APRIL and BAFF increase breast cancer cell stemness

Recent advances in cancer immunology revealed immune-related properties of cancer cells as novel promising therapeutic targets. The two TNF superfamily members, APRIL and BAFF even though were primarily studied in lymphocyte maturation, they have also been associated with tumor growth and aggressiveness in a number of solid tumors, including breast cancer. In the present work we studied the effect of APRIL and BAFF on epithelial to mesenchymal transition and migration of breast cancer cells, and their action on the sub-population of cancer stem cells identified by autofluorescence and ALDH activity. Their action on an number of pluripotency genes was examined and breast cancer stem cell ability to form mammospheres was also utilized. The receptor and the signaling pathway involved as well as the role of steroid hormones in their action were also investigated. Our findings show that both APRIL and BAFF increase epithelial to mesenchymal transition and migratory capacity of breast cancer cells, as well as cancer stem cell numbers, by inducing pluripotency genes such as KLF4 and NANOG. These effects are mediated by their common receptor BCMA and the JNK signaling pathway. Interestingly, androgens enhance APRIL transcription and subsequently its pluripotency effect. In conclusion, our data support the significant role of APRIL and BAFF in breast cancer disease progression and provide evidence for a new possible mechanism of therapy resistance, that could be particularly relevant in aromatase inhibitors-treated patients, were local androgen is increased.


Introduction
The breast cancer 5-year survival has significantly increased, edue to early detection, and the introduction of therapies, such as specific estrogen receptor modulators, aromatase inhibitors, specific chemotherapeutic agents and more comprehensive recent precision medicine strategies. However, long (10-years) survival decreases dramatically to 78% (http://www.cancerresearchuk.org/health-professional/cancerstatistics/statistics-by-cancer-type/breast-cancer/survival#heading-Zero), suggesting that, in spite of our progress in breast cancer cell biology, other elements are currently uncovered or understood.
Cancer stem cells are key players in tumor development, progression and metastasis as well as resistance to therapy. This cell population displays potential to proliferate infinetely, to initiate and/or reform a tumor and an enhanced resistance to therapy 1 .
Adult stem cells have been identified in the normal mammary gland development and function see 2, for a recent review , but also exert specific dynamics in in various mammary pathologies 3 . Breast stem cells interact in their niche with surrounding stromal and epithelial cells, through specific molecular crosstalk pathways 4,5 . These cancerimmune cell interactions are mostly orchestrated by tumor microenvironment 6,7 .
However, established immune-related therapies target immune cells (resident or infiltrating the tumor stroma) 8 , leading to an immune checkpoint blockade 9 , while the notion of immune-related properties of the cancer cell per se and its possible regulation as a possible therapeutic target is less well defined 10,11 .
A number of immune-related molecules are involved in the above mentioned interactions and being targeted in tumor immunotherapy approcahes. Among them, TNF superfamily members (including TNF, FAS and TRAIL and their receptors) 8 , have been actively investigated and targeted in a number of malignancies. The TNF superfamily includes 19 different ligands and 29 receptors, controls cell survival and differentiation and plays an important role in the growth, organization, and homeostasis of different tissues, by modulating major signaling pathways 12 . Our group has focused on two members of this superfamily, namely APRIL (A PRoliferation Inducing Ligand) and BAFF (B-cell Activating Factor of the TNF Family). These two ligands, act via two common receptors, B-Cell Maturation Antigen (BCMA) and Transmembrane Activator and CAML Interactor (TACI), while additionally, BAFF-Receptor (BAFF-R) is a specific receptor for BAFF. They had initially been reported to have a central role in lymphocyte maturation; however, they have been also identified as significant players in several other conditions including neoplasia 13 . BAFF and APRIL have been detected in solid tumors 14 , activating significant kinase signaling pathways, such as p38, JNK, NFκB and inducing, in the majority of cases, cell survival and growth.
Previously, we have shown BAFF and/or APRIL presence in many normal and cancer solid tumors, including breast cancer [15][16][17] , in which BAFF is constantly expressed in tumors, while APRIL is related to tumor grade 17 . Recently, higher APRIL expression was shown in human triple negative carcinomas and APRIL was reported to induce cell proliferation both in vitro and in vivo, suggesting an association of APRIL signaling pathways with tumor aggressiveness 18 . In the present work, we investigated the possible relationship of APRIL and BAFF with steroids, the important players in breast cancer, and the possible role of APRIL and BAFF in inducing breast cancer stemness.

BAFF and APRIL expression in breast cancer cell lines -Regulation by extranuclear-acting androgen
We have assayed the presence of these TNFSF members in an ERα-positive (T47D) and a triple negative breast cancer cell line (MDA-MB-231 cells, named thereafter MDA cells). BAFF, APRIL, BCMA, and BAFFR mRNA was present in both cell lines, while TACI transcript was constantly absent ( Figure 1A).
To explore whether there is a relation between steroids and APRIL and BAFF production we have, at first interrogated our previously reported transcriptome data on the effect of androgen (GSE18146) and estrogen (GSE32666 and GSE32668) on T47D and MDA cells. As depicted in Figure 1B a significant increase of APRIL transcription was observed in both T47D and MDA cells after 3h-incubation with testosterone.
Interestingly, this increase was observed with the membrane-only acting androgen (Testosterone-BSA) suggesting an extranuclear androgen action. In contrast, estrogen had a minimal effect on APRIL transcription in both cell lines. Conversely, no significant modification of APRIL-BAFF receptors was found (not shown).
Exploring the promoter region of APRIL, with the rVista V 2.0 web resource (https://rvista.dcode.org/), we have identified four putative androgen response elements in the promoter of the APRIL gene (Supplementary Figure 1), an element suggesting the implication of the classical activated AR. In view of this and the transcriptional activation of APRIL and BAFF by membrane-acting steroids, we explored the possible modification of APRIL transcription after cell incubation with testosterone-BSA.
Incubation of T47D cells with testosterone-BSA resulted in a significant increase of APRIL gene product at 3 and 6 hours ( Figure 1C). Interestingly, this effect was inhibited by the antiandrogen cyproterone acetate (which was ineffective per se).

BAFF and APRIL increase cell migration, epithelial-mesenchymal transition (EMT) and stemness in epithelial breast cancer cells
APRIL has been reported to be associated with increased breast tumor growth and metastasis 18 . In our cell lines both APRIL and BAFF enhance cell migration (the effect is more prominent after a 48h-incubation, Figure 2A). Additionally, the EMT cell status of breast cancer cells was enhanced after treatment with either APRIL or BAFF (100ng/ml) (Figure 2 B and C). This was accompanied by actin cytoskeleton rearrangements, characteristic of migrating cells (Supplementary Figure 2). It is to note that APRIL and BAFF induce similar modification of EMT markers and migratory activity, suggesting an interaction occurring through the same receptor or receptors.
In addition to EMT, another factor necessary for metastatic potential and disease progression is a combination of genetic alterations and epigenetic events that recapitulate normal developmental processes including stem cell self-renewal 19 and the acquisition of ''stem cell'' properties, through a ''dedifferentiation'' process. T47D and MDA cells, treated with APRIL or BAFF (100ng/ml) exhibited an increased ability to form mammospheres. More specifically, the number of primary mammospheres increased by 40% after 9 days of incubation ( Figure 3A). The effect of these two agents is more prominent in ERα-positive T47D than in triple negative MDA cells.
Interestingly, testosterone-BSA per se increases mammosphere formation in T47D cells. When cells were incubated with testosterone-BSA and APRIL, an additive effect was observed ( Figure 3B).
We have particularly focused in the T47D cell line, in which mammosphere formation was more prominent. After dispersion of primary mammospheres and culture of cells under the same conditions (formation of secondary mammospheres) the effect of both cytokines was further enhanced, with APRIL being slightly more potent than BAFF ( Figure 3C). The effect of BAFF and APRIL on the induction of stemness was further verified by the intrinsic autofluorescence of epithelial cancer stem cells, due to riboflavin accumulation in membrane-bounded cytoplasmic structures bearing ATPdependent ABCG2 transporters 20 . Incubation of breast cancer cells with APRIL or BAFF (100 ng/ml) for at least four days induced a significant increase in autofluorescence, from ~2% in control (untreated) cells to ~5% in APRIL or BAFFtreated cells Figure 3D). This increase in autofluorescence correlates and is of similar amplitude with the increase of ALDH1A1 activity, assayed by flow cytometry and the ALDEFLUOR® kit (see Supplementary Figure 3 for an example and Figure 3E).
Dispersed T47D cell mammospheres were assayed for a stem cell signature and they were found to have significantly more stem cells compared to cells from conventional monolayer culture (7.2% vs 2.1%). BAFF or APRIL treatment further increased this autofluorescent stem cell signature (17.9 vs 4.5 for APRIL and 11.2 vs 3.6 for BAFF respectively) suggesting that these cytokines have the capacity to induce pluripotency in T47D cells ( Figure 3F).
To further verify the emergence of stemness after 4 days incubation with APRIL or BAFF (100 ng/ml), the transcription of pluripotency-related factors (SOX2, c-MYC, KLF4, NANOG and ALDH1A1) was investigated in T47D cells. The expression of the above genes was examined both at the mRNA (by Real Time PCR, Figure 4A) and protein level (by immunocytochemistry, Figure 4B). As shown, APRIL and BAFF mainly increase the expression of ALDH1A1, KLF4, and NANOG.

APRIL and BAFF modulate stemness markers expression through BCMA and JNK signaling
Results presented so far suggest that APRIL and BAFF promote in a similar way the emergence of pluripotency in breast cancer cells. Taking into consideration that T47D cells express BCMA, on which both cytokines can bind, and BAFFR, an exclusive BAFF receptor (see Figure 1) 12 , we have assumed that this effect might be mediated by BCMA. To confirm this hypothesis, we transfected T47D cells with a specific anti-BCMA siRNA. Transfected cells were then incubated with BAFF or APRIL and pluripotent/stem cells were assayed by their autofluorescence. As depicted in Figure   5A, cancer stem (autofluorescent) cells were significantly decreased, suggesting the mediation of this effect through BCMA. Here, we have explored the two pathways mediating post-receptor effects of BAFF and APRIL [TRAF signaling and NFκB activation (31), and BCMA-specific signaling, mediated by JNK (21)]: As shown in Figure 5B, T47D cells, transfected with a specific plasmid, carrying NFκB response elements in front of the firefly luciferase gene, do not show any activation after a 24h BAFF or APRIL incubation; contrariwise, the JNK specific inhibitor, SP600125 or shRNAs against JNK1 and JNK2, significantly blocked the effects of APRIL and BAFF on ALDH1A1 and KLF4 ( Figure 5C and D), indicating that, in breast cancer cells, APRIL and BAFF binding to BCMA signal towards pluripotency is mediated by JNK1.

APRIL in aromatase-inhibitor treated patients' samples
As shown above, androgen enhances APRIL production in breast cancer cells regardless of estrogen receptor status. We hypothesized that increased local androgen (as could occur in patients treated with aromatase inhibitors that inhibit conversion of androgen to estrogen), could enhance APRIL production ultimately increase cancer pluripotency. We extracted APRIL data from the GEO-deposited GDS3116 study 21,22 .
In this series, transcriptome data of a series of 53 letrozole (an aromatase inhibitor) treated patients are included, prior and 3-months following therapy. 37/53 (69.8%) patients responded to the therapy (as estimated by changes in estrogen responsive genes and by ultrasound detected changes in tumor volume) while 16/53 (30.1%) were nonresponders. In responders, while in non-responders no change is observed ( Figure 6A and B).
Furthermore, cBioPortal® for Cancer Discovery 23-25 data analysis from a large breast cancer study (2509 samples, MetaBRIC study) 26,27 show that BAFF, APRIL, and BCMA have a significant tendency to co-express with androgen receptor AR (p<0.001 for BAFF and APRIL, p=0.012 for BCMA) further confirming the association with androgen ( Figure 6 C and D and Supplementary Figure 4). Significant correlations were also found for the stemness markers we have tested in the breast cancer cell lines.

Discussion
Our understanding of breast cancer physiology has expanded in recent years. In addition to the molecular analysis and identification of different breast cancer subtypes 28,29 , recent data reveal a biological evolution of breast cancer cells from the primary to the metastatic site of the disease, influencing the disease progression 30 . This evolution has been recently suggested as a possible novel therapeutic target in breast cancer 31 .
A major element of breast cancer molecular and phenotypic evolution is the differentiation of stem cells from their niches 3,5,30 and the cross-talk with the epithelial, stromal and immune components of the niche or surrounding microenvironment 2,4,5,30-32 . However, another element of this heterogeneity has been recently identified and is related to the effects of administered treatment to the transformation of cancer epithelial cells 33 , including the emergence of cell pluripotency. Recently, we have reported that short term exposure of breast cancer cells to tamoxifen may act as such an epigenetic stem-inducing stimulus 34 . Here, we provide evidence about such an induction of pluripotency by the epithelial APRIL and BAFF, through BCMA, a mechanism which is related to extranuclear androgen effects.
The various markers currently used for breast cancer stem cells (BCSCs) 35,36 identify slightly different subpopulations and it is, therefore, necessary to utilize more than one specific marker and/or to test for unique cancer stem cell properties such as tumor initiation (sphere formation or xenografted tumor development, etc) 37 . Therefore, in the present work, to investigate the effect of APRIL and BAFF, we utilized multiple approaches to detect BCSCs. Our results show that both BAFF and APRIL increase the percentage of breast cancer stem cells. Indeed, when breast cancer cells were treated for 4 days with either BAFF or APRIL a two to three-fold increase in the BCSC population was observed. This finding suggests that the epigenetic changes required for the induction of this phenotype need a prolonged exposure to these agents.
In fact it was found that breast cancer cells, treated with APRIL or BAFF display a more aggressive phenotype as they exhibit: (1) Increased migration and modifications of the actin cytoskeleton; (2) increased expression of mesenchymal markers and signs of the acquisition of a mesenchymal phenotype that is linked to stem-like cells 38,39 and is associated with metastasis and a higher migratory capacity 40,41 ; (3) increased stem cell population, as assayed by BCSC autofluorescence and ALDH1A1 activity, and (4) increased mammosphere formation, with a significant increase in BCSC population.
These results are in accordance with previously reported findings on APRIL promoting breast cancer lung metastasis 18 .
An important finding supporting the observed BCSC enrichment after treatment with APRIL or BAFF is the significant acquisition of pluripotency markers KLF4, NANOG, c-MYC, and ALDH1A1 [42][43][44][45] . These markers when properly modified, may act as transcription factors that promote a stem cell phenotype, maintain pluripotency and prevent differentiation. The fact that in the studied cell lines both BAFF and APRIL induced similar effects on these pluripotency markers at the mRNA and the protein level suggests that their action should be mediated through BCMA, a cognate receptor for both ligands 12 . This was further supported by the inhibitory action of the specific anti-BCMA siRNA on their effects and the fact that the BCMA specific signaling involving JNK activation was found to be involved. Indeed, their effect on ALDH1A1 and KLF4 was significantly reduced in the presence of the JNK inhibitor, SP600125 or shRNAs against JNK1 and JNK2 ( Figure 5C and D). Additionally, no NFκB activity was detected. Therefore, it seems that in breast cancer cells, APRIL-BCMA signal towards pluripotency is mediated by JNK, a signaling pathway also reported previously by our group for the effect of APRIL in hepatocellular carcinoma cell lines 46 . The complete inefficiency of BAFF to modify NFκB, further suggests that, in our settings, BAFFR signaling, might be ineffective in this cell line. Moreover, we report that this APRIL-BCMA regulated effect relays on local androgen. By analyzing our previously reported transcriptome data on T47D and MDA cells 16,47,48 , we confirmed that APRIL transcription was significantly increased after a 3h-incubation with testosterone while estradiol had only a minimal effect. Interestingly, a significantly higher change was induced by membrane-only acting testosterone (Testosterone-BSA) suggesting a membrane-initiated action, while the inhibition of the effect by anti-androgen suggests that this effect might be mediated by the classical androgen receptor, activated at the membrane level.
This androgen-induced APRIL production and subsequently, stemness, could possibly have a clinical importance in the case of breast cancer patients that are treated with aromatase inhibitors which can result at increased local androgen concentration. In order to provide a proof of principle of this hypothesis, we extracted APRIL data from a series of 37 letrozole-responders and 16 non-responders before and after therapy (GEO-deposited GDS3116 study) 21,22 . Indeed, APRIL expression was significantly higher in the responder's group (expected to have increased local androgen levels), while it was not changed in the non-responders, verifying our hypothesis on testosterone-control of APRIL in breast cancer.
We have also interrogated one large MetaBRIC breast cancer study 26,27 . APRIL mRNA downregulation was evidenced in 7% of cases, confirming our previous report on protein expression 17 . Moreover, it is evident that BAFF, APRIL, and BCMA tend to co-occur with AR and stemness markers. They are preferentially expressed in claudinlow tumors, that display a highly aggressive behavior.
Our study presents a number of limitations. First, only two breast cancer lines are tested.
Larger studies in more breast cancer cell lines and in vivo models are required to validate our data. However, evidence from a number of clinical studies, presented above, referring to treated patients provide a degree of validation in patients' data. Our omics approaches based on a retrospective analysis of publicly available data also supports our results. Future prospective studies could provide more refined data to the mechanism presented here. The clinical implications could be potentiated, as aromatase inhibition is widely used to treat ER-positive tumors and AR profiling is part of cuttingedge approaches in breast cancer.

Conclusions
Our findings clearly indicate that both BAFF and APRIL have the ability to increase the percentage of breast cancer stem cells through BCMA-JNK mediation, pointing out for the first time that APRIL and BAFF not only modify breast cancer cell proliferation but they can also contribute to the re-formation of the tumor. Additionally, they provide evidence for a new possible mechanism of therapy resistance that involves increased stemness by high APRIL levels as a result of the accumulation of androgen that could occur in aromatase inhibitors treated patients. They further provide evidence that, in order to establish optimal personalized, immune-related therapies in breast cancer patients, one should, in addition to the targeting of the stroma and cancer-infiltrating immune cells 8 , also investigate and target tumor cells, as was recently reported for another member of the TNFRSF, TNFR2 10,11 .

Cell cultures and chemicals
The T47D breast cancer cell line and MDA-MB-231 were purchased from DSMZ (Braunschweig, Germany), and were cultured in RPMI supplemented with 10% fetal bovine serum (FBS), at 37 °C, 5% CO2. All media were purchased from Invitrogen (Carlsbad, USA) and all chemicals from Sigma (St. Louis, MO) unless otherwise stated.

RT-PCR
RT-PCR was performed as described previously 46 . Positive controls were run in parallel with samples included in the study. We used adipose-derived mesenchymal stem cells as positive controls of BAFF and APRIL 15 , isolated human lymphocytes as positive controls for BAFFR and TACI, and HepG2 cells for BCMA 46

Real Time PCR
Total RNA was isolated with the Nucleospin RNA II isolation kit (Macherey-Nagel EURL, Fr). The absence of DNA was verified by PCR for G3PDH. One μg of RNA was subjected to ABI High Capacity cDNA reverse transcription kit (Applied

Mammosphere formation
Mammosphere formation was assayed as previously described 49  Cells were incubated for 7 additional days and secondary mammospheres were counted as described above.

Immunocytochemistry
For the detection of immunoreactive proteins in breast cancer cells, treated or not (control) with APRIL or BAFF (100ng/ml) for 4 days, cells were fixed with 2% paraformaldehyde. Afterwards, they were incubated at room temperature for one hour with 3% BSA in TBS, and then overnight at 4 °C with the primary antibody (c-MYC 1/600, ALDH1A1 1/800, SOX-2 1/100 all from Santa-Cruz, and NANOG 1/100 from eBiosciences). For the detection of antibody binding, the UltraVision LP Detection System: HRP Polymer Quanto (Thermo Scientific, Cheshire, UK) was used, with 3'3-

Diaminobenzidine (DAB) as chromogen. Stained cells were lightly counterstained with
Harris hematoxylin for 10 s, hydrated and mounted in Permount (Fisher Scientific, Fair Lawn, NJ). Internal controls for specificity of immunostaining included replacement of primary antibody with non-specific serum (negative). Slides were evaluated for the presence and the intensity of staining (quantified using ImageJ).

EMT transition
The EMT status of cells was determined by measuring the ratios of vimentin/keratins expression levels, using image analysis after laser scanning microscopy 50

JNK inhibition
T47D cells were cultured in six-well plates and treated with APRIL or BAFF (100ng/ml) for 4 days in the presence of the specific JNK inhibitor, SP600125 (10μΜ).

NFk-B activation assay
Cells were cultured in 24-well plate, and were transfected with 0.2μg/well of pNFκB-