Introduction

Metformin (Met, Fig. 1) is one of the most prescribed antidiabetic drugs in the world [1, 2]. Metformin is a derivative of guanidine, an active component in the extracts from French lilac (G. officinalis), the plant that has been used for centuries to combat diabetes. A major advantage of metformin as a drug is that it is relatively safe and diabetic patients typically take daily several grams of the drug. Recently metformin is being repurposed as an antitumor drug due to epidemiological findings, revealing an association between decreased incidence of pancreatic cancer and metformin use in diabetic individuals [3, 4]. However, at typical antidiabetic doses, the bioavailability of metformin and/or its uptake by cancer cells is probably too low to exert its antitumor effects in humans, although singificant antitumor activity was reported in the rodent models, at comparable plasma concentrations [5, 6]. It appears that the antineoplastic activity of metformin in humans will require much higher concentrations of the drug in the serum than what is typically administered in diabetics treated with metformin [5]. Thus, it is important to develop new and improved metformin analogs with increased bioavailability, resulting in more potent antitumor effects, although these analogs may show increased toxicity for routine treatment of diabetes. We hypothesized that attaching the triphenylphosphonium cation (TPP+) to metformin via an alkyl chain would greatly enhance its accumulation in mitochondria of cancer cells, leading to selective inhibition of proliferation of cancer cells [7]. Results obtained in this study substantiate the proposal that TPP+-modified metformins are more cytostatic to pancreatic cancer cells.

Fig. 1
figure 1

Chemical structures of metformin and mito-metformin analogs

Full size image

Metformin does not undergo metabolism like many other drugs but it is excreted out unchanged in the urine [8]. However, it induces extensive changes in the metabolic pathways and mitochondrial energy metabolism [9]. Although multiple mechanisms were proposed for metformin anticancer effects, mitochondria still remain as the major target of metformin, leading to inhibition of cell respiration, adenosine monophosphate (AMP)-activated protein kinase (AMPK) activation, mTOR inhibition, and inhibition of cell proliferation [10, 11].

Recent research implicates a role for mitochondrial complex I inhibition in the antitumor effects of metformin [12]. Interestingly, cancers with mutations in mitochondrial genes encoding proteins of the mitochondrial complex I of the electron transport chain are more susceptible to biguanides such as metformin [13]. Thus, cancers with oxidative phosphorylation deficits are likely to be more sensitive to biguanides than normal tissues and, consequently, patients with complex I mutated cancer phenotype may be more sensitive to metformin treatment [14]. Tumor microenvironment was proposed to be another factor for increased metformin potency in some tumor phenotypes [15]. More recently, metformin was reported to activate T cell-mediated immune response against cancer cells [15–17].

We have recently shown that attaching a positively-charged lipophilic triphenylphosphonium group to a nitroxide, quinone, or a phenolic group via an aliphatic linker chain vastly enhanced their ability to target mitochondria, and that such mitochondria-targeted compounds (e.g., Mito-CP, Mito-Q and Mito-Vitamin-E) significantly induced antiproliferative and/or cytotoxic effects in tumor cells without markedly affecting noncancerous cells [18–23]. Similarly, we hypothesized that attaching a positively-charged lipophilic substituent will improve mitochondrial targeting of metformin (e.g., mito-metformins, Fig. 1), thereby generating a new class of metformin analogs with enhanced antitumor potential. Results show that the mito-metformin analogs are much more potent (100–1000-fold higher) than metformin in inhibiting human pancreatic adenocarcinoma cells (PDACs) [7].

Antiproliferative Effects of Metformin and Mito-Metformin10

The effect of mito-metformin10 and metformin on MiaPaCa-2 cells was investigated by monitoring colony formation and cell proliferation. As shown in Fig. 2, in order to inhibit proliferation of pancreatic cancer cells, metformin was used at millimolar levels, whereas mito-metformin10 was used at micromolar concentrations. The IC50 values for inhibition of proliferation of human pancreatic cancer cells (MiaPaCa-2 and PANC-1) were more than 1000-fold lower for Mito-Met10 (0.2 µM), as compared to Met (0.6–0.9 mM). MiaPaCa-2 cells which were treated with 0.5 or 1 mM concentration of metformin, or 0.5–1 µM concentration of mito-metformin, showed the same effects in colony formation and cell proliferation experiments [7]. The IC50 values for inhibiting 50% of colony formation for metformin and mito-metformin10 were calculated to be 1.3 mM and 1.1 µM, respectively. Thus, mito-metformin10 is ca. 1000-fold more effective than metformin. The colony forming ability of the pancreatic cancers cells, while decreased by treatment with Mito-Met10, could be rescued be cell pre-treatment with compound C, an inhibitor of AMPK protein (Fig. 2b). This indicates that AMPK signaling pathway is involved in the antiproliferative effects of Mito-Met10. Interestingly, Mito-Met10 showed a good selectivity towards pancreatic cancer cells, as compared to non-tumorigenic cells (Fig. 2a), implicating a wider margin of therapeutic index for its use in vivo.

Fig. 2
figure 2

Effect of metformin and mito-metformin10 on PDAC proliferation. a Titration of pancreatic cancer (MiaPaCa-2 and PANC-1) and non-tumorigenic (N27, IEC-6 and HPNE) cells with Mito-Met10 (left) and Met (right). b Effect of Mito-Met10 on colony formation in MiaPaCa-2 cells and the rescue effects of AMPK inhibitor, compound C [7]

Full size image

Inhibition of Pancreatic Cancer Cell Respiration: Complex I Involvement

Metformin is weakly cationic, and weakly targets mitochondria. Metformin’s antitumor effects were attributed in part to its ability to inhibit mitochondrial complex I-dependent cell respiration and ATP generation [7, 12]. In contrast to rotenone, a non-selective inhibitor of complex I in both normal and cancer cells, metformin reversibly and more selectively inhibits mitochondrial complex I in cancer cells [12]. As mito-metformin10 selectively accumulates into mitochondria, we compared the relative inhibitory effects of metformin and mito-metformin10 on complex I activity in MiaPaCa-2 cells. The real time monitoring of oxygen consumption rate (OCR) was used as a measure of mitochondrial function in intact cells. The mitochondrial respiration or OCR was monitored in MiaPaCa-2 cells treated with metformin and mito-metformin10 using the Seahorse XF extracellular flux analyzer [7]. Figure 3 shows the OCR measurement of mitochondrial complex I activity of MiaPaCa-2 cells pretreated for 24 h with metformin or mito-metformin10. The IC50 values for inhibition of mitochondrial complex I for metformin and mito-metformin10 were determined to be ca. 1.1 mM and 0.4 µM. These results show that the complex I inhibition may be mechanistically related to the antiproliferative effects of these compounds. Again, Mito-Met10 exhibited selectivity towards cancer cells in inhibiting mitochondrial complex I (Fig. 3).

Fig. 3
figure 3

Effects of metformin and mito-metformin10 on mitochondrial complex I activity in pancreatic cancer (MiaPaCa-2 and PANC-1) and non-tumorigenic (HPNE and IEC-6) cells [7]

Full size image

Mito-Metformin-Mediated Stimulation of Reactive Oxygen Species (ROS) and Redox Signaling

One of the consequences of inhibiting mitochondrial complex I is stimulation of one-electron reduction of oxygen to superoxide radical anion (O2 •–) and other oxidants (hydrogen peroxide and higher oxidation heme species) derived from superoxide [7]. As mito-metformin10 treatment (at the concentration inhibiting proliferation) did not exert any cytotoxic effects, it is likely that ROS generated from mitochondria could be involved in redox signaling [7]. We used the cell-permeable probe, hydroethidine (HE), to detect superoxide by monitoring 2-OH-E+, the specific marker product of the reaction between superoxide and HE [24–26]. A marked increase in 2-OH-E+ formation was observed in mito-metformin10-treated MiaPaCa-2 cells (Fig. 4). Evidence for other one-electron oxidation species formed from mito-metformin10 interaction with mitochondrial proteins arose from studies demonstrating the formation of ethidium and a dimeric product of HE (E+ and E+-E+, Fig. 4) in isolated mitochondria and upon incubation of HE with heme proteins [27, 28]. Under similar experimental conditions, mito-metformin10 did not stimulate superoxide formation in non-cancerous human pancreatic epithelial renin-expressing (HPNE) cells (Fig. 4).

Fig. 4
figure 4

Intracellular superoxide stimulation by mito-metformin10. a Chemical scheme of superoxide-dependent and independent pathways of oxidation of hydroethidine (HE) probe. b Effect of Met (1 mM, 24 h), Mito-Met10 (1 µM, 24 h) or rotenone (ROT, 1 µM, 1 h) on HE oxidation in MiaPaCa-2 pancreatic cancer cells. Upper panel shows the HPLC chromatograms and lower panels shows the quantitative analysis of intracellular levels of HE and its oxidation products [7]

Full size image

We used the probe o-MitoPhB(OH)2 to detect H2O2 generated in mito-metformin-treated cells [29–31]. Results show that mito-metformin10 treatment of MiaPaCa-2 cells in the presence of o-MitoPhB(OH)2 led to an increase in o-MitoPhOH formation (Fig. 5). Although this product can be formed from either H2O2, peroxynitrite, or hypochlorous acid reaction with o-MitoPhB(OH)2, the lack of detection of peroxynitrite- or hypochlorous acid-specific products in addition to o-MitoPhOH suggests that H2O2 was responsible for oxidation of o-MitoPhB(OH)2 to MitoPhOH [29–31].

Fig. 5
figure 5

Hydrogen peroxide formation induced by mito-metformin10 in PDACs. a Chemical scheme of H2O2-dependent oxidation of the o-MitoPhB(OH)2 probe. b Intracellular levels of o-MitoPhB(OH)2 (left panel) and o-MitoPhOH (right panel) in MiaPaCa-2 pancreatic cancer cells treated for 24 h with Mito-Met10

Full size image

Mito-metformin10 activated AMPK phosphorylation at a 1000-fold lower concentration than metformin [7]. AMPK is a master regulator of cellular energy homeostasis and is activated via phosphorylation of its threonine-172 residue [32]. AMPK activation was linked to increased formation of hydrogen peroxide [33, 34]. Based on the results showing that mito-metformin10 stimulated mitochondrial superoxide, hydrogen peroxide and AMPK activation, we proposed that mito-metformin10-mediated antiproliferative effects are linked to inhibition of mitochondrial complex, enhanced ROS formation and AMPK phosphorylation (Fig. 6). These observations have been recently confirmed by an independent study on antiproliferative effects of triphenylphosphonium-linked metformin derivatives against pancreatic cancer cells [35].

Fig. 6
figure 6

Proposed signaling pathway induced by mito-metformin10 in PDACs. Block arrows indicate the changes observed upon cell treatment with Mito-Met10

Full size image

Potential use of Ex Vivo Low-Temperature EPR in Elucidating Mitochondrial Mechanism

Whereas the Seahorse technique can be used to measure the activities of mitochondrial complexes in cells, it is challenging to use this technique to measure mitochondrial activities in tissues. The low-temperature electron paramagnetic resonance (EPR) spectroscopy is uniquely capable of providing a “snapshot” of mitochondrial status in intact tissues [36]. To our knowledge, the EPR technique has not been used to monitor changes in the redox status of mitochondrial iron-sulfur complexes in cancer cells. The effect of mito-metformin10 and metformin on complex I activity in tumor tissues can be determined by performing EPR measurements at liquid helium temperatures. The signal at g = 1.94 attributable to the reduced form of [2Fe-2S]+ center of mitochondrial complex I is monitored [36, 37]. Inhibition of this signal is equated to inhibition of complex I. In a recent publication, Bennett et al. discussed the use of low-temperature EPR in mitochondrial diseases [38].

ROS generated from inhibiting complex I will react with other iron sulfur-containing proteins (e.g., mitochondrial and cytosolic aconitases), and the levels of oxidatively-inactivated aconitases can be monitored at g = 2.03 and 2.01 EPR signals due to the [3Fe-4S]+ form of aconitase [37, 39]. Oxidatively-inactivated mitochondrial aconitase has a distinctly different EPR signal from oxidatively-inactivated cytosolic aconitase [37, 39]. Other signals from reduced and oxidized iron sulfur centers associated with downstream complexes I and II in the electron transport chain are also detectable using the low-temperature EPR. These experiments can be done in both preclinical and human tumor samples, and may provide additional mechanistic insights into response to therapy.

Conclusions

We have demonstrated that more potent analogs of metformin (a.k.a. mito-metformins) can be synthesized by attaching a mitochondria-targeting triphenylphosphonium group tethered to an alkyl side chain. Mito-metformin10 was determined to be the most potent analog tested in inhibiting mitochondrial respiration, ROS activation, AMPK phosphorylation, and inhibition of proliferation in pancreatic cancer cells. We propose that AMPK activation induced by complex I-mediated ROS signaling is responsible for the antiproliferative effects. More recently, hydrogen peroxide was reported to induce redox signaling via a redox-relay involving peroxiredoxins and STAT3 [40]. One of the major targets of mito-metformins is the mitochondrial complex I in the electron transport chain. The exact mechanism by which metformin analogs inhibit complex I is not yet ascertained. It has been reported that the occurrence of mitochondrial complex I A/D transition could regulate ROS production [41]. The D-form of the enzyme has been shown to generate more O2 •– than the A-form, and irreversible locking of complex I in the D-form may contribute to more O2 •– formation [41]. Establishing this mechanism in the in vivo preclinical pancreatic cancer xenografts may be accomplished using the low-temperature EPR technique. This ex vivo EPR technique may also be extrapolated to analysis of human pancreatic tumors.