Chemotherapy resistance is a well-known phenomenon resulting in cancer cell insensitivity to the available treatments (Minassian et al., 2019). Several factors can cause cancer resistance. Among others, the most common are 1) metabolic changes of tumors under pressure of chemotherapeutics and development of phenotype overcoming the toxic effects of therapy; 2) tumor microenvironment, especially oxygen level and interaction with adjacent cells (Gatenby and Brown, 2018; Al-Akra et al., 2019; Minassian et al., 2019); 3) tumor heterogeneity and clonal evolution together with mutations and immunological system interactions (Kubuschok and Trepel, 2017); 4) numerous adaptive responses and alternative pathways activated under pressure with specific chemotherapeutics (Gatenby and Brown, 2018); and 5) epigenetic effects (Lue et al., 2015). Several tumors develop multi-drug resistance (MDR) characterized by irresponsiveness to distinct drugs (Gatenby and Brown, 2018) that limits the treatment options of leukemia patients. In chronic myeloid leukemia (CML), the therapeutic efficacy of first-line tyrosine kinase inhibitors (TKIs) such as imatinib, dasatinib, and nilotinib is significantly reduced by resistance mechanisms (Cortes and Deininger, 2006). One of the main causes of imatinib resistance is mutation in the breakpoint cluster region- Abelson murine leukemia viral oncogene homolog 1 fusion gene (BCR-ABL1) kinase (Siegfried and Karni, 2018). Some of the pathways that can be targeted to overcome this resistance are Ras-Raf-MAPK (mitogen-activated protein kinase) and PI3K/Akt/mTOR activated by BCR-ABL1. In respect to this, multi-targeted combinatorial therapy should include inhibitors of heat-shock protein 90 (HSP90), MEK 1/2, phosphorylated protein kinase B (p-Akt), and mTOR (Cortes and Deininger, 2006; Jabbour et al., 2011).

Tumor aggressiveness and resistance development often depend on cancer stem cells (CSCs) and tumor-initiating cells (TICs) (Neuzil et al., 2007; Clarke, 2019). They were first identified in leukemia patients and represent a small subpopulation (0.05–1% of total cancer mass) existing in almost all cancers (Neuzil et al., 2007; Clarke, 2019). In CML, CSCs are resistant to TKIs. It is estimated that they cause relapse in almost 50% of all patients who discontinue the therapy (Chereda and Melo, 2016). While current therapies mainly destroy non-dormant cells (Kangwan et al., 2014), targeting CSCs would be an important approach to overcome tumor resistance and metastasis (Skoda et al., 2019). This could be accomplished through mitochondria targeting because CSCs have very high demands on energy and highly depend on mitochondrial activity (Skoda et al., 2019; García-Heredia and Carnero, 2020). With this aim, anti-diabetic drug metformin (inhibitor of mitochondrial respiratory chain I) could be utilized. It has inhibitory effects in CSCs originating from different cancers (Bednar and Simeone, 2012; Bao et al., 2013). Metformin is the most prescribed anti-diabetic drug in the world used for more than 40 years, thus having a proven safety profile. Numerous studies have shown that metformin decreases cancer risk (Saraei et al., 2019; Leng et al., 2021). Metformin exerts anti-cancerous effects by two distinct mechanisms, including an indirect, insulin-dependent mechanism (Dowling et al., 2011), where metformin sensitizes the cells to insulin, leading to a decrease in glucose and insulin levels in circulation, indirectly diminishing pro-survival activity through insulin receptors. This inhibits commonly activated downstream pathways in cancers, including PI3K/Akt/mTOR and MAPK pathways (Taniguchi et al., 2006). However, many previous studies proposed controversial results regarding metformin effects on p-Akt levels in different cell lines— indicating no effects (Vakana et al., 2011; Shi et al., 2012; Scotland et al., 2013), inhibition (Grimaldi et al., 2012; Rosilio et al., 2013), or even stimulation of Akt (Janjetovic et al., 2011; Grimaldi et al., 2012; Leclerc et al., 2013; Scotland et al., 2013). The second mechanism is insulin independent, through the inhibition of the mitochondrial respiratory chain complex I (Fontaine, 2018) that leads to the activation of AMP-activated protein kinase (AMPK) and inhibition of the mTOR pathway (Rosilio et al., 2014). Metformin interferes with cancer cells’ metabolism and mimics glucose deprivation causing an accumulation of inadequately folded proteins and endoplasmic reticulum stress (Leclerc et al., 2013). This leads to decreased survival of leukemia cells (Rosilio et al., 2013). Despite the various pathways targeted by metformin, certain cancer cells develop resistance to this drug and combinatorial therapy could be a key approach to overcome this resistance (Scotland et al., 2013).

Combinatorial therapy utilizes compounds targeting different signaling pathways with the aim of overcoming cancer resistance and increasing the effectiveness of therapy while decreasing its adverse effects. This approach is already used in CML treatment (Bhaskar et al., 2014; Mu et al., 2021). Combinatorial therapy with metformin and dasatinib in chronic lymphocytic leukemia (CLL) is one of the promising approaches (Marignac et al., 2013). Chemicals with proven safety profiles, such as drugs already used in therapy and active substances isolated from natural sources, could find their applications in combinatorial cancer therapy (Rosilio et al., 2014). Drug repurposing is crucial in the development of combinatorial therapies. With well-known safety profiles supported by data from completed clinical studies and data from clinical practice, approved drugs offer an opportunity for fast development of new combinations. Combining metformin with specific Akt (Scotland et al., 2013; Jang et al., 2021) and nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) inhibitors (Mauro et al., 2011) could help decrease the drug concentrations used in therapy and overcome resistance.

Phytochemicals with a known safety profile can be used in combinatorial treatments. For instance, thymoquinone has an inhibitory activity against Akt (Koka et al., 2008; Yi et al., 2008; Arafa et al., 2011; Hussain et al., 2011; Rajput et al., 2013; Khan et al., 2019) and NF-κB (Sethi et al., 2008), representing a good candidate for combinatorial treatment with metformin. TQ transcriptionally up-regulates PTEN in doxorubicin-resistant MCF-7/DOX breast cancer cells leading to a decrease of phosphorylated Akt (Arafa et al., 2011). TQ suppresses NF-κB activation in a time- and dose- dependent manner resulting in a down-regulation of NF-κB–regulated gene products (Zhang et al., 2016). TQ inhibits NF-κB through direct interaction with the p65 subunit and suppression of TNF-induced IKK activation (Sethi et al., 2008). Inhibitory effects of TQ on the fitness of leukemic cells were previously shown in various in vitro (El-Mahdy et al., 2005; Alhosin et al., 2010; Dergarabetian et al., 2013; Salim et al., 2013; Khalife et al., 2014; Soltani et al., 2017) and in vivo models (Salim et al., 2014; Pang et al., 2017).

However, the mechanism by which metformin and/or TQ interfere with intercellular signaling is not elucidated in the context of leukemic cell resistance.

Metformin and TQ combinatorial targeting of complementary pathways in tumor cells represents a novel approach for increasing effectiveness and overcoming resistance to current leukemia therapies (Figure 1).

Considering promising therapeutic potential of metformin and TQ by targeting crucial players in leukemia tumorigenesis, we aimed to evaluate whether metformin and TQ, as monotherapies or in combination will have inhibitory effects on the fitness of primary leukemia cells and CML cell lines sensitive and resistant to imatinib.

In this study, mono and combinatorial treatments with metformin and TQ caused a concentration-dependent decrease of viability and proliferation and induction of apoptosis in leukemia cells. Stronger synergistic effects on proliferation of imatinib-resistant compared to imatinib-sensitive CML cells were observed. Also, stronger effects of combination on the induction of apoptosis were seen in resistant compared to sensitive CML cells. Primary cells were more sensitive to combinatorial treatments compared to cell lines.

Metformin and TQ monotherapies had dose-dependent effects on leukemia cells. Similar to previous studies, metformin had millimolar (mM) IC₅₀ values (Table 3) while TQ had micromolar (µM) IC₅₀ values (Table 4). Results obtained in cell lines were confirmed in primary cells from one AML and one CLL patient. AML and CLL represent two of the most common leukemia types in adults (Salim et al., 2014) and it was of particular interest to evaluate the effects of metformin, TQ, and their combination in primary patient samples. The IC₅₀ value for metformin in primary CLL cells and leukemia cell lines was between 13.9 and 30.3 mM (Table 1). The exception was observed with K562 cell lines which was less sensitive to metformin. Voltan et al., 2016 have found that metformin treatment for 48 h had similar effects in leukemic cell lines (EHEB and JVM-2) and primary CLL cells with IC₅₀ values 11.6 and 10.2 mM, respectively (Voltan et al., 2016). Primary AML cells were more sensitive to metformin treatment with IC₅₀ 13.9 mM. Energy metabolism plays an important role in the therapy response of AML cells (Scotland et al., 2013). AMPK is functional in AML and can be activated by metformin, leading to mTOR inhibition and a significant decrease of proliferation (Green et al., 2010). The inhibitory activity of metformin in leukemia primary cells was confirmed by different authors. Shi et al., 2015 have found that primary Ph + ALL blasts are more sensitive to metformin treatment than K562 cell lines (Shi et al., 2015). Vakana et al. have found that metformin caused the inhibition of colony formation of primary CML cells. The effects were already seen at 1 mM concentration while 10 mM concentration had similar effects to imatinib (Vakana et al., 2011). TQ IC₅₀ in primary AML cells (27.1 µM) is in accordance with results previously obtained on U937 AML cell line (21.4 µM) (Khalife et al., 2014). This is the first study to show TQ effects in CLL primary cells, where we have observed a complete loss of viability at 10 µM concentration.

Similar effects of metformin in 1 µM imatinib-sensitive and imatinib-resistant CML cell lines (LAMA-84s and LAMA-84r) indicate that metformin could be used in overcoming resistance to current CML therapies. Vakana et al., 2011 were first to show that metformin have an inhibitory activity against Ph + cell lines and primary cells, including those carrying T315I-BCR-ABL mutation rendering them resistant to imatinib therapy (Vakana et al., 2011). Shi et al., 2015 have found that K562 cells resistant to 1 µM imatinib were more sensitive to metformin treatment when compared to imatinib-sensitive K562 cells. Metformin had synergistic effects with imatinib in Ph + cells (Ph + ALL SUP-B15 cell line, Ph + ALL primary blasts, K562 cells resistant, and K562 sensitive to 1 µM imatinib) (Shi et al., 2015). Metformin had similar effects in KU812 CML cells sensitive and resistant to imatinib therapy (Reddy et al., 2013) while in K562-Lucena multi-drug resistant cells, metformin decreased the levels of P-glycoprotein (P-gp) and had synergistic effects with imatinib (Curvello et al., 2013). P-gp is a surface glycoprotein which plays an important role in the development of multi-drug resistance (Dewanjee et al., 2017). LAMA-84r cell line used in this study is characterized by an over-expression of BCR-ABL and increased the activity of P-gp (Radujkovic et al., 2014), so metformin could decrease P-gp levels and have a cytotoxic activity in this cell line. Effects of metformin against resistant Ph + cells need to be further investigated with a focus on leukemia initiating cells. The obtained data so far, including results of this study, are supportive for the potential application of metformin in abolishing leukemia resistance and inhibiting leukemia initiating cells. LAMA-84r cells were less sensitive to TQ than LAMA-84s cells with IC₅₀ values 28.6 ± 2.1 μM and 11.8 ± 0.9 μM, respectively. However, the TQ resistance index in LAMA-84r cells is only 2.4. Resistance index is a value obtained by the division of IC₅₀ of resistant cells and IC₅₀ of same cells sensitive to certain agents (Scappini et al., 2004). Usually, cells designated to be resistant to certain therapies have much higher resistance index such as 744.0 for bosutinib, 692.0 for nilotinib, or 360.0 for dasatinib in CML cells harboring T315I mutation (Sang et al., 2016), 9.1 for imatinib in resistant K562 cells (Sang et al., 2016), 13.3 and >20 for imatinib in 1 µM resistant KBM7 and KBM5 CML cell lines, respectively (Scappini et al., 2004), and 9.6 for doxorubicin in certain lymphoma subpopulations (Zhang et al., 2018). Previously, it has been shown that TQ has a potential to overcome resistance in different cancer cells, especially when combined with other therapeutics (Arafa et al., 2011; Siveen et al., 2014). Our results demonstrated that TQ and metformin have synergistic effects in the inhibition of metabolic activity and proliferation of LAMA-84r cells.

K562 was significantly less sensitive to treatment with metformin and TQ. Although LAMA-84 and K562 cell lines are both Ph + CML cells, their phenotypes are different. LAMA-84 has elevated proteins associated with an invasive behavior, while K562 has elevated proteins connected to the development of resistance. Some of the expressed proteins found in K562 but not in LAMA-84 cells are Annexin A1 involved in exocytosis, glutathione S-transferases involved in detoxification, and heat-shock proteins (HSP27 and HSP70) with cytoprotective effects (Fontana et al., 2007). Those proteins could contribute to a more resistant phenotype against metformin and TQ treatment. HSP70 is recognized as a protein involved in imatinib resistance, and indeed, in preliminary experiments we have found that in LAMA-84s cells viability decreased dramatically after2 days in 1 µM imatinib, while K562 cells were less sensitive and about 20% could survive 40 days in 1 µM imatinib (data not shown).

Strong inhibitory effects of metformin on AML cell lines’ proliferation have previously been shown (Scotland et al., 2013; Gopalakrishnapillai et al., 2014). Grønningsæter et al. found that metformin inhibited proliferation in each of the 17 patients’ primary AML cells in vitro (Grønningsæter et al., 2020). In our experiments, metformin caused a significant inhibition of proliferation on both imatinib-sensitive and imatinib-resistant cell lines, as shown by the BrdU assay (Figure 2). On the other side, TQ effects were depending on dose and time of incubation. A mild induction of LAMA-84s cells’ proliferation by a low dose of TQ was probably due to compensatory mechanisms that were overcome when higher concentrations of TQ were applied. Opposite effects of lower and higher concentrations (hormetic effects) are known for various compounds (Calabrese and Mattson, 2017). Thymol, a compound that is transformed to TQ by catalytic oxidation of essential oils (Jukić and Miloš, 2005), was found to be protective or cytotoxic depending on the cell line used and applied concentration (Günes-Bayir et al., 2020). Previous studies have demonstrated anti-proliferative effects of TQ in the T-ALL Jurkat cell line, starting from a 10 µM concentration (Alhosin et al., 2010) and malignant T-lymphocytes (Diab-Assaf et al., 2018). Combinatorial treatment with metformin and TQ significantly inhibited proliferation of LAMA-84r cells as shown by the BrdU assay (Figure 2B) and decreased levels of cyclin D1, phospho-p85 S6 kinase (p-P85 S6k), and phospho-p70 S6 kinase (p-P70 S6k) (Figure 4A). In LAMA-84s cells, lower combinatorial concentrations increased levels of proliferation proteins (Figure 4A), while a higher combinatorial concentration inhibited proliferation (Figure 2A). The observed hormetic effects of TQ applied in CML cell lines as monotherapy or in combination with metformin need to be further evaluated. Those effects were not observed in primary cells.

CI values obtained in combinatorial studies depended on the cell line, concentrations of used compounds, as well as the procedure for their application. Synergistic effects were observed in all cell lines regarding a decrease of viability and inhibition of proliferation. The strongest synergistic effects were observed when cells were pre-treated with TQ for 12 h, and then co-treated with metformin for an additional 48 h (Table 2). Previous studies have found the strongest synergism when TQ was applied before topotecan (Khalife et al., 2014) and gemcitabine (Mu et al., 2015). In the U937 AML cell line, TQ in combination with topotecan showed synergistic anti-proliferative and pro-apoptotic effects with pre-exposure to TQ more effective than simultaneous application (Khalife et al., 2014). Also, in the same AML cell line, pretreatment with Akt inhibitor resulted in strong synergistic effects with metformin (Scotland et al., 2013). What we found particularly interesting is the fact that sequential treatment had strong synergistic effects (CI value 0.515 ± 0.131) in LAMA-84r cells, resistant to 1 µM imatinib. The strongest synergistic effects were observed in LAMA-84r cell line proliferation where CI values for 48 and 72 h’ treatments were 0.314 ± 0.100 and 0.151 ± 0.075, respectively. Less intensive synergistic effects were also observed in proliferation decrease of imatinib-sensitive cell lines (Table 2). It is important to note that LAMA-84r cells were grown in a medium containing imatinib and stronger synergistic effects could potentially be due to this additional compound. This result needs to be further explored.

Metformin and TQ mono and combinatorial therapies induced an apoptosis in tested cell lines. It was previously shown that TQ had pro-apoptotic effects in CML (Sethi et al., 2008), ALL (Alhosin et al., 2010; Dergarabetian et al., 2013; Salim et al., 2013; Soltani et al., 2017), myeloblastic leukemia (El-Mahdy et al., 2005), and AML (Khalife et al., 2014) cells. The metformin pro-apoptotic effect depends on the dominant energy production pathway and compensatory capacity of treated cells (Scotland et al., 2013). In LAMA-84s cells, a slight increase in the percentage of apoptotic cells was observed with combinatorial therapies, and this was confirmed by decreased levels of Mcl-1 protein in western blot experiments. In LAMA-84r cells, similar effects were seen, but were more pronounced when compared to LAMA-84s cells. The relatively high level of apoptosis (15.0%) in controls of LAMA-84r cells was probably due to 1 µM imatinib present in the growth medium for this cell line. Stronger effects of combination on induction of apoptosis were seen in resistant (2.5 mM metformin + 5 µM TQ vs. control, 15.0% vs. 24.3%) compared to sensitive CML cells (2.5 mM metformin + 5 µM TQ vs. control, 3.7% vs. 6.5%). In LAMA cells resistant to imatinib therapy, 25 mM metformin monotherapy or a ten times decreased concentration (2.5 mM) of metformin in combination with 5 µM TQ reduced levels of Mcl-1 to 0.53 and 0.60 compared to control, respectively (Figure 4B). Inhibition of Mcl-1 by metformin was previously shown in AML cells (Zhou et al., 2021) and other tumors (Park et al., 2018; Ye et al., 2020; Chen et al., 2021).

Our results show that metformin and TQ treatments for 48 h decrease levels of p-NF-κB p65 in LAMA-84 cells sensitive and resistant to imatinib, as well as in primary CLL cells (Figures 4A,C). In contrast to this effect, TQ in a low concentration slightly increased the levels of p-NF-κB p65 in LAMA-84r cells (Figure 4A). Sethi et al., 2008 have found that TQ inhibited NF-κB in CML cells in experiments with incubation times up to 6 h (Sethi et al., 2008). Metformin have shown an NF-κB inhibitory activity in various cancer cells (Nguyen et al., 2019; Sultuybek et al., 2019; Yenmis et al., 2021). However, there are controversial results regarding metformin activity against phosphorylation of Akt in leukemia cells and this activity might depend on the mitochondrial energetic status and basal levels of Akt in each cell line tested (Scotland et al., 2013). In our study, metformin decreased levels of p-Akt in LAMA-84r cell lines while in LAMA-84s cell lines, this effect could not be observed because of very low or undetectable basal levels of p-Akt. Specifically, TQ in lower concentrations induced phosphorylation of Akt while higher concentrations inhibited p-Akt. TQ stimulatory effects on p-Akt were previously observed in MDA-MB-231 breast cancer cells (Yu and Kim, 2012) while the inhibition of p-Akt by TQ was previously confirmed in various cancer models (Koka et al., 2008; Yi et al., 2008; Arafa et al., 2011; Hussain et al., 2011; Rajput et al., 2013; Khan et al., 2019). The observed TQ hormetic effects where low concentration increased levels of p-NF-κB p65 in LAMA-84r cells and p-Akt in both LAMA-84 cell lines were probably due to compensatory mechanisms activated by cells. Increased Akt phosphorylation is recognized as a compensatory mechanism in cancer cell lines (Bergholz and Zhao, 2021). In some western blot experiments, we have added doxorubicin, which induces p-Akt in several cancer cell lines such as T-lymphoblastic leukemia (Maraldi et al., 2011), breast cancer (Wallin et al., 2010), gastric cancer (Zhou et al., 2013), and osteosarcoma (Wang et al., 2015). We have introduced this specific experimental design since it was previously seen that TQ increases anti-tumor effects of doxorubicin (Effenberger-Neidnicht and Schobert, 2011). In LAMA-84s cell lines, doxorubicin did not induce the phosphorylation of Akt. Primary CLL cells were more sensitive to metformin and TQ treatment compared to cell lines, and no hormetic effects were observed. Proteins involved in proliferation, p-Akt, and p-NF-κB p65 were decreased while cleaved PARP was increased when treated with very low concentrations of metformin and TQ, especially in combination. Compensatory response is not often seen in primary cells and it is more characteristic for cancer cell lines due to various changes during adaptation to in vitro growth conditions (Goodspeed et al., 2016).

In conclusion, this study has shown that metformin and TQ monotherapies possess a significant anti-leukemic effect that is more pronounced when combinatorial therapies are applied. Synergistic effects on the inhibition of proliferative capacity, particularly seen in imatinib-resistant leukemic cells, represent an encouraging direction for prospective in vivo studies.