Single and double modified salinomycin analogs target stem-like cells in 2D and 3D breast cancer models

Abstract

Breast cancer remains one of the leading cancers among women. Cancer stem cells (CSCs) are tumor-initiating cells which drive progression, metastasis, and reoccurrence of the disease. CSCs are resistant to conventional chemo- and radio-therapies and their ability to survive such treatment enables tumor reestablishment. Metastasis is the main cause of mortality in women with breast cancer, thus advances in treatment will depend on therapeutic strategies targeting CSCs. Salinomycin (SAL) is a naturally occurring polyether ionophore antibiotic known for its anticancer activity towards several types of tumor cells. In the present work, a library of 17 C1-single and C1/C20-double modified SAL analogs was screened to identify compounds with improved activity against breast CSCs. Six single- and two double-modified analogs were more potent (IC₅₀ range of 1.1 ± 0.1 to 1.4 ± 0.2 μM) toward the breast cancer cell line MDA-MB-231 compared to SAL (IC₅₀ of 4.9 ± 1.6 μM). Double-modified compound 17 was found to be more efficacious than SAL against the majority of cancer cell lines in the NCI-60 Human Tumor Cell Line Panel. Compound 17 was more potent than SAL in inhibiting cell migration and cell renewal properties of MDA-MB-231 cells, as well as inducing selective loss of the CD44⁺/CD24^−/low stem-cell-like subpopulation in both monolayer (2D) and organoid (3D) culture. The present findings highlight the therapeutic potential of SAL analogs towards breast CSCs and identify select compounds that merit further study and clinical development.

Graphical abstract

1. Introduction

Breast cancer remains one of the leading cancers among women, with an estimated 2.1 million cases worldwide in 2018 [1]. Approximately 90% of breast cancer deaths are due to metastasis and the consequent impaired functioning of vital organs (e.g., lung, liver, brain), regardless of whether the cancer was metastatic at diagnosis or developed later [2][3]. Despite the widespread availability of screening mammography for early detection, the rate of metastatic breast cancer at initial diagnosis in the U.S. has remained unchanged since 1975 [4]. Despite recent progresses in understanding mechanisms leading to breast cancer recurrence, there is no known cure for metastatic breast cancer.

Recently, it has become apparent that tumors are a heterogeneous population of cell types that in many instances display a distinct cellular hierarchy. At the apex of this hierarchy are tumor-initiating cancer stem cells (CSCs) that represent a subpopulation of cells with the dual capacity of self-renewal and differentiation [5][6][7][8]. They drive initiation, progression, metastasis, and tumor reoccurrence [9][2][3]. CSCs exhibit resistance to conventional chemo- and radio-therapies and their ability to survive treatment facilitates tumor reestablishment [10]. Therefore, therapeutic strategies targeting CSCs hold great potential for novel advances in cancer treatment [11].

With more than 120 structures reported to date, polyether ionophores represent a class of natural products displaying a broad spectrum of pharmacological activities, including cytotoxic activity against various types of cancer cells [12][13][14]. Of particular interest is salinomycin (SAL, 1, Figure 1), a carboxyl ionophore with selective activity toward multidrug resistant (MDR) cancer cells and CSCs [15]. Gupta and co-workers [16] used high-throughput screening of about 16,000 bioactive molecules to show that SAL markedly and selectively reduced the viability of breast CSCs, with a potency nearly 100-times that of the conventional drug paclitaxel [16]. In addition, SAL has attracted attention as a selective inhibitor of CSCs from other cancer types, including leukemia, colorectal cancer, osteosarcoma, and uveal melanoma [17][18][19][20]. Mechanistically, many stemness-associated transcription factors, such as β-catenin, c-Myc, Snail, SOX2, and Twist1, were found to be critical for SAL-mediated CSC elimination [16][17][21][22][23][24][25][26]. Changes in phenotype composition after SAL treatment [27] appears to be due to modulation of the Hedgehog, K-Ras, and Wnt signaling pathways [18][26][28]. In the recent study by Mai et al. SAL, and it’s analog – ironmycin (AM 5), were shown to selectively target breast CSCs through accumulating and sequestering iron in lysosomes in vitro and in vivo [24]. In response to cytoplasmic iron deficiency, the SAL-treated cells induced the degradation of ferritin in the lysosomes, resulting in further iron loading in these organelles [24]. Iron-mediated generation of reactive oxygen species induced the permeability of the lysosomal membrane, triggering ferroptosis [24]. This is important, since iron endocytosis mediated by CD44 is enhanced during epithelial-mesenchymal transition, making mesenchymal cells more vulnerable to ferroptotic cell death, such as induced by SAL [5].

Figure 1.

Synthesis of salinomycin analogs

It is important to note that in limited clinical trials SAL was well tolerated by cancer patients without any long-term acute adverse effects when used at doses of 200–250 μg kg⁻¹ [25].

The chemical modification of the SAL biomolecule through the rational design and synthesis of semisynthetic derivatives has been explored in an effort to obtain compounds with superior therapeutic properties. A series of SAL analogs has been synthesized to date via synthetic manipulation of either the C1 carboxyl [29][30][31][32][33] or the C20 hydroxyl of SAL [24][34][35][36][37][38][39] which in turn has led to the synthesis of a library of C1/C20 doubly modified products [40][41][42][43][44]. Among these SAL analogs, we identified select C1 amides and esters, especially compound 4 (Figure 1), as effective anticancer agents against drug-sensitive and drug-resistant cancer cells of various origins, and demonstrated their ability to overcome MDR [31][32][33][44][45][46]. Furthermore, we have recently identified select derivatives of SAL or C20-oxosalinomycin (19) (compounds 15–18, Figure 1) as lead structures for further antibacterial and antiparasitic drug development [41][43]. Interestingly, compounds that exhibit significant activity towards bacteria or parasites are often also active on cancer cells [32][47][48]. While there have been reports of C20 singly modified SAL analogs as selective agents towards CSCs [24][34][49][50], there has not been a comprehensive study to examine the ability of a large selection of C1-single and C1/C20-double modified analogs to target breast CSCs. A library of 17 C1-single and C1/C20-double modified SAL analogs was screened against breast cancer stem-like cells using both 2D and 3D models and several compounds have been identified that merit further study and clinical development.

2. Materials and methods

2.1. Compounds

Salinomycin (SAL) derivatives were synthesized as shown in Figure 1 and characterized by Huczyński and co-workers as reported previously [31][32][41] [43] [51][52]. SAL conjugates with hydroxamic acids (compounds 15–16) were resynthesized following the protocol published by Wu and co-workers [29], while C20-oxosalinomycin (compound 19) was generated by the chemoselective oxidation of the C20 allylic hydroxyl of SAL, as previously reported by Rodriguez and co-workers [24]. The NMR data of all semisynthetic analogs of SAL were in good agreement with those found in the reference literature [24][29][31][32][41][43][51][52].

2.2. Cells and culturing conditions

Human breast adenocarcinoma cell line MDA-MB-231, human primary acute lymphoblastic leukemia ALL-5 cells [53], and human breast epithelial cell line MCF 10A, were employed in the study. MDA-MB-231 cells were cultured in DMEM/Ham’s Nutrient Mixture F12 1:1 (cat. no. 10-090-CV, Corning, Manassas, VA, USA) supplemented with 5% (v/v) heat-inactivated fetal bovine serum (FBS) (cat. no. FP-0500-A, Atlas Biologicals, Fort Collins, CO, USA) and 1% penicillin/streptomycin solution 100x (cat. no. 30-002-Cl, Corning, Manassas, VA, USA). Primary ALL-5 cells were derived from the bone marrow of a 37-year old patient and can be cultured for up to 6 months retaining their original properties, as previously described [53][54]. They were cultured in IMDM Modified (HyClone) media supplemented with 10 μg ml⁻¹ cholesterol, 6 mg ml⁻¹ human serum albumin, 2 mM L-glutamine, 2% v/v amphotericin-B/penicillin/streptomycin, 1 μg ml⁻¹ insulin, 200 μg ml⁻¹ apo-transferrin, and 50 μM β-mercaptoethanol, and were subcultured to maintain a density of 1–3 x 10⁶ cells ml⁻¹. MCF 10A cell line was cultured in Mammary Epithelial Cell Basal Medium (cat. no. C-21215, Promo Cell, Heidelberg, Germany) supplemented with Mammary Epithelial Cell Supplement Pack (cat. no. C-39110, Promo Cell, Heidelberg, Germany), 100 μg ml⁻¹ gentamicin (cat. no. G1397, Sigma-Aldrich, Deisenhofen, Germany) and 0.05 μg ml⁻¹ amphotericin B (cat. no. A2942, Sigma-Aldrich, Deisenhofen, Germany). All cells were routinely maintained at 37°C in a humidified 5% CO₂ incubator. Mycoplasma testing and cell validation were performed via short tandem repeat profiling in February 2020 by Genetica DNA Laboratories (Burlington, NC, USA). Cell lines were reported to be negative for mycoplasma contamination and verified as authentic, giving a 100% match when compared to the known reference profile [55].

2.3. Cell viability assay

In order to assess cell viability, a 3-(4,5-dimethylthiazol-2-yl)-2,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT)-based assay was employed [56][57]. Compounds were dissolved in cell culture grade DMSO (cat. no. BP231-1, Fisher). MDA-MB-231 and MCF 10A cells lines were seeded in 96-well plates (TPP, Trasadingen, Switzerland) at a density of 2 x 10³ or 2 x 10⁴ cells/well, respectively, in 100 μl of corresponding complete medium. After 24 h of incubation cells were treated with compounds at concentrations up to 10 μM for 72 h (MDA-MB-231) or 96 h (MCF 10A) with control cells receiving vehicle (0.1% DMSO) alone. After treatment, 10 μl of MTT solution (5 mg ml⁻¹) was added to each well, and the plate was incubated at 37°C for 4 h in a humidified 5% CO₂ incubator. Medium was then aspirated and 150 μl of DMSO was added to each well and the plate agitated on a shaking platform (150 rpm) for 10 min. Absorbance was read at 540 nm using a BioTek Plate Reader. Inhibition of formation of colored MTT formazan was taken as an index of cytotoxicity activity. ALL-5 cells were seeded at a density of 10⁵ cells/well in a 96-well plate and treated with different concentrations of compounds or vehicle (0.1% DMSO) for 120 h. Notably, ALL cells have a relatively long doubling time of ~60 h [53] and the assay was designed to encompass two doubling times for untreated control cells. At the end of the incubation, 10 μl of MTT solution (5 mg ml⁻¹) was added to each well, and the plate was incubated at 37°C for 24 h in a humidified 5% CO₂ incubator. Then 100 μl of 10% SDS in 0.01 M HCl was added to each well and the plate was incubated at 37°C for a further 24 h. Absorbance was recorded at 540 nm using a BioTek Plate Reader. IC₅₀ values were determined by non-linear regression analysis using GraphPad Prism 6 for Windows (GraphPad Software).

2.4. NCI-60 human tumor cell line screen

Anticancer screening assay was performed by the National Cancer Institute (NCI) employing previously published methodology of Rubinstein and co-workers [58]. Tumor cell lines were cultured in RPMI 1640 medium containing 5% fetal bovine serum and 2 mM L-glutamine. After 24 h of culture, two microtiter plates of each cell line were fixed in situ with trichloroacetic acid (TCA) to obtain readings at time-zero (Tz) prior to compound addition. Additional plates of cells were treated with compounds for 48 h and after sulforhodamine B staining absorbance was read at 515 nm to determine control growth in the presence of vehicle (C) and test growth in the presence of compound at the different concentrations (Ti). The percentage growth was calculated for every compound concentration according to the following equations: [(Ti-Tz)/(C-Tz)] x 100 for concentrations for which Ti≥Tz, or [(Ti-Tz)/Tz] x 100 for concentrations for which Ti<Tz. GI₅₀ is defined as the drug concentration resulting in a 50% decrease in the net absorbance increase during 48 h growth (C-Tz), and LC₅₀ as the drug concentration that reduces Tz, the initial absorbance, by 50%, reflecting net loss, i.e. death, of cells.

2.5. Cell cycle analysis by flow cytometry

MDA-MB-231 cells (2 x 10⁵/well for compound treatment and 10⁵/well for DMSO treatment) were seeded in 6-well plates (TPP, Trasadingen, Switzerland) and allowed to attach for 48 h in a 37°C humidified 5% CO₂ incubator. The medium was replaced and cells were treated with 0.1% DMSO or SAL (compound 1), or compounds 2, 15, 16, 17 or 18, at concentrations equal to 5 x IC₅₀ values (Table 1), for 24, 48 or 72 h. Cells were then harvested, washed with Dulbecco’s phosphate-buffer saline (DPBS) (cat. no. 21-030-CM, Corning, Manassas, VA, USA), fixed with 3 ml of 70% ice-cold ethanol and stored at 4°C prior to flow cytometric analysis. Cells were centrifuged, treated with 400 μl propidium iodide/RNase Staining buffer (BD Biosciences, San Jose, CA, USA) and stored in the dark for 1 h at room temperature (RT). DNA content was quantitated with a FacsAria IIIu Flow Cytometer (BD Biosciences, San Jose, CA, USA) and data analysed using FlowJo software.

Table 1.

Antiproliferative activity (IC₅₀) values of salinomycin (1), and its analogs.

Compound	MDA-MB-231	ALL-5	MCF 10A
	IC50 (μM)	IC50 (μM)	IC50 (μM)
SAL (1)	4.9 ± 1.6&	5.8 ± 1.7	0.1 ± 0.02
2	2.7 ± 0.1**	1.5 ± 0.5 [46]	2.9 ± 0.1
3	3.9 ± 0.4	1.9 ± 0.7 [46]	3.1 ± 0.1
4	3.6 ± 0.5	1.2 ± 0.02 [46]	2.3 ± 0.4
5	2.7 ± 0.1**	2.9 ± 0.7 [46]	1.2 ± 0.2
6	> 10	5.7 ± 0.5	2.0 ± 0.4
7	> 10	9.4 ± 0.3*	> 10
8	2.6 ± 0.5**	1.7 ± 0.5 [46]	0.9 ± 0.3
9	1.9 ± 0.5***	4.3 ± 3.1 [46]	0.2 ± 0.01
10	> 10	3.6 ± 0.4	> 10
11	> 10	1.8 ± 0.9*	> 10
12	> 10	9.7 ± 0.3*	8.7 ± 5.0
13	> 10	9.6 ± 3.0	3.7 ± 0.8
14	> 10	2.3 ± 0.4*	0.4 ± 0.1
15	1.7 ± 0.2***	1.1 ± 0.2*	0.2 ± 0.05
16	1.4 ± 0.1***	1.5 ± 0.2*	0.2 ± 0.05
17	1.1 ± 0.2***	4.8 ± 0.3	0.2 ± 0.1
18	2.8 ± 0.2**	1.0 ± 0.1*	0.2 ± 0.1

^&The value is an average from two independent experiments (n=4 each).

IC₅₀ value is defined as the concentration of the compound which induces 50% growth inhibition. Values are given as mean ± SD (n = 4)

^*p ≤ 0.05

^**p ≤ 0.01

^***p ≤ 0.001.

2.6. Colony formation assay

MDA-MB-231 cells were seeded in 6-well plates (1.5 x 10⁵/well) and allowed to attach for 48 h in a 37°C humidified 5% CO₂ incubator. After the medium was replaced, cells were treated with 0.1% DMSO or SAL (compound 1) (2.5 μM), 2 (1.4 μM), or 17 (0.5 μM) (concentrations equal to ½ IC₅₀ values) (Table 1), for 6, 24, 48 or 72 h. Cells were harvested, re-seeded at the density of 250 cells/well (n=6) in 6-well plates in compound-free medium, and allowed to proliferate for 13 days. The medium was removed, and cells were washed twice with 1 ml DPBS prior to staining with 0.5% crystal violet in methanol for 40 min at RT with agitation. Colonies of at least 25 cells were counted manually and graphed as percent relative to the vehicle treated cells.

2.7. Wound healing assay

MDA-MB-231 cells were seeded in 60 mm petri dishes (3 x 10⁵ cells/dish) and allowed to attach in a 37°C humidified 5% CO₂ incubator for 72 h resulting in a confluent monolayer of cells. After removing the medium, cells were washed with DPBS and incubated overnight in the absence of FBS. Next, the medium was removed, and cross shaped scratches (wound areas) were made in the cell monolayer with a sterile 1250 μl pipette tip. After washing twice with DPBS, cells were incubated in medium containing 0.1% DMSO or SAL (compound 1) (2.5 μM), 2 (1.4 μM) or 17 (0.5 μM) at concentrations equal to respective ½ IC₅₀ values (Table 1). The scratch area was photographed by the inverted phase contrast microscopy with an EVOS FL Auto Cell Imaging System (Thermo Fisher Scientific) directly (time 0), as well as after 2, 4, 24, 26 and 48 h. Seven measurements for each condition and time point were taken and mean ± SD was calculated. The wound area at each time point was divided with the area at 0 h to obtain the measure of wound closure. The wound area was defined as 100% at 0 h for each treatment condition.

2.8. Organoids generation and culturing

Droplets of a 1:4 mixture of complete Mammocult medium (cat. no. 05620, Stemcell Technologies) and Matrigel (cat. no. 354277, Corning), containing 10⁴ MDA-MB-231 cells/droplet, were formed at 0°C by pipetting 20 μl of the cold mixture onto a sheet of Parafilm. Droplets were solidified for 1 h in a 37°C humidified 5% CO₂ incubator, and subsequently grown in complete Mammocult medium without agitation for four days. Half of the volume of fresh complete Mammocult medium was added every other day. On the fourth day, organoids were transferred to the shaking platform at 90 rpm (Orbi-Shaker Jr, Benchmark Scientific, USA). In order to propagate and maintain the organoid cultures, when organoids reached approximately 3 mm diameter they were minced into approximately 0.5–1 mm diameter pieces using Excelta scissors (cat. no. 17-467-493, 17-456-004, 17-467-497, Fisher Scientific) to avoid necrotic cell death in the inner core [59].

2.9. Cell surface markers identified by flow cytometry

MDA-MB-231 cells were seeded in 6-well plates (0.5 x 10⁵/well) and allowed to attach for 48 h in a 37°C humidified 5% CO₂ incubator. After the medium had been replaced, cells were treated with 0.1% DMSO or SAL (compound 1) (2.5 μM), 2 (1.4 μM) or 17 (0.5 μM) at concentrations equal to respective ½ IC₅₀ values (Table 1) for 48 h. For the organoid treatment, one dissected organoid was incubated in 1 ml of complete Mammocult medium in 24-well plates and treated with vehicle or test compound as described above for 48 h. After incubation cells were harvested with enzyme-free cell dissociation buffer (cat. no. 13151014, Gibco), and organoids with enzyme-free cell dissociation buffer containing 0.2% of anti-clumping agent, 2-naphtol-6,8-disulfonic acid dipotassium salt (NDA) (cat. no. 439013, Frontier Scientific) and dispersed by gentle pipetting with a 25 ml tip. Cells were then washed with flow cytometry buffer, and re-suspended in flow cytometry buffer containing FITC conjugated anti-CD24 (cat. no. ab30350, Abcam) 1:100 and PE conjugated anti-CD44 (cat. no. ab46793, Abcam) 1:200, and incubated for 30 min in the dark. Cells were washed twice with flow cytometry buffer and re-suspended in 300 μl propidium iodide/RNase Staining buffer and subjected to a FacsAria IIIu Flow Cytometer. Data were analysed using FlowJo software. The threshold lines were set according to the isotype control. The accuracy of the triple immunostaining was confirmed by the comparison with single immunostaining for CD44, CD24 and PI respectively.

2.10. Combination assay

In order to determine whether the activity of compound 17 was attributed to the molecule itself, or to the possible synergistic effect between its constituent components, C20-oxosalinomycin (19) and benzhydroxamic acid potentially released during enzymatic hydrolysis, MDA-MB-231 cells were treated with 19 at ¼, ½, 1, 2 or 4 x IC₅₀ concentrations in the absence or presence of benzhydroxamic acid at the following range of concentrations: 35.5 μM, 71 μM, 142 μM, 284 μM or 568 μM (where IC₅₀ value is 142 μM). The experiment was performed in triplicate and standard MTT viability assays were conducted as described above. Results were evaluated quantitatively by the median effect principle of Chou-Talalay [60]. This method uses the IC₅₀ values for individual compounds and the IC₅₀ for their combination to calculate a Combination Index, CI, where CI = D1/Dx1 + D2/Dx2; and where Dx1 = IC₅₀ of compound 1, Dx2 = IC₅₀ of compound 2, D1 = IC₅₀ of compound 1 in the presence of compound 2 at a concentration of D2, D2 = concentration of compound 2.

2.11. Immunoblot analysis

MDA-MB-231 cells (2.2 x 10⁶ cells/100 mm petri dish) were treated for 24 or 48 h with compound 1, 2 or 17 at 5 x IC₅₀ values concentration (see Table 1 for respective IC₅₀ values) or with vehicle (0.1% DMSO) in the absence or presence of 100 μM Z-VAD-FMK (cat. No. A1902, APExBIO). Cells were washed with PBS, released with 0.05% trypsin (cat. no. 25-052-CI, Corning), centrifuged and washed with PBS. Cell pellets were snap frozen in liquid nitrogen and stored in −80°C until processing. MDA-MB-231 organoids were generated as described in 2.8. Organoids generation and culturing. Two cut organoids/well were cultured in 24-well plates and treated for 24 or 48 h with compound 1, 2 or 17 at 5 x IC₅₀ values concentration or with vehicle (0.1% v/v DMSO). Organoids were washed with PBS, incubated with enzyme-free cell dissociation buffer with 0.2% (v/v) NDA and dispersed by gentle pipetting with a 1250 μl tip. Cells released from organoids were washed and frozen as described above. Cells were further thawed and lysed in lysis buffer (25 mM HEPES, pH 7.5, 300 mM NaCl, 0.1% w/v Triton X-100, 1.5 mM MgCl₂, 0.2 mM EDTA, 0.5 mM DTT, EDTA-free complete protease inhibitor tablets (Roche), 20 μg ml⁻¹ aprotinin, 50 μg ml⁻¹ leupeptin, 10 μM pepstatin, 1 mM phenylmethylsulfonyl fluoride, 20 mM β-glycerophosphate, 1 mM Na₃VO₄, and 1 μM okadaic acid). Protein content was measured by Bradford assay and equal amounts (20 μg) were separated by electrophoresis using 12% (w/v) acrylamide Mini-PROTEAN^® precast gels (Bio-Rad). Proteins were electrophoretically transferred onto a PVDF membrane (Immobilon-FL, Merck Millipore) and next stained with Ponceau S to assess transfer efficiency and verify equal loading. The membrane was blocked with 5% (w/v) non-fat milk in Tris-buffered saline containing 0.1% (w/v) TWEEN-20 (TBS-T) for 1 h at RT and incubated overnight at 4°C with primary antibodies against PARP (cat. no. 9532, Cell Signaling Technology) (1:1000 dilution), Bcl-2 (cat. no. sc-509, Santa Cruz Biotechnology) (1:200 dilution), Mcl-1 (cat. no. sc-12756, Santa Cruz Biotechnology) (1:200 dilution), Bak (cat no. 12105, Cell Signaling Technology) (1:1000 dilution), LC3B (cat. no. 2775, Cell Signaling Technology) (1:1000 dilution), and GAPDH (cat. no. 2118, Cell Signaling Technology) (1:10000 dilution). After washing with TBS-T for 5 x 5 min the membrane was incubated with secondary HRP-conjugated goat anti-rabbit IgG (H+L) antibody (1:5000 dilution) (cat. no. 170-6515, Bio-Rad) or goat anti-mouse (cat. no. 1706516, Bio-Rad) for 1 h at RT. After washing in TBS-T the membrane was exposed to ClarityTM Western ECL Substrate luminol enhancer solution and peroxide solution (Bio-Rad) for 5 min and visualized and quantified using Image J software. Uncropped immunoblots are shown in Supplemental Figures S1-S3.

2.12. qRT-PCR analysis

MDA-MB-231 cells (0.7–2.2 x 10⁶ cells/100mm petri dish) were treated for 24 h with compound 1, or 17 at ½ IC₅₀ values concentration (see Table 1 for respective IC₅₀ values) or with vehicle (0.1% v/v DMSO). Cells were washed in PBS, released with 0.05% trypsin (cat. no. 25-052-CI, Corning), spun down and washed with PBS. Cell pellets were snap frozen in liquid nitrogen and stored in −80°C until processing. Total RNA was isolated with RNeasy Mini Kit (cat no. 74104, Qiagen). Genomic DNA removal and reverse transcription ware performed with iScript gDNA Clear cDNA Synthesis Kit (cat no. 1725035, Rio-Rad). The level of specific transcripts was quantitated by qPCR using TaqMan assays: CDH1 (Hs01023895_m1), SNAIL-2 (Hs00161904_m1), Vimentin (Hs00418522_m1), Fibronectin (Hs01549976_m1), ZEB1 (Hs01566408_m1), 18s (Hs99999901_s1), GAPDH (Hs02786624_g1). The mRNA transcript levels were calculated relative to that of GAPDH (endogenous control) using the ΔΔCt method.

2.13. Statistical analysis

Unpaired t test with Welch's correction was performed for the significance and p values of <0.05 were considered significant. For the analysis of large data sets with multiple replicates (cell cycle analysis and colony formation assay) one-way ANOVA was employed. Data are presented as a mean ± SD. GraphPad Prism 9 for Windows (GraphPad Software) was employed for statistical analysis.

3. Results

3.1. Analog design and synthesis

The synthesis of salinomycin (SAL) analogs studied in this work, both C1 singly (compounds 2–9 and 15–16, Figure 1) and C1/C20 doubly modified products (compounds 10–14 and 17–18, Figure 1), is described thoroughly elsewhere [24][29][31][32][41][43][51][52]. Briefly, a library of SAL analogs (Figure 1) was devised based on the most bioactive compounds obtained in our previous studies; it included SAL C1 esters (compounds 2, 4–6, 8–9 and 15–16), C1 amides (compounds 3 and 7) as well as products that were obtained through the double modification of the SAL molecule, i.e. C1 esters/amides of SAL with the chemoselectively oxidized C20 hydroxyl to the corresponding ketone group (compounds 10–14 and 17–18). For better structure-activity relationship (SAR) studies, we selected for SAL derivatives that contained identical ester/amide moieties at the C1 position, such as propargyl ester 6 and amide 7, together with their C20-oxo counterparts (compounds 12–13). The spectroscopic data of all resynthesized SAL derivatives were in line with previously published data [24][29][31][32][41][43][51][52].

3.2. Activity of salinomycin and its analogs in cell viability screening

Most commonly, cancer stem cells (CSCs) are identified according to the expression of characteristic cell-surface markers, which vary in different subtypes of breast cancer. High expression of CD44 and low expression of CD24 (CD44⁺/CD24^−/low) in breast cancer is associated with cell proliferation and tumorigenesis [61][62]. The human triple negative basal-like MDA-MB-231 cell line expressing CD44⁺/CD24^−/low is one of the well-established in vitro models for studying breast CSCs [61][63].

It has been previously shown in JIMT-1 breast cancer cell line that SAL has the highest selective activity towards stem cells at ~IC₂₅ [66]. For this reason, and to investigate changes in stem-like properties without inducing significant cell death, all of the subsequent stem cell assays has been performed in concentrations equal to ½ of IC₅₀ (Table 1) of SAL and select analogs.

SAL and its ten single- and seven double-modified analogs were evaluated for their in vitro antiproliferative effect on primary ALL-5, human MDA-MB-231 and mammary epithelial MCF 10A cells. The data presented as IC₅₀ ± SD have been shown in Table 1 and the associated viability curves in Supplemental Figure S4.

In our previous study we identified six single-modified esters and amides of SAL potent against primary ALL-5 cells [46]. Here we expanded the library of investigated compounds and determined two additional single- (15 and 16) (IC₅₀ = 1.1 ± 0.2 – 1.5 ± 0.2 μM) and three double-modified (11, 14, and 18) (IC₅₀ = 1.0 ± 0.1 – 2.3 ± 0.4 μM) analogs of SAL (1) (IC₅₀ = 5.8 ± 1.7 μM) more active (indicated by lower IC₅₀ values) than parent molecule towards those cells (Table 1).

As SAL was previously shown to be a potent inhibitor of breast CSCs, it was of interest to investigate the activity of its analogs against MDA-MB-231 cell line which, due to the high ratio of CD24^low/CD44^high and expression of ALDH⁺, is an established model for studying breast CSCs [61]. Since SAL was employed as a positive control, we recorded two separate IC₅₀ values of 4.0 and 5.8 μM. In Table 1 we present the average from those measurements 4.9 ± 1.6 μM, which is in agreement with previously published data [29][67]. In the present work, we identified six single- (2, 5, 8, 9, 15, and 16) and two double-modified (17 and 18) analogs of SAL (1) more potent towards MDA-MB-231 cells than parent compound 1 (Table 1). The most active analogs towards both primary ALL-5 and MDA-MB-231 cells were conjugates of SAL with hydroxamic acids, making this modification particularly interesting for further studies.

The IC₅₀ value obtained for the parental SAL towards the non-cancerous, immortalized MCF 10A cell line was 0.1 ± 0.02 μM and stood in agreement with previously published data (Table 1) [42][68]. The MCF 10A IC₅₀ values recorded for single and double-modified SAL analogs were higher than for the unmodified scaffold and formed in the range of 0.2 ± 0.05 – 3.1 ± 0.1 μM for single- and 0.2 ± 0.1 – 8.7 ± 5.0 μM for double-modified compounds (Table 1).

3.3. In vitro growth inhibition against the panel of NCI-60 cancer cell lines

Taking into consideration the results from our preliminary cell viability assay (Table 1) and structural diversity, we identified seven analogs, 10, 11, 12, 13, 14, 17 and 18, for further screening in the panel of NCI-60 cancer cell lines (Supplemental Table S1). Compounds 10, 11, 12, 13, 14, 17 and 18 were tested at a single concentration of 10⁻⁵ M on the panel of 60 human tumor cell lines including leukemia, non-small cell lung, colon, central nervous system (CNS), melanoma, ovarian, renal, prostate, and breast cancer. Data are summarized in Supplemental Table S1 and represent percent of growth of the compound treated cells relative to the control, and relative to “time zero” cell count. The consequent growth percent is employed to calculate growth inhibition (values between 0 and 100) and lethality (values <0) (Supplemental Table S1). Compounds which exhibited at least 60% or more growth inhibitory property in at least eight cell lines during this single concentration screening were directed for the assessment against the same panel in 5-dose concentration range (for the details see 2.4. NCI-60 human tumor cell line screen).

After the initial screening two compounds, 13 and 17, were selected for further testing in the concentration range of 10⁻⁴ to 10⁻⁸ M (Supplemental Table S2). Dose-response curves were generated for each cell line by plotting cytotoxic activity against the log₁₀ of the compound concentration. Cytotoxic effect was determined as GI₅₀ value for each compound, which represents the concentration inducing 50% of growth inhibition. Due to its potent activity compound 17 was selected for the repeat of 5-dose testing (Supplemental Table S2). As presented in Table S2 compound 17 was more potent than parent SAL (1) towards the majority of cell lines included in the screening. This was especially true to panels of leukemia, colon, CNS, ovarian, renal and the entire panel of breast cancer (Supplemental Table S2).

3.4. Salinomycin and its analogs induce DNA fragmentation in the MDA-MB-231 cell line

In order to further examine the mechanism behind the promising activity of SAL and its analogs against the MDA-MB-231 cell line, we measured DNA content and DNA fragmentation via flow cytometry. Briefly, propidium iodide (PI) staining was employed to evaluate DNA content and cells with sub-G1 (<2N) DNA were scored as dead. MDA-MB-231 cells were treated with parent SAL (1), or its select and the most potent analogs (characterized by the lowest IC₅₀ values, see Table 1), 2, 15, 16, 17 and 18, each at concentrations equal to 5 x IC₅₀ values (Table 1) for 24, 48 or 72 h. MDA-MB-231 cells treated with 0.1% DMSO at an equivalent durations served as a control. The full set of representative cytograms is presented in Figure 2A. A graphical summary of cells in the different phases of the cell cycle as a mean from three independent experiments is shown in Figure 2B. Statistically significant increases (p ≤ 0.0001) in sub-G1 DNA content control vs. treatment were observed after the treatment with all of the compounds for 72 h. However, compounds 2, 17 and 18 induced DNA fragmentation as early as 24 h after treatment. Since the cytotoxic activity of parent SAL was tested in two independent cell viability experiments (see 3.2.), MDA-MB-231 cells were treated with two separate doses of at 20 and 30 μM (corresponding to 5 x IC₅₀ values). Only 30 μM of SAL (1) was able to induce statistically significant sub-G1 DNA content after 24 h. Importantly, since the concentrations were adjusted to five times of the respective IC₅₀ values, SAL analogs were able to induce a statistically significant amount of DNA fragmentation after 24 h in approximately two to five times lower doses than SAL (1). The findings from this experiment and the data from Table 1, prompted us to select compound 2 as the most promising representative of single-modified analogs and compound 17 as the best candidate from double-modified agents for further studies.

Figure 2.

Salinomycin analogs induced DNA fragmentation in MDA-MB-231 cells. MDA-MB-231 cells were treated with 0.1% DMSO (control), 20 μM (indicated by grey circle) or 30 μM (indicated by black circle) of salinomycin (1) or specified analogs at 5 x IC₅₀ values for 24, 48 or 72 h and subjected to propidium iodide staining and flow cytometry. A. Representative cytograms; B. Distribution of cells observed in different phases of the cell cycle or with Sub-G1 DNA content determined by PI staining. Data represent mean ± SD (n = 3) *p ≤ 0.05, **p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001.

3.5. Salinomycin and its analogs reduce colony formation potential of MDA-MB-231 cell line

The colony formation capacity of MDA-MB-231 cells was initially found to be significantly reduced by the treatment with compounds 2 (55%) and 17 (55%) as compared with both control (100%) and parent SAL (1) (92%) after 24 h (Figure 3A and B). In order to measure the effect on colony formation without inducing cell death, the dose of each compound was adjusted to 50% of the respective IC₅₀ value (see Table 1) and colony formation was assessed relative to vehicle-treated controls. Compounds 2 and 17 inhibited colony growth by ~50% after 24 h, but this was not seen for parent SAL (1). After 48 h of treatment SAL (1) dramatically inhibited colony growth by 84% (the same value was noted for compound 17), while compound 2 still had ~50% inhibitory effect (Figure 3A and B). 72 h of treatment lead to further decreases as follows: SAL (1) (90% inhibition); compound 17 (96% inhibition), and compound 2 (68% inhibition) (Figure 3A and B). Of note, the slight decrease in the number of colonies formed under control conditions after 72 h vs. earlier treatments may be related to the increased cell confluence and decreased amount of nutrients upon prolonged culturing. Based on these results compound 17 was shown to act more rapidly and was more potent than SAL or 2.

Figure 3.

Effect of salinomycin (1) and analogs 2 and 17 on colony formation potential of MDA-MB-231 cells. MDA-MB-231 cells were treated with 0.1% DMSO (control), 1 (2.5 μM), 2 (1.4 μM) or 17 (0.5 μM) for 6, 24, 48 and 72 h. At the end of the incubation, cells were harvested and re-seeded in 6-well plates at 250 cells/well. After 13 days, cells were fixed and stained with 0.5% crystal violet methanol solution. Colonies, defined by group of more than 50 cells, were counted manually for each well and normalized to the control. A. Representative images of colonies formed; B. Colonies quantification. Data represent mean values ± SD (n = 6). *p ≤ 0.05, **p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001.

3.6. Salinomycin and its analogs attenuate MDA-MB-231 cell migration

A wound healing assay was performed to assess the effect of our selected compounds on the migration of MDA-MB-231 cells. The wound was created on a confluent cell monolayer and cells were allowed to migrate for 48 h in the absence or presence of the compound of interest (Figure 4A). Compound 17 significantly reduced the migration of MDA-MB-231 cells as early as after 24 h (83% vs. 26% of wound surface area treatment vs. control respectively) (Figure 4B). Statistically significant inhibition of migratory properties of MDA-MB-231 cells was observed for treatment with SAL (1) and its two analogs: compounds 2 and 17 after 48 h, with compound 17 showing the biggest inhibitory effect (80% vs. 0% of wound surface area treatment vs. control, respectively) (Figure 4B). Due to the differences in IC₅₀ values (Table 1), it should be noted that compound 17 was used in this assay at a five-fold lower concentration then SAL (1). Taken together, these data suggest a superior anti-migratory effect of compound 17 over the parental SAL (1).

Figure 4.

Wound healing assay of MDA-MB-231 cells after treatment with 0.1% DMSO (control), 2.5 μM 1, 1.4 μM 2 or 0.5 μM 17 for 48 h. Wounds were created on a confluent cell monolayer and images were acquired at 0, 4, 24, and 48 h after wounding. The experiment was performed in duplicate. A. Representative images of wound healing; B. Quantification of wound healing. Data represents mean ± SD (n = 2), *p ≤ 0.05.

3.7. Salinomycin analogs reduce the CD44⁺/CD24⁻ stem-like population in MDA-MB-231 cell monolayer and organoid cultures

We subsequently investigated the effect of treatment with SAL (1) and compounds 2 and 17 on CD44⁺/CD24^−/low stem cell-like populations within the MDA-MB-231 cell line (Figure 5A). The quantification of CD44⁺/CD24^−/low populations relative to the control is shown in Figure 5B. All of the compounds induced small, but significant reductions in the percentage of viable CD44⁺/CD24^−/low cells, with a corresponding increase in number of cells with the CD44⁺/CD24⁺ phenotype (Figure 5A). Of these, compound 17 had a 2-fold greater effect than parent SAL (1) in reducing the stem cell-like population but at 1/5^th the concentration.

Figure 5.

Analysis of the expression of CD44^high/CD24^low stem-like populations in MDA-MB-231 cells and organoids. A. MDA-MB-231 cells were triple stained with anti-CD44-PE, anti-CD24-FITC and propidium iodide (PI) and viable PI negative cells are shown in the representative cytograms with CD24 on y axis and CD44 on x axis; B. The ratio of CD44^high/CD24^low MDA-MB-231 cells after treatment with 0.1% DMSO (control), 2.5 μM 1, 1.4 μM 2 or 0.5 μM 17 for 48 h. Bar graph was prepared from the percentage of viable CD44^high/CD24^low MDA-MB-231 cells in the flow cytometry analysis. Data represents means ± SD (n = 3), **p ≤ 0.01, *** p ≤ 0.001; C. Schematic representation of organoids acquisition and treatment. Scale bar, 2 mm; D. Organoids dispersed to single-cells were triple stained with anti-CD44-PE, anti-CD24-FITC and PI and viable PI negative cells are shown in the representative cytograms with CD24 on y axis and CD44 on x axis; E. The ratio of CD44^high/CD24^low organoid cells after treatment with 0.1% DMSO (control), 2.5 μM 1, or 0.5 μM 17 for 48 h. Bar graph was prepared from the percentage of viable CD44^high/CD24^low organoid cells in the flow cytometry. Data represents means ± SD (n =3), *p ≤ 0.05,*** p ≤ 0.001.

In order to overcome the limitations associated with utilization of cell monolayers and to more appropriately model the 3-dimensional (3D) nature of tumors [69], we employed organoid culture. The culturing of organoids and treatment scheme is shown in Figure 5C (see also 2.8. Organoids generation and culturing). Representative cytograms of CD24/CD44 surface marker distribution of untreated (control), SAL (1) and compound 17 treated organoids are shown in Figure 5D. As evidenced in Figure 5E, which represents quantification of the percentage of viable CD44⁺/CD24^−/low cells from treated organoids, both SAL (1) and compound 17 induced a modest, but significant reduction in the percentage of CD44⁺/CD24^−/low cells. Similarly, as in 2D culture treatment (Figure 5A), an increase in the amount of cells with CD44⁺/CD24⁺ phenotype can be observed (Figure 5D). We conclude that compound 17 was superior over SAL in reducing CD44⁺/CD24^−/low stem-like sub-population of MDA-MB-231 cells in monolayer culture and retained this activity in 3D organoid culture.

3.8. C20-oxosalinomycin shows antagonistic effect in combination with benzhydroxamic acid

In order to determine whether the activity of compound 17 was attributed to the molecule itself or the two individual synthetic precursors released during possible enzymatic hydrolysis in vitro, we assessed IC₅₀ values in cell viability assay for C20-oxosalinomycin (19) and benzhydroxamic acid. The IC₅₀ values found were 12.1 and 142.0 μM respectively, which corresponds to 11 and 129 X the IC₅₀ value of compound 17 (Table 2), and stays in agreement with our previous finding [42]. To investigate the possibility of a synergistic effect between those two compounds leading to the decreased IC₅₀ value seen with compound 17, we performed a Chou-Talalay combination assay. The results are shown in Table 2, and as previously described for a pair of drugs, CI > 1.3 indicates antagonism, CI = 1.1–1.3 indicates moderate antagonism, CI = 0.9 to 1.1 indicates an additive effect, CI = 0.8 to 0.9 indicates slight synergism, CI = 0.6 to 0.8 indicates moderate synergism, CI = 0.4 to 0.6 indicates synergism, and CI = 0.2 to 0.4 indicates strong synergism [60][70][46][71]. We found both compounds to have antagonistic effect, resulting in the increase of IC₅₀ value (Table 2), therefore proving that the activity of compound 17 is attributed to the molecule itself rather than the unbounded components.

Table 2.

Combination Index (CI) and IC₅₀ values of C20-oxosalinomycin (19) in combination treatment with benzhydroxamic acid against MDA-MB-231 cells.

Compound	Benzhydroxamic acid, μM	IC50, μM	CI	Effect
19	-	12.1	-	-
19	35.5	18.3	1.7	Antagonism
19	71.0	14.8	1.5	Antagonism
19	142.0	16.0	1.9	Antagonism
19	284.0	17.3	2.7	Antagonism

3.9. Salinomycin and its analogs induce features of apoptotic cell death in MDA-MB-231 cells and organoids

To examine the apoptotic effect of compounds 2 and 17, as well as the parental SAL (1), on MDA-MB-231 cells in adherent monolayer or in 3D organoid cultures we investigated the levels of PARP, Bak, Mcl-1 and Bcl-2 via immunoblotting. With the exception of analog 2, all of the compounds induced significant decrease of total PARP in cell monolayer after 48 h treatment as compared to the untreated control (Figure 6A and C). Interestingly, in organoid culture this effect was observed as early as after 24 h, and loss of total PARP further increased after 48 h in SAL (1) and compound 17 treatment (Figure 6B and D). The level of anti-apoptotic protein Bcl-2 was slightly reduced in MDA-MB-231 cells after 48 h of treatment with all of the compounds studied as compared to the control (Figure 6E). Similarly, in the organoid culture significant Bcl-2 degradation was observed after 48 h treatment with all of the compounds, with more robust effect after 24 h of incubation with SAL (1) and analog 17 (Figure 6F). Examination of another anti-apoptotic protein Mcl-1 in MDA-MB-231 monolayer revealed a significant increase in its level after 48h, as compared to the untreated control, with an even more robust increase induced by analog 17 (Figure 6I). A contrasting effect was observed in 3D organoid culture, where Mcl-1 protein was significantly downregulated after 24 h treatment with all of the compounds, to further decrease after 48 h (except analog 2) (Figure 6J). Investigation of the expression of pro-apoptotic protein Bak revealed its significant increase vs. control after treatment of MDA-MB-231 cells with analog 17 after 24 h and 48 h as well as a very modest, but statistically significant, increase in Bak with SAL (1) (Figure 6G). In organoid culture, treatment with all of the compounds significantly increased expression of Bak after 24 h and 48 h (Figure 6H).

Fig. 6.

A-B. Effect of compounds 1, 2 or 17 on PARP, Mcl-1, Bcl-2 and Bak expression. MDA-MB-231 cells or organoids were treated with 0.1% DMSO (control), or 25 μM 1, 14 μM 2, or 5 μM 17 for the times indicated, and extracts were prepared and subjected to immunoblotting for PARP, Mcl-1, Bcl-2 and Bak. GAPDH was employed as a loading control; .C-J Bar diagrams showing the fold changes of proteins normalized to GADPH. Images were quantified by measuring the band intensity using ImageJ software. Data represent mean ± S.D (n = 3). *p ≤ 0.05, **p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001.

In order to determine whether loss of Bcl-2 expression was the event leading to cell death, or its consequence, MDA-MB-231 cells were incubated with SAL (1) or analog 17 in the absence or presence of caspase-3 inhibitor – Z-VAD. Z-VAD was previously internally verified to be a caspase-3 inhibitor (M. Delgado, R.R. Rainwater, A. Urbaniak, K. Butler, M. Davidson, R.M. Protacio, B. Heflin, G. Baldini, A. Edwards, K.D. Raney, and T.C. Chambers, Microtubule depolymerization in primary acute lymphoblastic leukemia cells induces death in G1 and M phases which occurs through distinct cell death pathways, manuscript in preparation). Treatment with SAL (1) or its analog 17 in the presence of Z-VAD significantly inhibited loss of Bcl-2 protein as compared to incubation with compounds alone and untreated control (Figure 7A and C). Significant upregulation of Bcl-2 expression in cells incubated with Z-VAD alone vs. untreated controls suggests a decrease of physiologically occurring cell death, due to the inhibition of caspase-3 (Figure 7C). PARP was used as a control, and as expected, loss of its expression was protected by Z-VAD (Figure 7A and B).

Figure 7.

A. Inhibition of the degradation of PARP and Bcl-2 in the presence of caspase-3 inhibitor (Z-VAD). MDA-MB-231 cells were treated with 0.1% DMSO (control), 25 μM 1, or 5 μM 17 in the absence or presence of 100 μM Z-VAD for 48 h, and extracts were prepared and subjected to immunoblotting for PARP or Bcl-2. GAPDH was employed as a loading control; B. Bar diagram showing the fold changes of proteins normalized to GADPH. Images were quantified by measuring the band intensity using ImageJ software. Data represented as mean ± S.D. of three independent determinations (n = 3). **p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001.

We also examined the effect of the treatment with SAL (1), and its analogs, compounds 2 and 17 on the expression of autophagy pathway associated marker LC3 in monolayer and organoid cultures. There was a significant upregulation of LC3-II (14 kDa) expression after incubation of cell monolayers with SAL (1) and compound 17 for 24 h as compared to the untreated control, with further significant increase after 48 h of treatment (Figure 8A and C). Similarly treatment of organoids for 24 h significantly induced the expression of LC3-II (14 kDa) as compared to the control (Figure 8B and D). By contrast, compound 2 only induced a very modest increase in LC3-II in adherent and organoid culture.

Figure 8.

A. Effect of compounds on LC3B expression in MDA-MB-231 cells and organoids. Cells and organoids were treated with control , 25 μM 1 , 14 μM 2 , or 5 μM 17 for the times indicated. Extracts were prepared and subjected to immunoblotting for LC3B (14 kDa). GAPDH was employed as a loading control; B. Bar diagram showing the fold changes of proteins normalized to GADPH. Images were quantified by measuring the band intensity using ImageJ software. Data represented as mean ± S.D. of three independent determinations (n = 3). *p ≤ 0.05, **p ≤ 0.01.

Taken together, we conclude that both SAL (1) and compound 17 induced apoptosis in a caspase-3 dependent manner, as well as upregulation of the expression of autophagy pathway associated marker LC3II, in both monolayer and 3D organoid culture.

3.10. Treatment with salinomycin and its analogs increased the expression of mesenchymal markers and decreased the expression of E-cadherin

We next examined the effect of SAL (1) and compound 17 on epithelial to mesenchymal transition (EMT) associated markers (E-cadherin, Fibronectin, Vimentin, Snail-2 and Zeb-1) in cell monolayer. Compared to untreated controls, E-cadherin was downregulated by SAL (1) and no significant changes in the expression of this gene were observed after treatment with compound 17 (Figure 9A). No statistically significant changes in control vs. treatment level of Fibronectin were observed (Figure 9B). Vimentin was upregulated by both SAL (1) and compound 17, with a bigger effect of the latter (Figure 9C). Zeb-1 was upregulated only by compound 17 (Figure 9D) while Snail-2 was upregulated only by SAL (1) (Figure 9E). We conclude that, other than a common upregulation of vimentin, treatment with SAL (1) and compound 17 had differential effects on EMT marker gene expression.

Figure 9.

Effect of compounds on EMT markers in MDA-MB-231 cell line. (A–E) Effect of 24 h treatment with salinomycin (1) (2.5 μM), or 17 (0.5 μM) on E-cadherin, Fibronectin, Vimentin, Zeb-1 and Snail-1 mRNA abundance. Bar graphs represented as mean ± S.D. of three independent determinations (n = 3). *p ≤ 0.05, **p ≤ 0.01.

4. Discussion

We and others demonstrated that the modification of the C1 carboxyl alone or in combination with derivatization of the C20 hydroxyl may improve the biological activity of the resulting semisynthetic derivatives of salinomycin (SAL) [29][32][41][42][43][44][45][46][72]. For instance, SAL C1 esters and amides, especially 3 and 4, were highly selective agents in vitro and met the activity criteria for hit compounds for parasites that are responsible for African trypanosomiasis in humans (sleeping sickness) and animals (nagana disease) [72]. Recently, conjugates of SAL and C20-oxosalinomycin (19) with benz- and salicylhydroxamic acid (compounds 15–18) were found to exhibit excellent inhibitory activity against planktonic Gram(+) bacteria and methicillin-resistant staphylococci, but also towards biofilm-forming bacterial strains [41]. As far as activity against cancer cells, certain C1 singly and C1/C20 doubly modified analogs of SAL were shown to possess selective anticancer effects towards a panel of human cancer cell lines, and the potential to overcome MDR at micromolar concentrations [29][32][42][44][45][46]. However, the ability of this class of compounds to target CSCs remained unclear.

In the current study, we investigated the anti-breast cancer stem-like cell activity of select C1 singly and C1/C20 doubly modified SAL analogs using several independent CSC assays. Cancer cell renewal and migration have a great impact on tumor metastasis. SAL was previously shown to affect self-renewal of MDA-MB-231 [73] and HCC1937 [63] breast cancer cell lines. Additionally, SAL inhibited cell migration in MDA-MB-231 [74] [73] [75], MDA-MB-436 [76], JIMT-1 [50] and MCF-7 [74] breast cancer cell lines. Herein, we have found a double-modified analog of SAL, compound 17 to be more potent than the parent SAL molecule in colony formation and wound healing assays, by inducing significant inhibition of self-renewal and migration faster (as early as 24 h) at only one fifth of the concentration of SAL (Figures 3 and 4).

Current models for studying CSCs typically include using mice in vivo which has inherent issues because adaptation to the mouse can alter the fundamental properties of human CSCs. The use of 3D tumor organoids derived from human cancer cells has provided a robust alternative experimental system to study CSC function and responses to drug treatment [69]. Organoids are stem-cell derived structures that self-organize and as such more closely resemble the tissue or tumor of origin [69]. Additionally, they offer advantages for pre-in vivo testing over mammosphere assays, including maintenance of cell-cell interactions, heterogeneity, microenvironment, morphology, histopathology, receptor status and drug response of the tumor sample they originated from [69][77][78]. Therefore, we employed a breast cancer organoid model to validate anti-breast cancer stem-like activity of our experimental compounds.

SAL was previously shown to reduce the CD44⁺/CD24^−/low stem-like populations in breast cancer cells including HMLER [16][24] and JIMT-1 [66], as well as in MDA-MB-231 mammospheres [75]. Here we report that SAL (1) and its analog 17 significantly reduced the CD44⁺/CD24^−/low stem-like population in both adherent (monolayer) and 3D organoids derived from MDA-MB-231 cells (Figure 5). Indeed, analog 17 was considerably more potent then parent SAL (1) in this regard. The compensatory increase in the non-stem CD44⁺/CD24⁺ population may be explained by recent findings of Kamlund and co-workers, where treatment of JIMT-1 cells with 0.5 μM of SAL resulted in selective inhibition of the CD24⁻ subpopulation, consequently increasing the CD24⁺ population [79].

We were concerned whether the effects of compound 17 was a consequence of possible metabolic breakdown after enzymatic hydrolysis in vitro to its two synthetic precursors: 19 and benzhydroxamic acid. The Chau Talalay method was employed and it was found that the two compounds used simultaneously act in an antagonistic fashion, leading to the conclusion that the activity of analog 17 is attributed to this novel scaffold itself and not its breakdown products (Table 2).

Evasion of apoptosis is one of the major features by which a cancer cell population expands [80]. SAL was shown to induce an apoptotic cascade in various types of cancers including breast, hepatic, lung, prostate, leukemia and ovarian [81]. We therefore investigated whether SAL and its analogs 2 and 17 induced apoptosis in MDA-MB-231 cells and organoids by analyzing changes in apoptosis-related proteins such as PARP, Bcl-2, Mcl-1 and Bak (Figure 6). Our study shown that SAL (1) and its analogs not only down-regulated the expression of anti-apoptotic Bcl-2 and Mcl-1 (in organoid) and enhanced the expression of pro-apoptotic Bak but also increased the degradation of PARP in both 2D and 3D culture, which is consistent with induction of apoptosis. Interestingly, contrasting changes in Mcl-1 levels occurred between 2D and 3D cultures as a consequence of compound treatment. The increase of Mcl-1 in cell monolayers after treatment with SAL and its analogs may indicate the induction of initial defense mechanisms protecting cells from the toxic effects of these compounds. Such a protective effect is not evident in the 3D organoid model.

Autophagy is a self-degradable catabolic mechanism related to the breakdown of unnecessary or dysfunctional components [82]. It is also considered a survival mechanism to counteract stressors such as starvation, endoplasmic reticulum (ER) stress, hypoxia or infection [83]. Autophagy may participate in cell death when apoptotic processes are compromised or when cells are subjected to severe stressors [84]. Several studies have highlighted the role of autophagy in chemoresistant tumors, and provided an insight into the crosstalk between apoptosis and autophagy [85][86]. During starvation, cancer cells undergo autophagy, which may delay activation of the apoptotic pathway [87]. Despite its importance in cancer therapy, molecular pathways leading to autophagy are not well understood and precise therapeutic strategies to control autophagy have not yet been developed. SAL was previously shown to induce activation of autophagy in human cancer cells [88][89]. Therefore, it was of interest to investigate the status of the autophagy-associated marker LC3II after treatment of MDA-MB-231 cells and organoids with SAL (1) and its analogs, and this was found to be upregulated after treatment with SAL (1) and compound 17 in 2D and 3D cultures (Figure 8).

Breast CSCs have been shown to possess increased metastatic capacity that correlates with an ongoing EMT to promote cell migration [90]. SAL has been previously shown to upregulate epithelial marker E-cadherin in the MDA-MB-231 cell line [73] and downregulate various mesenchymal markers for example Fibronectin and Vimentin in MDA-MB-231 [73] and Vimentin in JIMT-1 cells [50]. We observed an inhibition of cell migration but an inconsistent effect on EMT marker gene expression (with some genes inhibited but others induced) after treatment with SAL and its analogs (Figures 4 and 9). These results could be explained by recent findings of Kamlund and co-workers, where treatment of breast cancer JIMT-1 cells with 0.5 μM of SAL resulted in increased cell motility, but at the same time it decreased average cell migration [79]. Since cell migration is related to the distance traveled by cells from a starting point, and motility reflects movement around starting point, therefore SAL treated cells appear to be “dancing” in circles in place to a greater degree than vehicle treated cells [79]. It should be noted however that in the JIMT-1 study [79] the authors observed an increase in E-cadherin (the opposite of our findings in MDA-MB-231 cells), which may reflect differences in the cell line used and/or the concentration of SAL employed.

Our results indicate that double-modified SAL analog 17 exhibited improved activity against breast cancer stem-like cells when compared to the native molecule using several well-accepted traits of CSCs. Although the single-modified analog 2 was less potent than SAL in CSCs assays, it still induced a significant effect as compared to the vehicle treatment at approximately half lower concentrations than the parent molecule. The differences in activity between double-modified analog 17 and single-modified compound 2 may be explained by an increased stability of molecule 17 achieved by the replacement of hydroxyl group by ketone in C20 position. This increase in stability translated to improved biological activity. Additionally, in the studies of Li et al. the conjugation with hydroxamic acid, such as we used with compound 17, was shown to increase membrane permeability, which corresponded with enhanced activity [29]. The same authors also pointed out to the ion-binding activity of some hydroxamic acids, which would lead to better biological properties [29].

Future studies will focus on the investigation of the metabolism and toxicity of these compounds in clinically relevant models, such as hepatic [91][92][93], cardiac [94][95] and kidney organoid [96] cultures. In addition, it will be important to address the selectivity of each compound on cancer vs. normal cells. To this end, an appropriate in vitro model which could be adapted for this study would be the HMLER (Human Mammary Epithelial Cells overexpressing hTERT, SV40 T/t and H-RasV12) cell line, which was also characterized by high CD44⁺/CD24^−/low ratio and the ability to form non-adherent mammospheres [64]. The HMLER cell line (CD24^low) and its isogenic control (CD24^hlgh) was previously successfully used to study the mechanism of anti-breast cancer activity of SAL and its analogs [24][65].

5. Conclusions

In the present study we investigated anti-breast cancer stem-like cell activity of select single- and double-modified analogs of SAL in cell monolayer and organoid cultures. Several analogs were found to be more potent than the parent molecule towards breast cancer cell line MDA-MB-231. These compounds induced DNA fragmentation suggestive of apoptotic cell death. Further study revealed double-modified compound 17 to be more effective than SAL at inhibiting cell migration and renewal, as well as reducing thee CD44⁺/CD24⁻ stem cell subpopulation in both monolayer and organoid cultures. Analog 17 was also found to be more potent than SAL in a majority of NCI-60 cancer cell lines, including the entire panel of breast tumor cell lines. Further investigation revealed that compound 17 induced features of apoptosis and upregulation of autophagy-associated marker LC3II in both cell monolayer and organoids. Our findings highlight the therapeutic potential of SAL scaffold analogs in the treatment of metastatic breast cancer and support further research towards clinical development of these compounds.

Highlights.

• Select single- and double-modified salinomycin analogs were more potent towards MDA-MB-231 breast cancer cell line than salinomycin.

• Double modified analog 17 induced DNA fragmentation and showed premises of apoptotic cell death in MDA-MB-231 cell monolayer and organoids.

• Double modified analog 17 reduced colony formation potential and migration of MDA-MB-231 cells to the greater extent than salinomycin.

• Double modified analog 17 more effectively than salinomycin reduced percentage of CD44⁺/CD24⁻ breast cancer stem-like cells in MDA-MB-231 monolayer and organoid cultures.