Targeting TRAF3IP2, Compared to Rab27, is More Effective in Suppressing the Development and Metastasis of Breast Cancer

Here we investigated the roles of Rab27a, a player in exosome release, and TRAF3IP2, an inflammatory mediator, in development and metastasis of breast cancer (BC) in vivo. Knockdown (KD) of Rab27a (MDAKDRab27a) or TRAF3IP2 (MDAKDTRAF3IP2) in triple negative MDA-MB231 cells reduced tumor growth by 70–97% compared to wild-type tumors (MDAw). While metastasis was detected in MDAw-injected animals, none was detected in MDAKDRab27a- or MDAKDTRAF3IP2-injected animals. Interestingly, micrometastasis was detected only in the MDAKDRab27a-injected group. In addition to inhibiting tumor growth and metastasis, silencing TRAF3IP2 disrupted inter-cellular inflammatory mediator-mediated communication with mesenchymal stem cells (MSCs) injected into contralateral mammary gland, evidenced by the lack of tumor growth at MSC-injected site. Of translational significance, treatment of pre-formed MDAw-tumors with a lentiviral-TRAF3IP2-shRNA not only regressed their size, but also prevented metastasis. These results demonstrate that while silencing Rab27a and TRAF3IP2 each inhibited tumor growth and metastasis, silencing TRAF3IP2 is more effective; targeting TRAF3IP2 inhibited tumor formation, regressed preformed tumors, and prevented both macro- and micrometastasis. Silencing TRAF3IP2 also blocked interaction between tumor cells and MSCs injected into the contralateral gland, as evidenced by the lack of tumor formation on MSCs injected site. These results identify TRAF3IP2 as a novel therapeutic target in BC.

The acellular fraction of the TME comprises of exosomes 7 . Exosomes are small vesicles (40-100 nm in diameter) released into extracellular space and play a role in cell-cell communication 8 . It has been shown that exosomes of non-malignant cells, such as MSCs, play a crucial role in tumor progression and increased tumor cell migration [9][10][11] . MSCs on the other hand can be transformed by exosomes of breast cancer cells in tumor-like cells, with the ability of tumor growth in vivo 6 . One key characteristic of exosomes is that their origin lies within the endosome. The early endosome matures into a multivesicular endosome (MVE, also called multivesicular body) and accumulates hundreds of intra-luminal vesicles due to inward budding of the endosomal membrane 12 . The Rab GTPase protein family and Rab effector molecules, specifically Rab27a (Rab effector molecule regulating exocytosis of exosomes), play a central role in promoting fusion of MVE to the cell membrane and the release of exosomes into the extracellular space 13 .
Cytokines, the soluble mediators of the acellular fraction of the TME, contribute to tumor progression by promoting angiogenesis and amplifying inflammation 14,15 . The dimeric nuclear transcription factor NF-κB is well known to transcriptionally upregulate several pro-tumorigenic mediators 16 , including cytokines. Under basal conditions, the NF-κB dimer is localized in cytoplasm due to its binding to an inhibitory subunit called IκB. The IKK signalosome induces phosphorylation of IκB, resulting in its dissociation and degradation in cytoplasm. The free NF-κB then translocates to the nucleus 17 . Because of their crucial role in the transcriptional regulation of several pro-tumorigenic and proinflammatory mediators, both IKK and NF-κB have been extensively investigated as potential targets in BC growth and metastasis, but with discouraging results. The adaptor molecule TRAF3IP2 (TRAF3 Interacting Protein 2) is an upstream regulator of NF-κB, AP-1 and stress-activated kinases, and its sustained activation contributes to tumor progression. The activation of TRAF3IP2 is mainly associated with IL-17 signaling. IL-17 mediates cellular immune responses and a dominant "signature" cytokine of TH-17 cells, which upregulates cytokines, neutrophil-mobilizing chemokines, and tissue-degrading matrix metalloproteases 18 .
Until today, it is not clear whether the insoluble or soluble fraction of the TME has the bigger impact on tumor development and progression. To further investigate this question, we focused on Rab27a as a representative of the insoluble (exosomes) fraction and TRAF3IP2 as the representative of the soluble (proinflammatory mediators) fraction of the TME. Since Rab27a and TRAF3IP2 contribute to exosome release and inflammation, respectively, we hypothesized that targeting Rab27a or TRAF3IP2 will reduce inflammation, tumor formation, growth and metastasis of BC in a preclinical breast xenograft model.  38 G > A; correlates to protein sequence p.G13D). The cells were pathogen-free (negative for HIV, HepB, HPV, EBV, and CMV by PCR). They have epithelial-like morphology and appear as spindle shaped cells. During the course of the experiments, we routinely verified their morphology under phase contrast microscope. The cells were able to grow on agarose (anchorage-independence), an indication of transformation and tumorigenicity, and display a relatively high colony forming efficiency. We monitored their tumorigenic potential every 6 months by intramammary injection (5 × 10 5 cells in Matrigel). The cells were cultured in MEM-Alpha Growth Medium (Cat#15-012-CV, CellGro, Manassas, VA) containing 10% FBS (Cat#511550, Atlanta Biologicals, Lawrenceville, GA), 1% Penicillin/Streptomycin (Cat#30-002-CI, Cellgro) and 1% L-Glutamine (Cat#25-005-CI, Cellgro). The culture medium was free of endotoxin as analyzed by Limulus Amoebocyte Lysate Assay (Pierce LAL Chromogenic Endotoxin Quantitation Kit, Catalog# 88282, Thermo Fisher Scientific, Waltham, MA).
Human adipose tissue-derived mesenchymal stem cells (MSCs). Following the guidelines of International Conference on Harmonization (ICH) E-6 Good Clinical Practice and approval by the Tulane University Institutional Review Board (IRB#140571) and informed written consent, adipose tissue specimens were collected from healthy subjects undergoing cosmetic surgery. The MSCs were isolated as previously described 19 . As recommended by the International Society of Cellular Therapy (ISCT), MSCs were characterized for the surface expression of CD4, CD11b, CD34, CD45, CD44, CD73, CD90, CD105, and HLA-DR by flow cytometry using a Beckman-Coulter Epics FC500 [19][20][21] . The MSCs were negative for CD4, CD11b, CD34, CD45 and HLA-DR, but positive for CD44, CD73, CD90, and CD105. The multilineage potential of MSCs was examined by adipogenic, osteogenic, and chondrogenic differentiation assays according to established methods 22,23 .      Statistics. All data relating to study specific was summarized using descriptive statistics such as mean, standard deviation and standard error. Estimates of mean difference and their 95% confidence intervals were calculated. The analysis of variance method was used to compare the mean differences. Where meaningful, the results were presented graphically. The study hypotheses were tested at 5% level of significance throughout the analysis. (*P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001).

Results
In vitro studies. MDA-MB231 cells express high levels of Rab27a and TRAF3IP2. The expression levels of both Rab27a and TRAF3IP2 were significantly higher in MDA-MB231 cells compared to 184A1, a normal BC cell line, and naïve MSCs (Fig. 1A). Interestingly, when cocultured with MDA-MB231 cells, the expression levels of Rab27a and TRAF3IP2 were markedly elevated in both 184A1 cells and MSCs (Fig. 1A), strongly suggesting that soluble mediators from MDA-MB231 cells affect gene expression in non-malignant epithelial and stromal cell populations in a paracrine manner.
Silencing TRAF3IP2 differentially regulates cytokine expression. In comparison to Rab27a, which plays a critical role in exosome release, TRAF3IP2 contributes to inflammation by regulating multiple proinflammatory and pro-tumorigenic mediators. Transcriptomic analysis of MDA KDTRAF3IP2 cells by RT² Profiler PCR Array identified a cluster of 12 pro-inflammatory genes that was differentially expressed (>2-fold) in MDA KDTRAF3IP2 cells compared to MDA w cells (Fig. 1F). For example, the expression of IL11 (Interleukin 11), a gene involved in promoting migration and invasion in BC cells, is reduced in MDA KDTRAF3IP2 cells 31 . The expression of BMP2 (Bone morphogenetic protein 2) and BMP3 were decreased, while BMP4 expression was increased. In addition, silencing TRAF3IP2 differentially regulated the expression of TGFβ (Transforming Growth Factor β) superfamily members TGFβ2 and TGFβ3; while the expression of TGFβ2 was increased, expression of  Table 1).
In addition, electron microscopy showed significant morphological changes in both MDA KDRab27a and MDA KDTRAF3IP2 cells (Fig. 2B); while MDA KDRab27a showed porous cell surfaces (Fig. 2BII), MDA KDTRAF3IP2 cells exhibited rough and shrunken surfaces (Fig. 2BIII) compared to MDA w (Fig. 2BI). However, the cell size was not markedly altered in the silenced cells (Fig. 2BI-III).
In vivo studies in a breast xenograft model. The overview of the in vivo experiments is displayed in Fig. 3. KDRab27a or MDA KDTRAF3IP2 results in reduced tumor growth. Nude mice injected with MDA w cells into 4 th left mammary gland developed large tumors in the primary location (average tumor weight: 4.2 g) at 8 weeks post-injection and displayed major metastases in abdomen, liver, bone, brain and spleen (Figs. 4AI-3AVI, Fig. 4F). However, no visible tumors were detected in animals injected with MDA KDRab27a or MDA KDTRAF3IP2 cells at a similar time period (Fig. 4B,C). However, a very small tumor was detected after 15 Scientific RepoRtS | (2020) 10:8834 | https://doi.org/10.1038/s41598-020-64781-z www.nature.com/scientificreports www.nature.com/scientificreports/ weeks in 6 out of 6 animals injected with MDA KDRab27a cells, but only 3 out of 6 mice injected with MDA KDTRAF3IP2 cells developed a small tumor. We then euthanized a subset of animals within each group at 15 weeks to analyze the tumors. The tumors exhibited limited growth in both MDA KDRab27a  At necropsy, while MDA w injected animals showed massive metastases in various organs (at week 8; Fig. 4AIII-AVI), there was no detectable metastasis in MDA KDRab27a injected animals at 30 weeks (Fig. 4BIII-BV) or MDA KDTRAF3IP2 injected animals at 52 weeks post-injection (Fig. 4CIII-CV). Both tumor weight (Fig. 4DI) and tumor volume (Fig. 4DII) were significantly suppressed when Rab27a or TRAF3IP2 was silenced. However, qPCR for human specific sequence confirmed the presence of metastatic cells (micro-metastasis) in major organs of MDA w (at week 8 post-injection) and MDA KDRab27a injected animals (at week 30 post-injection), but not in MDA KDTRAF3IP2 injected animals after 52 weeks (Fig. 4F).

Experimental series I: Engraftment of MDA
Histological analysis showed decreased expression of CytokeratinAE1/AE3 in MDA KDTRAF3IP2 compared to MDA w . However, the MDA KDRab27a tumors showed higher expression of CytokeratinAE1/AE3 compared to MDA w and MDA KDTRAF3IP2 . The expression of Caspase-3 is slightly increased in MDA KDRab27a and MDA KDTRAF3IP2 tumors (Fig. 4E).  Fig. 5BIII-IV]. The weight of both tumors was comparable after 15 weeks, however, compared to the MDA KDRab27a injected side, there was a significantly smaller tumor on the naïve MSC injected side after 30 weeks (Fig. 5BIII). No signs of macroscopic metastasis were found in major organs such as liver, kidneys or lungs (Fig. 5BV-VII).

Experimental series II: Silencing Rab27a or TRAF3IP2 affects communication between BC and stromal cells.
In the third group of animals injected with naïve MSCs into 4th mammary gland and MDA KDTRAF3IP2 cells into the contralateral mammary gland, no tumor growth was detectable at the naïve MSC injected site after 15 (Fig. 5CI) and 52 weeks post-tumor induction (Fig. 5CIII). However, after 15 weeks a small tumor was detected in 2 out of 6 animals injected with MDA KDTRAF3IP2 [~0.002 g (P < 0.001 vs. MDA w ), Supplementary Table 3; Fig. 5CII]. None of the mice showed detectable tumor growth (Fig. 4CIV) after 52 weeks. Also, at necropsy, no metastasis was detected in major organs (Fig. 5CV-VII).
In summary, tumors formed by MDA w after 8 weeks were significantly heavier in comparison to MDA KDRab27a , MDA KDTRAF3IP2 or any tumor-like masses produced by MSCs at any time point. MDA KDRab27a after 30 weeks was significantly heavier than any other tumor or tumor-like mass except MDA w . No tumor growth was seen in MDA KDTRAF3IP2 after 52 weeks as well as MSC coninjected with MDA KDTRAF3IP2 after 15 or 52 weeks. The detailed statistical analysis is displayed in Supplementary Table 3. The data indicate that regardless of the combination of injections without or with naïve MSCs, the tumorigenesis of MDA KDRab27a and MDA KDTRAF3IP2 cells are significantly reduced (Fig. 5D).
Experimental series III: Regression of pre-existing breast tumors by lentiviral TRAF3IP2 shRNA. We next determined whether treating pre-existing tumors with lentiviral TRAF3IP2 shRNA regresses their size. The results in Fig. 6A,B show a marked reduction in tumor size after 8 weeks in TRAF3IP2 KD shRNA-LV treated mice versus scrambled shRNA-LV (Fig. 6A,B). While tumors grew continuously in scrambled shRNA-LV treated animals, the tumor growth was reduced in TRAF3IP2 KD shRNA-LV treated mice. At necropsy, the TRAF3IP2 KD shRNA-LV treated mice showed a small residual tumor (Fig. 6C), which upon analysis revealed a marked reduction in IL-8, CytokeratinAE1/AE3 and Ki67 expression (Fig. 6D). A detailed summary of the results is displayed in Fig. 7.

Discussion
The results show that (1) silencing Rab27a or TRAF3IP2 in the triple negative breast cancer (BC) cell line MDA-MB231 results in the formation of a significantly smaller tumor in a breast xenograft model and significantly extends survival. Moreover, (2) while targeting Rab27a significantly reduces metastasis, targeting TRAF3IP2 prevents it. (3) Silencing Rab27a and TRAF3IP2 each inhibit interactions between breast cancer cells and naïve MSCs, resulting in reduced tumor growth in MSC-injected contralateral breasts. (4) More importantly, treatment of pre-existing tumors formed by the wildtype malignant BC cells with lentiviral TRAF3IP2 shRNA (2020) 10:8834 | https://doi.org/10.1038/s41598-020-64781-z www.nature.com/scientificreports www.nature.com/scientificreports/ not only regressed tumor size, but also prevented metastasis. Together, these novel data suggest that TRAF3IP2 is a potent pro-tumorigenic mediator, and thus a potential target in breast cancer.
The role of Rab27a and TRAF3IP2 in many tumors, including breast cancer, is still unclear. Silencing of Rab27a has been shown to suppress gastric, pancreas and lung tumors [32][33][34] . In glioblastoma, we reported that silencing TRAF3IP2 is linked to reduced tumor growth and metastasis 35 . On the other hand, TRAF3IP2 overexpression in keratinocytes is shown to induce proliferation and tumor growth 36 . Interestingly in the present study, silencing Rab27a or TRAF3IP2 in the breast cancer cell line MDA-MB231 differentially regulated the expression of genes involved in tumor development, growth and metastasis ( Fig. 2A). CDH2 (N-cadherin) is a calcium-dependent adhesion molecule, and its increased expression suppresses pancreatic tumor growth 37 . The expression of SERPINB5 is linked to high tumor grade, nodal metastasis and perineural invasion in invasive ductal carcinomas of the breast 38 . Our results show that targeting Rab27a and TRAF3IP2 each decreased SERPINB5 expression, but enhanced that of Cadherin 2, in both MDA KDRab27a and MDA KDTRAF3IP2 cells. The expression of MCAM/CD146, a suppressor of breast cancer cell invasion 39 , is higher in MDA KDTRAF3IP2 compared to MDA KDRab27a and MDA w cells. The expression of CXCL12, which is involved in breast cancer development and progression 40 , is also inhibited in both MDA KDRab27a and MDA KDTRAF3IP2 cells. Furthermore, the proliferation of MDA KDTRAF3IP2 cells is significantly reduced, as evidenced by a significant reduction in the expression of TGFβ1 and MAP2K5, mediators involved in cell proliferation, metastasis, and death 41 .
Persistent expression and activation of MMPs play a role in inflammation and invasion. The expression of the endogenous MMP inhibitors TIMP1 and TIMP2, which are involved in proliferation and inhibition of apoptosis 42 , is increased in MDA KDRab27a , but suppressed in MDA KDTRAF3IP2 cells. MMP7 promotes tumor growth in many different cancers, including breast cancer 43 . Silencing TRAF3IP2 downregulated its expression in MDA-MB231 cells compared to MDA w and MDA KDRab27a cells. A correlative downregulation was found in the transcription factors, ANGPTL4, which is critical for invasion of MDA-MB231 cells 44 , and ALDH3A1, which promotes a multi-modality resistance in human breast adenocarcinomas 45 . ANGPT1 (Angiopoetin 1), which is downregulated in MDA KDTRAF3IP2 , is a key pro-angiogenic factor, like VEGFA, that enhances endothelial cell migration and the formation of capillary-like structures 46 . This broad alteration in gene expression suggests the upstream involvement of Rab27a and TRAF3IP2 within the cell. In summary, targeting TRAF3IP2 appears to have a bigger impact on downregulation of genes involved in inflammation and pro-tumorigenic pathways, compared to silencing Rab27a.
Interestingly, silencing Rab27a and TRAF3IP2 each induced morphological changes in MDA-MB231 cells. Electron microscopic analysis indicated that silencing Rab27a results in the formation of a porous cell surface in www.nature.com/scientificreports www.nature.com/scientificreports/ MDA KDRab27a cells, while silencing TRAF3IP2 resulted in a rough surface, suggesting that Rab27a and TRAF3IP2 might be involved in remodeling or destabilization of tumor cell membrane. In addition, silencing Rab27a or TRAF3IP2 increased doubling time compared to the control MDA w cells (Fig. 1B), possibly contributing to smaller tumors. However, silencing Rab27a or TRAF3IP2 had no significant effects on non-malignant cells like MSCs, suggesting that targeting Rab27aor TRAF3IP2 affect only the tumor cells but not the normal mesenchyml cells. Furthermore, the breast cancer cells can induce the expression of Rab27a or TRAF3IP2 in non-malignant breast epithelial 184A1 cells and naïve MSCs via a paracrine mechanism (Fig. 1A), and demonstrates the impact of breast cancer cells on surrounding non-malignant stroma.
TRAF3IP2 is an upstream regulator of NF-κB-dependent inflammatory signaling 47 . To understand the role of TRAF3IP2 on inflammatory responses, we silenced its expression in MDA-MB231 cells. The data show that TRAF3IP2 knockdown markedly suppressed IL11 expression. Since IL11 is involved in migration and invasion in MDA-MB231 cells (Fig. 1F) 31 , our results suggest that it could be one of the key factors involved in tumor growth seen in our in vivo data. Additionally, the expressions of several genes of the TGFβ superfamily, such as BMP2, BMP3, BMP4, TGFβ2, TGFβ3, TNFRSF11B as well as INHα, are differentially regulated by TRAF3IP2 silencing. The TGFβ superfamily has been shown exert dual effects; it is shown to either suppress or promote tumor growth 48 . Our data show downregulation of BMP2 and BMP3. Expression of BMP2 is closely related to invasion of breast cancer cells by cytoskeletal reorganization and decreased adhesion 49 . Expression of BMP3 has been shown to positively correlate with MSC proliferation 50 . The function of BMP4, whose expression is higher in MDA KDTRAF3IP2 , is not fully understood. While a number of studies have shown its pro-growth effects in other cancers 51 , its increased expression has shown to inhibit breast cancer growth 52 . The expression of TGFβ3, which is lower in TRAF3IP2-silenced MBA-MB231 cells has been linked to enhanced metastasis in breast carcinoma 46 . The expression of TNFRSF11B and INHα are enhanced in both MDA KDRab27a and MDA KDTRAF3IP2 cells. Downregulation of TNFRSF11B (TNFR1) correlates with tumorgenicity and poorer prognosis in patients with breast cancer 53 . INHα has been suggested to have a tumor suppressor effect by suppression of cell growth and being associated with apoptosis 1 . PDGFA is related to tumor progression in breast cancer 54 , and its expression is reduced in MDA KDTRAF3IP2 . However, a small increase in the expression of FIGF (VEGFD) and CSF1 (Colony stimulating factor 1) was observed in MDA KDTRAF3IP2 cells. FIGF, a VEGF family member, is, if expressed, involved in lymphangiogenesis in breast cancer 55 but also has been shown to be downregulated in a metastic breast cancer cells line 56 . CSF1 is involved in breast cancer progression 57 . The LTα gene, associated with tumor progression and angiogenesis in cutaneous lymphomas, shows a significantly reduction in its expression in MDA KDTRAF3IP2 58 .
In accordance to our previous experiment here we display the profound impact of TRAF3IP2 within the cells. Many genes involved in cancer development and progression are altered in a tumor suppressive matter in MDA KDTRAF3IP2 .
Supporting our in vitro data, the in vivo data demonstrate that silencing Rab27a delays tumor development but not micrometastasis. However, targeting Rab27a significantly extended survival by more than 3-fold compared to the wild type. on the other hand, targeting TRAF3IP2 suppressed both tumor growth as well as macro-and micrometastasis. Furthermore, targeting Rab27a or TRAF3IP2 each reduced the potential of the tumor cells with the surrounding stroma, indicated by reduced growth of MSCs coinjected with MDA KDRab27a . Coinjection of MDA KDTRAF3IP2 and naïve MSCs resulted no tumor growth on MSC injected side, indicating lack of interaction between the MDA KDTRAF3IP2 and naïve MSCs. More importantly, targeting TRAF3IP2 not only regressed pre-formed tumors (Fig. 6), but also prevented metastasis. One potential explanation of reduced effect of Rab27a could be that silencing Rab27 blunts, but does not abrogate exosome release 8,59 . In fact, histologic analysis revealed a higher expression of Cytokeratin AE1/AE3 in MDA KDRab27a in comparison with MDA w or MDA KDTRAF3IP2 . This is in accordance with previous findings where exosomes of MDA-MB231 cells have been shown to contain cytokeratin 9 and that the reduction of Rab27a results in the intracellular accumulation of enlarged MVEs 59,60 . As indicated by our data, several genes altered by silencing TRAF3IP2 are individually the subject of clinical trials. For example, BMP4 is currently being tested as potential treatment for glioblastoma 61 , FIGF in end stage coronary heart disease 62 , and a MAP kinase inhibitor in non-small cell lung cancer 63 .
A limitation of the present study is that we used a single breast cancer cell line to investigate potential roles of silencing Rab27a and TRAF3IP2 on tumor growth. Our future studies will involve the use of patient-derived xenotransplants to further validate these fundamental first in vivo and in vitro results.

Conclusions
Both Rab27a and TRAF3IP2 play a causal role in breast cancer growth and metastasis. We showed that targeting Rab27a decreases exosome release and delays progression of tumor growth, but fails to affect micrometastasis. Targeting TRAF3IP2, as a representative of the soluble fraction of the TME, however, suppresses tumor growth as well as macro-and micro-metastasis. More importantly, treatment with lentiviral TRAF3IP2 shRNA regresses pre-formed tumors and prevents metastasis. These results indicate that TRAF3IP2 is a more potent inhibitor of tumorigenesis, and thus a novel therapeutic target in breast cancer.