TRIM58 Restrains Intestinal Mucosal Inflammation by Negatively Regulating TLR2 in Myeloid Cells

This article is featured in In This Issue, p.1407

Acute inflammation is a protective response to extreme challenges to homeostasis, such as pathogens, injuries, and other insults (1). TLRs signal the presence of microbes and tissue stress, triggering the host inflammatory process. Tight control of innate immune signaling via TLRs represents an essential defense mechanism to avoid exaggerated immune responses that may otherwise exacerbate mucosal damage in acute inflammation, resulting in impaired resolution and delayed wound healing. Aberrant TLR signaling has been shown to contribute to the development of inflammatory bowel diseases (IBD) (2). Within a healthy intestine, basal TLR signaling drives immune mechanisms essential for protecting intestinal epithelial barrier integrity. Within a susceptible individual, altered TLR signaling may perturb mucosal homeostasis between gut microbiota and mucosal immunity, leading to chronic inflammation in IBD (3). However, the cellular mechanisms of how TLR signaling is balanced before and during intestinal mucosal inflammation remain to be fully resolved.

TLR2, one major member of the TLR family, recognizes conserved molecular patterns associated with Gram-negative and Gram-positive bacteria, such as lipopeptides (4). Endogenous ligands (either released passively from injured/inflamed tissues and dying cells or actively secreted by activated cells, including proinflammatory cytokines such as TNF-α) may also activate TLR2 signaling (5, 6). TLR2 is differentially expressed by various cell types throughout the gastrointestinal tract, including intestinal epithelial cells (IECs) (7) and monocytes/macrophages (8). However, the functional outcome of TLR2 signaling varies considerably between cell types. Although TLR2 activation exerts several important cytoprotective responses in the intestinal epithelium required for cell survival, barrier stability, and mucosal healing (9–11), TLR2 activation may turn intestinal monocytes into proinflammatory effector cells in the lamina propria (8, 12). The mechanisms that maintain cell-specific balance between TLR2 activation and inactivation remain unclear. During colonization, TLR2 sensing negatively controls its signaling pathway by downregulation of MyD88 through a yet-unknown mechanism (13). Only few negative regulatory controllers of TLR2 have so far been identified in the intestine, such as Tollip (14), TNFAIP3 (A20) (15), NOD2 (CARD15) (16), and IRAK-M (17).

In this study, to discover proteins that may associate with TLR2 and influence its function, we carried out a yeast two-hybrid (Y2H) screen and identified tripartite motif containing 58 (TRIM58) as a novel TLR2-interacting protein. We show that TRIM58 associates with TLR2, promoting its proteolytic degradation. Thus, TRIM58 decreases excessive TLR2 activity in myeloid cells, resulting in rapid resolution of acute inflammatory stress-induced damage of the intestinal mucosa. Our findings identify TRIM58 as a previously undescribed negative regulator of TLR2 in the gut, which may be dysfunctional in intestinal inflammatory disorders, such as IBD.

TLR2 Abs were from Enzo (TL2.1), Cell Signaling (E1J2W), and Imgenex (IMG-319; IMG-410A). Abs to arginase-1 (catalog no. 9819), E-cadherin (24E10), hemagglutinin (HA; 6E2), phospho-histone H3 (D2C8), phospho-histone H2A.X (20E3), Myc (9B11), and ubiquitin (P4D1) were from Cell Signaling. CD11b Ab was from Abcam (EPR1344). CD3e (SP7) Ab was from Thermo Fisher Scientific, BIP/GRP78 (40/BiP) from BD Biosciences, IL-1β from Abcam (catalog no. ab9722), and HA (H6908) was from Merck. TRIM58 Ab (HPA023637) from Merck was used for immunohistochemistry. Alexa Fluor 488– or 647–conjugated goat anti-rabbit, anti-mouse, or anti-rat IgG Abs were from Thermo Fisher Scientific, and HRP-conjugated anti-rabbit and anti-mouse Abs were from GE Healthcare. Isotype-matched Igs were used as controls for protein (co)immunoprecipitation (Santa Cruz Biotechnology and R&D Systems). Synthetic lipopeptide Pam₃Cys-SKKKKx3-HCl (PCSK; lots A15 and A10/03) was obtained from EMC Microcollections and dissolved in sterile water (vehicle control). The inhibitor MG-132 was from Peptide Institute and dissolved in ethanol (vehicle control). Recombinant human Myc-tagged TRIM58 protein was obtained from OriGene. All other reagents were obtained from Merck, unless otherwise specified.

Bait cloning and Y2H screening were performed by Hybrigenics (Paris, France; http://www.hybrigenics.com). In brief, TLR2 (intracellular domain: aa 611–785) was PCR amplified and cloned in pB27 and pB31, two Y2H vectors optimized by Hybrigenics. The bait constructs were checked by sequencing the entire insert and subsequently transformed in the L40ΔGAL4 yeast strain (18). A human leukocyte and activated mononuclear cell cDNA library, transformed into the Y187 yeast strain, was used for mating. Interactions totaling 131 million (pB31) and 37 million (pB27) were actually tested with the TLR2 baits, respectively. After selection on medium lacking leucine, tryptophan, and histidine, positive clones were picked, and the corresponding prey fragments were amplified by PCR and sequenced at their 5′ and 3′ junctions. Sequences were then filtered, contiged as described previously (19), and compared with the latest release of the GenBank database using BLASTN (20).

Human TRIM58 protein (Q8NG06) was plotted using ExPASy Proteomics Tools Web site (http://www.expasy.org) for predictions of secondary structure and the area including aa 363–376 (CQDTLPRKGETTTPS) in the SPRY domain was subsequently chosen for peptide synthesis with C-terminal amidation and conjugation to Limulus polyphemus hemocyanin (no. 8925). Possible homologies with other proteins were excluded by using the ExPASY BLAST2 interface. The peptide no. 8925 was synthesized (purity 95%), and two rabbits (no. 3738 and no. 3739) were immunized by BioGenes (Berlin, Germany). The antisera were tested (serum titer per ELISA: each > 1:200,000), affinity purified (cyanogen bromide–activated Sepharose), and stored (with or without 0.02% thimerosal) in aliquots at −20°C. Additional peptides (no. 8924: CRATLQRLRESKSRL-amide; no. 8926: CLKELAEELEERSQR-amide) were produced as controls of the antisera. The newly raised TRIM58 Abs (no. 3738 and no. 3739) were used for immunoprecipitation, Western blotting, and immunofluorescence.

The cell lines HEK293, THP-1, Caco-2, SW480, and T84 were purchased from American Type Culture Collection (Manassas, VA), HEK-Blue/human TLR2 (hTLR2) from InvivoGen and HL-60 from Deutsche Sammlung von Mikroorganismen und Zellkulturen, respectively, and maintained as described previously (7, 11, 21) or as recommended by the distributors. PBMC and bone marrow cells were isolated from healthy donors (anonymous residual material kindly provided by the Institute of Transfusion Medicine, University Hospital Essen, Essen, Germany) using standard protocols. Peritoneal myeloid cells were prepared, as described previously (10), from age-matched female or male mice (as indicated in figure legends) and cultured for 24.5 h in RPMI medium 1640 supplemented with 10% v/v FBS (Thermo Fisher Scientific), 1% l-glutamine, and penicillin/streptomycin (Life Science) at 37°C with 5% CO₂ on tissue culture plastic treated with polystyrene (Costar).

Human HA-tagged full-length TLR2 (HA-TLR2), TLR4 (HA-TLR4), MyD88 (HA-MyD88), and TRAF6 (HA-TRAF6) plasmids in the same backbone vector (pUNO) were obtained from InvivoGen. Human Myc-tagged full-length TRIM58 (Myc-TRIM58-FL) in the pCMV6-entry-tagged cloning vector (pCMV6) was obtained from OriGene. Deletion mutant of TRIM58 lacking the really interesting new gene (RING) domain and site-specific phosphorylation mutations (Y55 and Y392) were generated within the Myc-TRIM58-FL expression construct (Myc-TRIM58-ΔRING, Myc-TRIM58-Y55F, and Myc-TRIM58-Y392F) and confirmed by sequencing (Trenzyme). Human ubiquitin cloned in pcDNA3.1-GFP was kindly provided by Dr. J. H. Brumell (The Hospital for Sick Children, Toronto, ON, Canada). Plasmids (Endo-free Plasmid Maxi; Qiagen) were transiently transfected into HEK293 or HEK-Blue/hTLR2 cells using Lipofectamine 2000 (Thermo Fisher Scientific), according to the manufacturer’s instructions.

Proliferation was measured using a 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium) (MTS)-based metabolic assay (Cell Titer 96 AQ_ueous One Solution Cell Proliferation Assay; Promega). NF-κB/AP-1 activation of HEK-Blue/hTLR2 cells was measured by a kinetic secreted alkaline phosphatase (SEAP) reporter assay using detection medium Quanti-Blue (InvivoGen) by ELISA.

All mice were bred and cohoused in the same room on a 12 h light/dark cycle (room temperature [RT] 21.0 ± 1.0°C, humidity 55 ± 5%) under identical specific pathogen–free conditions. The animals were provided with the same autoclaved tap water and autoclaved chow (ssniff M-Z, ssniff Spezialdiäten, Soest, Germany) ad libitum in autoclaved filter-top cages (bedding: Lignocel-3-4-S, SAFE, Augy, France; nesting material: cellulose; Lohmann & Rauscher, Neuwied, Germany). Extensive animal health monitoring (criteria of the Federation for Laboratory Animal Science Associations) was conducted routinely on sentinels (Gesellschaft für innovative Mikroökologie, Wildenbruch, Germany) and representative mice from this room; no pathogens (such as Murine norovirus or Helicobacter) were detected. For studies, only age-/gender-matched mice were used, as indicated (figure legends). Prior to use in experiments, mice were placed individually into a fresh cage and allowed to acclimatize for at least 3 d. Protocols were in compliance with German law for use of live animals and regulations of the Society for Laboratory Animal Science and the Federation for Laboratory Animal Science Associations. The study was prospectively reviewed and approved by the local animal protection officer at the University Hospital Essen and the responsible district government. All mice were sacrificed by cervical dislocation. All efforts were made to minimize animal suffering and to reduce the number of animals used.

The Trim58^fl/+ mice were generated (strain nomenclature: C57BL/6-Trim58^tm2313Arte) at TaconicArtemis (Cologne, Germany). Briefly, exon 1 of the murine Trim58 gene contains the translation initiation codon; exons 3–5 were therefore flanked by loxP sites to generate a frameshift from exons 2 to 6 (premature stop codon in exon 6). The positive selection markers were flanked by FRT (neomycin resistance) and F3 (puromycin resistance) and inserted into intron 2 and intron 5, respectively. Homologous recombinant clones were isolated using double-positive (neomycin resistance and puromycin resistance) selection to increase the efficiency of cointegration of both loxP sites. The targeting vector was generated using BAC clones from the C57BL/6J RPCIB-731 BAC library, linearized with NotI, and transfected in the C57BL/6N Tac embryonic stem (ES) cell line by electroporation. Resistant ES cell colonies were selected as correctly targeted colonies and underwent Southern blot analysis. The conditional knockout (KO) allele was obtained after Flp-mediated removal of the selection markers. The Trim58^fl/+ mice (specific pathogen free) were then transferred to the Central Animal Facility, University Hospital Essen, Essen, Germany. To achieve a constitutive KO allele, Trim58^fl/fl mice were crossed with CMV/Cre-deleter mice (22), and Trim58^−/− mice and their littermate Trim58^+/+ controls were used in experiments. To achieve myeloid cell–specific loss of Trim58, Trim58^fl/fl mice were crossed with mice expressing Cre under the LysM promoter (23), and Trim58^MC−/− mice and their littermate Trim58^MC+/+ controls were used in experiments. Gene-specific primers were used for standard genotyping of the above-described mouse lines (Supplemental Fig. 1A): Trim58 (wild-type [WT], 175 bp; KO, 410 bp; and flox, 294 bp): Trim58.2 5′-CACAGCCTGCTTGTTCCTGG-3′, Trim58.3 5′-CTGTGGTTGTTTCTAATAGACAG-3′, and Trim58.5 5′-CTTCTAGATACAAGAGCTAGTCTC-3′; and LysM-Cre (WT, 350 bp; and knock-in, 510 bp): IMR3067 5′-CTTGGGCTGCCAGAATTTCTC-3′, IMR3068 5′-TTACAGTCGGCCAGGCTGAC-3′, and Cre4× 5′-TTAGCTGGCCCAAATGTTGCTG-3′ (24). In addition, Trim58^−/− were crossed with Tlr2^−/− [B6.129-Tlr2^tm1Kir/J; The Jackson Laboratory (9)] to generate double-KO homozygotes (Trim58^−/−Tlr2^−/−).

Material transfer agreements from E. Cario (Universität Duisburg-Essen, Universitätsklinikum Essen, Germany) are required for newly generated Trim58-related mice strains, Abs, and plasmids.

Female mice were treated with dextran sodium sulfate (DSS), as previously described (9). Briefly, colitis was induced by addition of 4.0% (w/v) for 6–7 d of DSS (m.w., 36,000–50,000; lots no. M4241, no. M5164; MP Biomedicals, Irvine, CA) dissolved in drinking water and followed by normal drinking water for 2, 3, 5, or 6 d, as indicated. The duration of DSS was adjusted in a lot-specific manner to achieve consistent and comparable colitis development (lot no. M4241: 7 d; lot no. M5164: 6 d). Body weight was assessed daily. Colon length was measured as marker of inflammation. Histopathological analysis of distal colon cross-sections (three per mouse; 60-μm interval) was performed by standard H&E staining, and images were captured (see 14Immunohistochemistry). The parameters epithelial injury (0, normal morphology; 1, loss of goblet cells in small areas; 2, loss of goblet cells in large areas), mucosal ulceration (0, absent; 1, selective with signs of reepithelization; 2, limited to one 60-μm segment; 3, affecting two 60-μm segments; 4, extensive or confluent, affecting at least three 60-μm segments), and inflammatory cell infiltration (0, no increased inflammatory cells in the lamina propria; 1, increased numbers around crypt basis; 2, inflammatory cells reaching the lamina muscularis mucosae; 3, extensive infiltration reaching the lamina submucosae and edematous thickening; 4, transmural inflammatory infiltrates reaching muscle/serosa) were scored in a blinded fashion and simultaneously, giving a total score of 0–10.

We performed a retrospective, single-center cohort study among patients with a diagnosis of active ulcerative colitis (UC), using formalin-fixed, paraffin-embedded (FFPE) tissue blocks from surgical resection specimens archived from 2001 to 2016 at the Institute of Pathology, University Hospital Essen, Essen, Germany. As normal controls, nonneoplastic intestinal mucosal tissues (R₀) from sporadic colorectal cancer cases were used. All human tissue materials and information were rereviewed by a board-certified pathologist (H.R.) for diagnosis and histopathological characteristics. This study (anonymous analysis of historical collection) abided by the Declaration of Helsinki and was approved by the local Ethics Committee of the Medical Faculty of the University Duisburg-Essen, Essen, Germany.

For preparation of whole-cell lysates, cells were rinsed once in cold PBS (without Ca²⁺/Mg²⁺) with 100 μM Na₃VO₄ and then lysed in ice-cold lysis buffer (1% TX-100, 150 mM NaCl, 20 mM Tris-HCl [pH 7.5], and 2 mM EDTA and supplemented with phosphatase and protease inhibitor mixture tablets (PhosSTOP, cOmplete Mini EDTA-free; Roche), and 1 mM PMSF, and immunoblotting was performed, as previously described (25); precleared supernatants were used for immunoprecipitation (Dynabeads Protein G Immunoprecipitation Kit; Thermo Fisher Scientific) according to the manufacturer’s instructions. Alternatively, the Dynabeads Co-Immunoprecipitation Kit (Thermo Fisher Scientific) was used with the provided cell extraction immunoprecipitation buffer. To confirm equal protein loading, gels were stained with SimplyBlue (Thermo Fisher Scientific), and Western blots were reprobed with anti-GAPDH. MagicMark XP (unstained) and SeeBlue Plus2 (prestained) markers were used as Western protein standards (Thermo Fisher Scientific).

For immunohistochemistry, FFPE sections (5 μm) were treated with 10 mM sodium citrate buffer (pH 6) or Ag-Retrieval Solution AR-10 (BioGenex) at 95°C for 10–20 min, blocked with 5% normal goat serum (NGS; Vector Laboratories)/TBST at RT for 1 h, incubated with primary Ab (1:100-1:12,500) in 5% NGS/TBST or Ab diluent (Cell Signaling) at 4°C for 1 h or overnight, and incubated with a HRP-conjugated detection reagent at RT for 30 min, and Ab binding was visualized using SignalStain DAB Substrate Kit (Cell Signaling). Sections were counterstained using Vector Hematoxylin QS (Vector Laboratories). High-resolution images were captured using the Aperio ScanScope system (Leica Biosystems) and visualized using ImageScope software (version 11.2.0.780; Leica Biosystems). In all experiments, at least four individual sites of image capture were chosen randomly for each sample. Morphological results were considered significant only if at least 70% of the scanned sections per field exhibited the observed effect.

THP-1 or transiently transfected HEK-293 cells were pelleted by centrifugation and treated with BD Cytofix/Cytoperm (BD Biosciences) at 4°C for 20 min. After two washes in BD Perm/Wash buffer, cells were incubated with primary Abs (anti-TRIM58 [no. 3738], anti–α-tubulin [DM1A], or anti-Myc [9B11]) diluted (1:20) in wash buffer at 4°C for 15 min. Following a further two washes, Alexa Fluor 488–/647–conjugated goat anti-rabbit/mouse Abs (1:20) were applied at 4°C for 15 min. Cells were then washed again twice before being resuspended in 0.5% PFA, spun onto (1 × 10⁶/ml) microscope slides with Shandon Single Cytofunnel with white filter cards (Thermo Fisher Scientific) using Shandon Cytospin 2 and fixed with 4% paraformaldehyde at RT for 20 min. Frozen sections of distal colons were cut (7 μm), fixed with ice-cold acetone for 5 min on dry ice, blocked with 10% NGS for 1 h at RT, and incubated with anti-CD11b (1:50) in 1% NGS/PBS at 4°C overnight. Alexa Fluor 488–conjugated goat anti-rabbit Ab was used as secondary Ab (1:50, 60 min, RT). After mounting with Vectashield Mounting Medium containing DAPI (Vector Laboratories), immunofluorescent sections were assessed by confocal laser microscopy (Axiovert 100 M with laser scan head LSM 510; Carl Zeiss, Jena, Germany). The multitrack option and sequential scanning for each channel were used to eliminate any cross-talk of the chromophores.

Mouse assorted total RNA (AM7800) and First Choice Human Total RNA Survey panels (AM6000) were obtained from Thermo Fisher Scientific. Total RNA from mouse middle colon was extracted (RiboPure; Ambion) and purified, including the RNase-free DNase digestion step (RNeasy Mini Kit; Qiagen). FFPE tissue samples were cut into 10–12-μm-thick sections on a microtome with a disposable blade (in part after macrodissection to enrich for region-of-interest content, as needed). RNA was extracted from three to five sequential sections from the same paraffin block using the RNeasy FFPE Kit with the deparaffinization solution (Qiagen), as previously described (26).

All RNA samples from murine middle colons (n = 4 per group) were analyzed independently and further processed at the KFB Center of Excellence for Fluorescent Bioanalytics (Regensburg, Germany), where hybridization was performed using GeneChip Mouse Gene 2.0 ST Arrays (Affymetrix), according to the manufacturer’s instructions. Summarized probe set signals in log2 scale were calculated by using the Robust Multichip Average algorithm with the Affymetrix GeneChip Expression Console v1.4 Software. Data have been deposited in the Gene Expression Omnibus (GSE127182). Data were further classified through Ingenuity Pathways Analysis (Ingenuity Systems; http://www.ingenuity.com).

Quantitative real-time RT-PCR analysis was performed using the one-step QuantiTect SYBR Green RT-PCR Kit (Qiagen) on the Mastercycler ep realplex (Eppendorf) real-time amplification system. QuantiTect Primer Assays (Qiagen) were used as the gene-specific primer pairs. Optimized gene-specific primers were designed using Primer BLAST software for mouse Trim58: exons 1–2: 5′-CTG TGA AGA GGA CCA GAC GA-3′ and 5′-CCA CTT TCT CCT TCC AGA TGA-3′, exons 2–3: 5′-TCT GGA AGG AGA AAG TGG AGA TG-3′ and 5′-CCT GTT CCT ACT GTC ACG CA-3′, exons 3–5: 5′-GCA GAG ACT GCG TGA CAG TA-3′and 5′-TGA TAG CCT CAC ATC GTG TCA A-3′, and exons 4–6: 5′-CTT GAC ACG ATG TGA GGC TAT C-3′ and 5′-ATG CAA GGC CAG GTG TCA AA-3′. For human FFPE-derived RNA samples, the following primers were designed: TRIM58 5′-CGC CGG CAG CTA CCA GGT AAA-3′ and 5′-GGC GTC CTC CAA CTC TTT CCT CA-3′; and GAPDH 5′-CCC ATC ACC ATC TTC CAG GAG CGA-3′and 5′-GCC AGC ATC GCC CCA CTT GA-3′. Copy numbers of individual transcripts were related to Gapdh/GAPDH as endogenous control (x/100,000 copies) and normalized against untreated cells or respective WT normal tissues (10), or as indicated.

The unpaired, two-tailed t test was used to calculate differences between means (GraphPad Prism version 5.04; GraphPad Software), if not indicated otherwise. A p value < 0.05 was considered as significant. All data are expressed as the means ± SEM.

By aiming to identify novel interacting proteins of TLR2, we performed a large-scale Y2H screen of a human cDNA library, as described in 2Materials and Methods. Two constructs were made with a portion of the cytoplasmic domain of TLR2 fused C-terminal to the LexA (−/+ inducible) binding domain and both baits were tested against a human leukocyte/activated mononuclear cell library. In this screening process (Fig. 1A), TRIM58 was found in two independent clones (one per each bait construct: pB27_D-4 and pB31_A-6). The interacting fragment corresponded to amino acid residues 73–283 (624–1382).

We profiled the basal gene expression pattern of TRIM58 in various normal tissues and cell lines. The highest mRNA expression levels were detected in murine and human bone marrow and the human myeloid-derived cell line THP-1 (Fig. 1B). TRIM58 mRNA expression was also detectable in the healthy gastrointestinal tract (Fig. 1B), selective IEC lines (Fig. 1B), and other organs (Supplemental Fig. 1B). To determine whether TRIM58 and TLR2 indeed associate in mammalian cells, we transiently cotransfected cDNA encoding HA-tagged TLR2 with Myc-tagged TRIM58 into HEK293 cells, and association was shown by coimmunoprecipitation (Fig. 1C). To confirm that the association of the two proteins is physiologically relevant, we raised novel rabbit Abs against synthetic peptide fragments of human TRIM58, as described in 2Materials and Methods. Specificity of the newly generated Abs (no. 3738, no. 3739) to TRIM58 Ag was confirmed by Western blotting (Fig. 1D). Using these Abs, we demonstrated (Fig. 1E) that endogenous TRIM58 coimmunoprecipitated with endogenous TLR2, and vice versa, in THP-1 cells. Collectively, these data imply that TRIM58 and TLR2 interact in mammalian cells.

Next, we asked whether the two partners may influence each other. Stimulation of THP-1 cells with the specific TLR2 ligand PCSK resulted in a dosage-dependent increase in TRIM58 protein expression levels within 2 h (Fig. 2A, 2B), whereas TRIM58 gene expression was subsequently downregulated within 6 h (Fig. 2C, 2D). In contrast, the TLR4 ligand LPS did not significantly affect TRIM58 mRNA/protein expression (data not shown). The observed protein increase may be due to posttranslational stabilization of TRIM58, as treatment with MG-132, an inhibitor of proteasome-mediated degradation, upregulated the basal protein level of TRIM58 in THP-1 cells (Fig. 2E). To analyze the subcellular localization of TRIM58, we performed immunofluorescence confocal laser microscopy. Endogenous TRIM58 protein was diffusely expressed in cytoplasmic compartments of basal THP-1 cells. After TLR2 stimulation with PCSK, TRIM58 protein redistributed to the cytoskeleton, as evidenced by colocalization with α-tubulin (Fig. 2F). These results show that TLR2 signaling affects synthesis and localization of TRIM58 in human myeloid cells.

Although TLR2 protein expression was detected when coimmunoprecipitated with anti-TRIM58, the association was found to be discrete in cotransfected HEK293 cells (Fig. 1C). To assess whether TRIM58 association with TLR2 may reciprocally affect abundance of TLR2 protein, we transfected HEK293 cells with cDNA encoding HA-TLR2 plus different concentrations (0.1–2.5 μg) of cDNA encoding Myc-TRIM58 (or corresponding empty vector [pCMV6]). After 24 h of transfection, we analyzed the cell lysates by immunoblot of HA-TLR2 expression. As shown in Fig. 3A, TRIM58 overexpression induced reduction of TLR2 protein in a concentration-dependent way. A weak reducing effect of TRIM58 overexpression on the adaptor protein MyD88 was also detected (Fig. 3B), but no effect on other TLRs or downstream signaling molecules, including TLR1, TLR4, or TRAF6 (Fig. 3B), implying a specific TRIM58-dependent regulatory mechanism of the TLR2 pathway. Of note, site-directed mutants of TRIM58 (Y55F and Y392F) did not prevent TRIM58-mediated degradation of TLR2 protein (Fig. 3C), implying that these phosphorylation sites are dispensable in this context. To examine the functional effects of TRIM58 overexpression on TLR2 signaling, we transiently transfected HEK-Blue/hTLR2 reporter cells (which stably coexpress the hTLR2 gene and a NF-κB/AP-1–inducible SEAP reporter gene) with cDNA encoding full-length TRIM58 (or empty vector) and added the TLR2 ligand PCSK. As shown in Fig. 3D, overexpression of TRIM58 reduced PCSK-mediated activation of NF-κB/AP-1, as evidenced by significantly lower secretion of SEAP. These data suggest that TRIM58 regulates degradation and activation of TLR2 in a selective manner.

The RING domain is believed to harbor the E3 ubiquitin ligase activity needed by TRIM proteins to ubiquitinate their substrates (27). When HEK-293 cells were transiently cotransfected with expression vectors for TLR2 and GFP-ubiquitin together with full-length TRIM58, TRIM58-mediated polyubiquitination of TLR2 was readily detected (Fig. 4A). Treatment with the proteasome inhibitor MG-132 increased the level of TLR2 protein expression (Fig. 4A). However, overexpression of a deletion mutant of TRIM58 lacking the RING finger domain abrogated TRIM58-mediated reduction of TLR2 expression (Fig. 4B) and signaling activity (Fig. 4C). RING domain–deficient TRIM58 protein was localized diffusely to cytoplasmic compartments (Fig. 4D), whereas full-length TRIM58 was present at the cellular membrane of transfected HEK293 cells, thus mimicking the surface morphology of endogenous TRIM58 protein in basal versus stimulated THP-1 cells (Fig. 2F). These findings indicate that TRIM58 promotes proteasomal degradation of TLR2 only in the presence of the RING domain.

Next, to assess the in vivo significance of Trim58 expression in the immune system, we generated mice completely lacking Trim58 (Trim58^−/−), as described in 2Materials and Methods and depicted in Fig. 5A. Quantitative RT-PCR demonstrated efficient knockdown (>98%) of Trim58 mRNA expression in spleens from representative Trim58^−/− mice, as we ruled out residual transcription using a collection of primers spanning all exon/exon boundaries (Fig. 5B). Trim58^−/− were fertile and born at normal Mendelian ratios. Adult Trim58^−/− animals exhibited no overt phenotype, even after 12 mo (data not shown). Macroscopic examination of Trim58^−/− organs did not reveal any major abnormality. Murine Trim58^−/− myeloid cells from the naive peritoneal cavity showed significantly elevated expression of TLR2 protein (Fig. 5C), which was associated with enhanced metabolic activity (Fig. 5D). Thus, we confirmed the involvement of endogenous TRIM58 in negatively regulating TLR2 protein expression in vivo.

Because TLR2 plays important roles in intestinal inflammation, we next determined the effects of Trim58 deficiency on acute DSS colitis. Trim58^−/− mice were highly susceptible to inflammatory stress-induced damage with significantly delayed intestinal mucosal healing (Fig. 6), as assessed by any of the several parameters [including body weight (Fig. 6A), spleen weight (Fig. 6B), and colon length (Fig. 6C)]. By day 8, all mice (Trim58^+/+ and Trim58^−/−) developed colonic inflammation with histologically comparable degrees of increased disease activity (Fig. 6D, 6E). However, by day 12, colonic mucosal architecture of Trim58^+/+ was mostly intact again, whereas Trim58^−/− colitis persisted (Fig. 6D, 6E). The mean histology scores of DSS-Trim58^−/− did not differ statistically between days 8 and 12 (p > 0.05). Sustained intestinal inflammation in Trim58^−/− during the restitution phase was characterized by extensive ulcerations with oedematous wall thickening and transmural inflammatory infiltrates, mostly consisting of CD11b⁺ myeloid cells, which was not evident in Trim58^+/+ colons (Fig. 6E). Immunohistochemistry (Fig. 6F) revealed increased IEC injury with impaired restitution (goblet cell loss) in Trim58^−/− mice. Phosphorylation of histone H2A.X (γH2A.X; a marker of genotoxic stress-induced apoptosis DNA fragmentation) and phosphorylation of histone H3 (a marker of chromatin condensation) were enhanced in the damaged Trim58^−/− intestinal epithelium.

DSS-Trim58^−/− colons showed significantly upregulated protein levels of TLR2 and proinflammatory IL-1β, when compared with DSS-Trim58^+/+ on day 12 (Fig. 7A). We identified 195 genes significantly (p < 0.05) regulated with >2-fold differential expression by microarray in Trim58-deficient colitis (Supplemental Table I). Within this dataset, 36 genes (Supplemental Table II) were associated with the major canonical pathways “acute phase response,” “leukocyte adhesion and diapedesis,” and “matrix metalloproteases” (Fig. 7B), as annotated the Ingenuity knowledge base. Real-time PCR analysis of a selection of representative genes validated the upregulated levels of mRNA expression of selective acute phase/inflammatory-response (Fig. 7C), chemotaxis, (Fig. 7D) and tissue-injury (Fig. 7E) markers in DSS-Trim58^−/− versus DSS-Trim58^+/+ colons. Collectively, these data imply that Trim58 deficiency upregulates TLR2 protein expression and inflammatory cyto-/chemokine production during acute DSS-induced inflammation and that Trim58 is necessary for the resolution of mucosal intestinal inflammation.

Next, we asked whether TLR2 may be functionally responsible for the excessive inflammatory response in Trim58 deficiency. We isolated myeloid cells from healthy mice and stimulated them with TNF-α and IFN-γ. These two proinflammatory cytokines were predicted by upstream regulator analysis through Ingenuity Pathways Analysis as the two top-activated mediators linked to 72 major targets in the gene array dataset of DSS-Trim58^−/− colons. As shown in Fig. 8A, Trim58^−/− myeloid cells responded to TNF-α/IFN-γ with significantly increased mature IL-1β protein production and enhanced expression of arginase-1, when compared with Trim58^+/+. This inflammatory hyperreactivity correlated with unchanged high levels of TLR2 protein in stimulated Trim58^−/− myeloid cells (Fig. 8A), but was abolished by genetic deletion of Tlr2 in Trim58^−/− myeloid cells (Fig. 8B), essentially implying a TLR2-dependent mechanism.

Finally, to test the specific role of myeloid-specific deletion of Trim58 under inflammatory conditions in vivo, we generated conditional Trim58 KO mice (Trim58^MC+/+ and Trim58^MC−/−) in which Trim58 expression was controlled by LysM, a myeloid cell-specific promoter. We administered DSS and observed peak intestinal injury shortly after treatment on day 9 in Trim58^MC−/− mice (Fig. 9), which was more severe compared with DSS-Trim58^MC+/+, as shown by significant weight loss (Fig. 9A), no change in spleen weight (Fig. 9B), decrease in colon length (Fig. 9C), and increase in histology score (Fig. 9D). Rapid mucosal damage was histologically evidenced (Fig. 9E) by greater ulcer formations and frequent transmural inflammatory cell infiltrates (mostly CD11b⁺) in DSS-Trim58^MC−/− mice. This was consistent with upregulated expression levels of various genes associated with inflammatory injury on day 9 (Fig. 9F). However, by day 12, most DSS-Trim58^MC−/− colons showed significant improvement with signs of IEC regeneration and tissue repair, comparable to DSS-Trim58^MC+/+ (Fig. 9A, 9C–E). In contrast, DSS-Trim58^−/− showed delayed mucosal healing during the recovery phase under the same protocol conditions (Fig. 9A, 9C, 9D), consistent with Fig. 6A–E. When comparing the two different mouse strains, the mean histology score was significantly increased in DSS-Trim58^−/− versus DSS-Trim58^MC−/− on day 12 (Fig. 9D; p < 0.011). These data suggest that myeloid cells critically orchestrate the early damage response in Trim58 deficiency.

Finally, we assessed the mRNA and protein expression patterns of TRIM58 in human specimens of active UC in comparison with normal control (non-IBD) tissues. As shown in Fig. 10A, TRIM58 mRNA expression levels were markedly decreased in colonic specimens from UC patients with nonactive/mild or active disease, compared to normal tissues. Similarly, TRIM58 protein was barely detectable in inflamed mucosae in tissues from UC patients, whereas intense staining for TRIM58 was consistently present on scattered lamina propria mononuclear cells and IECs in normal colonic specimens (Fig. 10B). There were no apparent differences in expression level in relation to histological degree of inflammatory activity in tissues from UC patients. These data show that UC, a main form of human IBD, is associated with significant reduction of TRIM58 expression.

Innate immune responses are essential for elimination of danger arising from tissue trauma or pathogenic invaders (28). The inflammatory process must be self-limited to avoid excessive tissue destruction and the development of chronic inflammatory or autoimmune diseases (29). TLR2 plays important roles in initiating diverse cell-specific immune responses in the gastrointestinal tract (8, 9, 12, 30), but the mechanisms that restrain aberrant TLR2 signaling remain to be resolved.

TRIM family proteins are characterized by the presence of the N-terminal tripartite motif, which consists of a RING domain, one or two B-boxes, and a coiled-coil domain (31). TRIM proteins function as RING E3 ubiquitin ligases, which have been implicated in diverse biological processes, including autophagy and antiviral activity (32). Some TRIM proteins have been implicated in the pathogenesis of autoimmune diseases, including IBD, such as TRIM21 (33) and TRIM28 (34), which inhibit T cell activation, thus preventing colitis development. Other TRIMs may be involved in positively or negatively regulating TLR signaling (35). For instance, TLR-induced TRIM38 targets TRAF6 by proteasomal degradation (36). However, limited knowledge exists on TRIM58. It has recently been shown that TRIM58 promotes proteasomal degradation of the cytoskeletal motor protein dynein, leading to erythroblast enucleation in vitro (37), but a potential role of TRIM58 in innate immune signaling or colitis has not been studied yet. In this study, we identify TRIM58 as a balancing checkpoint that prevents TLR2-mediated detrimental and inappropriate inflammatory responses in myeloid cells, thus ensuring resolution of acute inflammation and preserving mucosal homeostasis in the intestine.

First, our in vitro studies indicated that TRIM58 directly interacts with TLR2 and that expression of TRIM58 is modulated in myeloid cells by TLR2 stimulation. We found that TRIM58 is essential for termination of TLR2-induced NF-κB/AP-1 signaling. TRIM58 targeted TLR2 for proteolytic degradation, thus regulating its expression and activation. Further work will need to identify potential additional (endogenous or exogenous) factors that may modulate TRIM58 signaling, besides TLR2.

TRIM58 has a RING domain at its N terminus that is believed to relate to ubiquitin E3 ligase activity (38). Deletion of the RING domain or administration of MG-132 blocked TRIM58-mediated TLR2 degradation, implying a proteasome-dependent pathway. The RING domain seems to be essential for proper localization of some TRIM proteins in different cellular compartments (39). We found that deletion of the RING domain caused aberrant cytoplasmic localization associated with functional inactivation of TRIM58. It is possible that the RING domain provides the protein/protein interface for the recruitment of proteins to TRIM58 to allow for active versus inactive subcellular compartmentalization, as recently suggested for other TRIMs (39). Future studies will need to determine the structural domains in TRIM58 and TLR2 that are responsible for the binding, as well as the potential effects of TRIM58 on other TIR domain proteins.

Second, our in vivo findings demonstrated that Trim58 deficiency led to an inflammatory prone status in the gut. Trim58-deficient mice did not develop spontaneously colitis, but Trim58 deficiency predisposed mice to increased susceptibility to acute stress-induced damage of the intestinal mucosa. Trim58 deficiency triggered DSS-mediated disease exacerbation associated with an increase in CD11b⁺- myeloid cells, high protein levels of TLR2, and activation of multiple chemokines as well as inflammatory damage mediators in the intestinal mucosa. Studies using cell-specific KO mice revealed that activation of Trim58 in myeloid cells acts early in the course of mucosal disease to restrict inflammatory responses. Mechanistically, loss of Trim58 resulted in posttranslational derepression of TLR2, inducing hypersensitivity to a high TNF-α/IFN-γ-environment in myeloid cells with excessive production of proinflammatory IL-1β via TLR2.

Recent studies have suggested that the role of TLR2 in the intestinal mucosa must be ambiguous because of cell type–specific functional differences. On the one hand, TLR2 engagement has been shown to be critical for the induction of proinflammatory activities in distinct lamina propria mononuclear subsets in the inflamed gut (8). Blocking TLR2 dimerization in pathogenic Ly-6C^hi monocytes ameliorated acute DSS colitis (12). On the other hand, previous work from this laboratory demonstrated that TLR2 plays a key role in maintaining tight- and gap-junction-associated IEC barrier integrity (9, 40). In addition, TLR2 controls goblet cell differentiation by selectively upregulating TFF3 expression (41). Thus, deficient TLR2 signaling in the intestinal epithelium may facilitate mucosal injury, leading to increased susceptibility of acute and chronic colitis (9, 40).

It is likely that the action of TRIM58 is also strictly context- and cell type–dependent. We found TRIM58 to be strongly expressed by human and murine mononuclear cells. It remains to be shown whether the regulatory TRIM58/TLR2 interaction is uniquely present in mononuclear cells. Our results suggest that the myeloid cell compartment plays a critical role in driving acute damage in Trim58 deficiency during the early phase of mucosal repair (day 9). However, DSS-Trim58^MC−/−, unlike DSS-Trim58^−/−, did not show delayed mucosal regeneration during the later phase of the wound healing process (day 12). Although comparability is limited, as Trim58^−/− and Trim58^MC−/− represent two separate mice strains, our findings imply essential contributions of cell types other than myeloid cells for slow resolution of colitis in Trim58 deficiency, which remain to be determined. Besides lamina propria mononuclear cells, TRIM58 protein was also expressed by IECs in the normal human colonic mucosa. Ongoing studies in this laboratory elucidate the functional role of intestinal epithelial Trim58 in the context of acute and chronic colitis.

Initial human genome–wide association studies suggested that TRIM58 may be associated with erythroid development and erythrocyte traits (42–45), yet no link has so far been established between TRIM58 and IBD. We examined TRIM58 mRNA and protein expression in colonic specimens from patients with UC and observed reduced expression compared with healthy individuals, regardless of inflammatory activity. In active human IBD, TLR2 mRNA and protein expression have been shown to be significantly upregulated in inflammatory mononuclear cells of the lamina propria (2, 46). Thus, TRIM58 expression correlates inversely with the expression of TLR2, supporting our concept of a potential negative feedback loop between TRIM58 and TLR2 in vivo. The IBD risk variant TLR2-R753Q leads to more extensive colonic disease because of impaired innate immune host defense in a subgroup of IBD (41, 47). Future studies will need to further determine the cell-specific and spatiotemporal expression patterns of human TRIM58 in the healthy versus inflamed gastrointestinal tract and test the potential impact of the hTLR2-R753Q polymorphism on TRIM58 signaling.

In conclusion, our findings identify TRIM58 as a previously undescribed regulator of homeostatic control in innate immunity, protecting against acute intestinal mucosal inflammation by inhibiting aberrant TLR2 signaling in myeloid cells. Thus, TRIM58 might represent a novel therapeutic target in inflammatory disorders, such as human UC, of the gastrointestinal tract.

The authors have no financial conflicts of interest.