Activation of TLR7–9 has been linked to the pathogenesis of autoimmune diseases such as rheumatoid arthritis, systemic lupus erythematosus, and psoriasis. Thus, therapeutic applications of antagonists of these TLRs for such disorders are being investigated. Bortezomib (Velcade) is a proteasome inhibitor known to suppress activation of these TLRs. To identify novel TLR7–9 inhibitors, we searched the Gene Expression Omnibus database for gene expression profiles of bortezomib-treated cells. These profiles were then used to screen the Connectivity Map database for chemical compounds with similar functions as bortezomib. A natural antibiotic, thiostrepton, was identified for study. Similar to bortezomib, thiostrepton effectively inhibits TLR7–9 activation in cell-based assays and in dendritic cells. In contrast to bortezomib, thiostrepton does not inhibit NF-κB activation induced by TNF-α, IL-1, and other TLRs, and it is less cytotoxic to dendritic cells. Thiostrepton inhibits TLR9 localization in endosomes for activation via two mechanisms, which distinguish it from currently used TLR7–9 inhibitors. One mechanism is similar to the proteasome inhibitory function of bortezomib, whereas the other is through inhibition of endosomal acidification. Accordingly, in different animal models, thiostrepton attenuated LL37- and imiquimod-induced psoriasis-like inflammation. These results indicated that thiostrepton is a novel TLR7–9 inhibitor, and compared with bortezomib, its inhibitory effect is more specific to these TLRs, suggesting the potential therapeutic applications of thiostrepton on immunologic disorders elicited by inappropriate activation of TLR7–9.
Toll-like receptors are a family of evolutionarily conserved pattern recognition receptors expressed in innate immune cells that are essential for the detection of a wide variety of microbial pathogen-associated molecular patterns (PAMPs) and subsequent initiation of host immune responses (1, 2). Ten TLRs have been identified in human cells (3). Of these, TLR1, TLR2, TLR4, TLR5, and TLR6 are expressed on the cell surface to detect PAMPs such as lipoprotein, zymosan, LPS, and flagellin. In contrast, TLR3, TLR7, TLR8, and TLR9 localize in the endoplasmic reticulum and traffic to the endosomal/lysosomal compartment to initiate cellular responses upon activation by microbial nucleic acids (4–8).
TLRs are type I transmembrane receptors that contain a cytosolic region, transmembrane region, and ectodomain. The cytosolic region contains a Toll/IL-1R domain for protein–protein interactions with the Toll/IL-1R domains of MyD88 adaptor protein family members; these interactions initiate intracellular signaling. The ectodomain is characterized by multiple leucine-rich repeats (LRRs). A TLR generally contains 19–25 LRRs. These LRRs are arranged in a horseshoe-shaped solenoid structure for ligand binding (9, 10). Of all the TLRs, TLR7, TLR8, and TLR9 (TLR7–9) share a more close phylogenetic relationship and thus comprise a subfamily. Additionally, in contrast to other TLRs, these three TLRs have a unique undefined region between LRR14 and LRR15 that may play a role in ligand binding (11–14).
The MyD88 adaptor protein family contains five members: MyD88, TRIF/TICAM-1, TIRAP/Mal, TIRP/TRAM, and SRAM (15). All TLRs signal through a MyD88-dependent pathway with the exception of TLR3, which uses a TRIF-dependent signaling pathway. Ligand-induced TLR7–9 activation triggers the sequential recruitment of MyD88, IL-1R–associated-kinase (IRAK), and TNFR-activated factor 6 (TRAF6) to form a complex that in turn activates downstream TGF-β–activating kinase, leading to the activation of important transcription factors such as NF-κB and IFN regulatory factors (16–18). These transcription factors play key roles in regulating the expression of adherent and costimulatory molecules and regulating the production of different cytokines such as TNF-α, IL-1β, IL-6, IL-12, IL-23, and type I IFNs, which are crucial for the maturation, differentiation, and proliferation of dendritic cells (DCs), NK cells, and cytotoxic T cells. Because these immune responses promote the killing of virus-infected cells and tumor cells, TLR7–9 agonists are being investigated for their applications as vaccine adjuvants and for their uses in the treatment of various infectious diseases and cancers (19–21).
In general, TLR7–9 differentiates between pathogen-derived and self-derived nucleic acids. The nucleic acids derived from viruses during cytosolic replication can be transported into endosomes via autophagy. The expression of these TLRs in intracellular compartments prevents the activation of these TLRs by self-derived nucleic acids under physiological conditions because self–nucleic acids from dead cells within damaged tissues are unable to enter the cells and endosomes spontaneously (22, 23). However, self–nucleic acid tolerance can be broken under some pathological conditions. For example, LL37 is an endogenous antimicrobial peptide that can be overproduced and released into inflammatory sites in psoriatic skin. This antimicrobial peptide has been shown to form complexes with self–nucleic acids, thus rendering these nonstimulatory self–nucleic acids into potent immune stimuli (24–26). This finding suggests a role for self–nucleic acids in inducing psoriatic inflammation via TLR7–9 activation. Similarly, inappropriate TLR7–9 activations have been linked to the pathogenesis of other autoimmune diseases such as rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE). Therefore, antagonists for these TLRs are being investigated for their applications in autoimmune disease treatment (27–29).
Bortezomib (Velcade) is a proteasome inhibitor approved for the treatment of multiple myeloma. This drug is also an NF-κB inhibitor, and its inhibitory effects on autoimmune disorders such as psoriasis, RA, and SLE have been shown in animal models (30–32). Although the functional mechanism has not yet been fully elucidated, bortezomib has been shown to inhibit TLR9 activation by suppressing trafficking of this TLR to the endolysosomes and reducing the inflammation induced in lupus and psoriasis (33, 34). The Connectivity Map (CMap) database collects gene expression data from cultured human cells treated with bioactive small molecules. This database enables the discovery of functionally similar chemicals through comparisons of the gene expression profiles of cells treated with these chemicals (35, 36). In an attempt to identify a novel TLR7–9 inhibitor, we acquired a set of gene expression profiles from bortezomib-treated cells from the Gene Expression Omnibus (GEO) database. These gene expression profiles were then used to screen the CMap database for chemical compounds with similar functions as bortezomib. In this study, we report that the antibiotic thiostrepton (37) is a novel inhibitor of TLR7–9 and of the psoriatic inflammation induced by these TLRs. This antibiotic is less toxic and more specific than bortezomib with respect to the inhibition of TLR7–9 activation and has a functional mechanism distinct from currently known TLR7–9 inhibitors.
Materials and Methods
Reagents and Abs
Thiostrepton, bortezomib, MG132, MG262, parthenolide, phenoxybenzamine, and 15-δ-PGJ2 were purchased from Sigma-Aldrich (St. Louis, MO). TLR ligands, including Pam3Cys, polyinosinic-polycytidylic acid, flagellin, R848, and CL075, were purchased from InvivoGen (San Diego, CA). CpG-oligodeoxynucleotide (CpG-ODN) was purchased from Invitrogen (Carlsbad, CA) or Genomics BioSci & Tech (New Taipei, Taiwan). LL37 was purchased from GeneDirex (Gueishan Township, Taiwan). Cytokines, including IL-1β, TNF-α, IL-4, GM-CSF, and Flt3 ligand, were purchased from PeproTech (Rocky Hill, NJ). Anti-IκBα, anti–phospho-IκBα, and anti–β-actin Abs were purchased from Cell Signaling Technology (Beverley, MA). Anti-GAPDH, anti-LAMP2, and anti-Rab5 Abs were purchased from GeneTex (Irvine, CA). FITC-conjugated anti-CD11b and anti-CD11c Abs were purchased from eBioscience (San Diego, CA). Luciferase assay reagents were purchased from Promega (Madison, WI).
Cell culture and generation of DCs
Human embryonic kidney (HEK)293 cells were grown in DMEM supplemented with 10% FBS. DCs were derived from mouse bone marrow cells. In brief, bone marrow cells were isolated from 6- to 8 wk-old C57BL/6 mice. Two induction methods as previously reported (38, 39) were used to generate DCs. When DCs were generated by culture in RPMI 1640 supplemented with 10% FCS and induction with GM-CSF (20 ng/ml) plus IL-4 (20 ng/ml) for 10 d, the DC was defined as a GM-CSF–induced DC (GM-DC). When induction was achieved by Flt3 ligand (100 ng/ml) for 10 d, the DC was defined as a Flt3L-induced DC (Flt3L-DC).
Bioinformatic database analysis
The GEO database (http://www.ncbi.nlm.nih.gov/gds) was searched for microarray databased gene expression profiles of bortezomib-treated cells. A GEO dataset (GSE30931) was selected to analyze the expression of genes upregulated and downregulated by bortezomib. The CMap database (https://www.broadinstitute.org/cmap/) was screened using the bortezomib upregulated and downregulated gene profile as input data. Candidates with positive connections (connectivity score > 0.85) were selected for further analysis.
TLR activation assays
To perform TLR activation assays, HEK293 cells were plated on 24-well plates and allowed to adhere overnight. Using polyethylene glycol, these cells were cotransfected with a TLR expression vector and an NF-κB–driven luciferase reporter plasmid. On the next day, the cells were treated with different inhibitors for 30 min and then treated with 2 μM or various concentrations of various stimuli as indicated for 7 h. The cells were lysed and the luciferase activity in each sample was determined. Relative luciferase activities were calculated as fold inductions compared with an unstimulated control. Data are expressed as mean ± SD (n = 3 independent experiments).
The cytotoxicities of different inhibitors were measured with the CellTiter 96 AQueous Non-Radioactive Cell Proliferation (MTS) assay according to the manufacturer’s instructions (Promega). Briefly, cells were treated with various concentrations of thiostrepton or bortezomib for 24 h or different time periods as indicated. MTS/PMS solution was added into each well. After 2 h, the absorbances at 490 nm were measured using an EnVision multilabel plate reader (PerkinElmer, Waltham, MA).
Flow cytometric analysis
For flow cytometric analysis, cells were suspended in PBS containing 2% FCS and incubated with FITC-conjugated Abs as indicated at 4°C for 30 min. After washing, cells were analyzed on a FACSCalibur flow cytometer with CellQuest software (Becton Dickinson, San Jose, CA).
ELISA assay for cytokine productions
Mouse GM-DCs and Flt3L-DCs (3 × 105/ml) were treated with thiostrepton or bortezomib for 1 h, after which 2 μM R848, 1.25 μM CpG-1826, or 1 μg plus 1 μg/ml LL37 plus RNA and LL37 plus DNA complex, respectively, were added for 24 h. The cell culture media were collected for cytokine measurement. Mouse IL-12/23p40, TNF-α, and IFN-α were measured using ELISA kits from eBioscience according to the manufacturer’s instructions. Genomic DNA and total RNA were extracted from RAW264.7 cells using the QIAamp DNA mini kit and RNeasy mini kit, respectively, from Qiagen (Venlo, the Netherlands).
Measurement of 20S proteasome activity
The inhibitory effects of thiostrepton and other inhibitors on 20S proteasome activity were measured using a proteasome activity assay kit (Chemicon/EMD Millipore, Temecula, CA) according to the manufacturer’s instructions. Briefly, cell lysates were treated for 30 min with and without various concentrations of inhibitors as indicated and then lysed. The cell lysates were incubated with the labeled substrate LLVY–7-amino-4-methylcoumarin (LLVY-AMC) for 2 h. After proteasomal cleavage from the substrate, fluorescence of the free AMC was quantified via fluorometry with a 380/460 nm filter set.
Measurement of endosomal acidification
pHrodo Green dextran (Invitrogen) emits pH-sensitive fluorescence emission that increases in intensity with increasing acidity. To measure the effects of thiostrepton and other inhibitors on endosomal acidification, cells were treated with various concentrations of inhibitors as indicated for 30 min and were subsequently treated with pHrodo Green dextran for 30 min. The fluorescence intensity was analyzed with a FACSCalibur flow cytometer and CellQuest software.
Analysis of the endolysosomal localization of TLR9
To analyze the endolysosomal localization of TLR9, HEK293 cells were transiently transfected overnight with an expression vector for TLR9-GFP or TLR9-FLAG. These cells were treated with CpG-ODN and thiostrepton, as indicated. For immunofluorescence to visualize localization of TLR9 on endolysosomes, cells were stained with LysoTracker (Invitrogen) and observed using a fluorescence microscope (Leica TSC SP5II). For immunoblotting to detect the presence of TLR9 on endolysosomes, cells were homogenized and fractionated on sucrose density gradient (Pierce) for isolation of the endolysosome-enriched fractions, which were confirmed by detection of the endosomal and lysosomal proteins LAMP2 and Rab5 in the fraction with immunoblotting.
Preparation of thiostrepton gel
The 5% thiostrepton gel was formulated based on a previously described method (40). To prepare this gel (10 g), hydroxypropyl cellulose powder (0.3 g) was dissolved in distilled water with stirring. Thiostrepton (0.5 g) and azone (0.1 g) were mixed with 95% ethanol (1.7 g) until completely dissolved. The thiostrepton solution was then mixed with the aqueous hydroxypropyl cellulose solution. This mixture was stirred continuously until a gel formed.
The animal experiments were approved by the Institutional Animal Care and Use Committee of the National Health Research Institutes, Taiwan. C57BL/6J mice were maintained and handled in accordance with the stated guidelines. To evaluate the inhibitory effect of thiostrepton on psoriasis-like inflammation, three different murine psoriasis-like models were established. In the first model, LL37 (50 μg), LL37 plus thiostrepton (50 μg plus 5 or 10 μg), or a control vehicle was injected into the right ear root of a C57BL/6J mouse twice per week according to the indicated schedule. The ear thicknesses were measured on day 27. In the second model, LL37 (50 μg) or control vehicle was injected in the right ear root each week, and 0.25 mg 5% thiostrepton or control gel was smeared on the injection site each week according to the indicated schedule. The ear thicknesses were measured on day 56. In the third model, 0.25 mg 5% imiquimod gel (Aldara) was mixed in a 1:1 ratio with 5% thiostrepton gel or control gel. These gels were smeared on the right ear root every week according to the indicated schedule. Ear thicknesses were measured on day 56.
For the histological analysis, isolated ear tissues were immersed in 10% formalin. These samples were then embedded in paraffin wax, and sections were cut for H&E staining.
Cytokine induction in mouse sera and tissues
To investigate the inhibitory effects of thiostrepton on imiquimod-induced cytokine productions, mouse ears were intradermally (i.d.) injected with 5 μg imiquimod mixed with/without 10 μg thiostrepton in PBS, or PBS alone. Serum samples were collected at indicated times for analysis of cytokine levels with ELISA. To investigate the inhibitory effects of thiostrepton on LL-37–mediated cytokine induction, mouse ears were i.d. injected with 50 μg LL-37, 50 μg LL37 mixed with 10 μg thiostrepton in PBS, or PBS alone. After 48 h, whole mouse ears were collected and lysed with Isol-RNA lysis reagent for total RNA isolation according to the manufacturer’s protocol (5 Prime, Hilden, Germany). First-strand cDNA was then synthesized from total RNA samples using a SuperScript preamplification kit (Invitrogen). Cytokine induction was determined via quantitative RT-PCR with a Kapa SYBR Fast qPCR kit (Kapa Biosystems, Wilmington, MA). The respective PCR primer pair sequences (forward/reverse) for mouse IL-1β, TNF-α, IL-8, and β-actin were 5′-CAGGCAGGCAGTATCACTCA-3′ and 5′-AGCTCATATGGGTCCGACAG-3′, 5′-AGCCCCCAGTCTGTATCCTT-3′ and 5′-CTCCCTTTGCAGAACTCAGG-3′, 5′-CGTCCCTGTGACACTCAAGA-3′ and 5′-TAATTGGGCCAACAGTAGCC-3′, 5′-AGCCATGTACGTAGCCATCC-3′ and 5′-CTCTCAGCTGTGGTGGTGAA-3′.
A Student t test was used to analyze the results, and a p value <0.05 was considered significant. Data are expressed as mean ± SD of three independent experiments.
Exploration for a novel TLR7–9 inhibitor by screening the CMap database with a bortezomib-induced gene expression profile
The CMap database collects numerous gene expression profiles of cultured cells treated with various chemicals compounds (35, 36). To explore novel TLR7–9 inhibitors, we designed a strategy as shown in Fig. 1A. According to this strategy, we first searched the National Center for Biotechnology Information GEO database for the gene expression profiles of cells treated with bortezomib. A set of microarray data from MCF-7 cells treated with bortezomib (accession no. GSE30931) was identified and analyzed. The top 20 upregulated genes and top 20 downregulated genes (Supplemental Table I) were selected as the input file for a database screening CMap query. A number of compounds with positive connections were identified from this screening, and those with connectivity scores >0.7 are listed in the table shown in Fig. 1B.
Identification of thiostrepton as a novel TLR9 inhibitor
Of the identified compounds, MG132 and MG262 are proteasomes and NF-κB inhibitors (41). Thiostrepton is an antibiotic derived from several strains of streptomycetes (37). Phenoxybenzamine is an α-1 adrenoceptor inhibitor and is used as an antihypertensive agent (42). 15-δ-PGJ2 is a peroxisome proliferator–activated receptor γ agonist, and parthenolide is an NF-κB inhibitor with anti-inflammatory and antiproliferation activities (43, 44). These compounds with connectivity scores >0.85 were first examined for their TLR9 activation inhibitory activities. A cell-based TLR9 activation assay was established by cotransfecting a TLR9 expression vector and NF-κB controlled luciferase reporter gene into HEK293 cells and then stimulating these cells with CpG-ODN. In this assay, TLR9-induced NF-κB activation was effectively inhibited by 2 μM bortezomib, MG132, MG262, and thiostrepton and was inhibited to some extent by phenoxybenzamine, 15-δ-PGJ2, and parthenolide (Fig. 2A). Interestingly, when HEK293 cells and TLR2-overexpressing HEK293 cells were treated with 2 μM of these compounds, bortezomib, MG132, and MG262 effectively inhibited IL-1β– and TLR2 ligand–induced NF-κB activation; however, no thiostrepton-mediated inhibition was observed (Fig. 2B, 2C). Nevertheless, thiostrepton weakly inhibited IL-1β– and TLR2 ligand–induced activation when its concentration was increased to 10 μM (Fig. 2B, 2C). To investigate whether this was due to a cytotoxic effect, the cytotoxicity of thiostrepton against HEK293 cells was determined using an MTS assay. The results showed no significant loss in cell viability after 10 h of treatment with 10 μM thiostrepton and bortezomib, although some viability was lost after 25–30 h of treatment (Fig. 2D). Because the times required for the IL-1β and TLR2 activation assays were <10 h, the results of this MTS assay indicated that inhibition (Fig. 2B, 2C) generated by 10 μM thiostrepton was not due to a cytotoxic effect. The activities of thiostrepton and bortezomib were further compared at different concentrations. These results showed that at various different concentrations, thiostrepton was as effective as bortezomib with respect to inhibiting TLR9 activation (Fig. 2E).
Thiostrepton selectively inhibits TLR7–9 activation
The suppressive effects of thiostrepton on NF-κB activation induced by IL-1β, TNF-α, and various TLR ligands were further investigated. Parental and TLR2-, TLR3-, TLR5-, TLR7-, TLR8-, and TLR9-overexpressing HEK293 cells were treated with 2 μM thiostrepton and bortezomib and then stimulated with IL-1β, TNF-α, or cognate TLR ligands. The results demonstrated that bortezomib inhibited NF-κB activation in response to all of these stimuli. In contrast, thiostrepton selectively blocked TLR7–9-induced activation and, to a lesser extent, inhibited TLR5 activation; however, it had no suppressive effect on activation induced by IL-1β, TNF-α, or other TLR ligands (Fig. 3). These results suggest that the inhibitory activity of thiostrepton is more specific to TLR7–9.
Thiostrepton inhibits TLR7–9 activation in DCs
DCs play a crucial role in sensing PAMPs and self-derived danger-associated molecular patterns to link innate and adaptive immune responses. Accordingly, DCs play a role in sensing self-derived nucleic acids that lead to pathogenic inflammatory responses in autoimmune disorders. Therefore, DCs are considered a target in the treatment of these diseases (45–48). To further investigate the effect of thiostrepton on DCs, we first assessed the cytotoxicity of thiostrepton and bortezomib against mouse GM-DCs and Flt3L-DCs using an MTS assay. Most GM-DCs were myeloid DCs. The Flt3L-DCs contained conventional DCs and a minor population of plasmacytoid DCs (pDCs) (38, 39). Significant bortezomib- and thiostrepton-induced cytotoxicities were observed in the GM-DCs at concentrations of 0.005 and 10 μM, respectively, and in the Flt3L-DCs at concentrations of 0.005 and 2 μM, respectively (Fig. 4A, 4B). These findings indicate that distinct from their equivalent toxicity with respect to HEK293 cells (Fig. 2D), thiostrepton is less toxic than bortezomib in both GM-DC and Flt3L-DC populations. Mouse TLR8 is a mute receptor. Therefore, the effects of thiostrepton on TLR7 and TLR9 activation suppression in GM-DCs and Flt3L-DCs were further studied and compared with the effects of bortezomib. Both GM-DCs and Flt3L-DCs were treated with different concentrations of thiostrepton and bortezomib and then stimulated with the respective TLR7/8 and TLR9 ligands R848 and CpG-1826. The production of IL-12/23p40 and TNF-α were measured by ELISA. The results indicated that thiostrepton and bortezomib effectively blocked the TLR7- and TLR9-induced production of IL-12/23p40 and TNF-α in both GM-DCs and Flt3L-DCs at concentrations of 0.5–1 and 0.01 μM, respectively (Fig. 4C–J). Nevertheless, note that bortezomib is cytotoxic to DCs at a concentration of 0.01 μM, whereas thiostrepton is not cytotoxic at 1 μM (Fig. 4A, 4B). Moreover, the inhibitory activity of bortezomib began to decrease at concentrations <0.01 μM (Fig. 4E). Furthermore, although pDCs are a minor population of the Flt3L-DCs, they are the major producer of type I IFN in the Flt3L-DCs (38, 39, 49). We investigated the effect of thiostrepton and bortezomib on IFN-α production. Flt3L-DCs were treated with/without R848 and CpG-ODN in the presence of thiostrepton or bortezomib. Production of IFN-α from the population of pDCs in the Flt3L-DCs was measured by ELISA. The results indicated that similar to their effect on suppressing production of IL-12/23p40 and TNF-α, both thiostrepton and bortezomib blocked TLR7- and TLR9-activated IFN-α productions (Fig. 4K, 4L).
Thiostrepton inhibits LL37/DNA and LL37/RNA complex–induced activation in DCs
LL37 is an antimicrobial peptide that is produced in excess within the inflamed skin associated with psoriasis. This peptide forms complexes with self-DNA and self-RNA molecules released from dying cells and enables these self–nucleic acids to activate TLR7–9 in DCs at inflammatory sites (24–26). Therefore, we further investigated the effects of thiostrepton on the inhibition of LL37/DNA and LL37/RNA complex–induced activation in DCs. Both GM-DCs and Flt3L-DCs were treated with different concentrations of thiostrepton and bortezomib and stimulated with LL37/DNA and LL37/RNA complexes. The CD86 was analyzed by flow cytometry, and psoriasis-associated inflammatory cytokines, including IL-12/23p40 and TNF-α, were measured by ELISA. The results indicated that similar to their inhibitory activities on R848- and CpG-1826–generated activations (Fig. 4C–J), thiostrepton and bortezomib effectively blocked LL37 complex–induced CD86 expression and IL-12/23p40 and TNF-α production in both GM-DCs (Fig. 5A–H) and Flt3L-DCs (Fig. 5I–P) at the concentrations of 0.5 and 0.01 μM, respectively.
Thiostrepton does not act on signaling molecules downstream of TLRs and has no effect on cellular endocytosis
We next investigated the underlying mechanism by which thiostrepton suppresses TLR7–9 activation. The IL-1β receptor and TLRs share a common signaling pathway that utilizes MyD88, IRAK, TRAF6, and other downstream signaling molecules to trigger phosphorylation and degradation of IκB leading to the activation of NF-κB (50, 51). To investigate whether thiostrepton acts on the signaling molecules downstream of these receptors, HEK293 cells were treated with 2 μM thiostrepton and bortezomib and stimulated with IL-1β. IκBα phosphorylation and degradation were then detected by immunoblotting. The results indicated that distinct from bortezomib, thiostrepton did not inhibit IL-1β–induced phosphorylation of IκBα, nor did it block degradation of the phosphorylated IκB (Supplemental Fig. 1A). Overexpression of signaling molecules downstream of IL-1R and TLRs in cells results in NF-κB activation (52). The activities of thiostrepton and bortezomib on NF-κB activation induced by overexpression of MyD88, IRAK2, TRAF6, and TGF-β–activating kinase 1 in HEK293 cells were further investigated. The results indicated that whereas bortezomib inhibited the NF-κB activation, an inhibitory effect with thiostrepton was not observed (Supplemental Fig. 1B). These findings were consistent with the result shown in Fig. 3, whereby thiostrepton preferentially inhibited TLR7–9-induced NF-κB activation but had no effect on NF-κB activation induced by IL-1β and other TLRs, suggesting that thiostrepton does not suppress TLR7–9 activation by acting on downstream signaling molecules that lead to IκB phosphorylation and degradation. Next, we investigated whether thiostrepton had any inhibitory effect on endocytosis, which plays a role in the uptake of TLR7–9 ligands into cells. For this, both GM-DCs and Flt3L-DCs were incubated with FITC-conjugated dextran in the presence and absence of thiostrepton and bortezomib. FITC-conjugated dextran uptake was analyzed by flow cytometry. The results indicated that both thiostrepton and bortezomib had no effect on endocytosis in the cells (Supplemental Fig. 2).
Thiostrepton contains dual functions of preventing endosomal acidification and inhibiting proteasome activity
Bafilomycin A1 represents a class of inhibitors distinct from bortezomib that acts to prevent endosomal acidification, thus suppressing TLR7–9 activation (53). Because thiostrepton, similar to bafilomycin A1, is an antibiotic derived from streptomycetes, we were interested to know whether thiostrepton would also exhibit endosomal acidification inhibitory activity. pHrodo Green dextran emits pH-sensitive fluorescence that increases in intensity with increasing acidity (54). To evaluate the effects of thiostrepton and to compare these with the effects of bortezomib and bafilomycin A1, cells were treated with these compounds and subsequently with pHrodo Green dextran. The fluorescence intensities in the cells were analyzed by flow cytometry. Interestingly, the results showed that similar to bafilomycin A1, thiostrepton suppressed endosomal acidification in HEK293 cells, GM-DCs, and Flt3L-DCs, whereas bortezomib did not (Fig. 6).
Furthermore, because the CMap screening suggested a functional connection between thiostrepton and bortezomib, we investigated the proteasome inhibitory activity of this compound. Cell lysates prepared from thiostrepton- and bortezomib-treated cells were incubated with the fluorogenic peptide LLVY-AMC, a substrate of 20S proteasomal activity, and the inhibitory activities were measured via fluorometry. The results indicated that similar to bortezomib, thiostrepton acted to inhibit proteasomal degradation of the LLVY-AMC peptide in HEK293 cell, GM-DC, and Flt3L-DC lysates (Fig. 7).
Thiostrepton inhibits TLR9 trafficking to endolysosomes
Bortezomib has been shown to inhibit TLR9 activation by suppressing TLR9 trafficking from the endoplasmic reticulum to endolysosomes upon CpG-ODN stimulation (33). Therefore, we further investigated whether thiostrepton could inhibit the localization of TLR9 on endolysosomes. To that end, HEK293 cells that expressed a TLR9-GFP fusion protein were stimulated with CpG-2006 in the presence and absence of thiostrepton and then stained with LysoTracker, a lysosomal marker. Cellular localization of TLR9 was analyzed by confocal microscopy. The results showed that most TLR9 colocalized with the lysosomal marker when the cells were stimulated with CpG-ODN in the absence of thiostrepton, whereas colocalization was reduced in the presence of thiostrepton (Fig. 8A). This inhibitory effect of thiostrepton was further studied with FLAG-tagged TLR9 expressed in HEK293 cells. These cells were treated with/without CpG-ODN in the presence or absence of thiostrepton. After homogenization and fractionation, the presence of TLR9 in the endolysosomal fractions was then analyzed by immunoblotting. Consistent with the confocal microscopy analysis, the result indicated that thiostrepton inhibited CpG-ODN–activated trafficking of TLR9 to endolysomes (Fig. 8B). Taken together, the results from Figs. 6–8 suggest that thiostrepton utilizes two different mechanisms to inhibit TLR7–9 activation. One mechanism is similar to that of bortezomib and is mediated through its proteasome inhibitory activity to inhibit trafficking of these TLRs to the endolysosomes for activation. The other mechanism is similar to that of bafilomycin A1, in which endosomal acidification is inhibited, thus suppressing TLR7–9 activation.
Thiostrepton attenunates LL37- and imiquimod-induced psoriasis-like inflammation
The effects of thiostrepton on LL37- and imiquimod-induced psoriasis-like inflammation were further investigated in three different animal models. In the first model, mouse ears were i.d. injected with LL37 and thiostrepton twice per week according to the schedule shown in Fig. 9A. In the second model, mouse ears were i.d. injected with LL37 and topically treated with a 5% thiostrepton gel once per week according to the schedule shown in Fig. 9B. In the literature, a frequently used protocol for imiquimod-induced psoriasis-like inflammation is treating mice daily with imiquimod for 1 wk (55, 56). To make this model more closely resemble chronic inflammation and to compare with the effect of LL37, in the third model, mouse ears were topically treated with 5% imiquimod cream and 5% thiostrepton gel once per week following the schedule shown in Fig. 9C. In these models, LL37 and imiquimod elicit responses in the mouse ears that closely resemble those of human psoriasis, including the symptoms of skin thickening and erythema. The mouse ear thicknesses were measured to access the inhibitory effects of thiostrepton on the induced psoriasis-like responses. The results show that both i.d. injection and topical treatment with thiostrepton inhibited the LL37- and imiquimod-induced increases in ear thickness (Fig. 9A–C).
The suppressive effect of thiostrepton on LL37-induced inflammatory responses was further investigated. Ear tissues were taken from the injection sites of control-, LL37-, and LL37 plus thiostrepton–injected mice in the first type of animal model (Fig. 9A) and were then homogenized. The expression of IL-1β, TNF-α, and IL-8 were measured by RT-PCR, and the accumulation of CD11b+ monocytes and CD11c+ DCs in the inflammed sites was determined by flow cytometry. The results showed that LL37 increased the production of these inflammatory cytokines and the accumulation of monocytes and DCs in the injection sites; however, these inflammatory responses were significantly suppressed by thiostrepton treatment (Fig. 9D, 9E). The inhibitory effect of thiostrepton on imiquimod-induced inflammatory responses was also investigated. Mouse ears were i.d. injected with imiquimod and thiostrepton, and serum samples were collected at 2 and 4 h after treatment. Induction of TNF-α and IL-12p40 in serum was analyzed by ELISA. The results indicated that thiostrepton inhibited the imiquimod-elicited serum levels of TNF-α and IL-12p40 (Fig. 9F). Taken together, these results indicate that thiostrepton attenuates TLR7–9-induced psoriasis-like inflammation.
TLR7–9 involvement has been suggested in the pathogenesis of autoimmune diseases such as psoriasis, SLE, and RA because of the abilities of these receptors to sense self–nucleic acid-containing protein complexes and activate inflammatory responses. Therefore, TLR7–9 inhibition has emerged as an important research area with respect to the treatment of these autoimmune diseases (27–29). Currently, the applications of different inhibitors are being investigated. These inhibitors block TLR7–9 activation via different mechanisms. For example, IMO-3100, IMO-8100, and CpG-52364 are inhibitory oligodeoxyribonucleotides that can inhibit interactions between TLR7–9 and their cognate ligands. ST2858 and RDP58 are peptides designed to block assembly of the MyD88, IRAK, and TRAF6 complex required for TLR signaling. Bafilomycin A1 prevents endosomal acidification, a prerequisite for TLR7–9 activation, and bortezomib inhibits TLR7–9 activation by inhibiting their trafficking to endolysosomes and thus preventing ligand interactions (27–29).
Bortezomib is a proteasome inhibitor that has been used clinically for the treatment of multiple myeloma. Additionally, treatment applications of this small–molecular mass compound for autoimmune disorders are in preclinical investigation with animal models. Nevertheless, the adverse effect of neurotoxicity limits the therapeutic applications of bortezomib (30–32, 57). To explore the applications of novel TLR7–9 inhibitors with respect to autoimmune diseases, in this study we designed a strategy to screen the CMap database for chemical compounds with similar functions as bortezomib. In this strategy, a gene expression profile of bortezomib-treated cells was first acquired from the GEO database. This information was then used as input data for CMap database screening. This strategy was proven successful because functionally similar proteasome inhibitors such as MG132 and MG262 were identified during screening.
In this screening, thiostrepton, an antibiotic derived from streptomycetes, was suggested to have a similar function as bortezomib. Our studies and various previous studies (58, 59) indicated that this antibiotic also exhibits inhibitory activity against the proteasome. Although its applications with respect to autoimmune disorders have never been investigated, antitumor effects of thiostrepton have been reported. For example, thiostrepton was found to exhibit anti–breast cancer cell activity by inhibiting proteasome activity and consequently controlling the transcriptional activity of FOXM1 (37, 58, 59).
Bortezomib has been shown to inhibit TLR9 localization on endolysosomes by disrupting the interaction between TLR9 and UNC93B1, an accessory protein required to escort TLR9 from the endoplasmic reticulum to endolysosomes (33). Nevertheless, the link between the proteasome inhibitory activity of bortezomib and its suppression of TLR9 trafficking is not yet fully understood. This activity of bortezomib may disrupt the function of an endoplasmic reticulum–resident protein required for UNC93B1-coordinated TLR7–9 trafficking (33). Our studies indicate that similar to bortezomib, thiostrepton contains proteasome inhibitory activity and inhibits TLR7–9 activation in both cell-based activation assays and DCs. Additionally, thiostrepton blocks TLR9 trafficking to endolysosomes. Therefore, thiostrepton at least partially utilizes the same molecular mechanism as bortezomib with respect to proteasome inhibitory activity to block the trafficking and activation of TLR7–9 in endolysosomes, although as with bortezomib, the molecular target of this activity remains under investigation.
Interestingly, thiostrepton also exhibits inhibitory activity against endosomal activation, in which it differs from bortezomib but is similar to bafilomycin A1. Several different classes of inhibitors target endosomal acidification. Bafilomycin A1 represents a class of inhibitors that acts by inhibiting the vacuolar H+-ATPase. In contrast, lysosomotropic weak bases such as NH4Cl and chloroquine elevate the endosomal pH by binding to protons in the acidic compartment. Carboxylic ionophores such as monensin and nigericin prevent acidification by causing proton exchange across the membrane against other monovalent cations (60–62). The mechanism by which thiostrepton inhibits endosomal acidification remains under investigation. Nevertheless, the observation that thiostrepton exhibits dual functions by inhibiting proteasomal activity and preventing endosomal acidification to inhibit TLR7–9 activation suggests the possibility that thiostrepton targets multiple molecules to inhibit TLR7–9 activation.
In contrast to bortezomib, thiostrepton preferentially inhibits TLR7–9 activation and is less cytotoxic to DCs. Thiostrepton is as effective as bortezomib in terms of inhibiting TLR7–9 activation but is relatively weaker in terms of suppressing NF-κB activation in response to IL-1β, TNF-α, and other TLRs. Proteolytic degradation of phosphorylated IκB is a cellular process required for NF-κB activation (63, 64). The observation that IL-1β–induced degradation of phosphorylated IκB was blocked by bortezomib but not by thiostrepton could explain the specificity of thiostrepton. Additionally, this observation may be part of the reason why thiostrepton is less cytotoxic than bortezomib.
Psoriasis is a chronic, inflammatory, autoimmune skin disease that affects 2–3% of the global population. This disease can be caused by various factors, including microbial infections, skin injuries, immune disorders, genetics, environment, weather, and stress triggers (65–68). These factors trigger immune responses, including the production of TNF-α, IL-1β, IL-12, IL-17, IL-23, and IFN-γ, by immune cells at the inflamed sites. Of these cytokines, IL-23 is produced by DCs and other APCs and drives the differentiation of naive CD4+ T cells toward the Th17 subset. Th17 cells produce IL-17, which along with other cytokines such as TNF-α, can induce the production of chemokines, cytokines, and antimicrobial peptides from epidermal keratinocytes. LL37 is a cationic antimicrobial peptide that is overexpressed in psoriasis lesions. This antimicrobial peptide forms complexes with the self-RNA and self-DNA molecules derived from dead cells at inflamed sites; these complexes then activate TLR7–9, which in turn increases the inflammatory responses in psoriasis lesions. This forms a self-amplifying loop of TLR7–9-associated psoriasis pathogenesis and suggests that TLR7–9 may be molecular targets for psoriasis treatment (24–26, 65–68). Accordingly, the therapeutic value of targeting these receptors has been supported by observed reductions in skin inflammation in an IL-23–induced psoriasis animal model following treatment with the TLR7–9 antagonists IMO-3100 and IMO-8400 (69). Additionally, data from a phase 2 clinical trial have demonstrated the efficacy of IMO-3100 for reducing the severity of psoriasis in human patients (69). In this study, we employed three different animal models to test the efficiency of thiostrepton for psoriasis treatment. In these models, mouse ears were 1) i.d. injected with LL37, 2) topically treated with LL-37, and 3) topically treated with 5% imiquimod cream. These treatments caused inflammation, erythema, epidermal alterations, and skin thickening that closely resembled human psoriasis. Consistent with its role as an inhibitor of TLR7–9 activation, both i.d. injection and topical treatment with thiostrepton attenuated the LL37- and imiquimod-induced psoriasis-like responses by suppressing inflammatory cytokine production and by blocking the infiltration and proliferation of monocytes and DCs at the psoriasis-like sites.
In conclusion, in this study we identified a novel TLR7–9 inhibitor. In a manner distinct from those of other known inhibitors, thiostrepton exhibits dual inhibitory functions against TLR7–9 activation by inhibiting proteasome activity and preventing endosomal acidification. Thiostrepton can reduce psoriasis-like inflammation in animal models. Moreover, its antibiotic function further increases the potential for its application as a treatment for bacterial psoriatic responses.
This work was supported in part by National Health Research Institutes, Taiwan Grant IM-104-PP-02 and by Ministry of Science and Technology of Taiwan Grants MOST102-2320-B-400-009-MY3 and MOST104-3011-B010-001. D.-W.Y. carried out his thesis research under the auspices of the Graduate Program of Biotechnology in Medicine, National Tsing-Hua University and National Health Research Institutes.
The online version of this article contains supplemental material.
Abbreviations used in this article:
Flt3 ligand–induced DC
Gene Expression Omnibus
GM-CSF plus IL-4–induced DC
human embryonic kidney
pathogen-associated molecular pattern
systemic lupus erythematosus
TNFR-activated factor 6.
The authors have no financial conflicts of interest.