TRIM27 mediates STAT3 activation at retromer-positive structures to promote colitis and colitis-associated carcinogenesis

STAT3 is a transcription factor that plays central roles in various physiological processes and its deregulation results in serious diseases including cancer. The mechanisms on how STAT3 activity is regulated remains enigmatic. Here we identify TRIM27 as a positive regulator of II-6-induced STAT3 activation and downstream gene expression. TRIM27 localizes to retromer-positive punctate structures and serves as a critical link for recruiting gp130, JAK1, and STAT3 to and subsequent phosphorylation of STAT3 at the retromer-positive structures. Overexpression of TRIM27 promotes cancer cell growth in vitro and tumor growth in nude mice, whereas knockdown of TRIM27 has opposite effects. Deficiency of TRIM27 significantly impairs dextran sulfate sodium (DSS)-induced STAT3 activation, inflammatory cytokine expression and colitis as well as azoxymethane (AOM)/DSS-induced colitis-associated cancer in mice. These findings reveal a retromer-dependent mechanism for regulation of STAT3 activation, inflammation, and inflammation-associated cancer development.

C hronic inflammation plays important roles in pathogenesis of many types of malignancies from leukemia to various solid tumors. Recent studies have highlighted critical roles of the master transcription factors nuclear factor κB (NF-κB) and signal transducer and activator of transcription 3 (STAT3) in linking inflammation to cancer development 1,2 . Specific ablation of the IκB kinase β (IKKβ) or deficiency of the interleukin-6 (IL-6)-STAT3 axis inhibits colitis-associated cancer (CAC) development in mouse models 3,4 . In addition to CAC, the NF-κB-IL-6-STAT3 axis is also closely linked to inflammatory processes during development of liver, lung, and pancreatic cancers [5][6][7][8] . Although the critical roles of STAT3 in tumorigenesis are evident, the mechanisms on how activation of STAT3 is regulated remains enigmatic. Activation of STAT3 is tightly regulated during various physiological processes, such as cell proliferation, survival, and differentiation. Aberrant and persistent activation of STAT3 has been found in various types of cancers 9 . Despite several cytokines or growth factors such as IL-11, oncostatin M (OSM), leukemia inhibitory factor (LIF), epidermal growth factor (EGF), and hepatocyte growth factor (HGF) can activate STAT3, IL-6 is the most important STAT3 activator in various tumors 10 . Binding of IL-6 to its receptor IL-6Rα/gp130 at plasma membrane leads to activation of JAKs that are constitutively associated with gp130. Activated JAKs then mediate phosphorylation of gp130, leading to the recruitment of cytosolic STAT3 and its phosphorylation at Y705. Phosphorylated STAT3 then dimerizes and translocates into the nucleus to induce transcription of a set of downstream genes, which play crucial roles in promoting tumor cell proliferation and survival, tumor invasion, angiogenesis, and immunosuppression 11 .
Endocytosis is a cellular process by which cell surface components and extracellular molecules are internalized into cellular vesicles 12 . There is growing evidence that the endosomal system not only serves as a conduit for the degradation or recycling of cell surface receptors, but also functions as an essential site of signal transduction. It has been reported that receptor-mediated endocytosis is required for STAT3 activation 13,14 . In addition, it has been shown that phosphorylated STAT3 induced by EGF and HGF co-localizes with endosomal vesicles in the cytoplasm 13,14 . However, phosphorylated STAT3 induced by IL-6 is sequestered into unidentified membrane vesicles which are distinct from the endosomal vesicles at where the EGF-and HGF-induced phosphorylated STAT3 is localized and are negative for various endosomal markers such as EEA1, Rab5, Rab7, and Rab11 among others 15 . These studies imply that there exists remarkable differences in STAT3 activation in response to different stimuli and the detailed mechanisms on how vesicular trafficking regulates STAT3 activation remain elusive.
The TRIM family of proteins is characterized by its tripartite motif (TRIM), which is composed of a RING finger domain, one or two B-box domains, a coiled-coil domain 16 . The TRIM family proteins are involved in various physiological processes and their alterations result in diverse pathological conditions such as developmental disorders, immunological diseases, and tumorigenesis 16,17 . TRIM27 (also called RFP) was originally identified as a gene involved in oncogenic rearrangements with the RET proto-oncogene 18 . It has been shown that TRIM27 is highly expressed in various cancers including breast, endometrial, ovarian, lung, and colon cancers [19][20][21][22][23] . TRIM27 is a multifunctional protein which is involved in cell proliferation, transcriptional repression, negative regulation of NF-κB activation, apoptosis, and innate immune response [24][25][26][27][28][29][30][31] . In a screen for proteins that regulate STAT3 activity, we identify TRIM27 as an important mediator of IL-6-induced STAT3 activation. Interestingly, we find that TRIM27 was localized at retromer-positive structures. Following IL-6 stimulation, the retromer-localized TRIM27 recruited JAK1 and STAT3, leading to STAT3 phosphorylation and induction of downstream effector genes. Furthermore, deficiency of TRIM27 significantly impairs STAT3 activation, suppresses dextran sulfate sodium (DSS)-induced colitis, and azoxymethane (AOM)/DSS-induced CAC development in mice. Our findings reveal a retromer-dependent mechanism for STAT3 activation and inflammation-associated cancer development.
Previously, it has been reported that TRIM27 has E3 ubiquitin ligase activity 28,33 . Therefore, we determined whether TRIM27mediated STAT3 activation is dependent on its E3 ubiquitin ligase activity. We found that the enzymatic inactive mutants TRIM27(C/S) and TRIM27ΔRing were able to potentiate IL-6induced STAT3 activation as well as wild-type TRIM27 in both HEK293 and HeLa cells (Supplementary Figure 4b). In addition, TRIM27 had no obvious effects on ubiquitination of JAK1 and STAT3 (Supplementary Figure 4c, d). These results suggest that the E3 ubiquitin ligase activity of TRIM27 is dispensable for its ability to mediate STAT3 activation.
Since TRIM27 is associated with JAK1 and required for full phosphorylation of STAT3 after IL-6 stimulation, we investigated whether TRIM27 could affect complex formation of JAK1 and STAT3. Coimmunoprecipitation indicated that overexpression of TRIM27 promoted the association between JAK1 and STAT3 ( Fig. 2f), while its knockdown markedly reduced IL-6-induced JAK1-STAT3 association (Fig. 2g). These results suggest that TRIM27 mediates the association between JAK1 and STAT3 after IL-6 stimulation.
Retromer acts as a platform for IL-6-induced STAT3 activation. Previously, it has been reported that TRIM27 is localized to retromer-containing endosomes 33 . Retromer is a highly conserved heteropentameric complex important for recycling proteins from the endosomes to trans-Golgi network or plasma membrane. The mammalian retromer complex comprises a core cargo recognition trimer composed of VPS26, VPS29, and VPS35, and a sorting nexin heterodimer composed of SNX1 or SNX2 with SNX5, SNX6, or SNX27 34 . To confirm whether the TRIM27-containing cytoplasmic punctate structures observed in our earlier experiments (Fig. 2d, e) are indeed retromercontaining structures, we further performed confocal microscopic study. We found that TRIM27 specifically colocalized with the retromer component VPS26 but not other endosomal markers such as EEA1 and RhoB (Fig. 3a). This led us to investigate whether retromer-containing punctate structures are involved in IL-6-STAT3 signaling. Confocal microscopy indicated that phosphorylated STAT3 induced by IL-6 was well colocalized with the retromer marker VPS26 but not found in EEA1-, Rab5-, and RhoB-positive endosomes or LAMP1-positive lysosomes (Fig. 3b). In addition, phosphorylated STAT3 and TRIM27 were well colocalized with VPS26 (Fig. 3c). Coimmunoprecipitation experiments indicated that the retromer component VPS35 could interact with JAK1, JAK2, and STAT3, but not IL6-Rα or gp130 in overexpression systems (Fig. 3d). It has been reported that IL-6 stimulation induces endocytosis of the IL-6Rα/gp130 receptor complexes, we therefore examined whether endocytosis was required for IL-6-induced STAT3 activation. As shown in Supplementary Figure 4e, f, IL-6-induced activation of STAT3 as well as its phosphorylation at Y705 was inhibited by treatment with the dynamin inhibitor dynasore 35 , indicating clathrin-dependent endocytosis of the IL-6Rα/gp130 receptor complexes is required for IL-6-induced STAT3 activation. This led us to investigate whether VPS35 interacted with the signaling components following IL-6 stimulation. As the available antibodies against VPS35 could not enrich endogenous VPS35, we generated a HeLa cell line in which relative low level of Flag-tagged VPS35 was expressed. As shown in Fig. 3e, Flag-VPS35 associated with endogenous gp130, JAK1, and STAT3 upon IL-6 stimulation, Fig. 2 TRIM27 promotes JAK1-STAT3 complex formation. a Effects of various dominant-negative mutants on TRIM27-mediated STAT3 activation. HEK293 cells (5 × 10 4 ) were transfected with STAT3 reporter (10 ng), TRIM27 and the indicated mutant plasmids (0.1 μg each) for 24 h before luciferase assays. b TRIM27 interacts with JAK1, JAK2, and STAT3 in mammalian overexpression system. HEK293 cells (2 × 10 6 ) were transfected with the indicated plasmids for 24 h. Coimmunoprecipitation and immunoblot analysis were performed with the indicated antibodies. c Endogenous TRIM27 is associated with gp130, JAK1, and STAT3. HeLa cells (2 × 10 7 ) were starved overnight and then treated with IL-6 (50 ng/mL) or left untreated for the indicated times. Coimmunoprecipitation and immunoblot analysis were performed with the indicated antibodies. d Colocalization of TRIM27, gp130, JAK1, and STAT3. HeLa cells (1 × 10 5 ) were transfected with GFP-tagged TRIM27 (0.2 μg) and Flag-tagged gp130 (0.1 μg) or Flag-tagged JAK1 (0.1 μg) expression plasmids. Twenty hours after transfection, cells were starved overnight followed by stimulation with IL-6 (50 ng/mL) for 30 min. Immunostaining was performed with anti-Flag and anti-STAT3. Scale bars, 10 μm. e Colocalization of TRIM27 and pY705-STAT3. The experiments were performed similarly performed as in d, except that antibody against pY705-STAT3 was used. Scale bars, 10 μm. f Overexpression of TRIM27 promotes JAK1-STAT3 interaction. HEK293 cells (2 × 10 6 ) were transfected with Flag-tagged JAK1 (4 μg) and HA-tagged STAT3 (3 μg) plasmids together with empty or TRIM27 expression plasmid (2 μg). Twenty-four hours after transfection, cells were starved overnight followed by stimulation with IL-6 (50 ng/mL) for 30 min. Coimmunoprecipitation and immunoblot analysis were performed with the indicated antibodies. g Knockdown of TRIM27 impairs JAK1-STAT3 interaction. The control and TRIM27-RNAi HeLa cells (2 × 10 7 ) were starved overnight followed by stimulation with IL-6 (50 ng/mL) for the indicated times. Coimmunoprecipitation and immunoblot analysis were performed with the indicated antibodies. Data are representative of three experiments with similar results. Graphs show mean ± SD; n = 3. *P < 0.05, **P < 0.01, ***P < 0.001, unpaired t test NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-05796-z ARTICLE NATURE COMMUNICATIONS | (2018) 9:3441 | DOI: 10.1038/s41467-018-05796-z | www.nature.com/naturecommunications whereas Flag-VPS35 constitutively interacted with endogenous VPS26 and TRIM27. Confocal microscopy indicated that gp130 colocalized with VPS26 and TRIM27 following IL-6 stimulation (Fig. 3f). These results indicate gp130 underwent endocytosis and migrated to the retromer following IL-6 stimulation. Knockdown of VPS35 inhibited IL-6-induced STAT3 phosphorylation (Fig. 3g). In similar experiments, knockdown of VPS35 prolonged IFN-α-induced STAT1 phosphorylation (Fig. 3g), which is consistent with a previous report 36 . These results suggest that retromer is essential for IL-6-induced phosphorylation of STAT3.
We further investigated whether the retromer acts as a platform for assembly of gp130-JAK1-STAT3 complex. Coimmunoprecipitation experiments indicated that IL-6 stimulation increased associations of JAK1 with the retromer component VPS35, as well as with gp130, STAT3, and TRIM27, whereas knockdown of VPS35 decreased these associations (Fig. 4a). Knockdown of TRIM27 also impaired the associations of JAK1 with VPS35, VPS26, or STAT3 (Fig. 4b). Moreover, knockdown of TRIM27 impaired the associations of VPS26 with gp130, JAK1, or STAT3, but had no effects on the association between VPS35 and VPS26 (Fig. 4c)  I L 6 R α -F l a g g p 1 3 0 -F l a g F l a g -J A K 1 F l a g -J A K 2 F l a g -S T A T 3 IP: Ig HA Ig HA Ig HA Ig HA Ig HA retromer components VPS26, VPS29, or VPS35 significantly inhibited IL-6-induced transcription of SOCS3 and FOS genes (Fig. 4d). Collectively, these results suggest that the retromerpositive structure acts as a platform for TRIM27-mediated assembly of gp130-JAK1-STAT3 complexes and subsequent phosphorylation of STAT3 following IL-6 stimulation.
TRIM27 plays a role in cell malignancy. Since TRIM27 modulates IL-6-induced STAT3 activation, which has been reported to play important roles in tumorigenesis 10 , we attempted to investigate the roles of TRIM27 in tumorigenesis. Overexpression of TRIM27 markedly promoted anchorage-independent growth of RKO, HEK293, and HeLa cells ( Fig. 5a and Supplementary  Figure 5a, b) in vitro and tumor growth in nude mice ( Fig. 5b and Supplementary Figure 5c), whereas knockdown of TRIM27 had opposite effects (Fig. 5c, d and Supplementary Figure 5d, e, f). These results suggest that TRIM27 plays an important role in tumorigenesis.
Deficiency of TRIM27 attenuates DSS-induced colitis. Because STAT3 plays important roles in inflammation-associated tumorigenesis, we next examined the effects of TRIM27 deficiency on DSS-induced acute colitis. Mice were treated with 3% DSS in drinking water over a 10-day period to induce acute colitis. We found that Trim27 −/− mice displayed attenuated colitis, as determined by less weight loss (Fig. 6a), higher survival rate (Fig. 6b), and reduced colon shortening (Fig. 6c) compared to their wild-type littermates. Histopathological analysis revealed that the colonic mucosa of Trim27 −/− mice was more intact without apparent loss of crypt structures and mucosal ulceration and the colonic tissues of Trim27 −/− mice had less infiltration of inflammatory cells in comparison to that of their wild-type littermates after DSS treatment (Fig. 6d), which was also reflected in the pathological assessment of colitis severity scores (Fig. 6e). These results suggest that TRIM27 deficiency reduces the severity of DSS-induced colitis. Consistent with above phenotypic consequences, both of phosphorylated STAT3 and p65 were decreased in the DSStreated colonic lysates of Trim27 −/− mice in comparison with their wild-type littermates (Fig. 6f). The mRNA levels of inflammatory cytokine genes including Il6, Tnfa, and Il17a as well as the STAT3 target gene Socs3 were decreased in colonic lysates of Trim27 −/− mice in comparison to their wild-type littermates (Fig. 6g). Consistently, the levels of TNF-α and IL-6 in either the serum or colonic mucosa were dramatically decreased in Trim27 −/− mice in comparison to their wild-type littermates (Fig. 6h). Similarly, other inflammatory cytokines IL-1β and IL-17A released from the colonic mucosa were also decreased in Trim27 −/− mice (Fig. 6h). Collectively, these results suggest that TRIM27 deficiency reduces DSS-induced inflammatory responses.
Deficiency of TRIM27 in hematopoietic cells is responsible for attenuation of DSS-induced colitis. To determine which cell types that are crucial for attenuated DSS-induced colitis in TRIM27-deficient mice, four groups of reciprocal bone marrow chimeric mice were generated by adoptively transferring bone marrow to lethally irradiated recipients. Eight weeks after bone marrow transplantation, these mice were subjected to DSS challenge. As expected, TRIM27-deficient mice transplanted with hematopoietic cells from TRIM27-deficient mice (KO > KO) were less susceptible to DSS-induced colitis than wild-type mice receiving hematopoietic cells from wild-type mice (WT > WT) as determined by less body weight loss, higher survival rate, reduced colon shortening, less inflammatory cells infiltration, lower histological score, decreased level of phosphorylated STAT3, and p65 and lower expression of inflammatory cytokines (Fig. 7a-h). However, TRIM27-deficient mice transplanted with hematopoietic cells from wild-type mice (WT > KO) developed colitis as severe as that of WT > WT mice (Fig. 7a-h). Conversely, wildtype mice transplanted with hematopoietic cells from TRIM27deficient mice (KO > WT) showed decreased susceptibility to colitis as KO > KO mice (Fig. 7a-h). These results indicate TRIM27-deficient hematopoietic cells are responsible for attenuation of DSS-induced colitis.
TRIM27 deficiency suppresses AOM/DSS-induced CAC. We next investigated the roles of TRIM27 in inflammation-associated cancer development by utilizing the AOM/DSS model 3 . The mice were injected with AOM followed by three rounds of 2.5% DSS exposure to trigger CAC (Fig. 8a). This procedure did not cause notable differences on the body weights, production of proinflammatory cytokines, and type I IFNs between wild-type and TRIM27-deficient mice (Supplementary Figure 6). The results indicated that TRIM27-deficient mice developed less and smaller colon tumors compared with their wild-type littermates (Fig. 8b). In addition, most of the colon adenomas in Trim27 −/− mice were low-grade dysplasia, while the adenomas in their wild-type littermates were high-grade dysplasia and infiltrated with more inflammatory cells (Fig. 8c). Consistently, there was a significant decrease in proliferation rates in the colon cancers of Trim27 −/− mice as determined by Ki-67 nuclear staining (Fig. 8c). Immunohistochemical analysis indicated that there were less phosphorylated STAT3-positive cells in colon tumors of TRIM27- Fig. 3 The retromer acts as a platform for IL-6-induced STAT3 activation. a TRIM27 localizes to retromer-containing punctate structures. HeLa cells (1 × 10 5 ) were transfected with Flag-tagged TRIM27 (0.2 μg) and GFP-tagged VPS26, EEA1, and RhoB plasmids (0.1 μg), respectively. Twenty-four hours after transfection, immunostaining was performed using anti-Flag antibody. b pY705-STAT3 localizes to retromer-containing punctate structures. HeLa cells (1 × 10 5 ) were transfected with GFP-tagged VPS26, EEA1, RhoB and Lamp1, and YFP-tagged Rab5 plasmids (0.1 μg), respectively. Twenty hours after transfection, cells were starved overnight followed by stimulation with IL-6 (50 ng/mL) for 30 min. Immunostaining was performed using antibody against pY705-STAT3. c Colocalization of TRIM27 and pY705-STAT3 to retromer-containing punctate structures. HeLa cells (1 × 10 5 ) were transfected with Flagtagged TRIM27 (0.2 μg) and GFP-tagged VPS26 (0.1 μg) plasmids. Twenty-four hours after transfection, immunostaining was performed using antibodies against Flag tag and pY705-STAT3. d VPS35 interacts with JAK1, JAK2, and STAT3 in mammalian overexpression system. HEK293 cells (2 × 10 6 ) were transfected with the indicated plasmids for 24 h. Coimmunoprecipitation and immunoblot analysis were performed with the indicated antibodies. e VPS35 is associated with endogenous gp130, JAK1, STAT3, TRIM27, and VPS26. HeLa cells (2 × 10 7 ) stably expressing low-level Flag-tagged VPS35 were starved overnight and then treated with IL-6 (50 ng/mL) or left untreated for the indicated times. Coimmunoprecipitation and immunoblot analysis were performed with the indicated antibodies. f Colocalization of gp130 and TRIM27 to retromer-containing punctate structures. HeLa cells (1 × 10 5 ) were transfected with GFP-tagged VPS26 (0.1 μg), Cherry-tagged TRIM27 (0.2 μg) and Flag-tagged gp130 (0.1 μg) plasmids. Twenty-four hours after transfection, immunostaining was performed using anti-Flag antibody. g Effects of VPS35 knockdown on IL-6-induced STAT3 phosphorylation and IFN-αinduced STAT1 phosphorylation. The control and VPS35-RNAi HeLa cells (2 × 10 5 ) were stimulated with IL-6 (20 ng/mL) or IFN-α (20 ng/mL) for the indicated times before immunoblotting analysis was performed. Scale bars, 10 μm. deficient mice compared to their wild-type littermates (Fig. 8c). Immunoblot analysis further confirmed that the levels of phosphorylated STAT3 were markedly decreased in colon tumors in TRIM27-deficient mice, while there were no apparent differences in phosphorylated STAT3 levels in the tumor-adjacent normal colon tissues between wild-type and TRIM27-deficient mice (Fig. 8d). Consistently, the protein levels of STAT3 downstream target genes Bcl-xl, c-Myc, Pcna, and Ccnd1, which were responsible for tumor cell survival and proliferation, respectively 4,37 , were dramatically downregulated in colon tumors of TRIM27-deficient mice compared to their wild-type littermates (Fig. 8d). Moreover, there was also a decrease in levels of inflammatory cytokines TNF-α, IL-1β, IL-6, and IL-17A in the sera or colon tissues of Trim27 −/− mice (Fig. 8e), which was consistent with the decreased inflammatory cell infiltrations in colon tumors of Trim27 −/− mice (Fig. 8c). Collectively, these results suggest that TRIM27 plays important roles in CAC development.
TRIM27 is overexpressed in human colorectal cancer (CRC) tissues. To evaluate the clinical relevance of TRIM27 expression in human CRC, we analyzed the expression of TRIM27 in CRC tissues using the public data from the TCGA database (https:// portal.gdc.cancer.gov/). We found that TRIM27 level was significantly higher in the CRC samples compared with normal colon tissues (Supplementary Figure 7a). In addition, the patients with high expression of TRIM27 had significantly b a TRIM27 is required for recruitment of JAK1 and STAT3 to the retromer. a Effects of VPS35 knockdown on the association of JAK1 with STAT3, gp130, TRIM27, or VPS35. HeLa cells (2 × 10 7 ) were starved overnight followed by stimulation with IL-6 (50 ng/mL) for the indicated times. Coimmunoprecipitation and immunoblotting analysis were performed with the indicated antibodies. b Effects of TRIM27 knockdown on the association of JAK1 with VPS26, VPS35, or STAT3. HeLa cells (2 × 10 7 ) were starved overnight followed by stimulation with IL-6 (50 ng/mL) for the indicated times. Coimmunoprecipitation and immunoblot analysis were performed with the indicated antibodies. c Effects of TRIM27 knockdown on the association of VPS26 with JAK1, STAT3, gp130, VPS35, or TRIM27. HeLa cells (2 × 10 7 ) were starved overnight followed by stimulation with IL-6 (50 ng/mL) for the indicated times. Coimmunoprecipitation and immunoblotting analysis were performed with the indicated antibodies. d Effects of knockdown of the retromer components on IL-6-induced transcription of downstream genes. The control, VPS26-RNAi, VPS29-RNAi, and VPS35-RNAi stable HeLa cells (2 × 10 5 ) were starved overnight and stimulated with IL-6 (20 ng/mL) for the indicated times before qPCR experiments. Data are representative of three experiments with similar results. Graphs show mean ± SD; n = 3. *P < 0.05, **P < 0.01, ***P < 0.001, unpaired t test worse survival than those with low TRIM27 expression (Supplementary Figure 7b). These results suggest that increased TRIM27 expression is correlated with the pathogenesis of CRC in human.

Discussion
STAT3 activity is tightly controlled for proper physiological processes, and dysregulation of STAT3 activity results in tumorigenesis. Aberrant and persistent activation of STAT3 can be mainly attributed to either oversupply of certain cytokines in tumor microenvironment or abnormal expression and dysfunctions of positive and negative regulators. For examples, elevated expression of STAT3 activators, such as IL-6, IL-11, and S1P, has been observed in various types of cancers 37,38 . In addition, increased expression or aberrantly activated upstream tyrosine kinases, such as JAK1, JAK2, EGFR, and BMX, as well as other positive regulators, such as EZH2 and PASD1, also contribute to persistent activation of STAT3 5,32,39-41 . On the other hand, impaired function or reduced expression of negative regulators, such as PIAS3, PTPs, and GDX, also results in constitutive activation of STAT3 42,43 . In this study, we identified TRIM27 as an important mediator of STAT3 activation, which also plays a role in colitis and CAC development. More importantly, our study reveals that the retromer is an essential platform for STAT3 activation induced by cytokines such as IL-6 and OSM.
Overexpression of TRIM27 markedly potentiated IL-6-and OSM-induced STAT3 activation, whereas TRIM27 deficiency had opposite effects. Overexpression of TRIM27 marginally enhanced IFN-γ-induced transcription of IRF1. It has been previously reported that TRIM27 negatively regulates IKKα/β-mediated NF-κB signals and IKKε-mediated IRF signals 31 . It is possible that the distinct cellular localization, post-translational modification, or complex formation of TRIM27 is responsible for its differential functions.
We found that overexpression of TRIM27 promoted growth and tumorigenicity of tumor cells in xenograft models. In addi-

-
tion, TRIM27 levels are elevated in human CRC tissues, and the high expression levels correlate with adverse survival of CRC patients 44 . These results suggest that TRIM27 is a potential oncogenic protein, which is consistent with the observation that TRIM27 is highly expressed in other types of cancer, including seminomas, breast, endometrial, ovarian, lung, and colon cancers [19][20][21][22][23] . As a mediator of IL-6-induced STAT3 activation, TRIM27 plays important roles in DSS-induced colitis and AOM/DSSinduced CAC development. In the DSS-induced acute colitis model, Trim27 −/− mice displayed attenuated colitis in comparison to their wild-type littermates. Consistently, TRIM27deficient mice had decreased activation of STAT3 and levels of inflammatory cytokines including IL-6, TNFα, and IL-17Α in the sera and/or colonic mucosa. In addition, our results suggest hematopoietic cells are responsible for attenuation of DSSinduced colitis in TRIM27-deficient mice. Previously, it has been observed that TRIM27 levels are increased in colon tissues of Crohn's disease patients 25 and in the CD4 + T lymphocytes of mesenteric lymph nodes in DSS-induced colitis model 45 . These studies collectively point to an important role of TRIM27 in colitis.
Chronic or unresolved inflammation promotes tumorigenesis. In the AOM/DSS model, TRIM27-deficient mice developed less, smaller, and lower-grade colon tumors compared with their wildtype littermates. Consistently, the tumor tissues in TRIM27deficient mice had decreased STAT3 activation, lower levels of inflammatory cytokines, lower infiltration of inflammatory cells, and decreased induction of STAT3 target genes including Bcl-xl, c-Myc, Pcna, and Ccnd1. These results suggest that TRIM27 plays important roles in CAC development.
In addition, we found that the body weight changes and inflammatory cytokine production between WT and TRIM27deficient mice during AOM/DSS treatment were not as notable as that of during the process of DSS-induced acute colitis, indicating that loss of TRIM27 barely reduced colonic inflammation in AOM/DSS-treated mice. Moreover, it has been reported that TRIM27 acts as a negative regulator of type I IFNs production in response to virus infection 31,46 . However, there is no difference in type I IFN production between TRIM27-deficient mice and their wild-type littermates, indicating that the actual link between TRIM27 and STAT3 seems to be direct. As TRIM27 could function both in hematopoietic cells and IECs to regulate STAT3 activation, less mucosal inflammation orchestrated by hematopoietic cells and impaired activation of STAT3 as well as expression of pro-oncogenic STAT3 target genes in IECs may lead to development of fewer tumors in TRIM27-deficient mice in this model.
Our findings reveal unexpected functions of the retromer as an important platform for IL-6-induced STAT3 activation. Confocal microscopy indicated that TRIM27 was localized at the retromer-positive structures but not early endosomes. Interestingly, phosphorylated STAT3 induced by IL-6 stimulation was also colocalized with TRIM27 at the retromerpositive structures but not early endosomes. Knockdown of TRIM27 impaired the recruitment of STAT3 to JAK1 as well as gp130, JAK1, and STAT3 to the retromer but not the assembly of the retromer, suggesting that TRIM27 acts as a linker for recruiting gp130, JAK1, and STAT3 to the retromer. Consistently, knockdown of the retromer component VPS35 impaired the associations of TRIM27 with JAK1 and STAT3, and inhibited IL-6-induced transcription of downstream genes. These results suggest that the retromer-positive structure acts as a platform for TRIM27-mediated assembly of JAK1-STAT3 complexes and subsequent phosphorylation of STAT3 following IL-6 stimulation.
Previously, it has been reported that internalized IFNAR1 and IFNAR2 subunits of the IFNAR complex are differentially sorted by the retromer at the early endosome, in which IFNAR2 is recycled to the plasma membrane, whereas IFNAR1 is sorted to the lysosome for degradation 36 . Knockdown of VPS35 leads to abnormally prolonged residency and association of the IFNAR subunits at the early endosome, resulting in prolonged activation of STAT1 in response to type I IFNs 36 . In light of this observation, it seems that the retromer can act as a platform for activation of different members of the STAT family, though by different mechanisms. It has also been reported that TRIM27-MAGEL2 is required for retrograde transport via facilitating K63-linked ubiquitination of WASH, which is an essential regulator of retromer-mediated transport 33 . However, our results indicated that the E3 ligase activity of TRIM27 was dispensable for mediating STAT3 activation, suggesting that different signals can dictate how TRIM27 functions in the retromer.
It has been reported that c-Met can activate STAT3 at the endosome and phosphorylated STAT3 induced by c-Met is colocalized with EEA1, an early endosome marker. Our results indicated that phosphorylated STAT3 induced by IL-6 was localized at the retromer-positive structures but not the endosome, which is consistent with a previous study showing that phosphorylated STAT3 induced by IL-6 does not colocalize with various endosomal markers including EEA1, Rab5, Rab7, Rab11, and LAMP1 15 . These studies suggest that different signaling pathways may utilize different platforms for activation of STAT3. Since TRIM27 deficiency also inhibited STAT3 activation by OSM, which is another IL-6 family member, it is possible that the retromer is a common platform for STAT3 activation induced by the IL-6 family members which share the common co-receptor gp130. Our findings provide important insights to the mechanisms of STAT3 activation induced by the IL-6 family and identify TRIM27 as a potential therapeutic target for treatment of cancer. Fig. 6 Deficiency of TRIM27 attenuates DSS-induced colitis. a Trim27 +/+ (n = 11) and Trim27 −/− (n = 6) mice were treated with 3% DSS over a 10-day period, and their body weights were daily monitored. Results are represented as mean ± SD. b Survival of mice described in a was monitored until the 13th day. Results are represented as mean ± SD. c Trim27 +/+ and Trim27 −/− mice were treated with 3% DSS or left untreated for 9 days before their colon lengths were measured. Results are represented as mean ± SD, n = 3. d Representative images of haematoxylin and eosin staining of colon tissues of mice described in c. Scale bars, 100 μm. e Histological analysis of colon tissues described in d. The histological scores were determined in a double-blind manner. Results are represented as mean ± SD, n = 5. f Immunoblotting analysis of pY705-STAT3 and p-P65 levels in colon tissues of Trim27 +/+ and Trim27 −/− mice treated with 3% DSS or left untreated for 9 days. Lysates from three different mice were analyzed for each group. g qPCR analysis of colon tissues of Trim27 +/+ and Trim27 −/− mice treated with 3% DSS or left untreated for 9 days. Results are represented as mean ± SD, n = 3. h ELISA measurement of cytokine levels in sera and colon tissues of Trim27 +/+ and Trim27 −/− mice treated with 3% DSS or left untreated for 9 days. Results are shown as mean ± SD, n = 5-13. *P < 0.05, **P < 0.01, ***P < 0.001, unpaired t test (a, c, e, g, h) or log-rank test Constructs. Mammalian expression plasmids for Flag-or HA-tagged TRIM27 and its mutants, gp130, JAK1, and its mutants, JAK2 and its mutants, STAT3 and its mutants, VPS35, VPS29, VPS26, and GFP-or Cherry-tagged TRIM27, VPS35, VPS29, and VPS26 were constructed by standard molecular biology techniques. STAT3 and STAT1/2 luciferase reporter plasmids were purchased from Qiagen. IRF1 luciferase reporter plasmid was previously described 48 .
Transfection and luciferase reporter assays. HEK293 cells were transfected by standard calcium phosphate precipitation method. HeLa and RKO cells were transfected by Fugene HD (Roche). Empty control vector was added to ensure that each transfection receives the same amount of total DNA. For normalization of transfection efficiency, 0.01 μg of pRL-TK (Renilla luciferase) reporter plasmid was added to each transfection. Luciferase assays were performed using a dual-specific luciferase assay kit (Promega). All reporter assays were repeated at least three times.
qPCR. Total RNA was isolated from cells using TRIzol reagent (TAKARA). After reverse transcription, the cDNA products were subjected to real-time PCR analysis to measure mRNA expression levels of tested genes. Gene-specific primer sequences were listed in Supplementary Table 1.
Establishment of stable cell lines. HEK293 cells were transfected with TRIM27overexpressed or TRIM27-RNAi or corresponding control retroviral plasmids and two packaging plasmids (pGag-Pol and pVSV-G) by calcium phosphate precipitation. The culture medium was replaced with new medium without antibiotics at 12 h after transfection. After additional 24 h, the recombinant virus-containing medium was filtered and used to infect HeLa, RKO, or HEK293 cells in the presence of polybrene (8 μg/mL). The infected cells were selected using puromycin (2 μg/mL) for 2 weeks before additional experiments were performed.
For endogenous coimmunoprecipitation experiments, HeLa (5 × 10 7 ) cells were starved in DMEM without FBS overnight and stimulated with IL-6 (100 ng/mL) for the indicated times or left untreated. Coimmunoprecipitation and immunoblot experiments were performed as described above. Uncropped scans of the immunoblots are provided in Supplementary Figures 8-14.
Confocal microscopy. HeLa cells were transfected with the indicated plasmids by Fugene HD (Roche). At 24 h post transfection, the cells were starved with medium without FBS overnight and then treated with IL-6 (50 ng/mL) for the indicated times followed by fixation with 4% (w/v) paraformaldehyde for 15 min at 4°C. The fixed cells were then permeabilized in PBS containing 0.1% Triton X-100 for 5 min. After washing in PBS, cells were blocked with 1% BSA prepared in PBS for 1 h, and then incubated with the indicated antibodies overnight at 4°C. After washing with PBS, cells were incubated with the secondary antibodies conjugated with Alexa Fluor 594 or Alexa Fluor 647 dye for 1 h at room temperature before confocal microscopy.
Soft agar assay. TRIM27-overexpressed or TRIM27-RNAi or corresponding control HeLa/RKO/HEK293 cells were suspended in 2 mL of 0.35% agarose in DMEM and seeded in each well of 6-well plate containing 2 mL of 0.7% agarose in DMEM. Cells were grown at 37°C for about 2-3 weeks and stained with crystal violet (0.005% w/v).
Tumor xenografts. TRIM27-overexpressing or TRIM27-RNAi or corresponding control HeLa cells (1 × 10 6 ) or RKO cells (1 × 10 7 ) were injected into 6-week-old athymic BALB/c nu/nu mice. Tumor volume (V) was measured every 4 days. All mice were handled in accordance with the Wuhan University Medical Research Institute animal care and use committee guidelines.
Colitis and CAC models. For colitis model, sex-and age-matched Trim27 +/+ and Trim27 −/− mice (8-10 weeks of age) were provided 3% DSS in their drinking water for 10 days. For CAC model, mice were injected intraperitoneally with AOM (10 mg/kg) and after 5 days, DSS (2.5%) was given in drinking water for 7 days followed by regular water for 14 days. This cycle was repeated twice with 2.5% DSS, and mice were killed on day 80. Histological assessments of colitis and severity scores were made in a double-blinded manner after hematoxylin and eosin (H&E) staining as described 49 .
Bone marrow chimeras. To evaluate the contribution of hematopoietic cells to DSS-induced colitis, bone marrow chimeric mice were established as described 50 . In brief, 8-week-old recipient mice were lethally irradiated (single dose of γ irradiation, 950 rad) and intravenously injected with 10 7 bone marrow cells isolated from wild-type and TRIM27-deficient mice 24 h after irradiation. The bone marrow reconstitution was performed reciprocally, resulting in the generation of four Fig. 7 Deficiency of TRIM27 in hematopoietic cells is responsible for attenuation of DSS-induced colitis. a Four groups of mice (WT > WT, WT > KO, KO > WT, and KO > KO) generated by bone marrow transplantations were treated with 3% DSS over a 9-day period, and their body weights were daily monitored. Results are represented as mean ± SD, n = 10-12. b Survival of mice described in a was monitored until the 13th day. Results are represented as mean ± SD, n = 10-12. c Colon lengths of mice described in a were measured 9 days after 3% DSS treatment. Results are represented as mean ± SD, n = 7-12. d Representative images of haematoxylin and eosin staining of colon tissues of mice described in c. Scale bars, 100 μm. e Histological analysis of colon tissues described in d. The histological scores were determined in a double-blind manner. Results are represented as mean ± SD, n = 5. f Immunoblotting analysis of pY705-STAT3 and p-P65 levels in colon tissues of mice described in c, n = 3. g qPCR analysis of colon tissues of mice described in c. Results are represented as mean ± SD, n = 3. h ELISA measurement of cytokine levels in sera and colon tissues of mice described in a. Results are represented as mean ± SD, n = 5. *P < 0.05, **P < 0.01, ***P < 0.001, unpaired t test (a, c, e, g, h) or log-rank test (b). Data are representative of two experiments with similar results   Statistics. Unpaired Student's t test was used for statistical analysis with Microsoft Excel and GraphPad Prism Software. For the mouse survival study, Kaplan-Meier survival curves were generated and analyzed by log-rank test; P < 0.05 was considered significant. Fig. 8 TRIM27 deficiency suppresses AOM/DSS-induced CAC. a A schematic overview of the CAC model. Five days after initial AOM injection (10 mg/ kg), DSS (2.5%) was given in drinking water for 7 days followed by regular water for 14 days, and this cycle was repeated twice before the following analysis. b The colons of Trim27 +/+ and Trim27 −/− mice were removed and photographed. The tumor numbers (left panel) and tumor sizes (right panel) were measured. Results are represented as mean ± SD, n = 16. c Representative images of immunohistochemical staining of colon tumors of Trim27 +/+ and Trim27 −/− mice. Scale bars, 100 μm. d Immunoblotting analysis of colon tumors and adjacent normal tissues of Trim27 +/+ and Trim27 −/− mice. Samples from three independent mice for each group were analyzed. e ELISA measurement of cytokine levels in sera and colon tissues of Trim27 +/+ and Trim27 −/− mice. Results are represented as mean ± SD, n = 5-11. *P < 0.05, **P < 0.01, ***P < 0.001, unpaired t test. Data are representative of three experiments with similar results