Evaluation of chemotherapy and P2Et extract combination in ex-vivo derived tumor mammospheres from breast cancer patients

The main cause of death by cancer is metastasis rather than local complications of primary tumors. Recent studies suggest that breast cancer stem cells (BCSCs), retains the ability to self-renew and differentiate to repopulate the entire tumor, also, they have been associated with resistance to chemotherapy and tumor recurrence, even after tumor resection. Chemotherapy has been implicated in the induction of resistant phenotypes with highly metastatic potential. Naturally occurring compounds, especially phytochemicals such as P2Et, can target different populations of cancer cells as well as BCSC, favoring the activation of immune response via immunogenic tumor death. Here, we evaluated the presence of BCSC as well as markers related to drug resistance in tumors obtained from 78 patients who had received (or not) chemotherapy before surgery. We evaluated the ex vivo response of patient tumor-derived organoids (or mammospheres) to chemotherapy alone or in combination with P2Et. A xenotransplant model engrafted with MDA-MB-468 was used to evaluate in vivo the activity of P2Et, in this model P2Et delay tumor growth. We show that patients with luminal and TNBC, and those who received neoadjuvant therapy before surgery have a higher frequency of BCSC. Further, the treatment with P2Et in mammospheres and human breast cancer cell lines improve the in vitro tumor death and decrease its viability and proliferation together with the release of immunogenic signals. P2Et could be a good co-adjuvant in antitumor therapy in patients, retarding the tumor growth by enabling the activation of the immune response.


Results
Luminal and triple negative patients have a higher frequency of breast cancer stem cells (BCSC). 78 breast cancer (BC) patients and 7 healthy donors (HD) were evaluated for the frequency of BCSC, CD45 + cells, and multidrug efflux pumps expression (BCRP, Pgp and MRP1). Patients' age was ranged from 30 to 92 years; mean age at diagnosis was 61.3 ± 1.3 years. The number of patients in stage I were (n = 9), stage II (n = 34), stage III (n = 31) and stage IV (n = 3). The estrogen receptor (ER), progesterone receptor (PR), Her2 expression and Ki-67 percentage, were used to classify the samples, as follows: Luminal A (n = 17), Luminal B (n = 42), Triple negative (TN) (n = 14) and Her2 (n = 5). Additionally, 27 patients received neoadjuvant chemotherapy (NAT) before surgery and 51 patients did not receive NAT. Received NAT regimens are shown in Table 1.
The BCSC characterization was performed using primary tumor tissue collected during the resection surgery and processed as shown in Fig. 1a. Single cell suspensions were obtained by mechanical and enzymatic digestion as previously described 25 (Fig. 1a), and tumor-derived organoids (mammospheres) were generated, this system provide higher physiological relevance than 2D-cultures, allowing the propagation of mammary stem and progenitor cells. The cells were analyzed by flow-cytometry excluding lineage negative populations (CD45, CD140b, CD31), staining for putative BCSC markers (CD24, CD44, EPCAM and CD49f) and multidrug efflux pumps (BCRP, Pgp, MRP-1). Additionally, we measured the ALDH enzymatic activity.
Multi-dimensional reduction analysis revealed the expression of putative BCSC markers (Fig. 1b). Due to different acquisition dates it was impossible to generate a common tSNE plot for all the samples; however, the BCSC (CD44 + CD24 − EPCAM + CD49f + ) can be appreciated (Fig. 1b). The BCSC characterization was performed as shown in Supplementary Fig. S1a,b. Further analysis showed high infiltration of CD45 + cells in Luminal and TN patients compared with HD (Fig. 1c). Higher frequency of BCSC ALDH + or Lin -CD44 + CD24 -CD49f. + EPCAM + phenotype was evident in Luminal and TN patients (Fig. 1d,e), together with overexpression of BCRP + and Pgp + but not MRP1 + as compared with HD ( Fig. 1f-h). In accordance with our data, a recent study showed the elevated Scientific Reports | (2020) 10:19639 | https://doi.org/10.1038/s41598-020-76619-9 www.nature.com/scientificreports/ expression of ALDH1 in a cohort of breast cancer patients in TN samples measured by IHC 26 . The expression of ABCG2, ABCB1, ABCC1 and Nanog, Sox2 and Oct4 was evaluated by real time PCR, but differences were observed as compared with HD ( Supplementary Fig. S2).

Neoadjuvant chemotherapy (NAT) promotes enrichment of the BSCS into the tumor tissue. It
has been extensively documented that neoadjuvant therapy impact in the frequency of CSC 27 . We interrogated whether NAT impacted the frequency of BCSC into the tumor tissue of patients who received (or not) treatment before surgery. As expected, NAT before surgery induce a higher frequency of Lin -CD44 + CD24 − CD49f. + EPCAM + (BCSC) cells in comparison with the HD or the group of patients who had not received NAT (Fig. 2a). The highest statistically significant frequency was observed for Luminal B patients (Fig. 2b) compared to the other groups ( Supplementary Fig. S3). On the other hand, we interrogated the presence of ALDH + cells (BCSC) and significant differences were only observed between patients who received NAT before surgery and HD ( Fig. 2c and Supplementary Fig. S3), being more evident in TN samples (Fig. 2d). We interrogated the correlation between BCSC and drug resistance, a positive correlation between BSCS markers (Lin -CD44 + CD24 − CD49f + EPCAM + , r = 0.30 p = 0.02) and BCRP protein (Fig. 3a) or ABCG2 gene expression (Fig. 3b) was observed, regardless of the NAT therapy status. Further, a positive correlation between BCRP1 and ALDH was found in patients with NAT before surgery (Fig. 3c), suggesting that ALDH expression is usually accompanied with other chemo-resistant traits as efflux pumps as previously reported 28 . Likewise, we found a positive correlation between ALDH and ABCC1 gene expression (MRP1 pump) in the NAT group ( Supplementary Fig. S4d), but in non-NAT treated group www.nature.com/scientificreports/ ( Supplementary Fig. S4e,f). Also, a positive correlation between ALDH and BCRP expression in the group of TN patients was observed ( Fig. 3d-f). Together, these results suggested that cells with a highly resistant phenotype could be selected in the tumors of patients after NAT, and these population was particularly high on TN patients, similar results were showed before 29 .
P2Et extract decrease the viability of mammospheres derived from breast cancer patients after in vitro treatment. We have previously shown that P2Et induces immunogenic tumor cell death, with the consequent activation of adaptive immune response 21,22 . Moreover, P2Et inhibits sphere-formation of 4T1 ALDH + CSC, and act synergistically with DX, in vitro and in vivo 23 .
With this on mind, we interrogated the effect of P2Et, DX or the combination of both (P2Et + DX) on mammospheres generated from patient samples in 3D cultures (Fig. 4a). Using this setting, we observed that mammospheres are more sensitive to P2Et treatment compared to the control group (Ethanol, negative control), independently of the molecular subtype; however, a trend to increased sensitivity was observed in tumor cells from TN patients (Fig. 4b). All the treatments induced significant cell death in 3D-culture conditions in comparison with vehicle (Fig. 4c).
Then, a positive correlation between the percentage of dead cells induced by P2Et and the gene expression of ABCG2 was found (Fig. 4d) and not with the other variables studied ( Supplementary Fig. S5). These findings confirm our previous results 23 . Thus, we observed that the human-derived mammospheres are sensitive to P2Et extract, and the effect was independent of the molecular subtype.   Fig. S6), multiresistance pumps (ABCG2, ABCC1 and ABCB1) ( Supplementary Fig. S6) and the BCSC frequency (CD44 + CD24 + CD49 + EPCAM + ) ( Supplementary  Fig. S1c,d) in three human breast cancer cell lines. MCF-7 (ER + ) and MDA-MB-468 (TN) had the highest ALDH activity, 28 and 25%, respectively (Fig. 5a). The cell line MDA-MB-468 had the highest frequency of BCSC cells (Fig. 5a). The results suggested that MDA-MB-468 was the most sensitive to P2Et treatment, with an IC 50 of 136.7 μg/mL compared with 236.1 μg /mL for BT-594 cell line (Fig. 5b). Also, we observed lower cell number (48,72 and 96 h), and decreased proliferation ability of MDA-MB-468 cells upon the treatment with P2Et (Fig. 5c, Supplementary Fig. S7). Further, we interrogate the synergistic effects of P2Et extract with DX, we performed MTT assays over several combinations and a synergistic effect was observed using higher concentrations of P2Et and lower concentrations of DX (Fig. 5d). It has been showed that chemotherapy induces the expression of glutathione S-transferase omega 1 (GSTO1) and further knockdown of GSTO1 expression, abrogates carboplatin induced BCSC enrichment, decreases tumor initiation, metastatic ability, and delays tumor recurrence after chemotherapy 15  b. c.

d.
All samples All samples NAT A A before surger T y r r LA e. f. www.nature.com/scientificreports/  www.nature.com/scientificreports/ could suggest that P2Et and DX induced cell death by different routes, implicating a different management of intracellular ROS, independent of glutathione. Thus, we examined whether P2Et extract induce DAMPs in MDA-MB-468 cells. It was found that P2Et induced apoptosis (phosphatidyl serine externalization) (Fig. 5e) and CRT expression (Fig. 5f), which is significantly higher compared with cells treated with the ETOH (negative control) or DX. ATP release, another feature of immunogenic cell death, was observed upon both treatments (Fig. 5g), similar results were observed previously for DX and P2Et 31 in other models of cancer.

LB TN
P2Et extract delay tumor growth in triple negative human breast cancer. MDA-MB-468 showed the higher percentage of CSC, we decided to interrogate the effects of P2Et in vivo. Two groups of NSG mice were engrafted with MDA-MB-468 and treated twice a week with P2Et (IP: 18.7 mg/Kg) or PBS. (Fig. 6a). P2Et treatment delayed tumor growth compared with control group from 32 to 56 day. However, from day 60 these differences were lost, and the control group was euthanized on day 67 by endpoint criteria, while half of the mice from the P2Et group remained alive until day 87 they were finally euthanized (Fig. 6b). To note, P2Et changed the metastasis profile observed in both groups. First, the group treated with PBS had macro-metastasis in the opposite mammary gland, peritoneum, kidney, mesentery and intestine; while, the group treated with P2Et had macro-metastasis only in the peritoneum, kidney and intestine ( Supplementary Fig. S9). With these results, we have shown that the P2Et extract has a direct effect on the primary TN tumor, significantly retarding tumor growth allowing significantly longer survival. Despite early activity on the tumor, this control was lost over the time; possibly due to the fact that these animals are immunodeficient, which does not let to assess the true role of the immune response in tumor control.

Methods
Natural products. Caesalpinia spinosa pods were collected in Villa de Leyva, Boyacá, Colombia and identified by Luis Carlos Jiménez at the Colombian National Herbarium (voucher specimen number COL 523714, Colombian Environmental Ministry agreement number 1470 related to the use of genetic resources and derived products). The P2Et extract was obtained from Caesalpinia spinosa as previously described 23 .
Tissue acquisition and processing. All primary tumor samples (Table 1) were collected under informed consent from individuals being treated at the Hospital Universitario San Ignacio and Centro Javeriano de Oncología under a protocol approved by the ethics committees. Samples from healthy donors were collected from patients with benign breast pathologies (fibroadenoma) or reduction mammoplasty. Collected samples were minced into small pieces and digested with 2 mg/mL of collagenase (Worthington biochemical corporation, NJ, USA) and 100 U/mL of hyaluronidase (Sigma-Aldrich, St Louis, MO, USA) at 37 °C overnight to generate organoids, then cultured in modified M87 media 25    www.nature.com/scientificreports/ sample was then processed using a FACSAria II (Becton-Dickinson, NJ, USA) and analyzed with FlowJo software (Tree Star, Inc., Ashland, OR, USA).   25 . Stained organoids were analyzed using a confocal microscope FV1000 (Olympus, Conklin, NY, USA). 488 nm line and 515 nm line from 30mW argon laser were used to stimulated calcein and ethidium homodimer-1 fluorescence, respectively. 640 × 640 images were acquired with a 10X UPLFLN, NA: 0.30 objective. Between 3 to 6 organoids per treatment were imaged and analyzed. Each image was analyzed using FIJI (ImageJ, version 1.51, NIH, USA). All images were equally processed adjusting contrast and saturation associated to each fluorescence channel. Area in terms of pixels was calculated for each separated fluorescence channel (Calcein and ethidium homodimer-1), thus the percentage of cell death (area from ethidium homodimer-1) respect to total area (areas from calcein + ethidium homodimer-1) was obtained.

RT-PCR.
In vitro cytotoxicity assays. Cytotoxic effects on cell lines were evaluated using MTT (Sigma-Aldrich, Saint Louis, MO) and trypan blue dye assays. Cells (5 × 10 3 cells/well) were seeded in 96-wells plates with ethanol (0.02%) as negative control, P2Et (250-0.95 μg/ml) or DX (5-0.03 μM) as positive control for 48 h. The cytotoxic effect was estimated by MTT assay according to procedure previously described 34 . The IC 50 (50% inhibition of cell growth) value was calculated using a no linear regression log (inhibitor) versus response-variable slope graph in GraphPad Prism (GraphPad Prism 8 Software, La Jolla, CA, USA). Synergistic effects were assessed over a dose-response matrix that included eight concentrations of P2Et (ranging from 0-500 μg/ml) and DX (0 to 1.6 μg/ml). The effects of drug combination were estimated using R Synergy Finder pipeline 35 and the zero-interaction potency (ZIP) model 36 was used to generate synergy score matrix from a dose-response matrix. At least two independent experiments with triplicate datasets were performed for each treatment. For breast cancer patients' samples and in vitro experiments, multiple comparisons were calculated by oneway ANOVA with Dunnet T3 correction or Tukey post-test analysis and unpaired t-test, and p value of < 0.05 was considered statistically significant. The specific statistical test results are indicated in each figure: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.001. Correlations were assessed using nonparametric Spearman correlation, determination coefficient r and p-value are shown. For in vivo experiments multiple t-test with Holm-Sidak method correction was performed assuming significance of α = 0.05.

Discussion
Traditional Chinese Medicine (TCM) and particularly Phyto-therapy has been used for centuries to improve the response to treatment in cancer patients, and there is currently strong evidence about its positive effects in patient survival and improving the quality of life in response to chemotherapeutic or biological treatment 43 . The mechanisms involved in the activity of plant-derived drugs have been extensively studied, and it has been reported that they can modulate p53, decrease of oncogenic proteins activation or expression, induce cell differentiation via epigenetic modifications, and alters the tumor microenvironment and decreasing metastatic spread by targeting CSC 44,45 .
Among plant-based drugs, polyphenols, are a broad category of plant metabolites that has been implicated in cancer control. They are characterized for the presence of one or more benzene rings attached to hydroxyl groups. They have health benefits attributed to their antioxidant capacity, neuroprotective and anti-inflammatory effects 46 .
For some polyphenols, its antitumor activity and its effect on the reduction of metastases, are also related to the induction of autophagy by a non-canonical route, the decrease in the drug-efflux and the impairment of signaling pathways involved in self-renewal 37,47 . Additionally, its pro-oxidant effect at high concentrations may be involved in mitochondrial dysfunction and decreased cell proliferation 20,48 .
We have previously shown that P2Et extract, a polyphenol-rich extract, have a direct effect on murine melanoma and breast cancer tumors. Previous reports of P2Et, have shown that this extract is harmless in vivo and in vitro and do not affect T-cell activation in healthy mice 38 . We have also shown that P2Et induced mitochondrial-dependent apoptosis. Recently, we demonstrated that P2Et partially acts through a mechanism involving PERK-dependent endoplasmic reticulum stress and Ca2 + unbalance causing mitochondrial-dependent apoptosis 31 . P2Et extract also induced immunogenic signals release such as calreticulin, ATP and HMGB1 21,22 , which explains the effects that partially relies on T-cell activation 21,22 .
Furthermore, we demonstrated that P2Et impair stemness in ALDH + cells (4T1-H17 cells), decreasing their sphere-formation ability and drug-resistance 23 . Besides, the in vivo effects of P2Et against ALDH + enriched models required the activation of the immune response 24 .
To date, P2Et has not been tested in human models of cancer. Hence, the objective of this work was interrogating the activity of P2Et in organoids derived from human specimens and correlate it with the percentage of BCSC.
A large amount of evidence indicates that BCSCs drive tumor development and explains the regenerative capacity, dormancy, chemo-resistance, and metastatic ability 7 . BCSCs can be isolated from fresh surgical specimens, using the currently established stem cell surface markers EpCAM, CD44, CD49f (or CD29), CD24, and ALDH1 25 , and enriched ex vivo by their ability to form spheroids 25 . Here we isolated and studied BCSC of primary tumors from patients with breast cancer who received or not NAT before surgery. Our data indicated that luminal and TN patients have a higher frequency of intra-tumoral BCSC (Lin -CD44 + CD24 − CD49f + EPCAM + ), which are enriched after NAT therapy and express ALDH + , BCRP1 + and Pgp + but not MRP1 + as compared with mammary cells obtained from normal human donors. We also observed a positive correlation between BSCS markers and ABCG2 gene or BCRP1 protein expression, which was also positively correlated with the Scientific Reports | (2020) 10:19639 | https://doi.org/10.1038/s41598-020-76619-9 www.nature.com/scientificreports/ expression of ALDH in the group of patients with NAT before surgery. As well, the group of patients that received NAT, a positive correlation between ALDH and ABCC1 gene expression was observed. In fact, related to our findings, ALDH1 gene expression has also been shown to be higher in TNBC than in Luminal A, Luminal B and HER2 + subtypes 39 . Also, our results confirm previous observations showing increased CD44 + /CD24 − and ALDH1 + cells in basal-like tumors compared to luminal A and B 40 , and reinforce the relationship between CSC and the expression drug resistance related markers. Our results confirm previous work suggesting that BCSC, can be an important target and prognostic marker 41 . Previously, the lack of efficacy of NAT against the PIK3CA-defective BCSCs, one of the most commonly found genetic mutation in breast cancer, have been reported 42 . This means that NAT might select CSCs not only because of their intrinsic resistant to chemotherapy, but also because of the induction of factors that enhance tumor survival. In this sense, it has been observed that treating human or murine TNBC cells with chemotherapeutics, induces a coordinate transcriptional program of CD47, CD73, and PD-L1, leading to higher percentage of CD47 + CD73 + PDL1 + breast cancer cells. Moreover, chemotherapy induces the expression of GSTO1, which is dependent of HIF-1 and HIF-2, and knockdown of GSTO1 expression abrogates carboplatin-induced BCSC enrichment, decreasing tumor initiation and metastatic ability, and delaying tumor relapse 15,16 .
In our hand, P2Et and DX acts through different mechanisms, killing a TNBC cell line (MDA-MB-468) by decreasing intracellular ROS without an induction of GSTO1. Meanwhile, it has been reported that DX induce HIF activation 43 , yet, we do not have evidence of HIF activation by P2Et, supporting the fact that polyphenols have been related to a decrease HIF activation 43,44 . For the first time, we observed that P2Et disrupt the organoids derived from human patients, and these effects were improved in combination with standard chemotherapy (DX). Previously, we observed similar synergistic effects in vitro, in a mechanism involving pump efflux inhibition 23 . However, additional effects that could induce collateral sensitivity cannot be discarded 45 .
These results suggested that drug-resistance induced by NAT in TNBC, could be partially overcome by P2Et used as adjuvant. This hypothesis is based on the evidence that P2Et targets stemness 23 , induced immunogenic cell death 21,22,31 that triggers a T-cell immune response 22,24 and it is cytotoxic to 3D organoids derived from patients 46 .
In order to further evaluate in vivo the impact of P2Et in BCSC-enriched breast cancer models (triple negative and luminal), we analyzed the expression of CD44, CD24, EpCAM, CD49f and ALDH in human cell lines. Our results showed that MDA-MB-468 cells contained the highest frequency of BCSC (Supplementary Figs. S1c,d and S5a). Also, MDA-MB-468 cells had higher ALDH activity compared to BT-549, and it was according to previous reports that used the same cell lines 47,48 .
Targeting ALDH + BCSC is important for several reasons. First, because ALDH is an important marker of drug resistance as mentioned above 49 , and second, because the peptides from the intracellular processing of ALDH can be presented in the context of MHC, unleashing BCSC-specific cytotoxic T-cell 50 .
In light of this, we implanted the MDA-MB-468 cells into NSG mice and we administered P2Et. We observed a reduction in tumor growth and less metastatic spread, however the differences in median survival were not significant, this support the fact that P2Et need an intact immune system to exert a better control of the disease, as we previously showed 24 ; we are aware that the role of innate and immune system in humanized models must be tested in the future 24 .
In summary, P2Et has the ability to act on BCSCs, increasing their sensitivity to chemotherapy, and inducing signals that can lead the activation of an immune response, needed for tumor control. These hypotheses must be evaluated in clinical trials with cancer patients and other models that allows to interrogate the role of human immune system in the control of tumor and metastasis upon P2Et and chemotherapy combination.

Data availability
The datasets generated and/or analyzed during the current study are not publicly available as the informed consent does not cover such release and, further, in compliance with current data protection regulations. Contingent on ethical and data protection board approval, the access to the data are available from the corresponding author on reasonable request.