the news, as well as fairness. A broader, more sophisticated vision of the future of personalized recommenders needs to be formed - a vision that can only be developed as the result of a collaborative effort by different areas of academic research (media studies, computer science, law and legal philosophy, communication science, political philosophy, and democratic theory). The proposed workshop will set first steps to develop such a much needed vision on the role of recommender systems on the democratic role of the media and define the guidelines as well as a manifesto for future research and long-term goals for the emerging topic of fairness, diversity, and personalization in recommender systems., Dagstuhl Reports, Volume 9, Issue 11, pages 117-124