Why AMPK agonists not known to be stressors may surprisingly contribute to miscarriage or hinder IVF/ART

Abstract

Here we examine recent evidence suggesting that many drugs and diet supplements (DS), experimental AMP-activated protein kinase (AMPK) agonists as well as energy-depleting stress, lead to decreases in anabolism, growth or proliferation, and potency of cultured oocytes, embryos, and stem cells in an AMPK-dependent manner. Surprising data for DS and drugs that have some activity as AMPK agonists in in vitro experiments show possible toxicity. This needs to be balanced against a preponderance of evidence in vivo that these drugs and DS are beneficial for reproduction. We here discuss and analyze data that leads to two possible conclusions: First, although DS and drugs that have some of their therapeutic mechanisms mediated by AMPK activity associated with low ATP levels, some of the associated health problems in vivo and in vitro fertilization/assisted reproductive technologies (IVF/ART) may be better-treated by increasing ATP production using CoQ10 (Ben-Meir et al., Aging Cell 14:887–895, 2015). This enables high developmental trajectories simultaneous with solving stress by energy-requiring responses. In IVF/ART, it is ultimately best to maintain handling and culture of gametes and embryos in the quietest state with low metabolic activity (Leese et al., Mol Hum Reprod 14:667–672, 2008; Leese, Bioessays 24 (9):845–849, 2002) using back-to-nature or simplex algorithms to identify optima (Biggers, Reprod Biomed Online 4 Suppl 1:30–38, 2002). Stress markers, such as checkpoint proteins like TRP53 (aka p53) (Ganeshan et al., Exp Cell Res 358:227–233, 2017); Ganeshan et al., Biol Reprod 83:958–964, 2010) and a small set of kinases from the protein kinome that mediate enzymatic stress responses, can also be used to define optima. But, some gametes or embryos may have been stressed in vivo prior to IVF/ART or IVF/ART optimized for one outcome may be suboptimal for another. Increasing nutrition or adding CoQ10 to increase ATP production (Yang et al., Stem Cell Rev 13:454–464, 2017), managing stress enzyme levels with inhibitors (Xie et al., Mol Hum Reprod 12:217–224, 2006), or adding growth factors such as GM-CSF (Robertson et al., J Reprod Immunol 125:80–88, 2018); Chin et al., Hum Reprod 24:2997–3009, 2009) may increase survival and health of cultured embryos during different stress exposure contexts (Puscheck et al., Adv Exp Med Biol 843:77–128, 2015). We define “stress” as negative stimuli which decrease normal magnitude and speed of development, and these can be stress hormones, reactive oxygen species, inflammatory cytokines, or physical stimuli such as hypoxia. AMPK is normally activated by high AMP, commensurate with low ATP, but it was recently shown that if glucose is present inside the cell, AMPK activation by low ATP/high AMP is suppressed (Zhang et al., Nature 548:112–116, 2017). As we discuss in more detail below, this may also lead to greater AMPK agonist toxicity observed in two-cell embryos that do not import glucose. Stress in embryos and stem cells increases AMPK in large stimulation indexes but also direness indexes; the fastest AMPK activation occurs when stem cells are shifted from optimal oxygen to lower or high levels (Yang et al., J Reprod Dev 63:87–94, 2017). CoQ10 use may be better than risking AMPK-dependent metabolic and developmental toxicity when ATP is depleted and AMPK activated. Second, the use of AMPK agonists, DS, and drugs may best be rationalized when insulin resistance or obesity leads to aberrant hyperglycemia and hypertriglyceridemia, and obesity that negatively affect fertility. Under these conditions, beneficial effects of AMPK on increasing triglyceride and fatty acid and glucose uptake are important, as long as AMPK agonist exposures are not too high or do not occur during developmental windows of sensitivity. During these windows of sensitivity suppression of anabolism, proliferation, and stemness/potency due to AMPK activity, or overexposure may stunt or kill embryos or cause deleterious epigenetic changes.

Introduction. Recent examples of compounds that have some activity as AMPK agonists; diet supplements, and drugs that have negative effects on reproduction or development

Recent research shows that AMP-activated protein kinase (AMPK) mediates or is associated with stress-driven loss of potency factor proteins in cultured embryonic stem cells (ESCs), placental trophoblast stem cells (TSCs), and AMPK agonists that block reprogramming which produces induced pluripotent stem cells (iPSCs) by preventing switch to Warburg metabolism and blocking Oct4 transcription [1–3]. AMPK activity is associated with decreased potency in oocytes [4], and AMPK activity causes potency loss in two-cell embryos [5–7] and blastocysts [5, 7, 8]. AMPK inhibitors, compound C and AraA, block stress-forced potency loss in ESCs, TSC blastocysts, and two-cell embryos suggesting that AMPK mediates potency loss.

Several reports prove that non-stressful activators of AMPK such as drugs (aspirin, metformin) and diet supplements (BioResponse 3,3′-Diindolylmethane [BR-DIM]) cause growth arrest or decrease and potency decrease in cultured two-cell embryos and ESCs tested in high throughput screen (HTS) [6, 9–11]. The benefit of the HTS is that it enables higher numbers of doses and rapid testing of effects of single of multiple stimuli with simultaneous outcomes of growth and stemness or differentiation. It was also shown that diet supplements (DS) and drugs that have some activity as AMPK agonists retard and arrest stem cell growth soon after exposure of cultured two-cell embryos ~ 2–3 days before the blastocyst stage. However, it has not been shown yet which stage of embryo, two-cell or blastocyst, is most sensitive to stress-, DS-, or drug-caused potency or growth decrease. In the bovine model, cultured two-cell stage embryos are more sensitive to metformin than maturing oocytes for growth and developmental arrest [10]. In the mouse model, the two-cell stage embryo is more sensitive than the blastocyst stage for several AMPK-dependent outcomes [5]. This may support the hypothesis that early post-fertilization events are susceptible to precise energy management, but in vivo data are required to confirm relevance of in vitro outcomes. This hypothesis is supported by the observation that most AMPKα1 null mutants survive the two- to four-cell stage in vivo and survive to birth, whereas most cultured AMPKα1 null embryos die at the two- to four-cell stage [12]. The authors of this study also suggest that metformin, partially through AMPK activity, can have negative epigenetic and spermatogenic effects [13, 14]. Together, the data suggest a need to test whether embryos (especially two-cell embryos) use AMPK to adapt to stress, but when over-activated, AMPK can also be dangerous for embryo health. There may be conditions of AMPK activation in vivo that would have the same negative effects observed in vitro, but this remains to be tested.

Compounds with some benefits due to AMPK agonism: there is evidence for efficacy but not all therapeutic mechanisms are AMPK-dependent

The negative experimental data are surprising since most reports of DS and drugs that have some activity as AMPK agonists show beneficial exposures effects. What are the reasons AMPK agonist improve oocyte and embryo progression in vitro? Are there specific AMPK agonist dose and exposure ranges that are beneficial? Are the effects beneficial only when compared with a control that is unhealthy (i.e., when embryos are isolated from diabetic mothers or are cultured under intrinsically stressful conditions)? Are the embryos or oocytes that are improved during culture already compromised in vivo to the point where AMPK agonists are needed to rescue poor health?

Oocyte stress by culture alone or by several stressors such as arsenite and hyperosmotic sorbitol is rescued for meiotic resumption by AMPK agonists; however, these effects are a rescue and do not emulate the speed of magnitude of oocyte maturation in vivo [15–17] and these studies have not culminated with successful fertilization, implantation, and parturition from oocytes treated with AMPK agonists. Reports of improvement are mostly for treatments in vivo, but oocytes and embryos from diabetic or experimentally obese females are improved by metformin and other drugs that have some activity as AMPK agonists [18–20]. It should be noted that obese women have other problems such as reactive oxygen species that can disturb one carbon metabolism and the status of DNA methylation [21]. However, these oocytes and embryos from obese women are also likely to have AMPK activity at abnormally low levels (e.g., suppressed by high ATP when high glucose and triglyceride uptake occurs in obese mothers) or require AMPK function in increasing uptake of glucose and triglcyerides and fatty acids when they are insulin resistant. Moreover, oocytes and embryos from normal mothers may suffer if exposed to AMPK agonists from many sources, for example due to stress, drugs, or DS or due to high doses.

The reasons why there is a lack of a literature on AMPK agonist toxicity were discussed previously in the first report of AMPK agonist that affects on arrest of two-cell embryo development [6]. The drugs and DS discussed here are known mostly for other mechanisms than AMPK agonism. For example, metformin is used to improve ovulation in infertile women with polycystic ovarian syndrome (PCOS) [22–24], or type 2 diabetes (T2D). Metformin overcomes insulin resistance to enable ovulation in infertile women largely by blocking glucagon-induced cAMP and inhibiting protein kinase A [25–27], but rosiglitzone counteracts insulin resistance and increases androgens in women with PCOS [28]. Aspirin is an anti-inflammatory, antipyretic, and analgesic with inhibitory effects on prostaglandin production and irreversibly inhibits cyclooxygenase (COX) 1/2 activity and is used by fertile and infertile women [23, 29–39]. Thus, although metformin [40] and Asa [41] have therapeutic mechanisms partly through AMPK activity, there are many other effects that are AMPK-independent. In some drugs and DS, AMPK-dependent mechanisms may be additively or synergistically less important because other therapeutic mechanisms are more important. But, the possibility of AMPK agonism overexposure exists because of the preponderance of DS, with some drugs that have some AMPK agonism.

The drug DS BR-DIM can improve maternal fertility and diabetic metabolism. BR-DIM is a DS derived from yellow cruciferous vegetables and is a proprietary derivative of DIM that is absorbed better from the intestines [42]. DIM and BR-DIM act through mechanisms as an androgen antagonist and as a histone deacetylase inhibitor [43, 44]. However, BR-DIM diminishes growth or kills prostate cancer cells that have some stem cell properties, via AMPK-dependent mechanisms [45] in vivo and in vitro. Although diminishing growth in cancer stem cells is beneficial, diminishing growth in stem cells of early embryos would be detrimental if in vitro results repeat during gestational exposures. Most DS are mild mitochondrial poisons and lower ATP and increase AMP due to poorer mitochondrial function. Like metformin and aspirin, diet supplements like BR-DIM mediate some of their activities through AMPK as well as other, sometimes better known mechanisms.

Commonly used drugs that have some activity as AMPK agonists include metformin and phenformin, aspirin, rosiglitazone, and troglizitazone and diet supplements such as resveratrol, green/black tea, berberine, and BR-DIM [1]. These italicized drugs and DS, with the exception of aspirin which directly contacts and decreases activity of the AMPK-negative regulatory β-subunit (decreasing the off-rate), are mitochondrial inhibitors which decrease ATP, increase AMPK, and activate the γ-subunit [46] (Fig. 1). For example, metformin was shown to bind to and inhibit complex I of the electron transport chain (ETC) of mitochondrial oxidative phosphorylation resulting in highly efficient ATP production, which also leads to an increase in mitochondrial superoxide production [47], modulating redox-regulated signaling pathways. Additional mitochondrial targets exist including glycerophosphate dehydrogenase, which metformin binds leading to enzyme inhibition and thus contributes to an increase in the cellular redox state [48]. The guanide phenformin has an even broader inhibitory spectrum by targeting ETC complexes II and IV in addition to complex I [49] as well as ATP synthase [50]. As a result, ATP levels drop leading to an increase of the AMP/ATP ratio and consequent activation of AMPK. However, metformin may also activate AMPK through direct contact in some circumstances [51]. Thus, these drugs increase the on-rate. In cells with mutant AMPK γ-subunits, these underlined drugs and DS do not have activated AMPK or cause their AMPK-dependent effects. Interestingly, the DS CoQ10 is exceptional in increasing electron transport chain activity and thus mitochondrial ATP production and may improve reproductive efficiency [52] but obviate the need to use AMPK agonists.

Fig. 1.

The AMPK heterotrimer is activated at the γ-subunit indirectly by Met, BR-DIM. The negative regulator β-subunit is deactivated (off-rate) and AMPK activated directly by Asa or A769662 which bind ser108

Generally, metformin is considered to be safe during pregnancy and its role in correcting metabolic derangement makes it beneficial for patients with diabetes during pregnancy. Animal studies using dosages up to 600 mg/kg daily did not show teratogenic effects and extremely high dosages between 900 and 1500 mg/kg also failed to induce carcinogenicity [53]. Using ex vivo perfusion of the placental cotyledon model, Kovo et al. demonstrated that the transport of metformin through placenta is mediated by carrier and at 1 mg/ml, the transport of metformin is saturated [54]. The potential competition between metformin and other drugs in the use of transporter can also limit fetal exposure.

However, it should be noted that the beneficial effects of metformin are partly conveyed by activating AMPK to counteract stress [55]. AMPK can also be activated by multiple other stressors. Over-activation of AMPK by both environmental stress and drugs may contribute to adverse outcomes. It has been reported that stimulation of AMPK in embryos in diabetic rodents can lead to neural tube defects through suppression of PAX3 [56]. We do not have sufficient data to exclude the possibility that human embryos may be subjected to the negative effect of metformin. The current evidences support the efficacy and safety of metformin in regard to immediate pregnancy outcomes [57]. It is important to follow the offspring in order to determine whether metformin treatment of diabetes during pregnancy results in many changes in the risk of metabolic disorders through epigenetic programming [58].

So what are the mechanisms that prescribed or elective drugs and/or diet supplements use that modulate ATP and AMPK?

Of the > 500 protein kinases in the kinome [59, 60], AMPK is unique in regulating nearly all the catalytic and anabolic pathways of intermediary carbon metabolism that enable synthesis of carbon polymers: polynucleotides or DNA and RNA, fatty acids, starches such as glycogen, and proteins [61, 62]. As the cellular energy sensor, AMPK homologs are present in the earliest forms of life and are present in nearly all types of cells, including gametes and stem cells.

When ATP is low (and thus AMP is high), AMPK downregulates anabolic metabolism that consumes ATP and upregulates catabolic metabolism that produces ATP, thus rebalancing ATP equilibrium [62]. AMPK also increases cellular uptake of triglycerides and fatty acids, glucose, and other nutritional molecules to increase ATP. It is this later AMPK function that makes it important in the treatments of T2D, PCOS [12, 63], hypertriglyceridemia, and obesity that negatively affect maternal health and reproductive capacity.

The fact that AMPK also favors catabolic over anabolic function also makes diet supplements and drugs that are AMPK agonists attractive to women who wish to keep their weight low. AMPK favors oxidative phosphorylation over glycolytic metabolism as the former produces 38 ATP/glucose and the latter only 2 ATP/glucose [64, 65]. This means that AMPK brings glucose and triglycerides and fatty acids into cells, but tends to burn them to CO₂ in the mitochondria rather than use them in anabolic pathways that would lead to weight gain and production of glycogen and fat as well as RNA/DNA and protein.

This may be one global reason why all diet supplements listed at the US Office of Dietary Supplements (https://ods.od.nih.gov/) studied to date (i.e., > 50 of > 70 DS listed) activate AMPK, and many have substantial therapeutic activity through an AMPK-mediated switch from anabolic to catabolic pathways [66].

As mentioned previously, many drugs activate AMPK as a mitochondrial blocker (e.g., metformin) or by directly interacting with the AMPK heterotrimer and activating it (e.g., aspirin). This also suggests the potential for synergism between diet supplements and drugs may include the possibility that the AMPK heterotrimer may be hyper-activated by increasing enzyme on-rate through the γ-subunit and decreasing the off-rate through the β-subunit [66–68] (Fig. 1).

Many drugs and diet supplements that activate AMPK are in clinical trials to block cancer growth because AMPK’s properties lead to decreases in anabolic and increase in catabolic pathways. Also, cancer stem cells proliferate more rapidly because they favor aerobic glycolysis (glycolysis even when oxygen is present) which is also called “Warburg metabolism” [69]. These proliferating cells have suppressed Krebs’s tricarboxylic acid pathway (TCA) that in postmitotic cells produces high ATP through oxidative phosphorylation (OxPhos) but “wastes” carbon polymers from nutritional sources and converting these to CO₂. In summary, AMPK blocks Warburg and favors OxPhos as well as favoring catabolic over anabolic pathways. However, from the blastocyst stage onwards, stem cells of the implanting embryo use Warburg metabolism to enable rapid growth at the start of the exponential growth phase of embryogenesis [64].

Miscarriage is likely to have a metabolic link as stunting of embryonic and stem cell growth is linked to insufficient embryonic function to continue pregnancy [1, 70]. Most of miscarriage happens soon after implantation and about one third (35%) occurs after chemical (hCG; 20%) or clinical (ultrasound; 15%) detection of pregnancy. It is estimated that half of miscarriages are due to non-genetic outcomes. But genetic anomalies that slow anabolism and stem cell doubling rates are like stressors or AMPK agonists; they slow cell growth rates that may trigger compensatory mechanisms of forced differentiation. Thus, miscarriage may have a common mediator; genetically or environmentally caused insufficient stem cell proliferation in the stem cell “pipeline” required to produce early sufficient differentiated function.

The chances of overexposure in vitro and in vivo

Since DS and drugs are generally cleared in hours in vivo, in vitro stimulations tend to be overexposures in regard to biological effects measured in days [5, 6]. However, AMPK-dependent decreases in anabolism and potency occur within one hour before the peak blood dose is reached and therefore these exposures are not artifacts of in vitro overexposures. Consequently, these mechanisms may lead to growth arrest or decreases in proliferation.

There are two possibilities for underestimates in studies in vivo. These underestimates are based on in vitro studies discussed above; most studies test drugs or DS as single variables, but DS and drugs may be exposed in pairs or sets and at different times of the day increasing duration of DS and drugs that have some activity as AMPK agonists. In addition, the magnitude of AMPK activity can be increased by multiple DS and drugs and by activators of the on-rate of the γ-subunit simultaneously with decreased off-rate of the α-subunit. These two possible mechanisms for AMPK agonist toxicity have not been widely tested in vivo.

What should be done next?

In the IVF setting, metformin has a beneficial effect by increasing pregnancy rates and decreasing miscarriages [71]. We have discussed the concept before that augmenting energy production might be better than inhibiting it with drugs that act as inhibitors on mitochondrial function such as the biguanidine metformin. However, it should be noted that cellular energy levels can be well maintained as long as there is enough glucose. This is so because ATP production via substrate level phosphorylation in glycolysis, although less efficient, is faster than ATP production through oxidative phosphorylation [72]. Therefore, as an alternate possibility, it should be considered that cellular energy levels may not be that central for potency and growth during early embryogenesis at the blastocyst stage when glucose usage increases [73]. As long as glucose is plentiful, energy levels can be low; this concept has increasingly gained acceptance in the cancer field (reviewed in [74]), with cellular growth rates similar to those seen in the early embryo. Inhibition of mitochondrial catabolic activity—paralleled by a reprogrammed mitochondrial metabolism—may thus augment the generation of biosynthetic intermediates including NADPH, amino acids, cytosolic acetyl-CoA, and five-carbon sugars, all of which the proliferating cell heavily relies on. Future selective experimental manipulation of cellular energy versus metabolic intermediates may shed light on their respective importance for potency and embryonic growth. It should also be noted that it was recently reported that high levels of glucose maintain low AMPK activity and low glucose levels increase AMPK activity, independent of AMP levels [75, 76]. Thus, AMPK responds directly, but independently to glucose or AMP. The embryonic niche at 2% oxygen, coupled with sufficient glucose to sustain low glycolytic ATP production but high conservation of carbon for anabolic proliferation, and sufficient microfluidics flow at the blastocyst stage [64] should improve blastocyst culture.

At the two-cell embryo stage in mouse, bovine and human pyruvate usage is highest and supports development through the blastocyst stage [73, 77–79] better than glucose and this occurs at a time when pentose phosphate pathway usage is highest [79]. The pentose phosphate pathway is significant in producing reducing power through NADPH for fatty acid synthesis (as well as other macromolecular synthesis) and prevention of oxidative stress, as well as nucleic acid synthesis. However, the higher pyruvate uptake and lower level of glucose uptake may lead to increasing AMPK activation that is prevented by glucose [75].

As a possible outcome, direct partial metabolic inhibition at the level of the mitochondrial ETC by metformin and other compounds may prevail over mild AMPK-mediated activation of catabolism. In contrast, stress-induced AMPK hyperactivation would overcome this effect leading to potency loss as we and others have shown in vitro [3–8, 10, 64, 80–83]. The concepts of additive or synergistic negative AMPK effects in vitro [67, 84] and of repeated doses modeled in vitro and in vivo [85] need more in vivo testing with regard to AMPK-dependent miscarriage and lesser developmental pathogenesis. AMPK activity is a homeostatic mechanism that when optimized can return stressed embryos a developmental trajectory that emulates those of the “quiet embryo” [86, 87, 88]. However suboptimal manipulation of AMPK activity can create an artifactual “noise” that is not desired. Whether CoQ10 has greater potential to return stressed embryos and oocytes to a developmental trajectory with less noise than AMPK agonism needs further testing. The emphasis has been on modeling the capabilities of the response of embryos and their stem cells in vitro, and these hypotheses need to be tested in vivo where indirect maternal effects and accessibility to embryos or their stem cells may be limiting.

Funding

This study received funding from DAR and EEP from the Office of the Vice President for Research at Wayne State University and an R03 to DAR 1R03HD061431 and an R41 to DAR 1R41ES028991-01 from the REI fellows’ fund (AB) and from the funding of the Mary Iacobell and Kamran Moghissi Endowed Chairs.