colorectal cancer stem cell

C ANCER S TEM Concise Review: A RISTIDES G.V AIOPOULOSinjury,disease,or cancer[10–12].Colon shares this SC-basedhierarchical organization with many other tissues/systems thatare under constant renewal,including the hematopoietic systemand the skin[11].Originally,the Unitarian theory proposed themonoclonal nature of intestinal crypts[10].More recent evi-dence,however,implies that crypts are probably polyclonal atbirth and subsequently shift toward monoclonality by followinga pattern of neutral drift[7,13].An intestinal crypt containsapproximately16SCs and harbors two distinct pools of puta-tive SCs.The one pool is located at the crypt base and is char-acterized by the expression of leucine-rich repeat containing Gprotein-coupled receptor5(Lgr-5)and the other pool resides atþ4position and consists of B lymphoma Moloney murine leu-kemia virus(Mo-MLV)insertion region1homolog(Bmi-1)and telomerase reverse transcriptase(Tert)expressing cells[13,14].The contribution of each type of cell to the maintenanceof the SC pool is under debate.One possible suggestion is theexistence of a pool of equally contributing cells where eachcell’s behavior is defined by its entourage and in terms ofreplacement they follow a pattern of neutral drift[13].An al-ternative approach is that Lgr-5þSCs comprise the active pop-ulation of the crypt,whereas Bmi-1þor Tertþcells are quies-cent SCs that represent a reserve pool of SCs with the abilityto replace Lgr-5þcells in case of loss or injury[7,14].More-over,it has been proposed that Bmi-1þcells may not have animpact on Lgr-5þSCs,as Bmi-1knockout mice show normalcrypt morphology and have a normal intestinal epithelium[15].The functional importance of the intestinal microenvironment(niche)in the advanced control of SC life cycle is widelyaccepted.The niche consists of cellular and extracellular com-ponents that ensure the optimal conditions for SC mainte-nance through the secretion of various cytokines,growth fac-tors,and direct interactions[10,16].Interestingly,intestinalSCs may also be affected by components in the crypt lumen,derived from epithelial cells or from bacteria.Intestinal sube-pithelial myofibroblasts are key regulators of SC self-renewaland differentiation,mediate the crosstalk between epithelialand mesenchymal cells,and secrete a wide range of morpho-genetic factors[7].Epithelial–mesenchymal interactions regu-late the normal intestinal architecture and additionally definethe balance between proliferation and differentiation[10,16].Wingless/Int(Wnt),Hedgehog,bone morphogenetic protein(BMP),Notch,and platelet-derived growth factor pathwaysare involved in these interactions[7].Identification and isolation of SCs remain an issue of debate[8]due to the lack of specific molecular markers along withunsolved technical issues,forcing the use of a functional defi-nition based on the attributes of self-renewal and multipo-tency.Consequently,only assumptions can be made regardingtheir exact number and position[17].Several studies havebeen carried out,mostly on small intestine,which used DNAlabeling techniques to exploit the comparatively slower cy-cling rate of SCs[18]and led to the formulation of two dif-ferent models regarding their position[5].Theþ4positionmodel postulates that SCs are located at theþ4position ofthe small-intestine crypt just above the Paneth cells,whichoccupy the bottom.These cells express the markers Bmi-1and Tert.SCs divide into progenitor cells that migrate to thetop of the crypt and differentiate,while Paneth cells followthe exact opposite direction[7,16].Alternatively,theSC Figure1.Hierarchical organization of a normal colonic crypt—comparison between the stochastic and the CSC model.Colonic crypts are hier-archically organized.SCs divide into progenitor cells that migrate at top and produce terminally differentiated cells.In the stochastic model,all cells possess tumor initiation potential.In the CSC model,only CSCs are capable of tumor propagation and metastasis.Microenvironmental crosstalk, epigenetic,and genetic aberrations(yellow lightning bolt)may lead to clonal evolution(dark purple cells).Circumscribed cells represent the mutated types of their normal homologs(white polygons:A,asymmetrical division;S,symmetrical division).Abbreviations:BMP,bone morphogenetic pro-tein;CBCC,crypt base columnar cell;CSC,cancer stem cell;D,differentiated cell;P,progenitor cell;SC,stem cell;WNT,Wingless/Int.364Stem Cells inCRCzone model was proposed after discovering the presence of small immature cycling cells at the crypt base between Paneth cells.These cells were named crypt base columnar cells and express the Wnt target gene Lgr-5[19].Recently,many mole-cules,mostly located on the cell surface,have been proposed as putative stemness markers,at the same time allowing isola-tion by fluorescence-activated cell sorting [10].In summar-y,the molecules involved in critical stages of proliferation and differentiation are Musashi-1(Msi-1),CD29,Bmi-1,Lgr-5,aldehyde dehydrogenase 1(ALDH-1),Tert,and achaete scute-like 2[5,7,9,10,14,15].For a more detailed presenta-tion of the currently used markers,the reader is referred to Table 1.Colorectal carcinogenesis results from a series of genetic/epi-genetic alterations and interactions with microenvironmental and germ-line factors that transform the normal colonic mu-cosa into an aberrant phenotype [2,3].The questions which is the cell of origin and which cells propagate the tumor after initiation remain elusive despite extensive in vitro and in vivo experiments [7,8].The cell of origin is the cell that acquires tumorigenic mutations and becomes the first tumor cell.A cell of origin is different from the cells responsible for tumor propagation after its inception [8].Every normal cell (SC,progenitor,or differentiated)can accumulate mutations and become the cell of origin [8,10,24].Following its malignant transformation,it is possible to follow one of the two pro-posed models for the explanation of carcinogenesis and pro-gression,the stochastic or the CSC model (Fig.1).Even if a normal SC becomes the cell of origin,it can form a tumor according to the features of the stochastic or the CSC model [8].The stochastic model suggests that every cell within a tu-mor is capable of both initiation and propagation.After trans-formation,each cell can acquire further mutations.The heter-ogeneity within a tumor can be explained by the aggregation of random genetic aberrations as well as the divergent micro-environmental influence [8,25].Concomitantly,a diverse population of subclones with different properties is generated.Only selected clones can migrate and form metastases.The migrating clones can further accumulate mutations,resulting in a divergent metastatic tumor from the parent lesion [11].The newly introduced CSC model,on other hand,states that only a small portion of cells within a tumor is endowed with tumor propagation potential and thus named CSCs,whereas all other cells are not [26].According to the CSC model,sim-ilar to the normal tissue,cancers are also hierarchically organ-ized.Tumor heterogeneity results from the production by the multipotent CSCs (high proliferative)of a wide variety of progenitor (medium proliferative)and differentiated (nonproli-ferative)cells [11,27].In this model,only CSCs can migrate and as a result the metastatic tissue resembles the pattern of the original lesion.However,individual cell responses to microenvironmental stimuli,epigenetic modifications,and additional genetic aberrations may be observed within a can-cerous tissue,which in turn may lead to clonal evolution and gain or loss of CSC attributes.A new and improved CSC due to clonal evolution may arise,which will now drive tumor growth and metastasis instead of the initial CSC [8,28].CSCs are characterized by self-renewal,multipotency,limit-less proliferation potential,angiogenic,and immune evasion features [24].Intriguingly,CSCs are relatively highly resistant to traditional tumor therapeutic measures and thus responsible for tumor relapse due to the expression of DNA repair mecha-nisms,detoxifying enzymes,and drug transporters [29].Most likely,CSCs are derived from normal SCs due to their extended life span that alter their behavior as a result of genetic and epigenetic changes.It has also been discussed that progenitor,differentiated,or even cells from outside the tumor,for example,bone marrow-derived cells,might also serve as CSCs’ancestors [8,10,11,28,30].Concomitantly,through symmetric (CSC overproduction)or asymmetric (het-erogeneity)divisions,the entire crypt will be colonized from the mutated SCs and their descendants.Future alterationsmayVaiopoulos,Kostakis,Koutsilieris et al.365lead to a more aggressive and metastatic phenotype[5,31].Throughout CRC progression,the number of CSCs remainsapproximately1%of the number of total cells.However,theirfrequency appears to be highly variable[6,28].The CSCmodel modifies the classic Fearon and Vogelstein model,which is characterized by step-by-step genetic modificationsof the adenoma to carcinoma sequence,by placing the normalSC as the primary candidate for being the cell of origin,byunderlying the crucial importance of microenvironmental sig-nals,and by explaining tumor heterogeneity within the con-text of a clonally evolved CSC model[7,10].The intestinal homeostasis is under the surveillance of a com-plex crosstalk network of evolutionary conserved pathways,such as Wnt,Notch,and Hedgehog,which control the balancebetween proliferation,differentiation,migration,and renewal.The Wnt pathway is responsible for endoderm formation andexerts fundamental role in crypt development,maintenance,and proliferation,as it is depicted by the failure of Wnt knock-out mice to develop colonic crypts[8,18,32].C-myc,cyclinD1,and Lgr-5are included among its various targets.A Wntgradient loss of expression from the base to the top is closelyrelated with differentiation and expression of the ephrin familyproteins,which regulate positioning[5].Mutations in the APCgene(80%in sporadic cancer),b-catenin,or the regulatory pro-teins in the Wnt pathway result in constant activation[18].This may lead to uncontrolled proliferation,a shift from asym-metrical to symmetrical divisions,and augmented survival.Wnt signaling is also involved in the process of epithelial tomesenchymal transition(EMT)and invasion[7,8].Notch maydrive tumorigenesis,because it potentiates proliferation andinhibits differentiation[10].BMP and TGF-b belong to a fam-ily of ligands whose receptors interact with the intracellularcascade of the Smad proteins[33].BMPs are mostly expressedby stromal cells,have opposing actions to Wnt,and are mostlyactive at the top of the crypt.In the crypt bottom,they are alsoproduced,but they are kept inactive by binding to specificinhibitors[7,16].TGF-b is an inhibitor of intestinal epithelialcell proliferation and inducer of apoptosis[3,31,34].TheHedgehog proteins carry signals between stromal and epithelialcells.Although it seems that Hedgehog acts as a Wnt suppres-sor,probably through BMP,and is also expressed byþ4posi-tion stem-like cells,its role remains elusive[7].Finally,incolorectal carcinogenesis,activation of the phosphoinositide3-kinase(PI3K)pathway results in rapid and excessive prolifera-tion of intestinal SCs[12].Like the normal SCs,CSCs reside in a qualified microenvir-onment that has gained tumor-promoting traits[35].A recip-rocal interplay between SCs and their niche is behind malig-nant transformation.Genetic or epigenetic aberrations in theSCs compartment may lead to alteration of the niche and viceversa[28,36].This tumorigenic niche is composed of trans-formed myofibroblasts,recruited myeloid cells,other celltypes,and extracellular components,which produce variousgrowth factors and cytokines,including hepatocyte growthfactor(HGF),tumor necrosis factor a(TNF-a),and interleu-kin(IL)-6,which promote dedifferentiation,carcinogenesis,and invasiveness[7,36].Although Wnt activating APC muta-tions are common in CRC,tumors often present with a heter-ogeneous Wnt activity,indicating an alternative regulation.Myofibroblast-derived HGF is probably a major enhancer ofWnt activity[7,36].Nuclear b-catenin localization,whichindicates Wnt-triggering and cancer-stemness,is predomi-nantly observed in the invasive regions of colorectal carcino-mas[37].Experimental data revealed that mice carrying anAPC mutation combined with Smad4loss,and thus unrespon-sive to differentiation-inducing microenvironmental signals,showed increased susceptibility in the formation of highlyinvasive tumors combined with extensive stromal proliferation[38].In another study,it was shown that inactivation of stro-mal BMP may also lead to extensive proliferation of both thestroma and the overlying epithelium and eventually adenomaformation.Wnt activity was increased in epithelial cells andwas under the manipulation of other stromal pathways,suchas Forkhead Box L1(FOXL1)and phosphatase and tensinhomolog(PTEN)[12].In humans,evidence that supports thisrelationship is derived from observations in patients withinflammatory bowel disease,histopathological data of stromalimmune cells,and the beneficial role of nonsteroidal anti-inflammatory drugs in CRC.The chronic inflammatory mi-lieu,composed of infiltrating macrophages,lymphocytes,andstromal cells,which produce IL-1,IL-6,TNF-a,and vascularendothelial growth factor,disturbs epithelial homeostasis andthe SC reservoir[6].It has been postulated that a normal in-testinal niche can prevent tumor growth,even if transformedSCs are present[39].Metastasis is responsible for approximately90%of cancer-associated mortality and can be divided into translocation andcolonization basically[40].As the CSC population representsthe only cells that propagate tumors,it can be extrapolatedthat these cells are responsible for metastasis formation[41,42].It is therefore a prerequisite that these cells must be ableto detach from the primary tumor,invade,access,and surviveat the circulation,disseminate at distant sites,transmigrateacross the endothelial lining of the target tissue,and form sec-ondary tumors(Fig.2)[43].A crucial step in the metastaticprocess is EMT.In this event,the simultaneous downregula-tion of the epithelial phenotype,along with enabling offibro-blast-like traits,enhances motility,invasiveness,and resist-ance to apoptosis[40,44].Cells that undergo EMT are alsoassumed to gain a stem-like phenotype[45,46].This processseems to be regulated by pathways that are linked with SCs,such as Wnt and TGF-b[28].Studies in CRCs revealed amixed epithelial(tumor center)and mesenchymal(invasivefront,EMT/stemness)phenotype in both the primary tumorand the metastasis[37,45].Removal of contextual signalingmay reverse this process[36,47].The proposed plasticity ofCSCs brings new insights in thefield,suggesting that notonly CSCs produce more differentiated descendants but alsonon-CSCs can be converted to CSCs through EMT.Thereprogramming is mediated by microenvironmental signals[7,36,47,48].To enlighten the microenvironmental-driven plas-ticity of CSCs,two models have been proposed.The intrinsicmodel suggests that certain cell population possesses CSCtraits from tumor initiation.Meanwhile,extrinsic signals,derived from recruited active stromal cells,force cells toundergo EMT and to adopt CSCs aspects[40,48].Therefore,the existence of a circulating tumor cell(CTC)with stem-likeproperties is speculated[40,49].Data from breast,lung,and 366Stem Cells inCRCcolon cancer patients indicate that not only current CTC detection techniques tend to underestimate the population of CTCs with stem-like and EMT traits but also that the pres-ence of stemness markers in the peripheral blood of cancer patients is associated with worse prognosis and recurrence [50–52].It seems that through EMT,CTCs adopt a more aggressive and SC-like phenotype.However,understanding the exact relation between EMT,CSCs,and CTCs will pro-vide additional tools for the diagnosis and therapy of cancer patients.Although CSCs have been implicated in colon carcinogenesis,their existence had not been experimentally demonstrated until recently.However,due to the complexity of their biol-ogy and unsolved technical issues,an unequivocally approved identification and isolation strategy is still a matter of debate [9,31].Experimental studies exploit basic,but not exclusive,features of CSCs,such as the long-term retention of DNA label,morphological traits,epigenetic modifications,or differ-ential gene expression [53–55].Most recently used and prom-inent techniques,such as immunohistochemistry and cell sort-ing isolation,are based on the expression of possible stemness markers [56,57].The tumorigenic potential of iso-lated cells is often evaluated by their ability to reproduce the full heterogeneous repertoire of the parent tumor when xeno-transplanted in immunodeficient mice or through colony form-ing assays [28].Certain speculations arise,however,regarding the reliability of these assays.The environment of the host or-ganism does not resemble exactly the original conditions and,additionally,cytokines and growth factors from different spe-cies may show limited affinity for their related receptors [12,26,28].The site of transplantation may also influence the ex-perimental outcome,as an orthotopic assay for colon CSCs has not been described yet.Further issues may occur due to the alteration of certain markers caused by culturing and sort-ing conditions [12,26,28].The presence of residual immune cells in the recipient organism and an altered vascular envi-ronment may influence the tumorigenic potential of injected cells [28].Several molecules have been proposed as CSC markers including CD133,CD44,CD24,CD166,Lgr-5,and ALDH-1(Table 2).CD133,a pentaspan transmembrane gly-coprotein implicated in the organization of plasma membrane [9],was first reported as a putative colorectal CSC marker in the studies of O’Brien et al.and Ricci-Vitiani et al.by inject-ing CD133positive and negative (sorted)cells in immunodefi-cient mice.Only a small subset of CD133þcells was capable of initiating tumor growth,while negative cells were not.In both studies,CD133expression was observed in normal co-lon,although at lower numbers,suggesting a possible relation between normal and cancer SCs [58,59].Following this breakthrough,several other groups investigated the expression patterns of CD133in CRC cell lines or biopsy specimens.CD133expression was independently correlated with SC potential,worse prognosis,and low survival.Moreover,its combined expression with other putative CSC markers was also evaluated [20,60-67].It seems,however,that most CD133antibodies target glycosylation-dependent epitopes [20],whose presence is related to the differentiation stage of the cell.Recent experimental data from colon and glioblas-toma cells suggest that the differential glycosylation of spe-cific epitopes may mask the presence of CD133on cells pre-viously characterized as negative [68,69].Nevertheless,data revealing that CD133is not restricted to normal or cancer SCs raised questions regarding its efficiency as a colorectal CSC marker.CD133was ubiquitously expressed in differenti-ated colonic epithelium and in tumor cells.Both CD133þ/Àisolated cells were able to form tumors when injected in mice [70].The cell adhesion molecule CD44was proposed asanFigure 2.The stem cell model of metastasis.The primary tumor releases in circulation CSCs that have underwent EMT and possibly clonally evolved (dark purple and dark purple ovular flattened cells)due to accumulating genetic and epigenetic modifications (yellow lightning bolt).These cells travel to distant organs,where they formulate metastases.The secondary organs may provide a niche for the tumor cells.Steps 1–5(white polygons)indicate the consequent stages required for the transition from local to metastatic disease.Abbreviations:CSC,cancer stem cell;D,differentiated cell;EMT,epithelial to mesenchymal transition;P,progenitor cell.Vaiopoulos,Kostakis,Koutsilieris et al.367alternative marker.CD44þcells exhibited CSC properties,as a single cell could form a sphere and a xenograft tumor that resembled the original lesion [71].Its expression in CRC has been negatively associated with the depth of invasion,lymph node infiltration,prognosis,and survival [72].Albeit CD133and CD44seem to be promising markers,as cells expressing this protein display tumorigenic potential in vitro and in vivo,a combination of markers is likely to be more suitable for detecting colorectal CSCs [73].The high heterogeneity of human cancers forces the thorough investigation of each indi-vidual subpopulation [55].Additional markers include CD166,epithelial cell adhesion molecule,ALDH-1,CD29,CD24,CD26,Msi-1,Lgr-5,and Wnt activity/b -catenin.The presence of these molecules was associated with stemness characteristics both in vitro and in vivo,breeding tumors that resemble the original lesion and displaying increased clono-genic ability,tumorigenicity,and multilineage potential.Asso-ciations were also made with tumor stage,differentiation,invasiveness,metastasis formation,prognosis,and survival.These markers were also used for further enrichment of iso-lated CSCs to enhance their tumorigenic ability [18,21–23,36,74-90].The pluripotency genes Oct-4,Sox-2,Nanog,Lin-28,Klf-4,and c-myc are regarded as promising surrogate markers.These genes seem to facilitate a shift toward an un-differentiated state [45,91].Their expression has been associ-ated with poor prognosis,relapse,and distant recurrence,as well as with resistance to conventional chemotherapy/radio-therapy [91–93].Interestingly,the CD133negative cell lineNANK displayed tumor initiating abilities when injected in mice.Despite lacking the expression of conventional CSC markers CD133and CD44as well as the self-renewal genes Oct-4and Nanog,NANK cells expressed the developmental markers CDX2and Bmi-1[94].According to the CSC model,tumor growth is sustained by a small population of cells that current therapeutic measures fail to eradicate.Although antitumor strategies eliminate the vast majority of cancer cells,leading to a reduction in tumor size and disease remission,they fail to prevent relapse by leaving intact the CSC reservoir.On the contrary,although a CSC-tar-geted therapy will not cause tumor-size shrinkage in short term,it will prevent future regrowth and metastasis [11,95].Experimental data support the notion that CSCs are responsi-ble for resistance to chemoradiotherapy and disease relapse [51,89,93,96].Resistance is exhibited either through a shift from active state to quiescence or a wide spectrum of protec-tive mechanisms (DNA damage repair,altered cell-cycle checkpoint control,malfunction of apoptosis,drug transport-ers,and detoxifying enzymes)[29,40,97].The proposed plasticity of CSCs complicates therapeutic approaches because not only CSCs and cancer cells with CSCs traits thatoccur368Stem Cells inCRCthrough EMT are more resistant to conventional treatment but also a strategy that targets only CSCs will fail to efficiently cure CRC [47].EMT and CSC plasticity participate in the ac-quisition of both de novo and acquired drug resistance [44].It is therefore essential to renovate our therapeutic arsenal by designing new curative methods that specifically target CSCs and also solve issues that arise from CSC flexibility by elimi-nating at the same time the non-CSC population and interven-ing in the process of EMT [44,47].Successful therapeutic interventions should dispatch all proliferating cells,eradicate the CSC reservoir,and then eliminate disseminated/circulating cells including those that may have become dormant [26,95].Several issues should be taken into account during the design of CSC-directed interventions.Possible adverse reactions may occur from the impact on normal SCs,which bear resem-blance to their tumor relatives.This potential toxic effect can be minimized by targeting molecules or pathways that are preferentially active in CSCs [57].Moreover,the development of resistance along with the heterogeneous nature of tumors points to the combination of agents [97].Monoclonal antibod-ies could be directed against cell surface molecules,such as CD133,CD44,or even drug transporters.Inhibition of these molecules may result in reduction of tumor size,metastatic potential,and resistance to chemoradiotherapy [24,57].Another rational target includes blockage of various self-renewal pathways,including Wnt,Notch,PTEN,and Hedge-hog [4].Small molecular inhibitors of Wnt and c -secretase inhibitors of Notch have been suggested as novel agents against CRC [10].An alternative approach is the induction of differentiation and the disruption of EMT.Differentiation therapy forces cells to shift into a mature phenotype,lose their self-renewal abilities,and therefore become vulnerable to conventional treatment.Targeting EMT will contribute to the depletion of new CSCs that arise as products of EMT and in the blockage of drug resistance.This strategy includes:(a)alterations in signaling cascades,for example,BMP4or PI3K,(b)application of microRNAs that alter gene expression pro-files,and (c)epigenetic therapy [24,44,95].BMP promoted differentiation and apoptosis in colorectal CSCs through mod-ulations in the Wnt signaling cascade.Administration of oxa-liplatin and 5-flouroucil,which enhance BMP antitumor action,resulted in the remission of colorectal CSC-derivedxenograft tumors [10].Reversing chemoresistance and radio-resistance represents a promising proposal.This can be achieved through interference with a plethora of cellular com-ponents,including inactivation of drug transporters and DNA checkpoint kinases,depletion of reactive oxygen species scav-engers,and inhibition of signal transduction pathways [97].Recently,inhibition of the IL-4pathway with an anti-IL-4antibody or an IL-4receptor antagonist in CD133þcolorectal CSCs augmented the antitumor effects of conventional che-motherapeutics.IL-4functions in an autocrine manner,pro-moting antiapoptotic pathways [98].The ‘‘malignant’’micro-environment has been proven to be essential for the maintenance and development of CSCs.Interruption of the crosstalk network between the elements of the niche and CSCs will drive the latter in restriction of growth and loss of their metastatic capacity [4,57].Evidence from other malig-nancies supports the notion that tumor initiation and propaga-tion may be mediated,at least in part,through evasion of host immune responses [57].The application of the bisphonate zolendronate on colorectal CSCs promoted an efficient immune T-cell response [10].The understanding of the molecular mechanisms implicated in the biology of CSCs as well as the identification of specific markers,which will permit efficient isolation,may aid efforts toward targeted treatment,more efficient diagnosis,and fol-low-up of CRC patients.CSCs are rational targets for the design of interventions that will enhance responsiveness to traditional therapeutic strategies and contribute in the reduc-tion of local recurrence and metastasis.The authors indicate no potential conflicts of interest.1Jemal A,Bray F,Center MM et al.Global cancer statistics.CA Can-cer J Clin 2011;61:69–90.2Markowitz SD,Bertagnolli MM.Molecular origins of cancer:Molecu-lar basis of colorectal cancer.N Engl J Med 2009;361:2449–2460.3Lampropoulos P,Zizi-Sermpetzoglou A,Rizos S et al.TGF-beta sig-nalling in colon carcinogenesis.Cancer Lett 2012;314:1–7.4Huang EH,Wicha MS.Colon cancer stem cells:Implications for pre-vention and therapy.Trends Mol Med 2008;14:503–509.5Ricci-Vitiani L,Fabrizi E,Palio E et al.Colon cancer stem cells.J Mol Med (Berl)2009;87:1097–1104.6Boman BM,Huang E.Human colon cancer stem cells:A new para-digm in gastrointestinal oncology.J Clin Oncol 2008;26:2828–2838.7Medema JP,Vermeulen L.Microenvironmental regulation of stemcells in intestinal homeostasis and cancer.Nature 2011;474:318–326.8Vries RG,Huch M,Clevers H.Stem cells and cancer of the stomachand intestine.Mol Oncol 2010;4:373–384.9Papailiou J,Bramis KJ,Gazouli M et al.Stem cells in colon cancer.A new era in cancer theory begins.Int J Colorectal Dis 2011;26:1–11.10Todaro M,Francipane MG,Medema JP et al.Colon cancer stemcells:Promise of targeted therapy.Gastroenterology 2010;138:2151–2162.11Dalerba P,Cho RW,Clarke MF.Cancer stem cells:Models and con-cepts.Annu Rev Med 2007;58:267–284.12Davies EJ,Marsh V,Clarke AR.Origin and maintenance of the intes-tinal cancer stem cell.Mol Carcinog 2011;50:254–263.13Lopez-Garcia C,Klein AM,Simons BD et al.Intestinal stem cellreplacement follows a pattern of neutral drift.Science 2010;330:822–825.14Tian H,Biehs B,Warming S et al.A reserve stem cell population insmall intestine renders Lgr5-positive cells dispensable.Nature 2011;478:255–259.15van der Flier LG,van Gijn ME,Hatzis P et al.Transcription factorachaete scute-like 2controls intestinal stem cell fate.Cell 2009;136:903–912.16Potten CS,Gandara R,Mahida YR et al.The stem cells of small in-testinal crypts:Where are they?Cell Prolif 2009;42:731–750.17Willis ND,Przyborski SA,Hutchison CJ et al.Colonic and colorectalcancer stem cells:Progress in the search for putative biomarkers.J Anat 2008;213:59–65.18Haegebarth A,Clevers H.Wnt signaling,Lgr5,and stem cells in theintestine and skin.Am J Pathol 2009;174:715–721.19Simons BD,Clevers H.Stem cell self-renewal in intestinal crypt.ExpCell Res 2011;317:2719–2724.20Zhu L,Gibson P,Currle DS et al.Prominin 1marks intestinal stemcells that are susceptible to neoplastic transformation.Nature 2009;457:603–607.21Barker N,van Es JH,Kuipers J et al.Identification of stem cells insmall intestine and colon by marker gene Lgr5.Nature 2007;449:1003–1007.Vaiopoulos,Kostakis,Koutsilieris et al.369。