Skip to main page content
U.S. flag

An official website of the United States government

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2023 Mar 1;16(3):dmm050120.
doi: 10.1242/dmm.050120. Epub 2023 Mar 29.

HuR modulation counteracts lipopolysaccharide response in murine macrophages

Affiliations

HuR modulation counteracts lipopolysaccharide response in murine macrophages

Isabelle Bonomo et al. Dis Model Mech. .

Abstract

Lipopolysaccharide (LPS) exposure to macrophages induces an inflammatory response, which is regulated at the transcriptional and post-transcriptional levels. HuR (ELAVL1) is an RNA-binding protein that regulates cytokines and chemokines transcripts containing AU/U-rich elements (AREs) and mediates the LPS-induced response. Here, we show that small-molecule tanshinone mimics (TMs) inhibiting HuR-RNA interaction counteract LPS stimulus in macrophages. TMs exist in solution in keto-enolic tautomerism, and molecular dynamic calculations showed the ortho-quinone form inhibiting binding of HuR to mRNA targets. TM activity was lost in vitro by blocking the diphenolic reduced form as a diacetate, but resulted in prodrug-like activity in vivo. RNA and ribonucleoprotein immunoprecipitation sequencing revealed that LPS induces a strong coupling between differentially expressed genes and HuR-bound genes, and TMs reduced such interactions. TMs decreased the association of HuR with genes involved in chemotaxis and immune response, including Cxcl10, Il1b and Cd40, reducing their expression and protein secretion in primary murine bone marrow-derived macrophages and in an LPS-induced peritonitis model. Overall, TMs show anti-inflammatory properties in vivo and suggest HuR as a potential therapeutic target for inflammation-related diseases.

Keywords: Anti-inflammatory agents; ELAVL1; HuR; LPS; RIP-seq; Tanshinone mimics.

PubMed Disclaimer

Conflict of interest statement

Competing interests The authors declare no competing or financial interests.

Figures

Fig. 1.
Fig. 1.
Chemical structure and synthesis of tanshinone mimics (TMs). (A) Chemical structure of standard DHTS-I 1 and unsubstituted TM6a 2 (top row), and of early lead 3/TM6n, dimethyl quinone 4ox/TM7nox, dimethyl diphenol 4red/TM7nred and dimethyl diacetate 5/TM8n (bottom row). (B) Synthesis of dimethyl quinone 4ox/TM7nox, dimethyl diphenol 4red/TM7nred and dimethyl diacetate 5/TM8n. Steps were as follows: (a) 2,6-dimethylphenyl boronic acid, aqueous Na2CO3, Pd(PPh3)4, 1,4-dioxane, 90°C, 16 h, Argon, 55% yield; (b) 1 M BBr3, dichloromethane (DCM), –78°C to 0°C, 3.5 h, Argon, 90% yield; (c) 1-Hydroxy-1,2-benziodoxol-3(1H)-one 1-oxide (IBX), dimethyl formamide (DMF), room temperature (r.t.), 3 h, 88% yield; (d) 4-methoxythiophenol, DMF, r.t., 1 h, Argon, 70% yield; (e) IBX, DMF, r.t., 30 min., 80% yield; (f) Ac2O, pyridine, DCM, r.t., 24 h, N2, 97% yield.
Fig. 2.
Fig. 2.
TM7nox binds to HuR and disrupts HuR–RNA binding ability in vitro. (A) Ligand-binding poses found by AutoDock4.2 and submitted to molecular dynamics (MD) simulations. The ligand is schematically shown in pink sticks, and HuR is represented in green. (B) Representation of the TM7nox exploration of the HuR cavity for each simulated pose. HuR is represented in green, and the ligand center of mass evolution during the trajectory is shown as colored spheres. (C) Theoretical TM7nox-binding mode PII, as suggested by our MD simulation. The ligand is shown in pink sticks and the protein in green. Main residues involved in interactions with the ligand are shown as green sticks. Nonpolar hydrogens are hidden for clarity. (D) Superposition of the HuR−RNA complex crystal structure with the final state of the dynamized PII. The secondary structures are depicted in gray; the loops are shown in light blue for the initial state and light green for the last frame of the MD simulation. (E) Representative RNA electromobility shift assay (REMSA) showing HuR–RNA binding impairment induced by TMs. rM1M2_HuR (3.7 nM) was incubated for 30 min with 1 nM 5′-DY681-labeled RNA probe alone, or together with DMSO used as control, or TM6n, TM7nox or TM7nred at 1 µM doses. (F) Representative REMSA showing TM7nox dose–response inhibition of the binding between 100 nM rM1M2_HuR and 1 nM 5′-DY681-labeled RNA probe. (G) Concentration–response analysis of TM7nox tested in HuR:RNA probe interaction assay. In a dose-dependent manner, TM7nox (0.01 μM, 0.1 μM, 1 μM, 5 μM, 10 μM, 25 μM, 50 μM, 100 μM) interferes with the binding between His-tagged recombinant M1M2 HuR protein (20 nM) and 5′-Bi-TNF ARE probe (50 nM). The calculated half-maximal inhibitory concentration (IC50) is 0.7956 µM, and data have been normalized to control (DMSO). Data fit nonlinear regression fitting curves according to a one-site binding model in GraphPad Prism. Plotted are mean±s.d. of three independent experiments.
Fig. 3.
Fig. 3.
RNA-seq analyses reveal that TM7nox modulates lipopolysaccharide (LPS)-induced response. (A) Principal component analysis (PCA) of the 12 samples. PC1 shows 83% variance and PC2 7%. Each dot represents a DMSO sample, each triangle is a sample treated with DMSO+LPS, and each square is a DMSO+LPS+TM7nox-treated sample. Every condition groups together with the same type samples, and it can be observed that the effect of TM7nox separates DMSO+LPS from DMSO+LPS+TM7nox conditions. (B) Venn diagram of differentially expressed genes (DEGs); the numbers in each circle represent the number of DEGs between the different comparisons while the ones overlapping are for mutual DEGs (DMSO+LPS+TM7nox versus DMSO+LPS in red, DMSO+LPS+TM7nox versus DMSO in green, DMSO+LPS versus DMSO in blue). (C) Heatmap of z-score of differential genes across different samples, each grouping together with its own sample type (red, DMSO; green, DMSO+LPS; blue, DMSO+LPS+TM7nox). The track on the right shows differential gene expression between different comparisons (black, downregulated DEGs; gray, upregulated DEGs; white, no changes in DEGs). ‘Average’ was used as clustering method and ‘correlation’ for clustering distance of both rows and columns. (D) Bar plot of z-score of key genes across different samples (red, DMSO; green, DMSO+LPS; blue, DMSO+LPS+TM7nox). *P<0.05, **P<0.01, ***P<0.001; ns, not significant (Welch's t-test). The whiskers above/below the bars extend from the upper/lower quartile to the highest/lowest actual value that is within the [75th percentile±1.5×(interquartile range)]. (E) Tree plot of enriched terms deriving from the 249 downregulated genes. Each dot represents an enriched term, colored according to P-adjusted values, spanning from red to blue. Terms' dimensions are relative to the number of genes found to be enriched in that category. The subclusters and their names, visible on the right, are highlighted with a specific color. (F) Network visualization of the top-five-ranking Gene Ontology (GO) pathways for the 249 downregulated genes. Each pathway is represented by a gray dot and highlighted with a particular color; each gene is connected to the pathway it belongs to. Genes are represented in dots ranging from blue to white according to their log2FC values; the sizes of the pathway dots depends on the number of genes enriched for the pathway itself.
Fig. 4.
Fig. 4.
Ribonucleoprotein immunoprecipitation followed by sequencing (RIP-seq) analysis reveals that TM7nox displaces the RNA targets from HuR. (A) Venn diagram of DEGs; the numbers in each circle represent the number of DEGs between the different comparisons while the ones overlapping are for mutual DEGs (DMSO+LPS+TM7nox versus DMSO+LPS in green, DMSO+LPS+TM7nox versus DMSO in blue, DMSO+LPS versus DMSO in red). (B) Heatmap of z-score of differential genes across different samples, each grouping together with its own sample type (red, DMSO; green, DMSO+LPS; blue, DMSO+LPS+TM7nox). The track on the right shows differential gene expression between different comparisons (black, downregulated DEGs; gray, upregulated DEGs; white, no changes in DEGs). ‘Average’ was used as clustering method and ‘correlation’ for clustering distance of both rows and columns. (C) (Left) Boxplot of 3′-UTR length of the upregulated and downregulated DEGs in the TM7nox and LPS co-treatment versus LPS and DMSO RIP-seq jointly with non-differential genes (ND). Upregulated genes are in yellow, downregulated genes are in blue, non-differential genes are in gray. Wilcoxon test was performed between the three classes. (Right) Boxplot of the number of AU/U-rich elements (AREs) of the upregulated and downregulated DEGs in the TM7nox and LPS co-treatment versus LPS and DMSO RIP-seq jointly with non-differential genes. Upregulated genes are in yellow, downregulated genes are in blue, non-differential genes are in gray. Wilcoxon test was performed between the three classes. (D) Workflow of the gene-filtering process starting from the LPS versus DMSO and TM7nox+LPS versus DMSO comparisons in both RNA-seq (P-adjusted<0.05, log2FCe>1) and RIP-seq (P-adjusted<0.05, log2FC>3), identifying two different subsets of genes (421 genes with correlation value of 0.84 for LPS versus DMSO and 362 genes with correlation value of 0.73 for TM7nox+LPS versus DMSO comparison). 278 genes were thus identified to be commonly shared between these two subsets. Moreover, of the 82 genes resulting from the filtering with the TM7nox+LPS versus LPS+DMSO downregulated RIP-seq comparison (P-adjusted<0.05), 20 were found to be of particular interest (Table 1). (E) (Left) Correlation between the RNA-seq z-score values and RIP-seq z-score values for the LPS and DMSO comparison. Each black dot represents a gene present in both experiments with a log2FC>1 for RNA-seq and log2FC>3 for RIP-seq. Regression line is depicted in red. Density plots of the distributions for each subset of values are shown on the right for RNA-seq z-score values and on the top for the RIP-seq z-score values. (Right) Correlation between the RNA-seq z-score values and RIP-seq z-score values for the TM7nox and LPS co-treatment versus DMSO comparison. Each black dot represents a gene present in both experiments with a log2FC>1 for RNA-seq and log2FC>3 for RIP-seq. Regression line is depicted in red. Density plots of the distributions for each subset of values are shown on the right for RNA-seq z-score values and on the top for the RIP-seq z-score values. (Bottom) 3D-rendered visualization from different angles of the difference between the correlation densities for LPS and DMSO comparison and TM7nox and LPS co-treatment versus DMSO comparison. On the axes, RNA-seq z-score values, RIP-seq z-score values and distribution density differences are shown. Color varies from purple to yellow according to the difference in the distribution values. (F) Network visualization of the top-five-ranking GO pathways of a subset of 82 genes of interest. Each pathway is represented by a gray dot and highlighted with a particular color; each gene is connected to the pathway it belongs to. A subset of 20 genes of interest is highlighted in red.
Fig. 5.
Fig. 5.
TM7nox/TM7nred decrease the binding between HuR and identified targets, reducing their expression level and Cxcl10 secretion in RAW 264.7 cells and bone marrow-derived macrophages (BMDMs). (A) RAW 264.7 cells were co-treated for 6 h with DHTS (5 µM), TM6n and TM7nox (10 µM), LPS (1 µg/ml) and DMSO as control. mRNA levels of Cxcl10, Cd40, Fas, Nos2 and Il1b were assessed using qRT-PCR with Rplp0 as housekeeping gene; data were normalized to LPS+DMSO condition. Data plotted as mean±s.d. of biological quadruplicate (*P<0.05, **P<0.01, ***P<0.001, ****P<0.0001; Welch's t-test). (B) TM7nox impairs the binding between HuR and identified targets. RIP-seq results were validated by RNA immunoprecipitation (RIP) followed by quantitative real-time PCR (qRT-PCR). RAW 264.7 cells were treated for 6 h with DMSO alone and LPS (1 µg/µl)+DMSO as controls, and LPS (1 µg/ml)+TM7nox (10 µM). Subsequently, cells were lysed, and RNA was precipitated with anti-HuR antibody [immunoprecipitation (IP)] and IgG isotype (IgG) as negative control. Changes in the mRNAs bound to HuR in the control or treatment were evaluated through qRT-PCR and normalized to the corresponding values obtained with IgG as negative control. The obtained numbers indicate the fold enrichment; experiments were performed in biological triplicate (*P<0.05, **P<0.01; ns, not significant; Welch's t-test). (C) Pull-down assays performed in RAW 264.7 cells pre-treated with TM7nox or DMSO as control for 6 h. Cell lysates were incubated for 1 h at 4°C with either biotinylated probe containing HuR consensus sequence, or probe not supposed to bound by HuR as a negative control. Precipitations of the probes were carried out with streptavidin beads, and HuR levels were detected by western blot analysis. HuR signal was quantified as the input (10%) and normalized to the DMSO sample. Data plotted as mean±s.d. of three independent experiments (*P<0.05; Welch's t-test). (D) BMDMs were co-treated for 6 h with TM6n and 7nox at 10 µM doses, LPS (1 µg/ml) and DMSO as control. mRNA levels of Cxcl10, Cd40, Fas, Nos2 and Il1b were assessed using qRT-PCR with Rplp0 as housekeeping gene; data were normalized to the LPS+DMSO condition. Data plotted as mean±s.d. of biological quadruplicate (**P<0.01, ***P<0.001; Welch's t-test). (E) TM7nox, TM6n, DHTS and TM7nred treatment reduce Cxcl10 secretion in RAW 264.7 cell supernatants. Cxcl10 protein levels were measured with ELISA. RAW 264.7 cells were treated for 6 h with DMSO as control, LPS (1 µg/ml) plus DMSO or TM7nred (10 µM). Relative quantity of Cxcl10 pg/ml for each sample was measured according to the number of cells quantified through Crystal Violet assay. Data were normalized to LPS+DMSO as control, and numbers are expressed as a percentage. Data represent as mean±s.d. of biological triplicate (**P<0.01, ***P<0.001; Welch's t-test). (F) TM7nred treatment reduces Cxcl10 secretion in BMDM supernatants. Cxcl10 protein levels were measured with ELISA. RAW 264.7 cells were treated for 6 h with DMSO as control, LPS (1 µg/ml) plus DMSO or TM7nred (10 µM). Relative quantity of Cxcl10 pg/ml for each sample was measured according to the number of cells quantified through Crystal Violet assay. Data were normalized to LPS+DMSO as control, and numbers are expressed as a percentage. Data represent as mean±s.d. of biological quadruplicate (**P<0.01, ***P<0.001; Welch's t-test).
Fig. 6.
Fig. 6.
TM7nox/TM7nred recapitulate partially HuR silencing without changing NF-κB translocation and HuR subcellular localization. (A) TM7nox and TM7nred treatment reduce Cxcl10 (top) and Il1β (bottom) intracellular levels in RAW 264.7 cells. Protein levels were measured with ELISA. RAW 264.7 cells were treated for 6 h with DMSO as control, LPS (1 µg/ml) plus DMSO or TMs (10 µM). Respectively, 30 µg and 5 µg of cellular lysates were loaded to measure Cxcl10 and Il1β (pg/ml). Data were normalized to LPS+DMSO as control, and numbers are expressed as a percentage. Data represent mean±s.d. of biological triplicate (*P<0.05, **P<0.01, ***P<0.001; Welch's t-test). (B) qRT-PCR of target mRNAs in siSCR and siHuR, after 6 h of co-treatment with LPS 1 µg/ml plus DMSO, DMSO alone to control LPS stimulation, or 10 µM TM7nox or TM7nred in RAW 264.7 cells. Data plotted as mean±s.d. are from three independent experiments (*P≤0.05, **P≤0.01, ***P≤0.001, ****P<0.0001; ns, not significant; Welch's t-test). (C) TM7nox affects the transcription of LPS-induced cytokines. RAW 264.7 cells were co-treated with DMSO, LPS+DMSO, LPS+TM7nox for 3 h. Act-D (2.5 µM) was then added/administered for 1.5 h or 3 h. qRT-PCR was performed to quantify the remaining Cxcl10, Il1b, Cd40, Fas, Nos2 and Gapdh mRNA levels. Data plotted as mean±s.d. of biological triplicate (*P≤0.05, **P≤0.01, ***P≤0.001, ****P<0.0001; Welch's t-test).
Fig. 7.
Fig. 7.
TMs modulate Cxcl10 and Il1β secretion in LPS-induced peritonitis mouse models. (A) Left and middle panels show representative immunofluorescence HuR localization after LPS administration, single-compound treatment and in combination with LPS (1 µg/ml). The right panel shows representative immunofluorescence showing that HuR cytoplasm accumulation induced by actinomycin D (ActD) does not change upon treatment with different TMs (10 µM). Cells were treated for 3 h in combination with ActD (2.5 µM). DMSO alone or in combination with ActD and LPS was used as control. In the graph, the ratio of HuR fluorescent signal between nucleus and cytoplasm (N/C) is plotted. For image acquisition (40× high-NA objective), Operetta was used, and evaluation was carried out by selecting 13 fields/well. The N/C ratio represents the mean±s.d. of single cells for every well (****P<0.0001; Welch's t-test). (B) Representative immunofluorescence showing that NF-κB nuclear translocation inside the cells induced by LPS does not change upon treatment with different TMs (10 µM). Cells were treated for 3 h in combination with LPS (1 µg/ml). To obtain a positive control given by high induction of NF-κB related to massive shuttling in the nucleus, we treated cells with 2.5 µM ActD; DMSO, either alone or with LPS, was used as negative control. In the graph, the ratio of NF-κB fluorescent N/C signal is plotted. For image acquisition (40× high-NA objective), Operetta was used, and evaluation was carried out by selecting 13 fields/well. The N/C ratio represents the mean±s.d. of single cells for every well (*P<0.05; ***P<0.0001; ns, not significant; Welch's t-test). (C) Levels of identified cytokines through Luminex analysis in sera from C57BL/6j wild-type mice after administration of LPS (150 μg/25 g of body weight) and TMs (40 mg/kg) or DMSO for 2 h. Data were normalized to LPS+DMSO and are expressed as mean±s.d. as a percentage of sex-balanced mouse group in which n=6-8 (*P<0.05, ***P<0.001; Welch's t-test). (D) Cxcl10 and Tnf levels measured with ELISA in the sera from C57BL/6j wild type mice after administration of LPS (150 μg/25 g of body weight) with TMs (40 mg/kg) or DMSO for 2 h. DMSO alone was used as a control for LPS inflammatory response insurgence, and LPS+DMSO was considered as the drug vehicle. Bar graphs show mean±s.d. from five mice per group (*P<0.05, **P≤0.01; Welch's t-test). Scale bars: 20 μm.

Similar articles

Cited by

References

    1. Abdelmohsen, K., Lal, A., Kim, H., Gorospe, M., Hyeon, H. K. and Gorospe, M. (2007). Posttranscriptional orchestration of an anti-apoptotic program by HuR. Cell Cycle 6, 1288. 10.4161/cc.6.11.4299 - DOI - PubMed
    1. Abdelmohsen, K., Gorospe, M., Abdelmosen, K. and Gorospe, M. (2010). Posttranscriptional regulation of cancer traits by HuR. Wiley Interdiscip. Rev. RNA 1, 214-229. 10.1002/wrna.4 - DOI - PMC - PubMed
    1. Abraham, M. J., Murtola, T., Schulz, R., Páll, S., Smith, J. C., Hess, B. and Lindahl, E. (2015). GROMACS: high performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1-2, 19-25. 10.1016/j.softx.2015.06.001 - DOI
    1. Anderson, P. (2010). Post-transcriptional regulons coordinate the initiation and resolution of inflammation. Nat. Rev. Immunol. 10, 24-35. 10.1038/nri2685 - DOI - PubMed
    1. Assoni, G., La Pietra, V., Digilio, R., Ciani, C., Licata, N. V., Micaelli, M., Facen, E., Tomaszewska, W., Cerofolini, L., Pérez-Ràfols, A.et al. (2022). HuR-targeted agents: an insight into medicinal chemistry, biophysical, computational studies and pharmacological effects on cancer models. Adv. Drug Deliv. Rev. 181, 114088. 10.1016/J.ADDR.2021.114088 - DOI - PubMed

Publication types