Pharmacological Mimicking of Caloric RestrictionElicits Epigenetic Reprogramming of Differentiated

Cells to Stem-Like Self-Renewal States

Cristina Oliveras-Ferraros,1,2 Alejandro Vazquez-Martin,1,2 and Javier A. Menendez1,2

Abstract

Networks of oncogenes and tumor suppressor genes that control cancer cell proliferation also regulate stem cellrenewal and possibly stem cell aging. Because (de)differentiation processes might dictate tumor cells to retro-gress to a more stem-like state in response to aging-relevant epigenetic and/or environmental players, werecently envisioned that cultured human cancer cells might be used as reliable models to test the ability ofantiaging interventions for promoting the initiation and maintenance of self-renewing divisions. Cancer cell linesnaturally bearing undetectable amounts of stem/progenitor-like cell populations were continuously cultured inthe presence of the caloric restriction mimetic metformin for several months. Microarray technology was em-ployed to profile expression of genes related to the identification, growth, and differentiation of stem cells.Detection of functionally related gene groups using a pathway analysis package provided annotated geneticsignatures over- and underexpressed in response to pharmacological mimicking of caloric restriction. By fol-lowing this methodological approach, we recently obtained data fitting a model in which, in response to chronicimpairment of cellular bioenergetics imposed by metformin-induced mitochondrial uncoupling as assessed bythe phosphorylation state of cAMP-response element binding protein (CREB), tumor cells can retrogress from adifferentiated state to a more CD44þ stem-like primitive state epigenetically governed by the Polycomb-groupsuppressor BMI1—a crucial ‘‘stemness’’ gene involved in the epigenetic maintenance of adult stem cells. Thesefindings might provide a novel molecular avenue to investigate if antiaging benefits from caloric restrictionmimetics might relate to their ability to epigenetically reprogram stemness while prolonging the capacity ofstem-like cell states to proliferate, differentiate, and replace mature cells in adult aging tissues.

Plant Stress and Dedifferentiation to Stem-LikeStates: Beyond Xenohormesis

Molecular responses to environmental stresses,including pathogen infection, famine, and nutrient

starvations, can promote genomic reprogramming of tran-scriptional activity required for plant cells to undergo dedif-ferentiation and assume a stem cell-like state.1 The notion that,upon certain stress conditions, cell responses converge ondedifferentiation processes whereby cells first acquire the stemcell-like state prior to acquisition of a new cell fate (e.g., re-entryinto the cell cycle, or death), inherently offers a novel view onthe xenohormesis hypothesis.2 The latter suggests that therehave been selection such that heterotrophs (animals and fungi)can efficiently detect chemical cues about their environmentfrom plants and other autotrophs, and that stress-inducedplant molecules (i.e., caloric restriction mimetics [CRMs]) pro-

vide the heterotroph advance warning about the deteriorationof the environment by inducing defense responses that ulti-mately lead to an extended life span. If stress-induced rejuve-nated, stem cell-like states occurring in dedifferentiatingprotoplast (plant-derived) cells could be induced by similarstress conditions in animal cells (e.g., CRM-induced repro-gramming of differentiated adult somatic cells to stem-likeself-renewal states), a ‘‘stemness’’-directed perspective of thexenohormetic hypothesis might have previously unrecognizedimplications for reversal/retarding of human aging.

CRMs and Self-Renewal Divisions in Human Aging:A Need for Model Systems

Self-renewing divisions of somatic stem cells are largely re-sponsible for the maintenance and regeneration of tissues and,consequently, exhaustion of their replicative capacity likelycontributes to tissue aging. In this regard, it is widely accepted

1Catalan Institute of Oncology (ICO), Girona, Catalonia, Spain.2Girona Biomedical Research Institute (IdIBGi), Girona, Catalonia, Spain and University Hospital of Girona Dr. Josep Trueta, Girona,

Catalonia, Spain.

REJUVENATION RESEARCHVolume 13, Number 5, 2010ªMary Ann Liebert, Inc.DOI: 10.1089/rej.2010.1022

519

520 OLIVERAS-FERRAROS ET AL.

that networks of oncogenes and tumor suppressors haveevolved to co-ordinately regulate stem cell function throughoutlife.3–5 The activity of oncogenes works to promote regenerativecapacity, thus enhancing stem cell function. Obviously, thispromoting activity should be balanced with tumor suppres-sor activity to avoid neoplastic proliferation. The activity oftumor suppressors works to inhibit regenerative capacity bypromoting cell death or senescence in stem cells. Therefore, (1)tumor surveillance mechanisms could reduce regenerativecapacity and frequency of normal stem cells, thus contributingto tissue aging, and (2) an imbalance within tumor suppressor/oncogene networks could cause cancer or premature declinesin stem cell activity that resemble accelerated aging. Changesin the expression of significant numbers of genes have beenlinked repeatedly to the lifespan prolonging and anticancereffects of lifespan interventions (i.e., caloric restriction [CR] andCRMs).6–9 However, there have been few studies addressingthe ability of CR and/or CRMs to alter stem cell self-renewalgene networks. Furthermore, we can anticipate that this issuecould be difficult to study, in part because of the presupposedlack of appropriate model systems.

Human Cancer Cell Cultures: Models to AssessIf Antiaging Interventions Promote Initiationand Maintenance of Self-Renewing Division

Although the term ‘‘cancer stem cell’’ does not imply thatthe cell is derived from a normal stem cell, breast cancer celllines retain the hierarchy characteristic of primary tumors andthey contain subpopulations of replenishing stem-like cellswith the ability both to self-renew and give rise to pheno-typically diverse progeny.10 Indeed, it remains to difficult todetermine whether a cancer stem cell is derived from a so-matic stem cell, from a (de)differentiated progenitor or even aterminally differentiated cancer cell. Thus, apart from bona fidegenetic lesions, (de)differentiation processes can also dictatetumor cells to retrogress to a more stem-like state in responseto relevant epigenetic and/or environmental players. In this

regard, recent studies have suggested that the percentage ofstem-like cells within a line can vary depending on conditionssuch as cell density, the presence/absence or growth factors,and even the frequency of passaging.10 Altogether, this ex-perimental evidence suggests that the dynamic nature ofstem-like cell populations can be actively regulated by theirimmediate microenvironment.11 Recently, we envisioned thatbreast cancer cell lines might be used as reliable models to testindirectly the ability of antiaging interventions to promoteinitiation and maintenance of self-renewing divisions.

We employed p53 wild-type MCF-7 cells to test ourworking model. Flow cytometry–based experiments con-firmed that MCF-7 luminal breast cancer cells naturally bearundetectable amounts of CD44pos/CD24neg/low stem/progenitor cell populations (data not shown),10,12 thus cir-cumventing the possibility that metformin treatment wouldselect for metformin-resistant proliferating breast cancerstem cells following metformin-induced killing of (differen-tiated) daughter breast cancer cells. MCF-7 cell cultures wereexposed continuously to noncytotoxic concentrations ofmetformin (5 mmol/L) in routine culture medium that wasreplaced every day for about 2 months.

Initially, MCF-7 cell numbers were somewhat reducedand metformin-treated cultures displayed signs of transientsenescence, as determined by the presence of senescence-associated b-galatosidase (b-Gal) staining at pH 6.0 (SA-b-Galþ cells; data not shown). For the next 2 months, MCF-7cells were passaged in the presence of up to 10 mmol/Lmetformin approximately every 10 days with a seeding ratioof 1:2. Cell proliferation significantly increased to allow apassage every 5–7 days with a seeding ratio of 1:4 over thenext 2 months. A stable growth rate (slightly higher thanmetformin-naıve MCF-7 parental cells) was reached after atotal of 6 months with routine maintenance of the LongTerm-Metformin adapted MCF-7 cells (LT-Met MCF-7) in-volving passage every 4 days with a seeding ratio of 1:5 of&75% confluent cell cultures growing in the presence of10 mmol/L metformin (Fig. 1, top panel). We then employed

FIG. 1. Chronic culture of breast epithelial cancer cells with the caloric restriction mimetic (CRM) metformin and microarrays-based global assessment of changes in stem cell–related gene expression: A reliable experimental model to indirectly andrapidly test the ability of anti-aging interventions to promote initiation and maintenance of self-renewing divisions. (Top panel)Scheme for development of Long Term-Metformin adapted MCF-7 cells (LT-Met MCF-7). (Middle panel) Total RNA isolatedfrom MCF-7 and LT-Met MCF-7 cells was isolated with TRIzol reagent (Invitrogen, Carlsbad, CA), according to the manu-facturer’s instructions. RNA quantity and quality were determined using the RNA 6000 Nano Assay kit on an Agilent 2100BioAnalyzer (Agilent Technologies, Palo Alto, CA), as recommended. Oligo GEArray� Human Stem Cell Microarrays(HybPlate Format Cat. No.: EHS-405, SABiosciences Corportation) were then hybridized as per the manufacturer’s instruc-tions. Both data acquisition and data analysis were performed using GEArray Expression Analysis Suite, a web-based inte-grated software package for the GEArray. The center line (the line of unity) indicates no changes in gene expression and parallellines move according a predefined boundary for gene expression changes (i.e., 2.0-fold). Black symbols indicate gene expressionchanges less than 2.0-fold change. Red symbols indicate increase in gene expression from x axis to y axis greater than 2.0-foldchange. Green symbols indicate decrease in gene expression from x axis to y axis greater than 2.0-fold change. The figure showsraw data from two representative experiments. (Bottom panels) Gene Networks were constructed using Ingenuity PathwaysAnalysis (Ingenuity� Systems, Redwood City, CA). Datasets containing gene identifiers of genes with>2.0-fold were uploadedinto the applications. These ‘‘focus genes’’ were overlaid onto a global molecular network developed from information con-tained in the Ingenuity Pathway Knowledge Base. Networks of these focus genes were algorithmically generated based on theirconnectivity. Briefly, genes or gene products are represented as nodes, and the biological relationship between two nodes isrepresented as an edge (line). All edges are supported by at least one reference from the literature stored in the IngenuityPathway Knowledge Base. The intensity of the node color indicates the degree of expression (green scale for downregulatednodes; red scale for upregulated nodes). Nodes are displayed using various shapes, each representing functional class of thegene product. A solid line indicates a direct interaction whereas a dashed line indicates an indirect interaction. The scoreindicates the likelihood of the genes in a network being found together because of random chance. Using a 99% confidenceinterval, scores of �3 are significant. (Color image is available online at www.liebertonline.com/rej).

‰

METFORMIN AND STEMNESS 521

Oligo GEArray� Human Stem Cell Microarray technology toprofile the expression of 263 genes related to the identifica-tion, growth and differentiation of stem cells. Thus, throughside-by-side hybridization experiments, we were able todetermine differential gene expression of stem cell–specificmarkers, stem cell differentiation markers, as well as sig-naling genes that function in the maintenance of stem cells

before and after chronic adaptation of MCF-7 breast epithe-lial cells to the CRM metformin.

To create genetic signatures that were over- and under-expressed in LT-Met MCF-7 cells, the data obtained wereimported and filtered into the GEArray Expression AnalysisSuite, a web-based integrated software package for theGEAarray (Fig. 1, middle panel). Sixteen stem cell-related

522 OLIVERAS-FERRAROS ET AL.

genes were significantly upregulated during chronic adap-tation of MCF-7 cells to metformin (fold change): FGFR1(þ37), BMI1 (þ31), PARD6B (þ25), BMP8A (þ19), BMP2(þ19), NES (þ18), NOTCH4 (þ14), NCAM1 (þ12), CCNE1(þ11), FGF13 (þ11), PPARG (þ9), CD3D (þ8), CDH4 (þ7),ALPPL2 (þ6), RBPJL (þ5), ALDH1A1 (þ5). Concomitantly,28 stem cell–related genes were significantly downregulated(fold-change): CDKN1A (�48), CDK4 (�27), HSPA9 (�19),RHOC (�19), ALDH2 (�14), COL1A1 (�13), COL2A1 (�11),GDF3 (�9), PDGFB (�9), COL9A1 (�8), HDAC1 (�7), CDH3(�7), SHH (�7), PPARD (�6), TUBB3 (�5), SOX3 (�4), UTF-1 (�4), STAU1 (�4), ALPL (�4), RHOA (�4), BMP10 (�4),PSEN2 (�4), RHOB (�3), ACTC1 (�3), RB1 (�3), TFRC (�2),PSEN1 (�2), ABCG2 (�2).

We further analyzed the data using a pathway analysispackage (Ingenuity) to detect groups of functionally relatedannotated genes. Networks of up- and downregulatedgenes were algorithmically generated based on their con-nectivity using Ingenuity Pathway Analysis (IPA) softwareand assigned a score. The score is a numerical value thattakes into consideration both the number of focus genes in anetwork and the size of the network to approximate howrelevant this network could be to the original list of focusgenes. Therefore, the score assigned by IPA analysis maynot be an indication of the quality or significance of thenetwork, but it ranks networks according to how relevantthey are to the genes in the input database. IPA analysisrevealed only one significant gene network of over-expressed genes in RNA samples obtained from LT-MetMCF-7 cells (score 26; focus genes 10; Fig. 1, bottom panel).The top functions of these upregulated genes (including keycomponents of the FGF signaling pathway—FGFR1 andFGFR13—and critical stemness genes that regulate self-renewal and developmental pathways—NOTCH4 andBMI1) were related to ‘‘cell-to-cell signaling and interaction,nervous system development and function, cellular devel-opment’’ and identified around mitogen-activated proteinkinase (MAPK). Accordingly, canonical pathway analysisrevealed that ‘‘factors promoting cardiogenesis in verte-brates’’ ( p value, 8.98E� 05) and ‘‘human embryonic stem cellpluripotency’’ ( p value, 2.88E� 04) were two importantpathways modulated by the upregulated genes in LT-MetMCF-7 cells. Among the three significant gene networks ofdownexpressed genes revealed by IPA analysis, the top

function with the highest score (27, focus genes 12; Fig. 1,bottom panel) related to ‘‘cancer, endocrine system disor-ders, reproductive system disease’’ and identified aroundCDKN1A (p21Waf1/Cip1), the expression of which wasdrastically inhibited along with other BMI1-related down-stream effectors including Rb (retinoblastoma) and CyclinD/cyclin dependent kinase (CDK) 4/6. Accordingly, ca-nonical pathway analysis revealed that ‘‘molecular mecha-nisms of cancer’’ ( p value, 3.21E� 10) and ‘‘cell cycle: G1/Scheckpoint regulation’’ ( p value, 3.84E� 06) were two im-portant pathways modulated by the downregulated genesin LT-Met MCF-7 cells.

Chronic Impairment of Cellular Bioenergeticsin Differentiated Tumor Cells: Evidence for EpigeneticReprogramming of Differentiated Cells to BMI1Epigenetically Governed Stem-Like States

There is some debate regarding whether uniformly acting‘‘stemness genes’’ exist at all. In addition, because cell-extrinsic signals are such important factors in the mainte-nance of the stem cell pool, it is crucial to analyze stem cellexpression profiles in the appropriate context of the stem cellniche. Nevertheless, it seems reasonable to designate at leastBMI1 as a gene intrinsically conferring stem cell character-istics to a cell.3,4 BMI1, a member of the Polycomb family ofgenes involved in chromatin remodeling and gene expres-sion, becomes elevated in response to extrinsic signals for astem cell to self-renew, and thus appears to be directly in-volved in decisions affecting stem cell fate, including self-renewal, senescence, and possible aging. This indispensablerequirement of BMI1 functioning for the maintenance ofadult stem cells relates to BMI1-repressing activity towardgenes that induce cellular senescence and cell death, thuspreventing senescence of stem cells. Importantly, recentstudies have revealed that BMI1 is a pivotal regulatorymolecule shared by breast cancer stem cells and their healthycounterparts, mammary stem cells.13 Therefore, BMI1-governed stem cell functions such as self-renewal and pro-liferation take place not just in healthy and disease tissue,but across different types of tissues. Indeed, CD44-overexpressing primitive human mammary cells display asignificant increase in BMI1 expression compared toCD44� isolated from the same mammary tissue.14

FIG. 2. Chronic culture of breast epithelial cancer cells with the caloric restriction mimetic (CRM) metformin: Epigeneticreprogramming to connect cellular energy homeostasis with stem cell markers. (Top) MCF-7 and LT-MCF-7 cells wereseeded at approximately 5,000 cells/well in 96-well, clear-bottomed imaging tissue culture plates (Becton Dickinson Bio-sciences, San Jose, CA) optimized for automated imaging applications. Triton� X-100 permeabilization and blocking andantibody staining (1:50 dilution of a mouse anti-human CD44 monoclonal antibody [mAb] obtained from BD Biosciences, SanDiego, CA), secondary antibody staining using Alexa Fluor� 594 goat anti-mouse IgG (Invitrogen, Probes, Eugene, OR), andDNA counterstaining (using Hoechst 33258; Invitrogen) were performed following BD Biosciences protocols. Images werecaptured in different channels for Alexa Fluor� 594 (pseudo-colored red) and Hoechst 33258 (pseudo-colored blue) on a BDPathway� 855 Bioimager System (Becton Dickinson Biosciences, San Jose, CA) with 20�objective (NA 075 Olympus). Imagesshow representative portions of MCF-7 (left) and LT-Met MCF-7 (right) cell cultures captured in different channels for CD44(red) and Hoechst 33258 (blue) with a 20� objective and merged using BD Attovision� software according to the Re-commended Assay Procedure (n¼ 3). (Bottom) Five hundred micrograms of lysates from MCF-7 (left) and LT-Met MCF-7(right) cell cultures were diluted and incubated with the Human Phospho-Kinase Antibody Array (Proteome Profiler� Array,R&D Systems, Minneapolis, MN) to simultaneously detect the relative site-specific phosphorylation of 46 proteins as per themanufacturer’s instructions. Array data were developed on X-ray film following exposure to chemiluminescent reagents.Images show representative portions of arrays from MCF-7 and LT-Met MCF-7 cells showing differential phosphorylation ofCREB on Ser-133 (n¼ 3).

‰

METFORMIN AND STEMNESS 523

Because MCF-7 luminal breast cancer cells naturally bearundetectable amounts of CD44pos/CD24neg/low stem/progenitorcell populations,10,12 our current findings fit a model in whichbreast cancer cells retrogress from a differentiated state to amore stem-like, BMI1-maintained primitive state in responseto chronic treatments with the CRM metformin. Althougha definitive elucidation of a causal relationship betweenmetformin-induced chronic impairment of cellular bioener-getics, upregulation of BMI1, and acquisition of stem cell-likeproperties far exceeds the scope of this study, it is noteworthythat LT-Met MCF-7 cells exhibit a dramatic upregulation of thestem cell marker CD44 when compared to MCF-7 parental cells(Fig. 2, top). Enhancement of CD44 protein expression in the cellsurface of LT-Met MCF-7 cells was confirmed further at themRNA level (more than 7-fold increase; data not shown). Giventhe well-recognized role of CD44 on the homing and engraft-ment of normal stem/progenitor cells, if our findings in culturedcancer cells would be generally applicable to the homeostasis ofnormal tissues upon mild, but chronic, treatment with CRM,they may provide an additional antiaging mechanism becauseit has been reported that homing and engraftment of stem/progenitor cells decrease with age.15

In normal stem cells, regulation of cell cycle, apoptosis,and senescence by BMI1 occurs through indirect (p16Ink4a-and p19Arf-dependent) regulation of the CDK4/6–cyclin Dcomplex formation, pRB (retinoblastoma protein) phos-phorylation, and expression of p53 target genes involved incell cycle arrest and apoptosis (CDKN1A; p21Waf1/Cip1).3,4

Interestingly, metformin-refractory MDA-MB-231 basal-likebreast cancer cells express CDKN1A (p21Waf1/Cip1) proteinat greatly reduced levels compared to metformin-sensitiveMCF-7 luminal breast cancer cells.16 Thus, even thoughmetformin treatment efficiently downregulates cyclin D1levels in MDA-MB-231 cells, insufficient levels of seques-tered p21Waf1/Cip1 fail to inhibit the activity of cyclin E/CDK2.

Of note, the MDA-MB-231 breast cancer cell line is naturallyenriched in CD44pos/CD24neg/low stem/progenitor cell popu-lations (>90% as assessed by flow cytometry; data notshown).10,12 We now report that, upon long-term accommoda-tion to the CRM metformin, CD44pos/CD24neg/low-negativeMCF-7 luminal breast cancer cells can proliferate rapidly in thepresence of growth inhibitory concentrations of metformin (i.e.,they exhibit acquired autoresistance to metformin) while ex-hibiting drastic changes in the expression levels of BMI1-relatedstem cell self-renewal gene networks. Thus, LT-Met MCF-7 cellsupregulate the expression of BMI1 and cyclin E genes andconcomitantly suppress at the transcriptional level the expres-sion of both the p16Ink4a-related CDK4/6-Cyclin D/pRB axis andthe p19Arf-related p53/CDKN1A (p21Waf1/Cip1) axis. Con-ceptually, long-term accommodation to metformin appears totrigger a BMI1-related suppression of genes that are importantfor the G1/S transition checkpoint, thus preventing any cellcycle arrest and apoptosis of the new cell progenies to efficientlyreplenish originally metformin-naive MCF-7 cell cultures.

CR-Enhanced Stem Cell Ability/Functionand Enhanced Cancer Risk with Age:A Breakable Paradigm?

Learning how to maintain and/or increase stem cellsfunctioning in aging mammals while reducing cancer rates

will have important implications for human health. Thecurrent paradigm in the field, which balances functionalsenescence and cancer risk, establishes that loss of functionin aging stem cells is a consequence of reducing the proba-bility that stem cells will transform (thus generating tumor-initiating stem cells). Therefore, the maximum potentiallifespan of an organism may be determined, at least in part,by the increased risk of cancer concomitant with surveillancemechanisms that can efficiently maintain stem cell functionin aging subjects. Our current findings may support thenotion that CR is the sole intervention that contradicts thecurrent paradigm that increasing adult stem function tolevels found in the young, to avoid senescence, must bebalanced against the risk of higher incidences of cancer. It iswell established that stem/precursor ability and cancer canbe linked in the predicted manner to the expression of thetumor suppressor p16Ink4a (i.e., p16Ink4a expression in stemcells increases with age and p16Ink4a-induced reduction ofstem cell function may be the mechanism by which agingcells balance proliferative ability against the increasing riskof cancer with age).17,18 A pivotal study by Ertl et al. hasshown that lifelong dietary restriction (DR) efficientlymaintains or increases repopulating ability (i.e., stem cellproliferation) in three different mouse strains.19 Ertl et al.suggested that mice on lifelong DR may be able to maintainlow levels of p16Ink4a, and thus more normal levels of pro-liferative ability, by reducing the mutation rate, obviating theneed for tumor suppressors. In this scenario, it is plausible toexplain the counterintuitive ability of CR and CRMs tomaintain or reduce the loss of stem cell function, reducethe incidence of cancer, and extend maximal lifespan inrodents.19

The plasma concentrations of the CRM metformin in di-abetic patients treated with the drug are estimated to rangefrom less than 50 mmol/L and up to 100 mmol/L. Therefore,it could be argued that chronic exposure to millimolar con-centrations of the CRM metformin does not recapitulate themetabolic effects of CR at the cellular level because this doseexceeds metformin’s therapeutic levels.20 However, it shouldbe noted that metformin can accumulate in tissues at con-centrations several-fold higher than those in blood.21 Becausethe positive charge of metformin could promote its accu-mulation within the mitochondrial matrix by 1,000-fold, thuseliciting the primary mechanism of action of metformin(i.e., blockade of the complex I of the mitochondrial elec-tron transport chain), metformin concentrations as high as8 mmol/L represent physiologically relevant metformindoses in hepatic tissue,22 a crucial target tissue for under-standing the lifespan-lengthening effects of CR and CRMs.23

Indeed, it is obvious that compensatory mechanisms crucialto the basic survival of cells should take place upon chronicexposure to metformin because, as expected from chemicalmitochondrial uncoupling, drop in adenosine triphosphate(ATP) levels, and activation of adenosine monophosphate(AMP)-activated protein kinase (AMPK),24 low-scale phos-pho-proteomic analyses revealed that LT-Met MCF-7 cellsdisplay a striking phosphorylation of the cAMP-responseelement binding protein (CREB) on Ser-133 (Fig. 2, bottom).We are currently investigating if these compensatory mech-anisms might involve metabolic adaptation through upre-gulation of the transcriptional coactivator peroxisomeproliferator-activated receptor-coactivator 1a (PGC-1a) and

524 OLIVERAS-FERRAROS ET AL.

activation of the redox-sensing deacetylase sirtuin 1 (SIRT1),thus mimicking true metabolic aspects of CR that target se-lective nutrient utilization and mitochondrial oxidativefunction to regulate cellular energy homeostasis.25

It is well recognized that most adult stem cells in vivousually reside in a microenvironment (niche) and remainrelatively quiescent until they engage in active self-renewalupon injury signals or under certain physiological conditionsdemanding rapid production of new progeny. Unfortunately,we are lacking experimental evidence to suggest thatrobust lifespan-prolonging interventions (e.g., CR and someCRM)26–29 might prevent senescence of stem cells whileprolonging their capacity to proliferate, differentiate, andreplace the mature cells that turn over in adult aging tissues.Obviously, non–cell autonomous mechanisms of the CRMmetformin (e.g., stem/progenitor cell niches) cannot be ex-plored by using cultured cancer cells. Nevertheless, metforminrecently has been found to have little or no effect in controllingcertain stem/progenitor niches, such as inflammation-drivenlung metastatic sites in orthotopic syngeneic breast cancermodels.30 On the other hand, our findings do not disagreewith the well-established anticancer actions of metforminobserved in population studies and in rodents.20,31 Rather,they exclusively provide an experimental framework forrapidly investigating candidate molecular mechanisms un-derlying the effects of CRMs. We suggest that assessment ofglobal changes in stem cell-related gene expression usingmicroarrays could be used for the rapid identification of genesignatures that may account for the initiation and mainte-nance of self-renewing divisions in response to antiaginginterventions. In addition, this method might provide anovel molecular avenue to investigate whether the antiagingeffects of some CRMs might relate to their ability to epige-netically reprogram stemness and stem-like cell states asthe ultimate mechanism responsible for the longevity ofmulticellular organisms.

Acknowledgments

Alejandro Vazquez-Martin is the recipient of a ‘‘Sara Bor-rell’’ postdoctoral contract (CD08/00283, Ministerio de Sani-dad y Consumo, Fondo de Investigacion Sanitaria –FIS-,Spain). This work was supported in part by Instituto de SaludCarlos III (Ministerio de Sanidad y Consumo, Fondo de In-vestigacion Sanitaria –FIS-, Spain, Grants CP05-00090, PI06-0778 and RD06-0020-0028 to Javier A. Menendez). Javier A.Menendez was also supported by a grant from the FundacionCientıfica de la Asociacion Espanola Contra el Cancer (AECC,Spain) and by the Ministerio de Ciencia e Innovacion(SAF2009-11579, Plan Nacional de IþDþ I, MICINN, Spain).

Author Disclosure Statement

No competing financial interests exist.

References

1. Damri M, Granot G, Ben-Meir H, Avivi Y, Plaschkes I,Chalifa-Caspi V, Wolfson M, Fraifeld V, Grafi G. Senescingcells share common features with dedifferentiating cells.Rejuvenation Res 2009;12:435–443.

2. Howitz KT, Sinclair DA. Xenohormesis: Sensing the chemi-cal cues of other species. Cell 2008;133:387–391.

3. Park IK, Morrison SJ, Clarke MF. Bmi1, stem cells, and se-nescence regulation. J Clin Invest 2004;113:175–179.

4. Valk-Lingbeek ME, Bruggeman SW, van Lohuizen M. Stemcells and cancer: The polycomb connection. Cell 2004;118:409–418.

5. Pardal R, Molofsky AV, He S, Morrison SJ. Stem cell self-renewal and cancer cell proliferation are regulated bycommon networks that balance the activation of proto-oncogenes and tumor suppressors. Cold Spring Harb SympQuant Biol 2005;70:177–185.

6. Dhahbi JM, Mote PL, Fahy GM, Spindler SR. Identificationof potential caloric restriction mimetics by microarray pro-filing. Physiol Genomics 2005;23:343–350.

7. Ingram DK, Zhu M, Mamczarz J, Zou S, Lane MA, Roth GS,deCabo R. Calorie restriction mimetics: An emerging re-search field. Aging Cell 2006;5:97–108.

8. Spindler SR. Use of microarray biomarkers to identify lon-gevity therapeutics. Aging Cell 2006;5:39–50.

9. Lane MA, Roth GS, Ingram DK. Caloric restriction mimetics:A novel approach for biogerontology. Methods Mol Biol2007;371:143–149.

10. Fillmore CM, Kuperwasser C. Human breast cancer cell linescontain stem-like cells that self-renew, give rise to pheno-typically diverse progeny and survive chemotherapy. BreastCancer Res 2008;10:R25.

11. Horwitz KB, Sartorius CA. Progestins in hormone replace-ment therapies reactivate cancer stem cells in women withpreexisting breast cancers: A hypothesis. J Clin EndocrinolMetab 2008;93:3295–3298.

12. Sheridan C, Kishimoto H, Fuchs RK, Mehrotra S, Bhat-Nakshatri P, Turner CH, Goulet R Jr, Badve S, Nakshatri H.CD44þ/CD24- breast cancer cells exhibit enhanced inva-sive properties: An early step necessary for metastasis.Breast Cancer Res 2006;8:R59.

13. Shimono Y, Zabala M, Cho RW, Lobo N, Dalerba P, Qian D,Diehn M, Liu H, Panula SP, Chiao E, Dirbas FM, Somlo G,Pera RA, Lao K, Clarke MF. Downregulation of miRNA-200c links breast cancer stem cells with normal stem cells.Cell 2009;138:592–603.

14. Liu S, Dontu G, Mantle ID, Patel S, Ahn NS, Jackson KW,Suri P, Wicha MS. Hedgehog signaling and Bmi-1 regulateself-renewal of normal and malignant human mammarystem cells. Cancer Res 2006;66:6063–6071.

15. Liang Y, Van Zant G, Szilvassy SJ. Effects of aging on thehoming and engraftment of murine hematopoietic stem andprogrenitor cells. Blood 2005;106:1479–1487.

16. Zhuang Y, Miskimins WK. Cell cycle arrest in Metformintreated breast cancer cells involves activation of AMPK,downregulation of cyclin D1, and requires p27Kip1 orp21Cip1. J Mol Signal 2008;318.

17. Beausejour CM, Campisi J. Ageing: balancing regenerationand cancer. Nature 2006;443:404–405.

18. Janzen V, Forkert R, Fleming HE, Saito Y, Waring MT,Dombkowski DM, Cheng T, DePinho RA, Sharpless NE,Scadden DT. Stem-cell ageing modified by the cyclin-dependent kinase inhibitor p16INK4a. Nature 2006;443:421–426.

19. Ertl RP, Chen J, Astle CM, Duffy TM, Harrison DE. Effects ofdietary restriction on hematopoietic stem-cell aging are ge-netically regulated. Blood 2008;111:1709–1716.

20. Martin-Castillo B, Vazquez-Martin A, Oliveras-Ferraros,Menendez JA. Metformin and cancer: Doses, mechanismsand the dandelion & hormetic phenomena. Cell Cycle2010;9(6) [e-pub ahead of print].

METFORMIN AND STEMNESS 525

21. Wilcock C, Bailey CJ. Accumulation of metformin by tissuesof the normal and diabetic mouse. Xenobiotica 1994;24:49–57.

22. Carvalho C, Correia S, Santos MS, Seica R, Oliveira CR,Moreira PI. Metformin promotes isolated rat liver mito-chondria impairment. Mol Cell Biochem 2009;308:75–83.

23. Dhahbi JM, Mote PL, Fahy GM, Spindler SR. Identificationof potential caloric restriction mimetics by microarray pro-filing. Physiol Genomics 2005;23:343–350.

24. Rohas LM, St-Pierre J, Uldry M, Jager S, Handschin C,Spiegelman BM. A fundamental system of cellular energyhomeostasis regulated by PGC-1alpha. Proc Natl Acad SciUSA 2007;104:7933–7938.

25. Rodgers JT, Lerin C, Gerhart-Hines Z, Puigserver P. Meta-bolic adaptations through the PGC-1 alpha and SIRT1pathways. FEBS Lett 2008;582:46–53.

26. Colman RJ, Anderson RM, Johnson SC, Kastman EK, KosmatkaKJ, Beasley TM, Allison DB, Cruzen C, Simmons HA, KemnitzJW, Weindruch R. Caloric restriction delays disease onset andmortality in rhesus monkeys. Science 2009;325:201–204.

27. Harrison DE, Strong R, Sharp ZD, Nelson JF, Astle CM,Flurkey K, Nadon NL, Wilkinson JE, Frenkel K, Carter CS,Pahor M, Javors MA, Fernandez E, Miller RA. Rapamycinfed late in life extends lifespan in genetically heterogeneousmice. Nature 2009;460:392–395.

28. Cox LS, Mattison JA. Increasing longevity through caloricrestriction or rapamycin feeding in mammals: common

mechanisms for common outcomes? Aging Cell 2009;8:607–613.

29. Cox LS. Live fast, die young: New lessons in mammalianlongevity. Rejuvenation Res 2009;12:283–288.

30. Phoenix KN, Vumbaca F, Fox MM, Evans R, Claffey KP.Dietary energy availability affects primary and metastaticbreast cancer and metformin efficacy. Breast Cancer ResTreat 2009;Nov 22[Epub ahead of print].

31. Hirsch HA, Iliopoulos D, Tsichlis PN, Struhl K. Metforminselectively targets cancer stem cells, and acts together withchemotherapy to block tumor growth and prolong remis-sion. Cancer Res 2009; 69:7507–7511.

Address correspondence to:Javier A. Menendez

Catalan Institute of Oncology, Girona (ICO-Girona)Dr. Josep Trueta University Hospital

Ctra. Franca s/nE-17007 Girona, Catalonia

Spain

E-mail: jmenendez@iconcologia.net

Received: January 13, 2010Accepted: March 13, 2010

526 OLIVERAS-FERRAROS ET AL.