Identification of natural products and FDA-approved drugs for targeting cancer stem cell (CSC) propagation

Here, we report the identification of key compounds that effectively inhibit the anchorage-independent growth and propagation of cancer stem cells (CSCs), as determined via screening using MCF7 cells, a human breast adenocarcinoma cell line. More specifically, we employed the mammosphere assay as an experimental format, which involves the generation of 3D spheroid cultures, using low-attachment plates. These positive hit compounds can be divided into 5 categories: 1) dietary supplements (quercetin and glucosamine); 2) FDA-approved drugs (carvedilol and ciprofloxacin); 3) natural products (aloe emodin, aloin, tannic acid, chlorophyllin copper salt, azelaic acid and adipic acid); 4) flavours (citral and limonene); and 5) vitamins (nicotinamide and nicotinic acid). In addition, for the compounds quercetin, glucosamine and carvedilol, we further assessed their metabolic action, using the Seahorse to conduct metabolic flux analysis. Our results indicate that these treatments can affect glycolytic flux and suppress oxidative mitochondrial metabolism (OXPHOS). Therefore, quercetin, glucosamine and carvedilol can reprogram the metabolic phenotype of breast cancer cells. Despite having diverse chemical structures, these compounds all interfere with mitochondrial metabolism. As these compounds halt CSCs propagation, ultimately, they may have therapeutic potential.


INTRODUCTION
Cancer stem-like cells (CSCs) represent a small subpopulation of cancer cells [1] (< 1%) that are clinically responsible for resistance to radiotherapy and chemotherapy, the development of tumour recurrence, and the formation of metastases [2][3][4]. Key distinguishing features of CSCs are pluripotency, self-renewal and the ability to undergo anchorage-independent growth, favouring their propagation and metastasis [5,6]. Unfortunately, conventional therapies are frequently not able to eradicate CSCs. For this reason, there is a clinical urgency to intervene via the discovery of new drugs that can inhibit the propagation of cancer stem cells, that can perhaps be used in conjunction with more traditional therapies.
CSCs show metabolic plasticity and are able to respond rapidly to diverse environmental stimuli [7]. In fact, CSCs can switch quickly from glycolysis to oxidative phosphorylation and vice versa, to meet their diverse metabolic needs of ATP production and consumption. However, many recent studies have highlighted the fact that mitochondrial biogenesis plays an especially vital role in CSCs [8][9][10][11][12]. Therefore, halting mitochondrial biogenesis or respiration, may be a vulnerability that we can exploit for their more effective eradication.
Interestingly, natural products and dietary supplements may also show anti-cancer properties. Indeed, we previously investigated the effect of Matcha green tea on proliferation and metabolism in MCF7. Moreover, AGING we used proteomics analysis to dissect how this dietary supplement was able to alter glycolytic and mitochondrial pathways, as well as others related to stem cells and DNA damage/repair [13]. In addition, we also studied how vitamin C [14,15], bergamot [16] and berberine [17] affect the proliferation of CSCs.
In the current study, we investigated the potential therapeutic effects of several classes of compounds (dietary supplements, FDA-approved drugs, natural products, flavours, vitamins), by assessing their ability to halt the propagation of breast cancer stem cells, using the MCF7 cell line as a model system.
Finally, we focused on the metabolic effects in MCF7 cells of the most promising compounds, two dietary supplements, quercetin and glucosamine, and of the FDA-approved drug, carvedilol. We used the Seahorse XFe96 Analyzer to measure the oxygen consumption rate (OCR) and the glycolysis (ECAR). Our results show that these three compounds can significantly interfere with cancer cell metabolism, resulting in the suppression of CSC propagation. Therefore, we believe that these compounds should be investigated further.

Mammosphere assay
From adherent MCF7 cells, we prepared a single cell suspension using enzymatic (1x Trypsin-EDTA) and manual disaggregation (25-gauge needle) [18]. Three thousand cells were plated into mammosphere medium (DMEM-F12/B27/20ng/ml EGF/PenStrep), under nonadherent conditions, in 6-wells plates coated with poly-HEMA. We counted the number of 3D spheroids with a diameter >50 µm, after five days of culture. All experiments were performed in triplicate and repeated three times independently.

Seahorse analysis
To evaluate the extracellular acidification rates (ECAR) and the oxygen consumption rates (OCR), we used the Seahorse XF96 metabolic flux analyser (Agilent Technologies, Inc.). Fifteen thousand MCF7 cells were seeded per well, into XF96-well cell plates, and cultured at 37° C in an incubator with a 5% CO2 humidified atmosphere. MCF7 cells were cultured in DMEM supplemented with 10% FBS (Foetal Bovine Serum), 2 mM GlutaMAX, and 1% Pen-Strep. After twenty-four hours from plating, the cells were incubated in the presence or absence of quercetin, glucosamine hydrochloride or carvedilol. After forty-eight hours, cells were washed in pre-warmed XF assay media, as previously described [19]. ECAR and OCR measurements were normalized by cellular protein content (SRB). Data sets were analysed using XFe-96 software and Excel, then Student's t-test calculations were performed. All experiments were performed in sextuplicate and repeated three times independently.

SRB assay
SRB is a colorimetric assay for cytotoxicity, based on the measurement of cellular protein content. Briefly, MCF7 cells in monolayers were first fixed with 10% trichloroacetic acid and then washed with 1% acetic acid after incubation with SRB. The dye dissolved in 10 mM Tris base solution, and the OD determined at 565 nm, using a microplate reader [20].

Compound screening
Here, our goal was to identify key compounds that effectively inhibit the anchorage-independent growth and propagation of cancer stem cells (CSCs), using MCF7 cells as a model system. Briefly, these compounds can be classified within 5 sub-categories: 1) dietary supplements; 2) FDA-approved drugs; 3) natural products; 4) flavours; and 5) vitamins. See Table 1.
To assess their potential effect(s) on cancer stem cell activity, we cultured MCF7 cells under low-attachment conditions, in the presence or absence of a given compound. We evaluated CSC activity after five days of culture, by counting the number of mammospheres formed.
We first analysed CSC propagation after treatment with two dietary supplements: quercetin and glucosamine.
Quercetin is a flavonoid present in vegetables, fruits and beverages. It has been extensively studied as a chemoprevention agent in several cancer models [21][22][23]. It has anti-oxidant, anti-inflammatory and anti-cancer activities [24][25][26][27][28][29][30]. Glucosamine is a monosaccharide, precursor used for the glycosylation of proteins and lipids. It is naturally present, for example, in animal bones, bone marrow and the shells of shellfish.
We tested the quercetin at concentrations of 10, 20 and 40 µM. Figure 1A shows that at the concentration of 40 µM, quercetin was effective in halting CSC propagation by over 60%, and its IC50 fell in the range between 20 and 40 µM. In Figure 1B, results with glucosamine are shown, over the range of 5 to 20 mM. Note that the lowest concentration tested is already effective as an inhibitor of CSC propagation. Interestingly, glucosamine (2-amino-2-deoxy-D-glucose) is structurally related to another well-established metabolic inhibitor, namely 2-DG (2-deoxy-D-glucose). Based on our previous studies using 2-DG in the same MCF7 CSC assay [14,31], glucosamine appears to be approximately 4 times as potent.
Next, we investigated the effects of two FDA-approved drugs: the beta-blocker carvedilol and the antibiotic ciprofloxacin. Carvedilol, brand name Coreg, is a betablocker and is used to treat mild to severe congestive heart failure [32,33].
We tested carvedilol at the concentrations of 10, 25 and 50 µM. The IC50 was 25 µM and the highest dose was so potent as to completely block the mammosphere formation ( Figure 2A). However, ciprofloxacin was less potent, with an IC50 of approximately 100 µM ( Figure 2B).
Using this approach, we also focused on compounds that are found naturally in plants, or in vegetables and as additive in certain foods. Firstly, we tested two compounds related to each other, aloin emodin and the aloin. These are distinguished only by the fact that aloin emodin lacks a sugar compared to aloin.
Aloin (or barbaloin) is a natural anthraquinone extracted from the plant aloe latex and together with aloe emodin, that lacks a sugar group compared to the first, is widely used as an anti-inflammatory and shows anti-cancer activity [34]. Figure 3A shows that aloin emodin at the concentration of 15 µM was effective in reducing CSC propagation by over 70%. Aloin was also effective at all three concentrations tested of 50, 100 and 200 µM ( Figure 3B). AGING  Tannic acid is a polyphenol, a specific form of tannin, naturally found in the nutgalls made by insects on twigs of oak trees. It has been also used as embalming material of mummies in ancient Egypt [35,36]. Tannic acid is a potent anti-oxidant with anti-proliferative effects on diverse types of cancer [37]. Tannic acid was tested at the concentrations of 10, 25 and 50 µM, revealing an IC50 of approximately 25 µM ( Figure 4A).
Chlorophyll is present in green leaves of vegetables as spinach and is a food colouring agent. It has been shown exhibit anti-oxidant and anti-apoptotic effects [38,39]. Figure 4B shows the results obtained with chlorophyllin, which has its IC50 in the range between 50 and 100 µM.
The last two compounds in this category were azelaic acid and adipic acid ( Figure 5A, 5B). Azelaic acid is found in wheat, rye and barley. It inhibits mitochondrial enzymes of the respiratory chain and enzymes involved in DNA synthesis showing antiproliferative and cytotoxic effects in melanoma, bladder and breast cancers, and leukaemia [40][41][42]. Adipic acid is used mainly in the production of nylon and is also used as a food additive [43,44]. Azelaic acid displayed effectiveness starting at a concentration of 2.5 mM, with an IC50 between 5 and 10 mM; note that at 10 mM the propagation of CSCs was completely halted. Similarly, adipic acid also showed promising inhibitory effects.
We next investigated flavour-related compounds, such as citral and limonene ( Figure 6A, 6B). Citral (or lemonal) is naturally present in lemons, oranges and limes. Limonene is used as a flavouring in foods, beverages and chewing gum. Citral had an IC50 between 10-50 µM and limonene greater than 50 µM.
Finally, we assessed the effects of two common vitamins on CSCs proliferation: nicotinamide, which is the active form of vitamin B3, and nicotinic acid (a.k.a, niacin). Nicotinamide is an amide form of vitamin B3, and is found in foods like fish, poultry, eggs and is used as a dietary supplement/medication, to prevent and treat pellagra [45]. Nicotinic acid or niacin is the vitamin B3 and is used to reduce elevated levels of cholesterol [46]. Importantly, nicotinamide and nicotinic acid are both precursors of the co-enzymes nicotinamide adenine dinucleotide (NADH) and nicotinamide adenine dinucleotide phosphate (NADPH) [14]. Interestingly, treatment with nicotinamide significantly increased the CSC propagation, at concentrations of 10 and 20 µM. However, nicotinic acid did not show any significant effects at the doses tested (5, 10, or 20 µM) ( Figure 7A, 7B).

Metabolic validation via seahorse analysis
In the literature, it is well documented that propagation of CSCs depends on several factors including an increased mitochondrial metabolism and biogenesis [7,47,48]. In this regard, to address an effect of quercetin, glucosamine and carvedilol on cellular metabolic features, we performed analysis with the Seahorse XF Analyzer after a 48-hours treatment with the compound. We measured oxygen consumption rate (OCR) and extracellular acidification rate (ECAR). Interestingly, quercetin significantly increased the glycolysis at the concentration of 20 µM and reduced the glycolytic reserve and glycolytic reserve capacity (maximal capacity of the cells to respond to a higher ATP demand) at the concentration of 40 µM as compare to the untreated control cells ( Figure 8A). Moreover, all the OCR parameters were significantly decreased: basal respiration, proton leak, ATP production, maximal respiration and spare respiratory capacity ( Figure 8B). Next, we investigated the effect of glucosamine finding that it was able to significantly increase glycolysis at the concentration of 5 mM and decrease the glycolytic reserve capacity ( Figure 9A). Importantly, at 20 mM all OCR parameters were significantly decreased but the spare respiratory capacity which was already negatively affected at the dose of 10 mM      ) shows that quercetin treatment significantly decreases the basal respiration, ATP production, maximal and spare respiration, as compared to the control cells. In panels A and B, experiments were performed three times independently, with six repeats for each replicate. Bar graphs are shown as the mean ± SEM, t-test, two-tailed test. *p < 0.05, ***p < 0.001.
( Figure 9B). Lastly, we examined carvedilol to see if it could affect cellular metabolism. We treated MCF7 cells with 25 and 50 µM of carvedilol. Results in Figure 10A show that this drug dramatically negatively affected all ECAR parameters, at the maximal concentration of 50 µM. At the dose of 25 µM only glycolytic reserve and glycolytic reserve capacity were significantly decreased. On the contrary, glycolysis was increased more than three times perhaps as an attempt to compensate for the dramatic decrease in all the parameters related to the oxygen consumption rate. Indeed, OCR analysis highlighted that carvedilol had a powerful capability in decreasing the mitochondrial respiration ( Figure 10B).
Interestingly, in the end what stands out from this more detailed analysis conducted on quercetin, glucosamine and carvedilol, is that what these three compounds have in common is their effects on mitochondrial respiration, and that they are effective in inhibiting the propagation of MCF7 cancer stem-like cells (Figure 11).  ) shows that glucosamine treatment significantly decreases the basal respiration, ATP production, maximal and spare respiration, as compared to the control cells. In panels A and B, experiments were performed three times independently, with six repeats for each replicate. Bar graphs are shown as the mean ± SEM, t-test, two-tailed test. *p < 0.05, ***p < 0.001.

DISCUSSION
The eradication of cancer stem cells remains a focal point in the battle against cancer, regardless of the type of cancer. Cancer stem cells are considered to be responsible for the dissemination and formation of distant metastases [49], as well as resistance to anti-cancer therapies [50][51][52]. For this reason, it is vital to find the Achilles' heel of CSCs. Recently, many investigators have highlighted the importance of the metabolic microenvironment, as well as the metabolic features of CSCs [47,[53][54][55][56].
In this study, we investigated the effectiveness of different compounds in decreasing or blocking the anchorage-independent growth of MCF7 cancer stem cells, by using the mammosphere assay, as a rapid in vitro screening tool that exploits the ability of CSCs to grow under low-attachment conditions.
Here, we chose to examine the activity of different compounds, which can be classified into 5 different groups: 1) dietary supplements (quercetin and glucosamine); 2) FDA-approved drugs (carvedilol and ciprofloxacin); 3) natural products (aloe emodin, aloin, tannic acid, chlorophyllin copper salt, azelaic acid and adipic acid); 4) flavours (citral and limonene); and 5) vitamins (nicotinamide and nicotinic acid). Our results are summarized schematically in Figure 11.  ) shows that carvedilol treatment significantly decreases the basal respiration, ATP production, maximal and spare respiration, as compared to the control cells. In summary, at 25 μM, carvedilol enhanced glycolysis, but inhibited mitochondrial oxygen consumption. In contrast, at 50 μM, carvedilol inhibited both glycolysis and mitochondrial oxygen consumption. In panels A and B, experiments were performed three times independently, with six repeats for each replicate. Bar graphs are shown as the mean ± SEM, t-test, two-tailed test. *p < 0.05, ***p < 0.001. AGING Glucosamine is widely used in the medical field in the treatment of osteoarthritis [57], and it is well known that it does not have side effects in humans. Moreover, it has been reported as an attractive candidate in lung carcinogenesis, decreasing the lung cancer risk [58]. Interestingly, glucosamine has inhibitory effects on glycolysis [59,60] and drives general cell ATP depletion [61]. Moreover, it has been also found that glucosamine induced dysfunction of mitochondria as well as that of the peroxisome in human chondrocytes [62]. In our study, the results showed that glucosamine was able to reduce the mammospheres formation efficiency starting from the lowest tested concentration of 5 mM (IC50). This result is relevant because compared to 2-DG (IC50, 20 mM), glucosamine is a more potent glycolytic inhibitor. This evidence adds to the advantage that glucosamine is already used in the medical field, while the 2-DG cannot be administered to humans. In addition, we investigated the metabolic effect of glucosamine on MCF7 cells in adhesion, representing the bulk tumour cells, using the Seahorse Analyzer. Importantly, at 20 mM all OCR parameters were significantly decreased, but the spare respiratory capacity was already negatively affected at a dose of 10 mM. As such, glucosamine may have potential as a therapeutic to halt the proliferation of CSCs, as we are starting to understand the importance of metabolic flexibility, as an intervention point, for decreasing tumour recurrence and metastasis [47,55].
Here, we revealed a decrease in the proliferation of CSCs and moreover, by Seahorse analysis, we showed that this treatment was able to negatively affect the glycolytic reserve capacity. Importantly, almost all OCR parameters were significantly decreased, at both of the concentrations of quercetin tested.
Carvedilol or Coreg is a beta-blocker widely used as a cardio protector in cardiac dysfunction [70,71]. It is also used to prevent chemotherapy-related cardiotoxicity [72,73], and cardiac mitochondrial oxidative damage [74]. Studies on malignant breast cancer cells have been performed to investigate the ability of carvedilol in inhibiting their proliferation Quercetin is flavonoid found in many foods, glucosamine is a dietary supplement, and carvedilol is an FDA-approved beta-blocker. Intriguingly, although these three compounds are so different in their chemical structure, they share the ability to interfere with mitochondrial metabolism and block the propagation of CSCs. AGING and migration [75,76]. Here, we wanted to test its effectiveness in halting breast CSC propagation and its possible role in altering their metabolic pathways. Our findings show that the drug was effective in reducing mammospheres formation (IC50 at 25 µM and complete inhibition at 50 µM), and in reducing the glycolysis and the oxidative respiration parameters, as highlighted by the Seahorse analysis.
We also tested another FDA-approved drug such as the antibiotic ciprofloxacin, which has previously reported to effectively block cell proliferation of bladder and melanoma cancer cells [77,78]. In addition, this antibiotic has shown effectiveness in altering  mammalian mitochondrial DNA replication [79]. Here, ciprofloxacin displayed its IC50 of approximately 100 µM in the mammosphere formation assay.
Next, we investigated the mammosphere formation capacity of aloin, which has been reported to be cytotoxic against two human breast cancer cell lines [80]. Further, aloin was able to induce apoptosis in lung cancer cells causing disruption of mitochondrial membrane potential and inducing ROS production [81].
Then, we tested tannic acid, a potent anti-oxidant and anti-proliferative agent that is effective in inhibiting EGFR/STAT signalin, resulting in cell cycle arrest and apoptosis [82]. Chlorophyllin, another natural compound, also has as anti-cancer effects [83,84]. Interestingly, chlorophyllin inhibits oxidative phosphorylation in rat liver mitochondria [85]. Moreover, we tested the azelaic acid already described to be effective against tumour as well as in cutaneous disorders [40,86,87], beside its ability to inhibit mitochondrial respiration and promoting mitochondrial damage [88], and adipic acid which it is used in the industrial production of nylon [44]. Finally, we tested the effectiveness of citral and limonene which are already used in medicine [89]. Citral is well known to have an antifungal activity altering oxidative phosphorylation [90], by altering the mitochondrial membrane potential in MDA-MB-231 cells [91]. In addition, limonene plays a key role in regulation of oxidative stress mediated by ROS in a broad variety of organisms [92].
We have previously shown that the upregulation of NAD+ salvage pathways increases stemness [14], and here we confirm our results since the administration of nicotinamide increased the proliferation of breast CSCs [14]. Very recently, an independent group reported that decreased intracellular NAD, due to the up-regulation of miR-381, was able to induce apoptosis in breast cancer cells [93].
Intriguingly, several of the agents tested here shared the property of interfering with mitochondria and their function.

CONCLUSIONS AND FUTURE DIRECTIONS
Here and in previous reports, we have identified numerous chemical entities, with anti-mitochondrial activity, that can target and eradicate CSCs in vitro. Figure 12 shows a summary diagram that illustrates the workflow of this experimental screening and clinical strategy. A relatively comprehensive list of these compounds can be found in the following review article [94]. AGING Next steps would include: 1) their evaluation in preclinical animal models; and 2) clinical trials, as well. For example, using Doxycycline, we have previously shown that it eradicates CSCs and prevents metastasis in a preclinical animal model [95]. Moreover, a phase II clinical trial (window study) showed that Doxycycline eradicates CSCs in vivo, using CD44 and ALDH1 as CSC-markers [96].
Therefore, Doxycycline provides the first example that this strategic approach is indeed successful.

AUTHOR CONTRIBUTIONS
GB wrote the first draft of this article, which was then further edited by MPL and FS. GB prepared the final figures. All authors contributed to the article and approved the submitted version.

CONFLICTS OF INTEREST
MPL and FS hold a minority interest in Lunella Biotech, Inc.

FUNDING
This work was supported by research grant funding, provided by Lunella Biotech, Inc. (to FS and MPL). The funder was not involved in the study design, collection, analysis, interpretation of data, the writing of this article, or the decision to submit it for publication.