The RNA-binding protein TTP is a global post-transcriptional regulator of feedback control in inflammation

RNA-binding proteins (RBPs) facilitate post-transcriptional control of eukaryotic gene expression at multiple levels. The RBP tristetraprolin (TTP/Zfp36) is a signal-induced phosphorylated anti-inflammatory protein guiding unstable mRNAs of pro-inflammatory proteins for degradation and preventing translation. Using iCLIP, we have identified numerous mRNA targets bound by wild-type TTP and by a non-MK2-phosphorylatable TTP mutant (TTP-AA) in 1 h LPS-stimulated macrophages and correlated their interaction with TTP to changes at the level of mRNA abundance and translation in a transcriptome-wide manner. The close similarity of the transcriptomes of TTP-deficient and TTP-expressing macrophages upon short LPS stimulation suggested an effective inactivation of TTP by MK2, whereas retained RNA-binding capacity of TTP-AA to 3′UTRs caused profound changes in the transcriptome and translatome, altered NF-κB-activation and induced cell death. Increased TTP binding to the 3′UTR of feedback inhibitor mRNAs, such as Ier3, Dusp1 or Tnfaip3, in the absence of MK2-dependent TTP neutralization resulted in a strong reduction of their protein synthesis contributing to the deregulation of the NF-κB-signaling pathway. Taken together, our study uncovers a role of TTP as a suppressor of feedback inhibitors of inflammation and highlights the importance of fine-tuned TTP activity-regulation by MK2 in order to control the pro-inflammatory response.

(a) GFP and GFP-TTP were precipitated (IP) under native (N) and 2M urea (U) conditions by GFP-Nanotrap pulldown and analyzed by ECL-based Western blot. The input for each IP is shown left of the IP lanes. (b) GFP-TTP precipitates were treated with different RNases for 5 min at 37°C. After labelling the RNAs with radioactive 33 P, a 4-12% NuPAGE gradient gel was used to resolve GFP-TTP-RNA complexes that were transferred to a nitrocellulose membrane afterwards. The autoradiograph is shown. RNase A was the only RNase that digested TTP-associated RNAs to a minimal size under denaturing conditions. (c) Precipitated radiolabelled GFP-TTP-RNA and GFP-TTP-AA-RNA complexes after resolution by gel electrophoresis. Several concentrations of RNase A were tested in order to isolated TTP-RNA complexes above 100 kDa for iCLIP library preparation ("cut"). Importantly, immunoprecipitation using protein lysates from GFP-only expressing control macrophages did not result in co-purification of RNAs for cDNA library preparation (data not shown).

Supplementary Figure 3
Peak and density enrichment analysis of the genomic distribution of iCLIP crosslinks in dependence on the stimulation (a-c) The detailed distributions of iCLIP targets (peak enrichment) from GFP-TTP and GFP-TTP-AA expressing cells stimulated as indicated above the plots are shown as mean and standard deviation. For 6h/1h LPS (a) four (GFP-TTP) and three (GFP-TTP-AA) replicates were considered, respectively. For all other conditions (6h Dox (b) and 6h Dox/4h LPS (c)) the mean and standard deviations of data from two individual experiments are shown. (d,e) Complementary to the density enrichment for 6h Dox and 1h LPS shown in Figure 1d, density enrichments for the genomic features that are crosslinked to GFP-TTP and GFP-TTP-AA are depicted for 6h Dox (d) and 6h Dox/4h LPS (e).

Supplementary Figure 4
Confirmation of TTP and TTP-AA binding to novel targets (a) Examples of precipitates of GFP, GFP-TTP and GFP-TTP-AA used for RNA-IP experiments (Figs. 3a-c and 5a-c). (b) Insertion of 3´UTRs of Cxcl10 and Gdf15 mRNAs reduce the expression of Luciferase reporters in HeLa cells. * p  0.05. (c) Luciferase reporter expression using the 3´UTRs of Cxcl10 and Gdf15 was further decreased by cotransfection of either GFP-TTP or GFP-TTP-AA. A significantly enhanced reduction by GFP-TTP-AA compared to GFP-TTP was only observed for Gdf15. * p  0.05. (d) Reporter gene assay as for Cxcl10 and Gdf15 in (b) was done for the insertion of the Tnfaip3 3´UTR. * p  0.05. (e) Reporter gene assays as in (b) and (c) for the 3'UTR of Tnfaip3 mRNA. * p  0.05.

Supplementary Figure 5
Analysis of the translational regulation of TTP targets using ribosome profiling Fractionation of ribosome-associated mRNAs was performed using sucrose gradient centrifugation followed by gradient fractionation and quantification of specific mRNAs by RT-qPCR in the different fractions. The repression of polysomal loading differed from transcript to transcript. (a) Examples of polysome gradients for the different cells and the indicated conditions. Absorption at 254 nm is plotted against the fraction number.
(b) Association of TNF mRNA (left panels) with actively translating ribosomes (fractions 6-11) is shown by polysome profiling under the indicated conditions. For each fraction percentages in relation to relative total mRNA amounts of all fractions for each transcript were calculated and plotted against the fraction numbers. Expression of TTP or TTP-AA resulted in a clear reduction of polysomal loading of TNF mRNA and a shift to monosomal fractions (2-5). The corresponding distribution of -Actin control mRNA is shown on the right; it is not affected by TTP or TTP-AA expression. (c) The distribution of inflammation-related mRNAs for Cxcl10 (left) and Gdf15 (right) were analyzed in polysome profiling experiments. Reduction of polysomal loading by TTP or TTP-AA expression was evident for Gdf15. (d) Distribution of mRNAs for the following feedback inhibitors was analyzed by polysome profiling experiments under the indicated conditions: Dusp1 (left), Ier3 (middle) and Tnfaip3 (right panels) mRNAs. The most prominent effects by TTP or TTP-AA expression were observed for Ier3 mRNA after 1 or 2h LPS treatment.

Supplementary Figure 6 NF-κB-related proteins in dependence of TTP-and TTP-AA-expression
TTP-and TTP-AA-expression influence NF-κB activation and signaling. For quantification of this effect Western blot signals derived from Figure 4c and Figure S6 b, a minimum of two different exposures was normalized to at least three different exposures of the loading control blot that was derived from the same membrane. All background-reduced band intensities were determined using the Multi Gauge V3.2 software (FujiFilm). (a) Quantification of the Western Blot signals derived from Figure 4c, where cells of all three genotypes were pre-treated with doxycycline for 6h and consequently stimulated with LPS for 0, 15, 30, 60, 120 and 240 minutes. GAPDH served as loading control and was used for normalization of specific signals. (b) The NF-κB distribution upon differential lysis was analyzed by Western Blot. For this, nuclear and cytosolic fractions of the different cell types were analyzed for the indicated proteins in a timedependent manner. Tubulin served as a cytosolic and ATF-2 as a nuclear marker. (c) Effect of TTP/TTP-AA expression on the nuclear translocation of NF-κB. The nuclear to cytosolic ratio of NF-κB Western blot signals derived from (b) were calculated and plotted against the time of LPS treatment. Normalized signals are shown as mean with their standard deviations. * p ≤ 0.05. For quantification of Western blot signals a minimum of two different exposures was normalized to at least three different exposures of the loading control blot that was derived from the same membrane. All background-reduced band intensities were determined using the Multi Gauge V3.2 software (FujiFilm). The signals of GFP/EV-expressing cells were set to one in unstimulated and 6h Dox/2h LPS stimulated cells.

Supplementary Figure 9
A similar role of TTP and TTP-AA in Ier3-and Dusp1-expression of mouse embryonic fibroblasts (MEFs) (a) Sorted immortalized TTP-deficient MEFs transduced with the same vectors like the immortalized BMDMs were induced for expression of GFP, GFP-TTP and GFP-TTP-AA with doxycycline for a total of 6h. Cells were starved for 16h in serum-free medium and phosphorylation of TTP was induced by stimulating the cells for 90 min with 10% (v/v) FCS to activate the p38 MAPK /MK2 signaling pathway. Activated p38 MAPK and its downstream kinase MK2 are detected by phospho-specific antibodies as indicated. Phosphorylated TTP was detected by the pTTP S178 -specific antibody. GAPDH served as loading control. (b) Changes in Ier3 mRNA levels were measured in the TTP KO MEF cell lines. Cells were starved in serum-free medium overnight before inducing GFP/EV, GFP-TTP or GFP-TTP-AA expression for 4.5h with doxycycline. Then, cells were stimulated with 10% (v/v) FCS for 90 min, before the cells were harvested for mRNA expression analysis by qPCR. Values are shown as mean with their standard deviations. Changes in Dusp1 and Cxcl10 mRNA level were measured from the samples and showed similar TTP-dependence (data not shown). (c) Changes in Ier3 and Dusp1 protein levels were detected by Western blot under the same conditions as in (b). The arrow indicates for the position of Ier3 protein. Other bands in the same blot were unspecific. GAPDH served as a loading control. (d) and (e) Western blot signals from (c) were quantified. A minimum of two different exposures was normalized to at least three different exposures of the loading control blot that was derived from the same membrane. All background-reduced band intensities were determined using the Multi Gauge V3.2 software (FujiFilm). Protein levels for each genotype under unstimulated conditions were set to one. Quantified signals of Ier3 protein level (e) and Dusp1 (f) are shown as mean and standard deviations. * p ≤ 0.05. Quantification Supplementary Table 1 Distribution of cDNA counts, number of overlapping targets and percentages of them comparing GFP-TTP and GFP-TTP-AA 3´UTR iCLIP targets upon 6h doxycycline induction in combination with 1h LPS stimulation. Targets with >1, ≥5 or ≥10 cDNA counts in their 3´UTR were compared. Supplementary Table 2 Summary of the high-throughput sequencing approaches used in this study. For each method the samples that were analyzed and the different conditions for each sample are listed.

Supplementary Data Set 5
Composition of the compiled TNF-/NF-κB custom gene set and the associated RNASeq and RiboSeq DESeq2 analyses.

Supplementary Files 1 and 2
These two bed files summarize iCLIP mapped data for the groups of GFP-TTP (1) and GFP-TTP-AA (2) upon 6h Dox/1h LPS after peak enrichment (FDR<0.05) and allow the upload of the tracks in the UCSD Genome Browser to visualize crosslink events for both proteins.