TRIB1 regulates tumor growth via controlling tumor-associated macrophage phenotypes and is associated with breast cancer survival and treatment response

Molecular mechanisms that regulate tumor-associated macrophage (TAM) phenotype and function are incompletely understood. The pseudokinase TRIB1 has been reported as a regulator of macrophage phenotypes, both in mouse and human systems. Methods: Bioinformatic analysis was used to investigate the link between TRIB1 expression in breast cancer and therapeutic response to chemotherapy. In vivo models of breast cancer included immune-competent mice to characterize the consequences of altered (reduced or elevated) myeloid Trib1 expression on tumor growth and composition of stromal immune cell populations. Results: TRIB1 was highly expressed by TAMs in breast cancer and high TRIB1 expression correlated with response to chemotherapy and patient survival. Both overexpression and knockout of myeloid Trib1 promote mouse breast tumor growth, albeit through different molecular mechanisms. Myeloid Trib1 deficiency led to an early acceleration of tumor growth, paired with a selective reduction in perivascular macrophage numbers in vivo and enhanced oncogenic cytokine expression in vitro. In contrast, elevated levels of Trib1 in myeloid cells led to an increased late-stage mammary tumor volume, coupled with a reduction of NOS2 expressing macrophages and an overall reduction of macrophages in hypoxic tumor regions. In addition, we show that myeloid Trib1 is a previously unknown, negative regulator of the anti-tumor cytokine IL-15, and that increased myeloid Trib1 expression leads to reduced IL-15 levels in mammary tumors, with a consequent reduction in the number of T-cells that are key to anti-tumor immune responses. Conclusions: Together, these results define a key role for TRIB1 in chemotherapy responses for human breast cancer and provide a mechanistic understanding for the importance of the control of myeloid TRIB1 expression in the development of this disease.


Introduction
Breast cancer (BC), the leading cause of cancer death in females [1], is initiated by the formation of a tumor niche where cancer-initiating cells or breast cancer stem cells recruit healthy, non-transformed cells [2,3]. These cells are re-educated by signals released from cancer cells to promote the expression Ivyspring International Publisher of oncogenic cytokines and growth factors [4]. Tumor-associated macrophages (TAMs) are one of the most abundant cell types that can comprise up to 50% of the tumor microenvironment (TME) and facilitate tumor initiation and development [5]. A number of published studies reported high infiltration of TAMs in breast cancer, correlating with poor prognosis and clinical outcomes [6][7][8][9]. Triple-negative breast cancer (TNBC) tumors have been shown to have a higher number of CD68 + macrophages compared to other subgroups [10].
The plasticity of infiltrated TAMs is influenced by environmental signals and can be functionally classified into M1 (pro-inflammatory) and M2 (anti-inflammatory) cells, as two extremes [9]. Though TAMs are able to express markers of either polarization phenotype, pro-inflammatory (M1-like) macrophages are generally observed upon entering to the tumor site [11], but these macrophages, stimulated by the type 1 T helper cell (Th1) cytokines are known to exhibit anti-tumor capacity by generating anti-tumor cytokines (such as TNF, IL-2, and IL-12) and reactive nitrogen and oxygen intermediates [9,12,13]. Most TAMs are polarized to have a M2-like phenotype after infiltration and produce antiinflammatory cytokines (such as IL-4) and growth factors to inhibit immune response and promote proliferation [14]. These M2-like TAMs have been associated with unfavorable clinical outcomes and patient survival [15].
Hypoxia promotes the dephosphorylation of chemoattractant receptors and inhibits migrating stimulating factors that trap TAMs in the hypoxic area and is associated with aggressive breast tumor phenotypes [14,16]. The entrapped, hypoxic TAMs facilitate tumor vascularization and immune suppression by expressing angiogenic molecules and immunosuppressive factors, respectively [16,17].
Using Bayesian network inference modelling, TRIB1 expression was shown to be correlated with the levels of NF-κB and IL-8 in breast cancer, and was also considered as a potential biomarker for clinical outcomes [27]. However, unlike recent papers reporting an oncogenic role of TRIB1 in prostate cancer via regulating macrophage infiltration and inducing M2-like polarization [28], to date no study has examined the TAM-specific tumoral capacity dependence on Trib1 and how this may influence tumor development. Therefore, we analyzed the impact of TRIB1 on the survival of BC patients and showed that mutations or reduced expression of this gene are associated with a poor clinical prognosis, response to therapy and survival. We also found that TRIB1 highly expressed by TAMs, both in human BC and murine model. Prompting us to study mammary tumor development in mice where levels of myeloid-Trib1 (mTrib1) have been genetically altered. Based on our analyses, we demonstrate that both overexpression and knockout of mTrib1 promote tumor growth, albeit through distinct molecular mechanisms and at different stages of tumor growth, providing a novel mechanistic insight into the functional importance of TAM phenotypes.

Study approval
All animal studies were approved and conducted in accordance with the University of Sheffield code of ethics and Home Office regulations (project license No. PPL70/8670). Human monocytederived macrophages (MDMs) isolated from healthy participants were obtained with signed participantinformed consent and approval from the University of Sheffield Research Ethics Committee (project license No. SMBRER310) and in accordance with the Declaration of Helsinki. All participants gave written informed consent. The TRIB1-CD68 co-staining was studied by immunofluorescence in breast tumor samples from patients according to the Declaration of Helsinki. Studies were performed after approval of The Bioethics Committee at The MD Anderson Cancer Center Madrid (MD21/004). Written informed consent was obtained from each patient.

Bioinformatics analysis of human BC transcriptomes
In order to establish a correlation between mutations and patient clinical outcome, Genotype-2-Outcome (http://www.g-2-o.com) algorithms were used, as previously described [29]. This approach calculates the prognosis conferred by a specific transcriptomic signature linked with a mutation and patient survival. Briefly, 763 breast cancer patients with NGS data publicly available from TCGA were collected and classified in terms of TRIB1 mutation status (functionally annotated using SNPeff v3.5 [30] and including just somatic mutations), considering only the ones labelled as 'KEEP' by the MutTect judgment algorithm and present in at least four reads with a minimum of 20-fold read coverage. Two cohorts, the wild type and the mutant, are defined and the transcriptomic signature by univariate receiver operating characteristic (ROC) analysis performed separately for each gene and significantly altered genes are selected by their area under the curve value (calculated by the ROCR package), and the associated P value obtained from the ROC analysis, considering the null hypothesis where AUC value equals to 0.5 and only genes passing both AUC and P value thresholds are considered significant. The median expression values for different transcripts are used as a cut-off to discriminate "high" and "low" expression cohorts, which are compared using a Cox survival analysis (proportional hazards) and graphics are drawn using ggplot2 package running in R Studio Version 1.2.5033. To calculate prognosis under treatment, gene expression and therapy response are compared using receiver operating characteristics and Mann-Whitney test t or ROC test in the R statistical environment (www.r-project.org) using Bioconductor libraries. Statistical significance was set at p < 0.05 in both cases.

Mice
All mice were bred on a C57BL/6 genetic background under the University of Sheffield code of ethics, and Home Office regulations in the University of Sheffield Biological Service Unit. Trib1 fl/fl × Lyz2Cre (Trib1 mKO ), ROSA26.Trib1Tg × Lyz2Cre (Trib1 mTg ) and their corresponding WT controls have recently been described [19].

Tumor models
The mouse Basal-B BC cell line, E0771 [37] (obtained from Dr Jessalyin Ubellacker (University of Harvard, USA) was cultured in DMEM medium (Gibco) containing 10% (v/v) low endotoxin heat-inactivated fetal bovine serum (Biowest), and 1% L-glutamine (Lonza). Eight-week-old female Trib1 mKO and Trib1 mTg mice were inoculated with 3 × 105 E0771 cells into the right nipple via intra-ductal injection. Once the tumors formed, the size was measured every 2 days with calipers until it reached 15 mm in diameter. Data was accumulated from >5 independent experiments, each containing several litter-mates, including both wild type and mTrib1-altered mice. Samples were harvested at the end of each experiment and were processed/analyzed for either as a batch (FACS) or together (immunofluorescence, qRT-PCR), as appropriate.

Cancer cell culture and conditioned medium production
MDA-MB-231, BT474, SKBR3 and MCF7 cell lines were cultured in RPMI-1640 (Gibco) with 10% (v/v) low endotoxin heat-inactivated fetal bovine serum (LE-FBS) (Biowest), 1% (v/v) streptomycin/ penicillin (Gibco), 1% L-glutamine (Lonza). All cells were obtained from Dr. Penelope Ottowell and Dr. Munitta Muthana (University of Sheffield, UK) and subsequently maintained in our laboratory. Routinely testing for mycoplasma contamination demonstrated they were consistently negative. To obtain MDA-MB-231 conditioned medium (CM), cells were cultured in T75 flasks for 48 hours at 37 °C in a 5% CO2 and the supernatant centrifuged at 600 × g for 5 minutes to remove cells and cellular debris (but not extracellular vesicles released by these cells).

Isolation of human blood monocytes
Whole blood was collected in 3.8% trisodium citrate (Sigma) and used immediately to isolate cells. In 15 ml of Ficoll-Paque PLUS (GE Healthcare), 30 ml of blood was gently layered and centrifuged at 900 × g for 20 minutes at room temperature (RT) to separate peripheral blood mononuclear cells (PBMCs) from plasma. PBMCs were recruited in PBS-EDTA (Thermo Fischer) solution (PBSE) and centrifuged at 400 × g for 5 minutes at RT. After red blood cell lysis with 10 ml of RBC lysis buffer (155 mM NH4Cl, 10 mM KHCO3, 0.1M EDTA in H2O) for 5 minutes at RT, 40 ml of PBSE was added and centrifuged at 1500 rpm or 400 × g for 5 minutes. Cells were counted using a hemocytometer (Hawksley) and resuspended in 90 μl 4 °C MACS buffer (0.5% [w/v] bovine serum albumin [BSA, Sigma] -PBSE) and 10 μl CD14+ microbeads (Miltenyi Biotec) per 10 7 cells for 15 minutes at 4 °C. 2 ml of MACS buffer was added and centrifuged at 260 × g for 5 minutes. CD14+ monocytes were isolated with LS column (Miltenyi Biotec) and MidiMACSTM Separator (Miltenyi Biotec) for differentiation.

TRIB1 siRNA transfection
Viromer Green (Lipocalyx) was used to transfect TRIB1 siRNA (ON-TARGET plus siRNA, Dharmacon) and Non-Targeting Control siRNA (ON-TARGET plus siRNA, Dharmacon) in order to knockdown TRIB1 level in humans MDMs according to the manufacturer's instructions.

Isolation of BMDMs
The femur and tibias of mice were collected, and tissues were gently removed from the bones. Bone marrow was harvested by flushing the bones with PBS using a 2.5 ml syringe. Any clumps of cells were dispersed with a pipette and passed through 70 μm cell strainer (Fisher Scientific). The cell suspension was centrifuged at 500 × g for 5 minutes, and the pellet was cultured in fresh L929 cell-conditioned DMEM medium for 6 days to differentiate into BMDMs.

Protein extraction and quantification
Cells washed with PBS were collected into a 1.5 ml Eppendorf tube and thoroughly mixed with lysis buffer (RIPA buffer with 1% protease and phosphatase inhibitor). Cells were incubated at -80 °C for 30 minutes and sonicated for 15 seconds to allow further lysis. Cells were then centrifuged at 15,000 × g for 10 minutes at 4 °C to remove debris and supernatant was collected and stored at -80 °C. PierceTM BCA Protein Assay Kit (Thermo Scientific) was used to quantify proteins as manufacturer's instructions.

Western blot
Proteins were mixed with 5 × Laemmli buffer and incubated at 100 °C for 10 minutes. Samples were immediately transferred in the ice afterwards. All samples and prestained protein ladder (10-250 kDa, Thermo ScientificTM) were loaded into the columns of NuPAGETM 4-12% Bis-Tris Gel (Invitrogen) placed in the Invitrogen tank containing 1× NuPAGE MOPS SDS running buffer (Novex). The gel was run at 100v for 75 minutes and transferred to a PVDF (Polyvinylidene difluoride) membrane (Millipore) using NuPAGE transfer buffer (Novex) with methanol and antioxidant (Invitrogen) at 35v for 60 minutes. The membrane was blocked with 5% milk-TBST at RT for 1 hour and incubated overnight with TRIB1 (Millipore), and HSP90 (Abcam) diluted in 5% milk-TBST (1:1000 and 1:5000 respectively) at 4 °C. The membrane was then washed with 0.1 v/v TBST for 5 minutes 3 times and incubated with Polyclonal Goat anti-Rabbit Immunoglobulin/HRP, and Polyclonal Rabbit anti-Rat Immunoglobulin/ HRP (Dako) diluted in 5% milk-TBST (1:2500 and 1:5000 respectively) at RT for 1 hour. The membrane was then washed with TBST 3 times for 5 minutes, incubated with ECL, and imaged with Bio-Rad imager.

RNA extraction and quantification
Cells were gently washed twice in PBS and incubated at RT for 5 minutes in 700 μl of QIAzol lysis reagent to homogenate the cells, 140 μl of chloroform was added to cells and RNA extracted using the miRNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. The amount of RNA was quantified using a Nanodrop Spectrophotometer ND1000 and stored at -80 °C until used for RT-qPCR analysis.

cDNA synthesis and Real-time quantitative PCR analysis
cDNA was produced with iScript cDNA synthesis kit (Bio-Rad) according to the manufacturer's instructions. Quantitative RT-PCR was performed using primers designed with NCBI BLAST to target human macrophage polarization markers (Supplementary Table 1) and PrecisionPLUS SYBR-Green master mix (Primerdesign). SYBR green master mix with forward and reverse primers were added to each well of a 364-well RT-qPCR plates at a total volume of 5.6 μl, followed by the addition of 5 μl of cDNA (0.4 ng/ul). The plates was then centrifuged for 2 minutes at 2000 rpm and fluorescence measured in a Bio-Rad I-Cycler PCR machine using the protocol provided by the manufacturer. GAPDH and Β-actin were used as housekeeper genes, and the changes in gene expression were obtained using the 2 -ΔΔCT method.

Tissue dissociation
The tumor tissue collected from mice was shredded with scissors, and placed in 5 ml of tumor-dissociation medium (TDM) (IMDM medium, 0.2 mg/ml collagenase IV, 2 mg/ml dispase, 1.25 ug/ml DNase 1) in a 15ml bijou tube, and rotated at 37 °C for 30 minutes. 5 ml of 10% FBS-TDM was added into the tube and passed through a 70 μm filter (Fisher Scientific). The samples were placed directly on ice and centrifuged at 4500 rpm for 5 minutes. The cell pellet was washed three times with PBS and used for flow cytometry analysis.

TRIB1 is highly expressed in tumor-associated macrophages and its expression correlates with response to chemotherapy and patient survival in breast cancer
Whilst TRIB1 is oncogenic in several cancer settings, its potential importance in BC pathogenesis and response to therapy are largely unknown. To explore the potential role TRIB1 may play in BC we analyzed the correlation between transcriptomic signatures associated with somatic TRIB1 mutations and BC survival in a dataset of 6697 patients [29], using the G-2-O algorithm, as previously described [29]. This established the association between the prognosis of the specific transcriptomic signature linked with mutations in the protein coding region of TRIB1 mRNA and patient survival by comparing two cohorts, encoding for wild type or mutant forms of TRIB1. The mutant cohort includes only somatic mutations labelled as 'KEEP' by the MutTect judgment algorithm, and present in at least four reads with a minimum of 20-fold read coverage. Mutations are functionally annotated using SNPeff v3.5 [30]. Both cohorts were compared using a Cox survival analysis, that showed a highly significant reduction in patient survival in tumors with TRIB1 mutations ( Figure 1A, HR 0.56 CI 0.50-0.62; lograk p = 4.1×10 -26 ), indicating a worse prognosis for those patients with the transcriptomic signature associated with somatic mutations in the TRIB1 gene.
Next, we evaluated whether RNA levels of TRIB1 correlated with relapse-free survival (RFS) in BC patients by interrogating a dataset including 1329 patients with RFS information at 5 years [38]. We analyzed different types of therapy, ranging from endocrine treatment (tamoxifen and aromatase inhibitor) to specific anti-HER2 inhibitors (trastuzumab or lapatinib), and several chemotherapy treatments, including taxanes, anthracyclines, Ixabipelone, CMF (cyclophosphamide, methotrexate, fluorouracil), FAC (fluorouracil, adriamycin, citroxan) and FEC (fluorouracil, epirubicin, cyclophosphamide). We found that TRIB1 expression specifically correlates with 5-year relapse-free survival only in anthracycline-based chemotherapy in BC patients ( Figure 1B) and specifically in Basal ( Figure 1C) and Luminal-B ( Figure 1D) BC subtypes, suggesting it as a putative prognostic marker in this setting. Of note, the luminal-B and basal-like BC subtypes often contain tumors with enhanced proliferation rate and are usually treated with more aggressive chemotherapy combination, including anthracycline-based chemotherapies. Since TRIB1 is well known for its action to control cell proliferation [39][40][41][42], this suggests a potential mechanistic explanation for this association. Data presented as mean±SD. Cells in 5 random fields of views for each tumor were manually quantified using ImageJ. (F) Representative image of TRIB1 (red) and F4/80 (green) fluorescence staining of primary murine breast tumors from wild-type (C57/BL6) animals (Scale: 50 µm). Quantification of TRIB1 expressing cells and F4/80+ cells in the TME from "F" relative to total cell counts (n=5 mice/group). Data presented as mean±SEM. Cells in 5 random fields of views for each animal were manually quantified using ImageJ. (G) Human MDMs isolated from healthy donors were transfected with either control (M Control ) or TRIB1 siRNA for 48 hours (M TRIB1-KD ). Results of paired t-tests are presented; mean±SEM is plotted; *p < 0.05, **p < 0.01 (n= 6-9 donor/group). (H) Pathway analysis of TRIB1 associated genes in monocytes and MDMs in participants of the Cardiogenics Transcriptomic Study. MDM (n= 596) and monocytes (n= 758) were ranked according to TRIB1 RNA levels and genes that were differentially expressed between the top vs. bottom 25% of the rankings were analyzed with QuSAGE. The  Overall, these analyses suggest the involvement of TRIB1 in response to therapy in breast cancer, and both its loss of function mutations and reduced expression levels render worse prognosis.
Once identified the potential role of TRIB1 in breast cancer outcome and response to therapy, we aimed to characterize TRIB1 protein expression within mammary tumors, using both specimens from human breast cancer ( Figure 1E) and a murine model where BC growth was induced in immunecompetent, C57BL/6 mice with an orthotropic injection of a murine, Basal-B BC cell line, E0771 [37] ( Figure 1F). This analysis revealed that up-to 25% of cells in the tumor expressed high levels of TRIB1 protein and that about 42% (Figure E) of these cells were also positive for the macrophage marker CD68 ( Figure 1E). Similarly, 25% of cells in the murine tumor expressed high levels of TRIB1 and 85% ( Figure  1F) of these were also positive for the mouse macrophage marker F4/80, together suggesting a potential role for macrophage TRIB1 in regulating tumorigenesis in BC, both in human and murine tumors. Of note, the functional importance of monocyte/macrophages in this mouse model has been demonstrated previously by showing that selective depletion of these cells (but not neutrophils) with gemcitabile led to reduced E0771 tumor growth [43].
In order to gain an initial mechanistic insight into myeloid TRIB1-dependent alterations, relevant to tumor-biology, we used human monocyte-derived macrophages (MDMs) isolated from healthy human blood and transfected them with TRIB1 siRNA to reduce the expression (M TRIB1-KD ) to assess the expression of a range of genes known to play an important role in TAM function. Analysis of TRIB1 knockdown efficiency is presented in Figure 1M. This analysis revealed that the knockdown of TRIB1 in MDMs (for knockdown efficiency of approx. 50%, see Figure 3M) enhances their pro-inflammatory phenotype with a significant increase in levels of expression of IL-1β (p < 0.05), CD80 (p < 0.05), and TNF (p < 0.01) mRNA and reduced SCARB1 expression (p < 0.05) ( Figure 1G), in line with changes observed in MDMs stimulated towards an inflammatory (M1) phenotype (M LPS+INF-γ , Supplementary  Figure 1). These changes are also in line with reported transcriptomic changes in TAMs [44].
Next, we carried out a gene enrichment analysis with QuSAGE [34][35][36] to identify biological pathways associated with altered TRIB1 expression in human monocytes (n = 758) and MDMs (n = 596), using data from the Cardiogenics Transcriptomic Study [31][32][33]. Comparison of the 10 most significantly enriched pathways in MDMs (macrophage) and monocytes revealed that most of these pathways were only enriched in macrophages, confirming the distinct regulatory impact of TRIB1 between these cell types ( Figure 1H, Supplementary Table 2). From those pathways significantly associated with TRIB1 levels in macrophages, such as creation of C4 activators, translocation of ZAP 70 to immunological synapse, PD1 signalling, and phosphorylation of CD3 and zeta chains, have previously all been shown to be involved in the promotion of tumor growth and regulate T-cell activation and polarization [45,46]. In addition, increased PD1 signalling has been reported to increase macrophage proliferation and activation, and inhibit phagocytosis and tumor immunity in TAMs [47,48].
Finally, treatment of MDMs with cancer cell-conditioned medium (CM) also showed a significant overexpression of TRIB1 in these cells (M CM ), compared to control (M UN ) and M LPS+INF-γ cells (p < 0.05) ( Figure 1I), suggesting a potential two-way regulation of TRIB1 expression between BC cells and tumor macrophages.

Mammary tumor growth is accelerated by alteration of myeloid Trib1 levels
Based on the above evidence of potential association between TRIB1 and BC, we hypothesized that myeloid TRIB1 expression may influence the aggressiveness of BC, and thus modulation of Trib1 expression in these cells would alter tumor growth. The complex role of tumor resident macrophages has been studied extensively, and has recently been proposed that anthracycline-based chemotherapy may lead to an effective anti-tumor immunity via macrophage-mediated effects [49]. Our observations presented above may link high TRIB1 levels mechanistically to the enhanced responses to anthracycline-based chemotherapy, prompting us to characterize how altered Trib1 expression in myeloid cells may alter BC tumor growth. To test this, we used myeloid-specific Trib1 overexpressing (Trib1 mTg ) and knockout (Trib1 mKO ) mice and. their littermate controls, on an immune-competent, C57BL/6 background. We have recently reported details of the development and initial characterization of these mouse lines [19]. We modelled BC growth, using these myeloid-specific mouse lines and a murine breast cancer cell line, E0771, that has recently been reported to be a luminal-B subtype [37], thus closely resembling the human BC subtype, where TRIB1 expression levels are correlated with response to chemotherapy ( Figure 1D). Mammary fat pads of 8-week-old mice were injected with E0771 cells, and the rate of tumor growth measured (Figure 2A). Interestingly, whilst Trib1 mKO tumor growth rate accelerated from an early stage and significantly increased from day 18 (p < 0.001), reaching 15 mm in diameter (1770 mm 3 volume) at day 22, Trib1 mTg animals displayed a tumor growth rate similar to wild-type littermates until day 24 ( Figure 2B). At this point, the rate of tumor volume growth became slower in wild-type animals, compared to Trib1 mTg animals. In the Trib1 mTg group, tumor continued to grow and reached 15 mm in diameter at day 30 (p < 0.01) ( Figure 2B). Tumors from both cohorts were collected when they reached 15 mm in diameter for further analysis, together with corresponding WT littermate controls.

Myeloid-Trib1 knockout reduces macrophage infiltration and promotes oncogenic cytokine expression in TAMs
Tumors from Trib1 mKO animals were initially analyzed by flow cytometry to investigate populations of immune cells in the tumor microenvironment (TME) ( Figure 2C, Supplementary  Figures 2A and C). This analysis revealed that tumors developed in Trib1 mKO animals had a significantly reduced infiltration of both Ly-6C + monocytes and F4/80 + macrophages into the TME (p < 0.05) ( Figure  2D-E). In contrast, the percentage of Ly-6G + neutrophils, NK1.1 + NK cells, and CD3 + T-cells and its subtypes (CD4 + naïve and CD8 + cytotoxic T cells) in the tumor were not altered between Trib1 mWT and Trib1 mKO ( Figure 2F-H, Supplementary Figure 2B and D).
In order to explore the potential mechanisms of the observed accelerated mammary tumor growth and its links with the reduced monocyte and macrophage infiltration in Trib1 mKO mice, we further assessed the localization of macrophages and their phenotype using immuno-fluorescence staining and flow cytometry ( Figure 3A and 3C, and Supplementary Figure 2E). Perivascular TAMs (PV TAMs) are in close contact with blood vessels (within 250 µm radius) and play a crucial role in angiogenesis in mammary cancers as well as metastasis and intravasation of cancer cells [50,51]. Staining of TAMs and endothelial cells for F4/80 and CD31, respectively, revealed a significant reduction of PV TAMs in Trib1 mKO tumors ( Figure 3B). However, although an increase in pro-inflammatory macrophages has been reported previously both in full-body and myeloid-specific Trib1 knockout animals [19,21], inhibition of myeloid Trib1 expression did not alter the ratio of NOS2 + pro-inflammatory TAMs and mannose receptor (MR) + , anti-inflammatory TAMs in the TME ( Figure  3D, Supplementary Figure 2F, Supplementary figure  3), suggesting that the reduced TAM numbers, rather than their altered inflammatory phenotypes, may have contributed to the accelerated tumor growth.
In order to gain a mechanistic insight into how TRIB1 regulates monocyte-derived macrophages and the impact of reduced TRIB1 expression on re-educating macrophages towards TAMs, human MDMs were transfected with siRNA against TRIB1 (M TRIB1-KD ), followed by a treatment with tumor-conditioned medium (CM) (TAM TRIB1-KD ) which resulted in ~50% reduction in TRIB1 expression ( Figure 3M). Expression of key cytokines were assessed by RT-qPCR, revealing that expression of pro-inflammatory cytokines IL-1β (p < 0.05), IL-8 (p < 0.05), and TNF (p < 0.01) were significantly increased in M TRIB1-KD but were not altered in TAM TRIB1-KD , compared to non-targeting siRNA transfected MDMs ( Figure 3E

Overexpression of Trib1 reduces hypoxic TAM infiltration and inhibits pro-inflammatory TAM polarization
The above analysis of tumor growth in Trib1 mTg mice (Figure 2A) revealed that elevated myeloid-Trib1 levels lead to an increase in tumor size at advanced stages of tumor growth. To gain a mechanistic understanding of this effect, TAM localization and phenotypes in Trib1 mTg TME were investigated using immune-fluorescence staining. Carbonic anhydrase IX (CA9) is a cell-surface glycoprotein in the tumor, expression of which is induced by hypoxia and has been shown to be involved in cancer progression [52].

Trib1 mTg tumors display reduced T cell infiltration and reduced IL-15 expression
Recruitment of T-cells to the TME is a central mechanism for inhibition of tumorigenesis and cytokines secreted by TAMs play a key role in this process [53,54]. Our above data demonstrates that reduced vs. elevated expression of mTrib1 leads to distinct changes in TAM phenotypes and have also shown that recruitment of T-cells in Trib1 mKO tumors is unaltered ( Figure 2H and Supplementary Figure  2A-D). Thus, we next tested whether the observed alterations in TAM numbers and phenotypes affect T-cell recruitment in Trib1 mTg animals. Fluorescence staining was used to identify changes in CD3 + T-cell numbers and populations of CD4 + naïve and CD8 + cytotoxic T-cells ( Figure 5A and B, Supplementary  Figure 6), revealing a significant reduction in the overall number of CD3 + T-cells in the Trib1 mTg TME, compared to Trib1 WT (p < 0.01) ( Figure 5C). Furthermore, the proportion of both CD4 + CD3 + naïve T-cells and CD8 + CD3 + cytotoxic T-cells were significantly reduced in Trib1 mTg (p < 0.05) ( Figure  5D-E).
In order to identify mTrib1-dependent mechanisms that may explain an impaired T-cell recruitment to the tumor, we have assessed the correlation between TRIB1 levels and genes that have been shown to regulate T-cell recruitment in the Cardiogenics Transcriptomic Study [31][32][33]. This analysis of approx. 600 independent samples revealed that high TRIB1 levels correlate very significantly with a reduced IL-15 expression in human macrophages (but not in monocytes) ( Figure 5F). IL-15 is a cytokine expressed by myeloid cells crucial for the development, function and survival of T-cells. IL-15 stimulates tumor-specific T-cell responses, increases cellular growth, inhibits apoptosis, and enhances immune cell activation, and as a consequence, promotes anti-tumor responses [55]. Recent work reported by Pavlakis et al. demonstrated that peritumoral delivery of heteromeric IL-15 led to an effective suppression of E0771 orthopically implanted tumors [56,57], thus, demonstrating the direct anti-tumorigenic properties of IL-15 in this in vivo model of BC. Therefore, we next assessed IL-15 in Trib1 mTg TAMs by fluorescence staining ( Figure 5G, Supplementary Figure 6). Quantification of IL-15 + TAMs revealed that overexpression of Trib1 led to significantly reduced IL-15 expressing TAM numbers in the TME (p < 0.01) ( Figure 5H). To substantiate that TRIB1 is a direct regulator of IL-15 expression in myeloid cells, RT-qPCR analysis was performed in BMDMs isolated from Trib1 mTg showing that enhanced Trib1 expression led to a significant reduction in IL-15 expression ( Figure 5I). In line with this, a transient knockdown of TRIB1 with siRNA transfection in human MDMs from healthy human participants significantly increased IL-15 expression (p < 0.01) ( Figure 5J).

Discussion
It is now widely recognized that TAMs are a crucial component of TME and the number of these cells is associated with cancer cell resistance to therapy, poor patient survival and prognosis [6][7][8][9]. However, the molecular mechanisms that shape TAM phenotype, and thus determine whether they are pro-tumorigenic or promoting anti-tumor immune responses, are poorly understood.
The pseudokinase protein, TRIB1, has been reported as a potential regulator of macrophage phenotypes. It is highly expressed in the myeloid lineage and is associated with altered tissue macrophage phenotypes [19,21,58]. Although the effect of TRIB1 in TAMs has not been elucidated, previous studies investigated TRIB1 as an oncogene in several contexts [24][25][26], including prostate and colon cancer and also associated it with sensitivity of breast cancer cells to TNF-related apoptosis-inducing ligand (TRAIL) -induced apoptosis [27,59,60].
Putting together these published data, our observations that TRIB1 expression is associated with patient survival and therapy responses in breast cancer patients and that the majority of TAMs highly express TRIB1 in a murine model of BC, as well as in human BC specimens, we hypothesized that altered TRIB1 expression in myeloid cells may modulate TAM phenotypes. As a consequence, mTRIB1 would mechanistically contribute to breast cancer tumor growth, as well as to response to chemotherapy.
Of note, the TRIB1-dependent prediction of response to therapy is only observed in those tumors with a higher rate of proliferation, such as the basal-like and luminal B phenotypes. The latter is characterized by the dual expression of the estrogen and HER2 receptor that constitutes two druggable oncogenic vulnerabilities [61]. Chemotherapy, and particularly anthracyclines, is a backbone treatment in this disease and it's known that the immunologic state can modulate the efficacy to these agents through the presence of different immune populations and secreted factors [62].
We used E0771 cells in this study, that have been shown to express estrogen receptor (ER), progesterone receptor (PR) and ERBB2 and are therefore classified as a luminal B subtype> This subtype is found in 30-40% of BCs and generally known to be more aggressive than luminal A BCs [37]. Of note, our analysis of patient survival showed that TRIB1 expression is elevated in tumors responding to chemotherapy in Luminal B BC, compared to non-responders, further justifying the choice of this murine model. TAM phenotype in the solid tumor is critical for tumor growth, where proteins secreted by cancer cells (such as IL-4, IL-10, and CSF-1) drive TAMs towards an anti-inflammatory phenotype that promotes angiogenesis and immunosuppression [11,63]. However, pro-inflammatory TAMs can also play oncogenic roles, particularly at the early phases of tumor growth, linked to hypoxia. Abundant infiltration of pro-inflammatory TAMs has been observed in early tumor development, and the expression of TNF and activation of PGC-1α and AMPK was shown to promote glycolysis and exacerbate tumor hypoxia [54,64]. The importance of hypoxic signals has also been evidenced where knockout of HIF-1α reduced the proliferation of BC cells in vitro as well as primary breast tumor volume by 60% in vivo [65]. Hypoxic TAMs have also been shown to secrete angiogenic proteins, with HIF-1α stimulating pro-angiogenic functions in TAMs, thus facilitating tumor vascularization [17,63]. Perivascular (PV) TAMs express high levels of MRC1 and VEGF to facilitate tumor angiogenesis, and help formation of paracrine feedback loops (CSF1 from cancer cells, EGF from TAMs, and HGF from endothelial cells) to initiate metastasis and intravasation of cancer cells at the TME of metastasis [50,51,66,67]. Thereby, although Trib1 mKO and Trib1 mTg both demonstrated a significant reduction in TAM infiltration overall in our BC models, detailed analysis of Trib1 mTg revealed a significant and localized reduction of TAM numbers in hypoxic areas, as well as inhibition of pro-inflammatory TAM polarization in the TME, both of which mechanisms that may contribute to the observed late acceleration of mammary tumor growth. In contrast, Trib1 mKO reduced infiltration of PV TAMs, but did not alter the number of NOS2 positive macrophages in the tumor. Instead, in vitro TRIB1 knockdown in a model of human TAMs revealed that inhibition of TRIB1 enhances expression of oncogenic cytokines in TAMs, which are involved both in cancer cell survival and immune suppression. Increased IL-6 expression in TAMs was reported to promote cancer cell survival resistance to hypoxia [68]; IL-10 is known to suppress immune surveillance, inhibit apoptosis, and to enhance migration of cancer cells [69,70]; overexpression of PD-L1 disrupts T-cell proliferation and function [71] and VEGF enhances BC growth and angiogenesis [72]. These observations are in line with our previous work, where we have shown that myeloid Trib1-deficiency alters macrophage function (in that case, formation of foam cells in the atherosclerotic plaque), rather than a clear shift in inflammatory status of Trib1 mKO cells [19].
TAMs interact with T-cells in TME to suppress T cell-driven cytotoxic immune response and promote tumor growth. Previous studies reported that TAMs impair CD8 + T-cell activation and proliferation, and depletion of TAMs in the TME enhances the infiltration of both naïve and cytotoxic T-cells [54,73]. Tumor-infiltrating T-cells enter tumor at an early stage as naïve CD4 + T-cells, followed by macrophage infiltration and contribute to early tumor rejection and/or anti-tumor effects through promoting senescence and tumor apoptosis via secretion of cytokines (such as IFN-γ and TNF) and interact with macrophages, NK cells and CD8 + T-cells to enhance tumor eradication [53]. Interestingly, a recent study from Carrero et al. have shown that most myeloid cells in the tumor TME express high levels of IL-15 [74], proposing that these stromal cells may be a critical source of this anti-tumor cytokine. In this study, we identified a significantly reduced infiltration of both CD4 + naïve and CD8 + cytotoxic T-cells into the TME in Trib1 mTg animals, despite TAM infiltration being inhibited. Our mechanistic analysis revealed that regulation of T cell infiltration may be due to a previously unrecognized role of myeloid-TRIB1 as a critical regulator of IL-15 expression.
We have reported previously that Trib1 mTg mice also express the transgene in neutrophils and Lyz2-Cre expression is also expected to delete the Trib1 allele in the mKO animals in these cells, in addition to monocyte/macrophages [19]. However, we focused our analysis reported here on monocyte/macrophages only, for a number of reasons. First, we have shown that Trib1 mT and Trib1 mKO animals have unaltered neutrophil numbers [19]. Second, relevant to our specific in vivo model of BC, as well as the TRIB1-associated patient survival, most Luminal primary BC tumors in patients have been shown not to contain tumor-associated neutrophils (TAN) [75]. Whilst a number of analyses have investigated the prognostic value of neutrophil-to-lymphocyte ratio (NLR), these (mostly retrospective) reports yielded conflicting data [76,77] and prospective studies concluded that NLR has no prognostic value in most BC subsets after correcting for clinic-pathological factors [78]. More recently, a transcriptomic analysis of BC datasets has shown that the proportion of neutrophils was significantly higher in BC cases with a higher grade and of the luminal B, TNBC and HER2+ subtypes but was not associated with tumor size or axillary lymph node metastasis [79].
Given the data we present in this study, we propose that dysregulated levels of TRIB1 in myeloid cells leads to accelerated tumor growth via distinct molecular mechanisms ( Figure 6). Specifically, TRIB1 expression is associated with lower levels of anti-tumorigenic factors such as IL6 (that promotes hypoxia-induced apoptosis), IL10 and PD-L1 (that regulate T cell immunosuppression), CCL20 (that influences response to chemotherapy) and proangiogenic VEGF. Conversely, higher levels of TRIB1 are linked to increased tumor cell survival (via NOS2) and decreased T cell-mediated immunosuppression (via IL15). More generally, this study exemplifies how alterations in the expression of the same gene in TAMs may have opposing consequences at different stages of tumor development. Whilst it is to be formally tested in future studies, we speculate that TRIB1 expression changes in TAMs could be associated with the initiation and/or with the growth of the tumor and adaptation to lack of nutrients, as well as to hypoxic environment. Nevertheless, knockout of myeloid-TRIB1 upregulates the expression of oncogenic cytokines in TAMs whilst its overexpression modifies TAM phenotype and T-cell composition in the TME, both enhancing tumor growth. Such data reinforce the general concept for the complex role of TAMs in BC and analysis of consequences for altered TRIB1 expression highlight potential diagnostic/prognostic markers and therapeutic markers for anti-cancer immunotherapy. In addition, our findings also support the idea that enhanced TRIB1 expression could be explored as a potential biomarker in BC that might help to predict response chemotherapy. Figure 6. Schematic representation of TRIB1 mediated TAM regulation. Alteration of TRIB1 distinctly regulates TAMs to enhance tumor growth. Reduction of TRIB1 accelerates oncogenic cytokine IL-6, IL-10, CCL20, PD-L1, and VEGF expressions in TAMs which take part in the inhibition of cell apoptosis, immune cell dysfunction, angiogenesis, and develop resistance to therapies. Whilst the overexpression inhibits pro-inflammatory TAM in the TME to suppress the immune response and enhance cancer cell survival. TRIB1 conversely influence IL-15 expression, which reduces T cell infiltration and potentially disrupts T cell-induced immune responses.