Our studies were designed to determine if there might be a biological, non-cytotoxic mechanism to explain several epidemiological and experimental in vitro and in vivo studies, suggesting an anti-cancer effect of metformin. By using human breast cancer cell lines, with and without the estrogen receptor, grown in 3-dimension to try to mimic some in vivo conditions, we show that several known growth promoters or endocrine disruptors, 17-beta estradiol, phenol red, TCDD, and bisphenol A, did stimulate human breast cancer stem cells, as evidenced by both the number and sizes of MCF-7 mammospheres with the estrogen receptor and that metformin could suppress this stimulated MCF-7 cancer stem cell growth at non-cytotoxic concentrations.
The major findings of our studies demonstrated that (a) the 3-dimensional mammospheres could be used to detect both agents that stimulated or inhibited the numbers and growth of these estrogen receptor- positive human breast carcinoma cells; (b) that 17-beta estradiol, and the endocrine disruptors, phenol red, TCDD, bisphenol A, stimulated both the numbers and growth of the MCF-7 cancer stem cells; (c) that metformin could inhibit the mitogenic stimulus of estrogen and the endrocrine disruptors at non-cytotoxic concentrations; (d) that at low concentrations, estrogen stimulated the growth of the mammospheres, probably by an estrogen-dependent mitogenic signaling mechanism, whereas at higher concentrations, growth inhibition occurred, probably by some estrogen receptor -independent oxidative stress-induced mechanism, that blocked the estrogen-dependent signaling; (e) estrogen signaling increased OCT4 expression, while metformin interrupted estrogen-induced OCT4 expression; and (f) the mechanism by which bisphenol A enhanced MCF-7 cancer stem cell self-renewal, as evidenced by mammosphere numbers and growth were different than the mechanisms by which estrogen and TCDD worked.
One of the markers used to monitor the effect of both the estrogen and estrogenic-like compounds, as well as the effect of metformin, was the OCT4 gene. The OCT4
gene is a member of POU family and functions as a transcription factor 
. This gene is expressed in embryonic stem (ES) cells, germ cells 
, and adult human stem cells 
, while it helps to maintain an undifferentiated state and to prevent differentiation 
. In this study, we showed that OCT4
expression can be induced by 10 nM 17-beta-estradiol in MCF-7 mammospheres. Usually estrogens act through two kinds of pathways, namely, an estrogen receptor-dependent pathway and an estrogen receptor-independent pathway in the cells 
Estrogens bind to the estrogen receptor of the nucleus to form ER-estrogen complexes in the ER-dependent pathway. These complexes might affect, directly, OCT4
expression by binding to the OCT4
gene promoter region, thereby, activating gene transcription. ER-estrogen complexes might also affect, indirectly, OCT4
expression in relation to histone stability of OCT4
gene promoter. When ER-estrogen complexes bind to the estrogen responsive element (ERE) of target genes, p160 and p300 are recruited to the ER-estrogen complexes and then the PBP/TRAP220/DRIP205 subunit interacts with complexes 
. As these actions facilitate histone acetylation, the OCT4
promoter region could be exposed to other transcription factors, thereby, inducing OCT4
promoter activation. On the other hand, in the ER-independent pathway, estrogens might be metabolized to metabolites in cytoplasm. As a result, ROS are created. These ROS are the cause of oxidative stress. ROS induction of various intra-cellular signal transductors, for example, NF-κB, might be activated through this pathway 
. Activated NF-κB could lead to histone deacetylase (HDAC) activation, inhibiting OCT4
gene transcription. Recently, Itoh et al. reported that estrogen could dissociate physical incorporation of ER and HDAC2 which, in turn, could increase accessibility of ER-estrogen complex to promoter region of target genes 
. Moreover, they reported that treatment of E2 increased transcriptional activity of Sp1, Sp3 transcription factors against GC- rich Sp1, Sp3 site in IL-1α promoter region. Given that Sp sites are also present in OCT4
promoter region 
, it is reasonable to speculate that estrogen might affect OCT4
gene transcription directly, or indirectly.
In this study, 17-beta-estradiol (E2) might affect OCT4 expression through both pathways. In low concentrations, up to 20nM E2, the ER-dependent pathway might be activated to increase the OCT4 expression and a mitogenic response. On the other hand, ROS production might be increased through an ER-independent pathway rather than the ER-dependent mechanism in high concentration 100 nM. The oxidative stress-induced signaling could inhibit the mitogenic signals of the estrogen-dependent pathway. Indeed, repression of OCT4 expression in mammosphere treated with high concentration of E2 was seen. This suggests that ROS production, induced by 17-beta-estradiol metabolism, suppressed OCT4 expression.
This dramatic increased expression of the OCT4 after exposure of these cancer stem cells in mammospheres to estrogen suggests that estrogen stimulated the symmetrical cell proliferation of MCF-7 breast cancer stem cells. If this interpretation is correct, it becomes obvious that a potential human in vitro assay, using 3-dimenisional MCF-7 mammosphere culture, could be developed to screen for breast tumor promoters that might mimic what estrogen does to increase the expression of the OCT4 gene and lead to human breast cancer.
The normal human breast stem cell expresses OCT4
, does not express connexin43, and expresses the estrogen receptor, as well as other markers 
, similar to the MCF-7 cells. Therefore, in order to “target” the human breast “cancer stem cells”, one must design new chemopreventive and therapeutic strategies that will affect the expression of the (a) OCT4
(a gene needed to maintain the “stemness” of both the normal breast and breast “cancer” stem cells) and (b) connexin 43
gene (a gene required for allowing differentiation to occur in the normal human breast stem cells). Rather than trying to “kill” tumor cells or even the “cancer stem cells”, altering the expression of these two genes could induce these “cancer stem cells” to terminally differentiate. However, up until now, there has been no systematic approach to screen for agents that might directly affect the OCT4
and connexin 43
genes, with the possible exception of the study with SAH 
. Therefore, suspected human breast “carcinogens” could be detected by an increase of the expression of OCT4
and estrogen receptor in the mammospheres, while continuing to suppress the expression of connexin 43
The results of these studies, based on the use of OCT4 as a normal or cancer stem cell marker and the three-dimensional mammospheres, implies many additional basic mechanistic experiments should be done to understand the biology of breast cancer stem cells and to screen for human breast tumor promoters and preventive/therapeutic agents for breast cancer. However, these studies do provide some mechanistic support for the epidemiological observations that metformin could be a useful anti-breast cancer chemopreventive treatment.