All-trans retinoic acid and protein kinase C α/β1 inhibitor combined treatment targets cancer stem cells and impairs breast tumor progression

Breast cancer is the leading cause of cancer death among women worldwide. Blocking a single signaling pathway is often an ineffective therapy, especially in the case of aggressive or drug-resistant tumors. Since we have previously described the mechanism involved in the crosstalk between Retinoic Acid system and protein kinase C (PKC) pathway, the rationale of our study was to evaluate the effect of combining all-trans-retinoic acid (ATRA) with a classical PCK inhibitor (Gö6976) in preclinical settings. Employing hormone-independent mammary cancer models, Gö6976 and ATRA combined treatment induced a synergistic reduction in proliferative potential that correlated with an increased apoptosis and RARs modulation towards an anti-oncogenic profile. Combined treatment also impairs growth, self-renewal and clonogenicity potential of cancer stem cells and reduced tumor growth, metastatic spread and cancer stem cells frequency in vivo. An in-silico analysis of “Kaplan–Meier plotter” database indicated that low PKCα together with high RARα mRNA expression is a favorable prognosis factor for hormone-independent breast cancer patients. Here we demonstrate that a classical PKC inhibitor potentiates ATRA antitumor effects also targeting cancer stem cells growth, self-renewal and frequency.

Mammosphere size determination. After 96 h treatment, mammospheres were observed under an inverted microscope (Nikon Eclipse TE 2000-S) and 10 random fields were digitally photographed using a digital camera.
Mammosphere diameters were measured on the long axis using Image J software and average size was calculated.
Secondary mammosphere assay. Primary mammospheres were dissociated using 0.05% trypsin for 15 min at 37 °C in order to obtain a single-cell suspension. Then, cell suspension was seeded and cultured as described above for mammosphere assay, in order to obtain a new generation of mammospheres. Sphere formation was assessed after 5 days and mammospheres size and number was recorded.
Clonogenic assay. After 96 h treatment, primary mammospheres were enzymatically dissociated as described above and a suspension containing 1 × 10 3 cells was plated in adherent conditions. After 7 days culture, colonies were fixed with methanol:acetic (3:1) and stained with crystal violet.
Detection of apoptosis by annexin V assay. Cell monolayers were treated with ATRA (0.5 µM), Gö6976 (0.5 µM), their combination or vehicle alone for 48 h. Then, cells were collected, and apoptotic cells were quantified as described by the manufacturer. Briefly, cells (1 × 10 6 ) were washed and resuspended in 100 μl 1X binding buffer. Then, cells were incubated 15 min in darkness at room temperature with 5 μl of Alexa488-conjugated Annexin V. Cells were washed with 1X binding buffer and finally 5 μl of propidium iodide was added. Cells were mixed in darkness at room temperature for 10 min, then 300 μl of 1X binding buffer was added, and cells were mixed in an ice bath at dark. Cell suspension was examined under 488 nm excitation wavelength by flow cytometry using an Epics Elite ESP coulter cytometer (Beckman coulter, Fullerton, CA).
Western blot. Western blot (WB) assays were performed as previously described by Berardi et al. 14 , employing cell lysates prepared from monolayers treated for 48 h with ATRA (0.5 µM), Gö6976 (0.5 µM), their combination or vehicle alone as control.
RT-qPCR. Subconfluent cultures of each cell line were treated for 48 h with ATRA (0.5 µM), Gö6976 (0.5 µM), their combination or vehicle alone. Total RNA was prepared using Tri Reagent (Merck, Darmstadt, Germany). cDNA was prepared with the iScript cDNA synthesis kit (Bio Rad) and amplified by real-time PCR using a CFX96 Real-Time PCR detection systems kit (Bio-Rad) and SYBR green PCR master mix (Applied Biosystems, Carlsbad, CA). PCR products were obtained using primers indicated in Table S1. GAPDH was used as housekeeping gene. Relative changes in gene expression were calculated with the 2-∆∆CT or 2-∆RAR∆CT method 28 .
Wound migration assay. Subconfluent LM38-LP and SKBR3 monolayers were treated with Gö6976 (0.5 µM) and/or ATRA (0.5 µM) or vehicle as control for 48 h. Then, wounds of approximately 400 μm width were performed and cells were allowed to migrate to the cell-free area for a period of 12 h for LM38-LP or 24 h for SKBR3 cells. The same spot was photographed at different times and migratory area was quantified with ImageJ software. Cell migration was expressed as percentage of area occupied by the migratory cells in the original cell-free wounded area. Cell viability was not affected by ATRA or Gö6976 doses used in this experiment. We established humane endpoint when mice met one of the following signs: Bristling coat and/or hemorrhagic diarrhea, loss of > 20% of the initial weight or lethargy. Animals were euthanized by CO 2 inhalation.
Orthotopic tumor growth and spontaneous metastatic ability. Tumor growth and spontaneous metastatic ability were evaluated as previously described in detail 14,24 . In brief, mice were inoculated orthotopically into the fat pad of the 4 th mammary gland with 2 × 10 5 LM38-LP cells (n = 5 for group, 20 animals in each experiment). Five days after cell inoculation, mice were anesthetized injecting a combination of ketamine (100 mg/kg) and xylazine (5 mg/kg) intraperitoneally. Then each mouse received a subcutaneous silastic pellet containing ATRA (10 mg) or an empty pellet as control. Gö6976 (60 mg/kg) was locally administered twice a week. The control group received the same volume of solvent (0.2 ml of 0.1% DMSO physiologic solution). Mice were monitored daily. Twice a week, tumor diameters were measured with a sliding caliper and tumor volume was calculated using the following formula: Dxd 2 /2, where D is the longest and d is the shortest diameter. Twenty-one days after tumor treatment, mice were sacrificed as described above and necropsied. Lungs were removed and fixed in Bouin's solution to investigate the presence of spontaneous metastases. The number of surface lung nodules was recorded. Liver, kidney, and spleen were also fixed and examined for the presence of metastatic nodules.
LM38-LP tumor cell suspension preparation and limiting dilution assay. LM38-LP tumors harvested post-treatment were minced and digested in digestion media as previously described in detail 29 . Next, LM38-LP tumorderived were plated at 10 and 100 cells per well into a ultra-low attachment 96 well plate. 5 days after plating, mammospheres number found in each well was quantified under microscope. Cancer stem cell frequency and p-values were calculated by using ELDA software 30 .
Statistical analysis. All assays were performed in triplicate, and independent experiments repeated at least twice. Statistical differences between groups were calculated by applying ANOVA, Student's t or Kruskal-Wallis tests, as indicated. A p value < 0.05 was considered statistically significant.

Studies in silico.
Bioinformatic analysis from Kaplan Meier Plot database of PRKCA and RARA expression. A Kaplan-Meier survival database that contains survival information of 801 estrogen receptor negative breast cancer patients and gene expression data at diagnosis obtained by using Affymetrix microarrays 31 was employed. Probes set were 213093_at (PRKCA) and 203750_at (RARA) and split patients by auto select best cut off into a low-expression group and a high-expression group. Relapse free survival (RFS) curves were plotted according to the Kaplan Meier method and evaluated by the log-rank test.
Ethics approval and consent to participate. All animal studies were conducted in accordance with the standards of animal care as outlined in the NIH and ARRIVE Guidelines for the Care and Use of Laboratory Whole cell lysates prepared from LM38-LP and SKBR3 cell lines were resolved on sodium dodecyl sulfate-polyacrylamide gel electrophoresis and blotted with antibodies against PKC α, β and γ. Actin expression levels was used as protein loading control. (b) LM38-LP and SKBR3 cell number was assessed 96 h after treatments with ATRA (0.25-1 µM) and/or Gö6976 (0.25-1 µM) or vehicle as control. (c) LM38-LP mammospheres diameter was measured 96 h after treatments with ATRA (0.5 µM) and/or Gö6976 (0.5 µM) or vehicle as control. (d) LM38-LP cells were treated with ATRA (0.5 µM) and/ or Gö6976 (0.5 µM) or vehicle as control for 48 h and then RNA was isolated. Nanog and Sox2 expression was analyzed by RT-qPCR. The fold of change of mRNA levels was calculated using the ΔΔCt method with GAPDH used as an internal control. Histograms represent mean ± S.D. (e) Representative photographs of LM38-LP mammospheres after 96 h treatments. (f) HCC38 mammospheres diameter was measured 96 h after treatments with ATRA (0.5 µM) and/or Gö6976 (0.5 µM) or vehicle as control. (g) HCC38 cells were treated with ATRA (0.5 µM) and/or Gö6976 (0.5 µM) or vehicle as control for 48 h and then RNA was isolated. Nanog and Sox2 expression was analyzed by RT-qPCR. The fold of change of mRNA levels was calculated using the ΔΔCt method with GAPDH used as an internal control. Histograms represent mean ± S.D (h) Representative photographs of HCC38 mammospheres after 96 h treatments. Scale bar 100 µm. Data represent the mean ± S.D. *p < 0.05 versus control, **p < 0.01 versus control, ***p < 0.001 versus control, # p < 0.05 versus Gö6976, ## p < 0.01 versus Gö6976 (ANOVA test). Three independent experiments were performed.

ATRA and Gö6976 treatments reduce breast cancer and breast cancer stem cells proliferation.
First, we have analyzed the expression of PKC α, β and γ by WB in LM38-LP and SKBR3 cell lines, since this classical PKC isoforms are the main target of Gö6976. As shown in Fig. 1a, PKCα showed high expression levels in both cell lines, while PKCβ showed moderated expression. PKCγ was almost undetectable, as previously reported for mammary tissues 32 .
LM38-LP and SKBR3 cells were treated with ATRA or Gö6976 at a range of doses from 0.25 to 1 μM for both drugs. Additionally, ATRA and Gö6976 combination employing those doses was also tested. After 4 days treatment, the effects of single agents or drug combinations on LM38-LP and SKBR3 cell number was examined. As shown in Fig. 1b, we could determine that ATRA and Gö6976 treatment alone led to a significant decrease in the proliferative potential of both cell lines. Moreover, ATRA and Gö6976 combination also reduced proliferative capacity but in a higher degree than each treatment alone. The stronger effect was achieved employing 0.5 μM of both compounds thus, this concentration was chosen for the next set of experiments.
In order to evaluate the effect of ATRA and Gö6976 on CSC proliferation, a mammosphere assay was performed. The stem/progenitor component characteristics of LM38-LP cell line was described elsewhere 24 . SKBR3 cells were not employed in these assays since they are not able to form mammospheres 33 , thus we used HCC38 human breast cancer cell line, where only combined ATRA and Gö6976 treatment was able to inhibit the proliferative capacity (Fig. S1). We observed that ATRA or Gö6976 treatment alone slightly reduced LM38-LP and HCC38 CSC proliferation between 20 and 30% (Fig. 1c,e,f,h). Moreover, ATRA and Gö6976 combination highly reduced proliferative capacity of LM38-LP and HCC38 CSC between 50 and 60% (Fig. 1c,e,f,h). Surprisingly, when we analyzed CSC markers by qPCR after 48 h of treatments, we observed that ATRA increase SOX2 expression in both cell lines (Fig. 1d,g). On the other hand, Gö6976 treatment led to reduce only NANOG or SOX2 depending the cell line (Fig. 1d,g). Remarkably, only combined treatment led to a significant reduce of both CSC markers in both cell lines.
ATRA and Gö6976 combination have a synergic interaction. Next, we analyzed whether the growth inhibition observed under combined treatment was due to a synergistic interaction between the drugs employed. Results obtained in proliferation assays were expressed as the percentage of control and drug interactions was analyzed by Chou-Talalay's method 25,26 . As shown in Table 1, each single agent displayed a dose-dependent inhibition of cell proliferation. All ATRA and Gö6976 combinations studied exhibited a synergistic effect shown by their Combination index (CI) lower than 0.7.
ATRA and Gö6976 combination impair cancer stem cell self-renewal and clonogenicity. In order to determine if the different treatments affected cancer stem cell self-renewal and clonogenicity, LM38-LP and HCC38 mammospheres were pre-treated for 96 h with ATRA and/or Gö6976 and a secondary mammosphere assay was performed. Although ATRA treatment increased the number of secondary mammospheres (Fig. 2a,c,d,f) and the clonogenic capacity (Fig. 2g,h), mammospheres diameters were smaller (Fig. 2b,e). On the other hand, Gö6976 pre-treatment was able to reduce both the number and diameter of secondary mammospheres ( Fig. 2a-f), as well as clonogenicity (Fig. 2g,h). Combined treatment impairs the increase of cancer stem cells self-renewal and clonogenicity induced by ATRA and led to a significantly diameter reduction of secondary mammospheres (Fig. 2a-h).
Since, retinoic acid drive many of its biology effects through RARs activation, we analyze which RARs isotype would be involved in cancer stem self-renewal and clonogenicity upon ATRA treatment. The RARγ antagonist (MM11253) was able to impair the effect of ATRA treatment in cancer stem cell population (Fig. 2i,j). On the other hand, RARβ antagonist (LE135) induced a potentiation of ATRA effect treatment on cancer stem cell Table 1. Proliferation assay, determination of the combination Index (CI). LM38-LP and SKBR3 cells number was assessed 96 h after treatments with ATRA (0.25-1 µM) and/or Gö6976 (0.25-1 µM). Each data point represents the mean ± standard error of three independent experiments. CI < 0.7 indicates synergism, CI > 0.7 and CI < 1 indicates additivity, and CI > 1 denotes antagonism. www.nature.com/scientificreports/ self-renewal (Fig. 2i,j) indicating that RARβ might function as a buffer for retinoic acid response on cancer stem cell population.

ATRA and Gö6976 combination modulates cell cycle progression, induces apoptosis and impairs autophagy. Cell cycle distribution of LM38-LP and SKBR3 cells was analyzed after treatments
with single agents or their combination. As compared to control cells, ATRA led to an increased accumulation of LM38-LP cells in the G 0 /G 1 phase coupled with a reduction of the S phase and an increase of the sub G 1 phase of cell cycle (Fig. 3a, left panel). In SKBR3 cells, ATRA treatment increased cells accumulation in G 2 and sub-G 1 phases in cell cycle, coupled with a reduction of cells in S and G 1 (Fig. 3a, right panel). On the other hand, Gö6976 treatment induced a significantly increase in the Sub-G 1 fraction on both cell lines (Fig. 3a). Although ATRA induced a slight but significant increase of this fraction, combined treatment showed a greater effect in accumulation of cells in the sub-G 1 phase (Fig. 3a), suggesting apoptosis events. To confirm apoptosis induction, LM38-LP cells were collected and annexin V staining was evaluated after ATRA and/or Gö6976 treatments. Combined treatment showed a greater effect on apoptosis induction as shown in Fig. 3b. Next, we wanted to elucidate whether Gö6976 could behave as an autophagy inhibitor, due the role of autophagy in the inhibition of the apoptosis process 34 . Through Western blot we analyzed LC3-II/LC3-I expression ratio and p62/SQSTM accumulation as a marker of autophagy process. We could determine that ATRA treatment significantly increased LC3-II/LC3-I ratio in both cell lines, without significant modulation in p62/ SQSTM, which is compatible with an autophagy activation profile (Fig. 3c). On the other hand, Gö6976 treatment and its combination with ATRA led to a significant decrease of LC3-II/LC3-I protein ratio with a significant increase of p62/SQSTM1expression in both cell lines. Altogether these protein expression profile correlates with autophagy inhibition.
ATRA and Gö6976 treatments impair migratory potential and soluble MMPs activity. The effect of ATRA and Gö6976 combination on migratory potential of both LM38-LP and SKBR3 cells were analyzed through a "wound healing" assay as described in "Materials and Methods" section. Both ATRA and Gö6976 treatments significantly impaired LM38-LP cell migration towards the wounded area as compared to control cells. Moreover, combined treatment enhanced this inhibition (Fig. 4a, upper panel). SKBR3 cells have a very low migratory capacity, nevertheless, both treatments decreased migratory potential (Fig. 4a, lower panel). Additionally, we had explored soluble MMPs activity in LM38-LP cells since these proteases are intimately related to migratory and invasive processes. Both ATRA and Gö6976 treatments decreased secreted MMP-2 activity, being undetectable under combined treatment (Fig. 4b).

ATRA and Gö6976 treatments modulate RARs expression. By RT-qPCR we could determine that
Gö6976 treatment alone did not alter RARs expression (Fig. 5a). On the other hand, ATRA induced a significant increase in RARβ and RARγ in both cell lines (Fig. 5a). When the combined condition was analyzed, both RARα and RARβ significantly increased their expression, while RARγ levels were affected only in SKBR3 cells (Fig. 5a). Given that RARs ratio is essential to elucidate the final cell's fate in response to a treatment 35 , we analyzed RARα/ RARγ and RARβ/RARγ ratio after ATRA and Gö6976 treatments. The combined treatment induced a significant increase of RARα/RARγ and RARβ/RARγ ratio in both cell lines, which is compatible with a differentiated cell profile (Fig. 5b).
ATRA and Gö6976 treatment impaired in vivo tumor growth, metastatic dissemination and CSC frequency. Next, we evaluate whether in vitro described results had an in vivo correlation. Twenty female BALB/c mice were orthotopically inoculated with LM38-LP cells and five days later, animals received treatments as described in "Materials and Methods" section (5 animals per group, assays were performed twice and all animals 20/20 presented good health status). We could determine that both ATRA and Gö6976 treatments reduced LM38-LP in vivo tumor growth (Fig. 6a,b). Once again, ATRA/Gö6976 combined significantly impairs in vivo tumor growth when compared with each treatment alone. (Fig. 6a,b). Although Gö6976 treatment alone induced a reduction in lung metastatic capacity, it is important to note that, in the combined treatment group, just one of tumor bearing mice developed one metastatic focus (Fig. 6c). Furthermore, only combined treatment was able to significantly reduce both NANOG and SOX2 expression (Fig. 6d). Finally, an ELDA was performed using 10 and 100 cells derived from LM38-LP tumors harvested post-treatment, in order to evaluate CSC frequency. ELDA results revealed that ATRA/Gö6976 combined treatment led to a significantly lower CSC frequency as compared to the vehicle (1/1710, p = 1.26E-07), as well as to ATRA (1/240, p = 0.000136) or Gö6976 (1/388, p = 0.00976) treatment alone (Fig. 6e) (Fig. 7c). These results reinforce the importance of proposed ATRA and Gö6976 combined treatment for estrogen receptor negative breast cancer patients.

Discussion
Among women, breast cancer is the most frequently diagnosed cancer in the vast majority of countries around the world and is also the leading cause of cancer death in over 100 countries 1 . Hormone-independent breast cancers are considered a high-risk group since patients present an unfavorable prognosis with high recurrence rates 36 . Targeted therapy has changed the course of breast cancer treatment, but blocking a single pathway is finally ineffective, due to the activation of redundant and/or alternative oncogenic pathways 37 . Therefore, it is imperative to develop new therapies, targeting different signaling pathways, in order to generate a great impact on the evolution of this disease. ATRA, as well as its natural and synthetic derivatives collectively known as retinoids, are promising agents for treatment or chemoprevention of different malignancies including breast cancer. Although in clinical settings, the use of retinoic acid as monotherapy has been controversial for solid tumors, some phase II clinical trials are still being evaluated 38,39 . However, several authors have demonstrated the effectiveness of retinoid therapy in combination with other drugs such as tamoxifen or trastuzumab 40 , suggesting that retinoic acid utility depends on its capacity to potentiate the effect of other compounds employed for cancer treatment 15,39,41 .
The importance of performing a combined treatment relies in the fact that ATRA can activate other kinases through a non-canonical pathway. In fact, several studies have reported that retinoids can activate some PKC isoforms [42][43][44][45] . So, our rationale was to potentiate the differentiator effect of ATRA by combining this retinoid with a classical PKC inhibitor thus avoiding ATRA undesired effects.
As shown in results section, ATRA and Gö6976 combined treatment highly reduced proliferative capacity of breast cancer cell lines in a synergistic manner. To elucidate mechanisms underlying this proliferation inhibition, we examined the effect of both single agents and their combination on autophagy, since it has been described that this mechanism impairs apoptosis 46 . Although, combined treatment induced a significant apoptosis increase when compared with each treatment alone, it has been reported that ATRA induced autophagy in hormoneindependent breast cancer cell lines employed 47 . Nevertheless, apoptosis induction caused by Gö6976 treatment prevailed over ATRA effect, demonstrating that blocking autophagy could be an interesting strategy to potentiate ATRA effect on apoptosis.
Regarding cancer stem cell population, we could observe that primary mammospheres pre-treatment with ATRA, favors the preservation of stem/progenitor cells with the potential to regenerate the original cell line. This was reflected both as an increase in secondary mammospheres number, as well as in the clonogenic capacity. However, secondary mammospheres obtained after ATRA treatment showed a reduction in growth rate, evidenced by the decrease in their diameter. In sum, while an increase in the self-renewal capacity of CSC could be detected, these cells display a lower growth rate.
To evaluate the involvement of retinoid receptors in these results, we employ specific RARβ and RARγ antagonists 15 . We observed that RARγ activation by ATRA was involved in CSC population growth. These results correlate with previous publications, where RARγ is described as a key receptor involved in hematopoietic cells self-renewal 48 . On the other hand, ATRA treatment also increases RARβ levels showing that retinoic acid receptors system is activated and functional. It is important to note that RARβ promoter contains RARE sequences, allowing this gene transcription after retinoid stimulus. Due to its role in cell differentiation, pharmacological inhibition of RARβ activity increased even more CSC self-renewal.
Regarding Gö6976 effects, we observed that the treatment with this PKC inhibitor blocks both self-renewal and clonogenic capacity of CSC, confirming that PKCα is a critical signaling component for CSC 18 , also impairing the negative effects exerted by ATRA over that cell population.
Several in vitro features associated with tumor progression were also analyzed. In this sense, ATRA/Gö6976 combined treatment reduced migratory capacity and MMP-2 secreted activity better than each treatment alone. Consistently, in vivo metastatic capability was also affected. In fact, combined treatment not only affected the metastatic potential but also produced an important impairment of in vivo tumor growth.
It is well known that RARα is one of the key members in the response to ATRA treatment. Acute promyelocytic leukemia patients have a malfunction of RARα protein due to a genetic fusion between RARα and PML     49 . This genetic abnormality causes a lack of response to basal plasma levels of retinoic acid, leading to undifferentiated immune cells 50 . In these malignancies, increasing Retinoic Acid levels in plasma led to the differentiation of immune cell, allowing total cure of this disease. For this reason, we focus on how combined treatment modulates RARs expression. Only combined treatment led to a significant increase of RARα levels, probably indicating the induction of a differentiated phenotype and therefore reducing malignant potential. Regarding RARβ, ATRA or the combined treatment, increased its expression. This receptor acts as a tumor suppressor 51 , and generally is downregulated or not expressed in breast cancers 52 . RARγ display a proliferative role in hepatocellular carcinoma 53 and in breast cancer 35 . At the same time this receptor is involved in hematopoietic stem cell self-renewal 48 . Our studies reveal an upregulation of RARγ induced by ATRA treatment. Nevertheless, the combination with Gö6976 impaired induction of this retinoid receptor in LM38-LP cells. Furthermore, it has been described that RARα/RARγ and RARβ/RARγ expression ratio is critical to elucidate the response to ATRA treatment 35 . Only combined treatment increased both ratios, leading to a response compatible with the reversion of the malignant phenotype driving to a differentiated state 35 . Finally, it is important to note that the cell lines used in the present study express PKCα. It has been reported that this PKC isoform could be considered as a poor prognosis marker in breast cancer 17 , thus we decided to perform an in-silico analysis in order to evaluate how PKCα and RARα mRNA expression levels affected the RFS probability in hormone receptor negative breast cancer patients. Our analysis showed that low PRKCA mRNA expression together with high RARA mRNA expression becomes a favorable factor of prognosis for these patients.
In sum, our findings show that Gö6976 treatment potentiates antitumor effect of ATRA by inducing apoptosis of breast cancer cells, inhibiting cancer stem cell self-renewal and clonogenicity and leading to RARs balance compatible with an anti-oncogenic response. Moreover, in vivo tumor growth, metastasis spreading, and CSC frequency were also inhibited by the combined treatment in the LM38-LP triple negative mammary cancer cell line.
The importance of our findings relies in the fact that experimental models employed are hormone-independent tumors. Although clinical treatment of these pathologies is chemo and radiotherapy, these malignancies present high recurrence rate and/or acquire treatment-resistance and to date, there is no effective or directed therapy. For this reason, the development of pre-clinical assays imperative in order to propose novel therapies, such as the presented in this work.
The rationale design of molecules that block classical PKCs activity, in combination with retinoids could led the development of new potential therapies for the treatment of hormone-independent breast cancer patients.