Background
ERBB2 is a member of the epidermal growth factor (EGF) receptor (ERBB1) family and is chaperoned by HSP90; unlike other ERBB members (ERBB1, ERBB3 and ERBB4), ERBB2 has no soluble ligands [
1]. ERBB2 is amplified and overexpressed in 20% to 30% of human breast cancers (BCs), and it is often associated with aggressive disease and poor prognosis [
2,
3]. Trastuzumab (Tz, Herceptin®), a humanized monoclonal antibody [
4], binds the extracellular region of ERBB2 and inhibits receptor signaling via several mechanisms, including the down-regulation of both the Ras/Raf/MAPK pathway, which contributes to cell proliferation [
5], and the PI3K/AKT pathway, which regulates cell survival [
6,
7]. Moreover, nuclear accumulation of the cell cycle inhibitor CDKN1B/p27
KIP1, and cell cycle arrest have been reported [
8,
9]. The introduction of Tz therapy and, more recently, of the first successful HER2-targeted antibody–drug conjugate (ADC), trastuzumab emtansine (T-DM1; Kadcyla; Genentech) have improved significantly progression free survival and overall survival in ERBB2-amplified metastatic BCs [
10]. Nevertheless, the presence of primary and acquired resistance, as well as serious adverse side effects to Tz and T-DM1 treatments remain a significant common challenge [
11,
12]. The catabolic process of autophagy is considered to be associated with resistance to chemotherapy and targeted inhibitors in cancers [
13]. Indeed, autophagy induced by anti-ERBB2 targeting agents like Tz and Lapatinib may allow cancer cells to survive the stress induced by the therapy [
14,
15]. Thus, the inhibition of autophagy after treatment with Tz resulted in apoptotic cell death [
16].
Rosemary (
Rosmarinus officinalis) and common sage (
Salvia officinalis) contain multiple bioactive components including carnosic acid (CA), rosmarinic acid, carnosol, caffeic acid, and ursolic acid, which have antioxidant, antinflammatory, anti-steatosis and anticancer properties [
17‐
19]. Whole extracts and/or purified components isolated from rosemary have been shown to inhibit the in vitro and in vivo cell growth from solid tumors including BCs and leukemia cells [
17]. The biological effects of CA on the modulation of cell cycle arrest, apoptosis and autophagy programs have been studied in different cancer cell types [
17]. CA-mediated inhibition of the PI3K/AKT/mTOR pathway and induction of autophagy have been observed in hepatocarcinoma cells [
20]. At difference, in myeloid leukemia cells the inhibition of AKT signaling has induced p27
KIP1 expression and a block in the G1 phase of the cell cycle [
20]. In colon adenocarcinoma cells, CA reduced leptin signaling through dephosphorylation of AKT, ERK and Insulin-like growth factor-I receptor, which have caused a reduced expression of both BCL-XL and Cyclin D1 resulting in cell cycle arrest [
21]. Induction of apoptosis through the production of reactive oxygen species (ROS) or mediated by the AKT/IKK/NF-kB axis has been reported in different carcinoma cell lines [
17,
22]. In contrast, in human glioblastoma (GBM) cell lines CA did not down-regulate the PI3K/AKT pathway, whereas it induced proteasomal down-regulation of Retinoblastoma, Cyclin B1, SOX2 and GFAP and cell cycle arrest with only a minor induction of apoptosis [
23]. Therefore, these evidences corroborate the notion that CA does not elicit a common response but down-modulates distinct cellular pathways depending on a cancer cell-type specific context. In particular, in BC cell lines, a previous study showed that CA inhibited cell growth and suggested that ERBB2 was required for this activity [
24]. Others reported that rosemary extract enhanced Tz effects on survival inhibition [
25]. However, neither the active compound nor the molecular mechanisms at the basis of this interaction were provided.
In the present study, we aimed at investigating the mechanisms of action and the potential benefit of the combined treatment of CA and Tz in vitro in ERBB2+ BC cell lines SKBR-3 and BT474, which are sensitive to Tz, and in Tz-resistant SKBR-3 cells. We have found that CA exerts a reversible cooperation with Tz in boosting the inhibition of cell survival and cell migration, by inducing cell cycle arrest in G0/G1 phase. These events are coupled with the down-regulation of PI3K/AKT/mTOR pathway, the inhibition of late autophagy and derangement of the endolysosomal compartment. In addition, we have shown for the first time that co-treatment with CA and Tz partially restored the sensitivity to Tz in Tz-resistant SKBR-3 BC cells. These results suggest that a combined therapy of CA and Tz could represent a new and potentially less toxic approach worth of further investigations for the treatment of ERBB2+ BCs.
Methods
Plant material
Fresh aerial parts of
Salvia somalensis Vatke were obtained from Centro Regionale di Sperimentazione ed Assistenza Agricola (Albenga, Italy). All reagents were analytical or high performance liquid chromatography (HPLC) grade. The isolation of the leaf constituents of
Salvia somalensis (1.15 Kg) was performed following a procedure previously described [
26]. Carnosic acid (CA), m.p. 190–192 °C was identified by comparison of its physical and spectroscopic data with those published in the literature [
27] and obtained with a HPLC purity of 95%.
Cell culture and drug treatments
BC cell lines SKBR-3, BT474, MCF7 and MDA-MB-231 were obtained from Banca Biologica and Cell Factory in IRCCS AOU San Martino – IST belonging to the European Culture Collection’s Organization. Tz resistant SKBR-3 (Tz-Res SKBR-3) cells were generated by continuous treatment of SKBR-3 cells with Tz 200 μg/ml for 10 months. MCF10A cells were obtained from NIH Institute and cultured according to the manufacturer’s instructions.
BC cells were cultured in complete medium (DMEM high glucose supplemented with 10% heat-inactivated fetal bovine serum, 1% glutamine, penicillin and streptomycin (Euroclone S.p.A., Milan, Italy). Tz (Genentech-Roche, South San Francisco, CA, USA) was donated by the UFA-Unità Farmaci Antiblastici of the IRCCS AOU - San Martino - IST. Tz was used at a concentration of 10 μg/ml for SKBR-3 (parental and Tz-resistant), MDA-MB-231 and MCF7 cells and at 0.24 μg/ml for BT474, respectively. CA was used at 27.5 μM for SKBR-3 (parental and Tz-resistant), MDA-MB-231 and MCF7 cells and 37.5 μM for BT474, respectively. Control cultures were challenged with DMSO (CA solvent) and human IgGs. Similarly, CA treated culture were also exposed to human IgGs and Tz treated cultures to DMSO, respectively.
Cell survival assay
All BC cells were plated in 24-well plates in complete medium (triplicate of SKBR-3, BT474 and MDA-MB-231 28,000 cells/well, MCF7 15,000 cells/well) and CA and/or Tz were administered every 48 h for up to 7 or 10 days (d) as indicated. Cell survival was measured at different time points using the 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide (MTT) or Alamar Blue (Thermo Fisher Scientific, Waltham, MA, USA) colorimetric assay as described before [
28], and as indicated in the manufacturer’s instructions, respectively.
Cell migration assay
Cell migration assay was performed in quadruplicate using 8.0 μm pore size inserts in 24-well plates (BD Bioscience, Franklin Lakes, NJ). Fifty thousands MDA-MB-231 cells and 150,000 BT474 cells were seeded in the upper chamber and overnight starved (DMEM supplemented with 1% glutamine, penicillin and streptomycin). The day after, starvation medium was replaced, in the lower chamber, with complete medium supplemented with CA for MDA-MB-231 cells or CA and/or Tz for BT474 cells. Migrated cells were stained with Crystal Violet after 48 h for MDA-MB-231 cells and after 7d for BT474 cells (treatments every 48 h). Cell migration was quantified using ImageJ [
29] as previously described [
30].
Flow cytometry (FCM) analysis
BT474 and SKBR-3 cells were treated with CA and/or Tz for 48 h. Both adherent and floating cells were then collected and centrifuged at 980 g for 5 min. Cell cycle analysis was performed through evaluation of DNA content in cell nuclei stained with DAPI by high resolution DNA flow cytometry (hr DNA-FCM) using a CyFlow ML flow cytometer (Sysmex-Partec Inc., Lincolnshire, IL, USA) [
31]. Metabolic active, apoptotic and necrotic cells were evaluated using a Cyan ADP flow cytometer (Beckman Coulter, Brea, CA, USA) and the Vybrant® Apoptosis Assay Kit (Thermo Fisher Scientific) with a minor modification as we used the nuclear staining fluorochrome sytox blue (Thermo Fisher Scientific) in place of the sytox green.
Antibodies
All primary antibodies used are listed in Additional file
1: Table S1.
Immunofluorescence analysis
SKBR-3, BT474 and MCF10A cells were treated for 7d with CA and/or Tz (treatments every 48 h), fixed in 3% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) pH 7.4 and then quenched with 30 mM NH4Cl. Cell permeabilization was performed with Triton for the Ki-67 staining and with saponin for LAMP 1 and LAMP 2 staining. Alexa-conjugated secondary antibodies were from Thermo Fisher Scientific. Image deconvolution and acquisition was performed with an Axio Imager A2M microscope equipped with an Apotome module (Carl Zeiss, Jena, Germany).
Immunoblot analysis
BC cells were lysed using lysis buffer (Hepes pH 7.4 20 mM, NaCl 150 mM, 10% Glycerol, 1% Triton X-100) with protease inhibitors cocktail Complete (Roche Applied Science, Penzberg, Germany) and sodium orthovanadate or Phostop (Roche). Proteins were resolved on SDS-polyacrylamide gel electrophoresis and blotted on nitrocellulose (Thermo Fisher Scientific) or PVDF (Merck Millipore, Darmstadt, Germany) membranes. Detection was performed with ECL Detection Reagent (GE Healthcare) according to manufacturer’s protocol. ECL signals were detected, recorded and measured by either the GS-800 Calibrated Imaging Densitometer and the Quantity One Software (BioRad, Hercules, CA) or the Li-Cor scanner and Image Studio software (LI-COR Biosciences Inc., Lincoln, NE, USA) or the Uvitec Cambridge gel doc system and software (Cambridge, UK).
Transmission electron microscopy (TEM)
For electron microscopy analysis, SKBR-3 and BT474 cells were seeded on glass chamber slides (Lab-Tek 177,380, Nalge Nunc int., Rochester, NY, USA) and treated for 7d. After drug treatments, cells were processed for electron microscopy [
32] and observed with a CM10 electron microscope (Philips, Eindhoven, The Netherlands). Digital images were taken with a Megaview 3 camera. Analysis of morphologically identified multivesicular bodies (MVBs), autolysosomes (AL), autophagic vesicle/lipid droplets (AV/LDs) and lipid droplets (LDs) diameters were assessed in 10 cells for each treatment. The diameter of each organelle was measured with iTEM software package (Olympus-SYS; Olympus Corporation, Shinjuku, Tokyo, Japan) and plotted as histograms.
RNA extraction and real-time and quantitative PCR (RT qPCR)
RNA was isolated using the Trizol reagent (Thermo Fisher Scientific), cDNA was synthesized and real-time quantitative PCR (RT qPCR) was performed in quadruplicate using 1 × IQTM SybrGreen SuperMix and CFX apparatus (Biorad, Milan, Italy). The relative quantity of target mRNA was calculated by the comparative Cq method using glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as housekeeping gene (GAPDH fwd: 5′-ACCCACTCCTCCACCTTTGACG-3′; GAPDH Rev. 5′- CTCTTGTGCTCTTGCTGGGGCTG-3′), and expressed as fold induction with respect to controls [
33]. PLIN2 primer pairs (PLIN2 Fwd 5′- TGTGAGATGGCAGAGAACGGT-3′; PLIN2 Rev. 5′-CTGCTCACGAGCTGCATCATC-3′) were designed ad hoc starting from the coding sequences of
Homo sapiens available on the GenBank database (
https://www.ncbi.nlm.nih.gov/genbank/) and synthesized by Tib MolBiol s.r.l. (Genova, Italy). PLIN1 primer pairs were purchased from Biorad (# qHsaCID0011127). Amplification conditions consisted of 2 min at 95 °C followed by 5 s at 95 °C and 30 s for PLIN1 or 45 s for PLIN2 at 60 °C for 40 cycles.
Statistical analysis
All measurements here reported are presented as mean ± standard deviations. For MDA-MB-231 and MCF7 MTT assay, MDA-MB-231 cell migration assay and SKBR-3 and BT474 Alamar Blue cell survival assay we used a two-tailed distribution Student’s t-test. For ultrastructural studies, we used the Mann–Whitney test. For BT474 cell migration assay, cell cycle analyses and Tz-Res SKBR-3 MTT assay, we used one-way ANOVA plus post-hoc Tukey’s test. For the Ki-67 measurements we used one-way ANOVA plus post-hoc Bonferroni’s test. For SKBR-3 and BT474 MTT assays we used two-way ANOVA plus post-hoc Bonferroni’s test. For SKBR-3 and BT474 RT qPCR analysis we used one-way ANOVA Kruskal Wallis’s test plus the post-hoc Tukey’s test. Mean differences were considered statistically significant (P value) at P < 0.05.
Discussion
The identification and characterization of the anticancer properties of natural products have aroused significant interest over the years. Several studies have shown that CA has antitumor properties in several cancer cell models [
18]. However, a comprehensive evaluation of CA as anticancer agent in ERBB2
+ BC cells is still lacking. For this reason, in this study we aimed at verifying whether CA was active in BC cells and specifically whether it could enhance the in vitro antitumor effects of Tz in ERBB2
+ cells.
Differently from previous reports [
24], our findings show that CA inhibits cell survival and migration in both ERBB2
− and ERBB2
+ BC cells and, therefore, that its function is independent from the expression of this receptor. In addition, we have found that CA and Tz cooperate in inhibiting both migration and survival in ERBB2
+ BC cells.
One of the major effects of both CA and Tz is to inhibit cell cycle progression [
23,
47,
48]. Therefore, we have investigated cell distribution in cell cycle phases and demonstrated that the combined treatment with these two agents was more effective in blocking ERBB2
+ BC cells in G0/G1 compared to CA and Tz alone. Accordingly, the decrease of Ki-67 expression, a cell proliferation marker, further showed that in ERBB2
+ BC cells the block observed upon the Tz and CA + Tz treatments occurs mainly in G0. Noteworthy, others studies and ours have shown that CA causes a late G2 [
23] or a G2/M block [
47] in other cancer cell types. Thus, it appears that CA effects on cell cycle are cell context dependent. In ERBB2
+ BC cells, our data strongly suggest that the cooperative mechanism exploited by CA and Tz causes cell cycle arrest in G0, which involves the induction of p27
KIP1, as both drugs also cooperatively increase the levels of this CDKs inhibitor. According to this result, the observation of nuclear localization of p27
KIP1 in ERBB2
+ BC cells treated with CA + Tz is consistent with this proposed mechanism. However, we cannot rule out that the concomitant induction of p21
WAF1, which we have also observed in GBM cells [
23], may contribute, albeit to a lesser extent, to the cell cycle arrest determined by the CA + Tz treatment in these cells.
The chains of molecular events activated or repressed by CA and Tz, upstream to the induction of CDK inhibitors, remain unclear. In particular, we point to the evidence that although both ERBB2 expression levels and AKT phosphorylation levels were reduced by both drugs when used individually, we have not found a consistent combined effect on ERBB2 and AKT phosphorylation in both ERBB2
+ BC cell lines. For example, CA and Tz cooperatively reduced ERBB2 levels in BT474 (estrogen receptor positive, ER
+) but not in SKBR-3 (estrogen receptor negative, ER
−) cells. As ER− status correlates with ERBB2 overexpression and poorly differentiated tumors [
49], further studies are needed to determine if CA preferentially targets ER
+ versus ER
− BC cells, therefore providing a conceivable explanation for the differences observed in the two ERBB2
+ cell lines.
Cancer cells tend to activate autophagy for their survival. Here we have reported for the first time that CA inhibits autophagy by impairing the autophagic flux in a peculiar way that may impact cell survival of cells also exposed to Tz. Importantly, we demonstrate that CA does not affect ATG5, an E3 ubiquitin ligase necessary for the early stages of autophagosome formation, but that it increases p62 and LC3 II levels. These events likely lead to the accumulation of AL and aberrant AV/LD structures in ERBB2+ BC cells. In addition, we show that p62 induction by CA is more evident in SKBR-3 Tz-Res than in parental Tz-sensitive SKBR-3. Overall, our data demonstrate that CA disrupts the autophagic flux, thus impairing an important cancer survival pathway, which is also relevant in resistance to Tz. Notably, our data showing induction of p62 not only in ERBB2+ BC cells but also in MCF10A cells, strongly suggest that CA is a general inhibitor of autophagy.
Furthermore, we have shown that in Tz-resistant SKBR-3 cells CA and Tz cooperate in reducing ERBB2 levels along with a dramatic AKT dephosphorylation. Most importantly, we have proven a statistically significantly decrease of cell survival in Tz-resistant cells treated with CA + Tz compared to CA alone. This observation implies a partial rescue of the Tz action. It is tempting to speculate that the derangement of the endocytic/lysosomal pathway caused by the CA may impair the recycle of the ERBB2 to the plasma membrane. Certainly, further studies are needed to clarify the molecular mechanism responsible for the observed rescue of Tz activity.
As obstruction of the autophagic flux might lead to endoplasmic reticulum stress and, ultimately, apoptosis, we have consequently investigated apoptosis, necrosis, endoplasmic reticulum stress and ROS production as possible cellular processes that could impair cell survival in our ERBB2
+ BC cell models (SKBR-3 and BT474). Though none of these processes were reported to have a major role in the Tz mechanism of action, several studies have highlighted their involvement in CA-induced inhibition of cell survival in cancer cell lines [
18]. In our experimental context, CA and Tz, alone or in combination, have not elicited significant cell apoptosis or necrosis. However, the involvement of endoplasmic reticulum and ROS stress is less clear, since we have not found a consistent modulation of the HSP70 chaperone, calreticulin and catalase markers in the two ERBB2
+ BC cell lines. Indeed, although CA and Tz up-regulated calreticulin, alone or in combination in both cell types, HSP70, which is also an endoplasmic reticulum stress marker, was up-regulated in BT474 but down-regulated in SKBR-3 cells.
Finally, we have evaluated whether an aberrant differentiation process could contribute to cell cycle arrest and survival inhibition in ERBB2
+ BC cells treated with CA and Tz. An increase of neutral lipid production, associated with cell differentiation, was indeed reported to occur in SKBR-3 and in BT474 cells treated with Tz [
45]. On the other hand, contrasting data exist on CA-mediated modulation of PPARγ, which is a key transcription factor up-regulated during adipogenesis [
50,
51]. In this study, we have evaluated PPARγ protein levels, which is a major pro-adipogenic transcription factor [
52], and the mRNA levels of PLIN1 and PLIN2, two proteins associated to the coat of lipid droplets that block lipolysis and increase during lipid accumulation [
53‐
56], respectively. As we have observed that PPARγ levels were unchanged in SKBR-3 and that PLIN1 and PLIN2 mRNA levels were not modified consistently in the two ERBB2
+ BC cells co-treated with CA and Tz, we conclude that the activation of an aberrant differentiation process, possibly activated by CA and Tz, is unlikely to have a major role in cell cycle arrest caused by CA and Tz in our experimental context.
Finally, we would point that CA inhibition of cell survival, the induction of both p21
WAF1 and p62 and derangement of the endocytic/lysosomal pathway have been observed both in BC cells and in normal mammary epithelial cells. It should be noticed, however, that MCF10A cells are spontaneously immortalized mammary epithelial cells [
35] with a doubling time that is actually shorter (16 h) [
57] than that of the SKBR-3 and BT474 cells (44 and 72 h, respectively) [
58,
59]. Therefore, it is very likely that MCF10A cells have a deranged control of the cell cycle perhaps explaining why we have observed p21
WAF1 but not p27
KIP1 induction mediated by CA.
Survival analysis in MCF10A treated with CA confirms and extends our previous observation in normal human astrocytes [
32] and appears to limit a putative in vivo use of CA for cancer treatment. However, the cooperative activities of CA with Tz in ERBB2
+ BC cells and the observation that 90 d administration of rosemary extract (equivalent to 20–60 mg/Kg/d of carnosol plus CA) in rats revealed no adverse effects [
17] warrant further studies on the possible therapeutic use of CA in ERBB2
+ BCs.
Currently, the attempt to block the cellular survival functions of autophagy in cancer by combining chemotherapy with the autophagy inhibitor cloroquine and/or hydroxychloroquine is undergoing promising clinical trials in several aggressive tumors [
16]. Yet, a possible alternative strategy in fighting resistance or boosting the anticancer activity of specific antibodies is to couple the targeting of ERBB2 with phytochemical compounds with low toxicity profiles, such as CA [
17], which may interfere with therapy-induced protective autophagy as standalone drugs.