Batlle-2017-Cancer stem cells revisited.pdf

1124 VOLUME 23|NUMBER 10|OCTOBER 2017 NATURE MEDICINETissues such as the intestinal epithelium and the hematopoietic system continuously self-renew through the activity of a dedicated population of tissue-specific stem cells,also known as adult stem cells1,2.Unlike the bulk of the cells that populate these tissues,adult stem cells are long-lived and generate cellular progeny throughout life to regenerate the multiple specialized,short-lived cells that ultimately perform tissue-specific functions.The CSC theory states that tumor growth is similarly fueled by small numbers of tumor stem cells hidden in cancers.It explains clinical obser-vations,such as the almost inevitable recurrence of tumors after initially successful chemotherapy and/or radiation therapy,the phenomenon of tumor dormancy,and metastasis.The first decade of this century has seen an avalanche of reports on the identification of CSCs in many common cancer types,including leukemia35,breast cancer6,colorectal cancer79,and brain cancer10.The CSC concept has inspired the design of innovative treatment strategies for these cancers,not aimed at shrink-ing tumor bulk,but rather at exterminating CSCs,the cell population that sustains long-term growth.The aim of this review is to discuss recent developments in the CSC field while bearing in mind newly emerging views on the biology of normal stem cells.Over the past few years,experiments involving lineage tracing and cell ablation in intact tumors have confirmed that many of these tumors harbor stem cells in dedicated niches1116.Yet,it has gradu-ally become clear that CSCs(similarly to normal stem cells)do not nec-essarily have to be rare and/or quiescent;multiple examples now show that they can be abundant and can proliferate vigorously.Furthermore,it is emerging that stem cell hierarchies may be much more plastic than previously appreciated17,a phenomenon that complicates the identi-fication and eradication of CSCs.This review will also focus on major advances in the development of CSC-based therapies,some of which are currently being implemented in the clinical setting.The shaping of the CSC conceptWe have given an extensive history of the CSC field previously18.Here we briefly highlight the major influence of hematopoietic stem cell(HSC)research in shaping the adult stem cell and CSC fields.Features of hematopoietic stem cellsHSCs were first identified 60 years ago.The prevailing view on the defining characteristics of HSCs and their downstream differentiation hierarchy can be summarized as follows2:the HSC is at the top of the hematopoietic hierarchy,is a rare cell type,and it divides infrequently,i.e.,it is quiescent,because DNA replication prior to division carries a risk of mutation.When the HSC divides,it does so in an asymmet-ric fashion,yielding one actively dividing daughter cell and one new,quiescent stem cell.Thus,a stem cell has the unique capacity to self-renew,and its lineage is long-lived.Individual HSCs can generate all blood lineages,that is,they are multipotent.As daughter cells continue to divide,they migrate down the hierarchy and become progressively lineage restricted.This occurs through a well-orchestrated series of dis-crete steps,and eventually,yields the various mature blood cell types.In homeostasis,the cell hierarchy is rigid.Although these insights were obtained decades before it became feasible to study other types of adult stem cell,the defining characteristics of HSCs have served as a dominant template for interpreting experimental observations of the renewal of other mammalian tissues and cancer.Xenotransplantation assays to assess CSC activityIn HSC research,the capacity of a given cell to reconstitute the hemato-poietic lineage upon transplantation into lethally irradiated immune-deficient mice has been used as a surrogate of stem cell potential2.Over the past two decades,this assay has been instrumental in identifying 1Institute for Research in Biomedicine(IRB Barcelona),The Barcelona Institute of Science and Technology,Barcelona,Spain.2Instituci Catalana de Recerca i Estudis Avanats(ICREA),Barcelona,Spain.3CiberONC,Centro de Investigacin Biomdica en Red de Cncer(CIBERONC),Barcelona,Spain.4Hubrecht Institute,Royal Netherlands Academy of Arts and Sciences,University Medical Center Utrecht and Princess Maxima Center,Utrecht,the Netherlands.Correspondence should be addressed to E.B.(eduard.batlleirbbarcelona.org)and H.C.(h.clevershubrecht.eu).Received 23 January;accepted 23 August;published online 6 October 2017;doi:10.1038/nm.4409Cancer stem cells revisitedEduard Batlle13&Hans Clevers4The cancer stem cell(CSC)concept was proposed four decades ago,and states that tumor growth,analogous to the renewal of healthy tissues,is fueled by small numbers of dedicated stem cells.It has gradually become clear that many tumors harbor CSCs in dedicated niches,and yet their identification and eradication has not been as obvious as was initially hoped.Recently developed lineage-tracing and cell-ablation strategies have provided insights into CSC plasticity,quiescence,renewal,and therapeutic response.Here we discuss new developments in the CSC field in relationship to changing insights into how normal stem cells maintain healthy tissues.Expectations in the field have become more realistic,and now,the first successes of therapies based on the CSC concept are emerging.REVIEW 2017 Nature America,Inc.,part of Springer Nature.All rights reserved.NATURE MEDICINE VOLUME 23|NUMBER 10|OCTOBER 2017 1125tumors results from its hierarchical organization,which is often,but not always,reminiscent of the hierarchy in the tissue of origin.Second,tumor hierarchies are fueled by rare self-renewingtypically quies-centCSCs,whereas the bulk of the tumor is composed of non-CSCs,which are capable only of transient proliferation and therefore do not contribute to long-term growth.Third,CSC identity is hardwired,as illustrated by the fact that non-CSCs seldom initiate tumors in xenograft assays.Thus,there is limited plasticity in the tumor hierarchy.Finally,CSCs are resistant to standard chemotherapy and radiation treatment,which preferentially target non-CSCsa phenomenon that explains relapse after treatment.Analysis of CSCs in intact tumorsThe xenotransplantation approach to investigating the properties of CSCs carries inherent technical and conceptual limitations(see Box 1 and refs.18,24).In search of more direct evidence,several groups have studied CSCs in intact tumors through genetic-lineage tracing,the gold standard for assessing adult stem cell activity in situ(see Box 1).Lineage-tracing studies in mouse tumorsAn early study using this approach in a chemically induced tumor model for squamous skin cancer traced clones using an inducible,basal cellspecific keratin-14-Cre driver allele11.Although most labeled tumor sets of cell-surface markers that allowed for fluorescence-activated cell sorting(FACS)of normal human HSCs and their various progenitors2.In pioneering studies of the properties of leukemia,it was found that most subtypes of human acute myeloid leukemia(AML)could be engrafted reliably into immune-deficient mice by transplanting a population of leukemic cells that expressed a combination of surface markers(CD34+CD38-)characteristic of normal HSCs.The frequency of these tumor-initiating cells was on the order of one per million tumor cells35.Using equivalent cell-sorting and xenografting approaches,a team showed that human breast cancer,i.e.,a solid tumor,is comprised of functionally heterogeneous populations of cancer cells with varying ability to form the xenograft6.As few as 100 CD44+CD24-/low breast can-cer cells could initiate tumor growth,even in a serial xenograft setting,whereas tens of thousands of cells with alternative phenotypes could not6.Similar studies of other solid tumors,such as brain cancer10,pros-tate cancer19,colon cancer79,pancreatic cancer20,ovarian cancer21,and lung cancer22,rapidly followed.We refer the reader elsewhere for a comprehensive overview of these efforts23.The standard CSC modelToward the end of the past decade,these studies supported a CSC model based on four premises that echoed the key features of the HSC hier-archy.First,a substantial fraction of cellular heterogeneity observed in Box 1 Transplantation compared to lineage-tracing approaches to study CSCsPrototypic CSC-transplantation assays involve sorting tumor cell populations on the basis of surface-marker heterogeneity,followed by inoculation of the isolated cell populations into immunodeficient mice in numbers sufficiently low to limit the formation of a xenograft by the bulk tumor cell population.The capacity of a given cell population to initiate a tumor under these conditions over serial passages in mice is interpreted as evidence for the presence of CSCs310,1922.These tumor-cell transplantation assays were originally designed to study hematological malignancies,which poses several caveats regarding their utility for analyzing CSCs in solid cancers:in general,leukemias have limited genetic changes and harbor restricted intratumor genetic heterogeneity,as compared to most solid cancers.Thus,the a priori assumption of the FACS or xenografting approach(i.e.,that functional differences are being scored between genetically homogeneous tumor-cell subpopulations)may be more appropriate for leukemias than for solid cancer.Although the sorting strategies in leukemias are founded on an extensive knowledge of marker combinations that define the various normal HSC and progenitor cell types,such knowledge is often absent for the tissue of origin of many types of solid cancer.The choice of testable markers for CSCs in solid cancers is often based on differential expression between different tumor-cell subpopulations and/or on knowledge of stem cellspecific expression of the marker in an unrelated tissue.Finally,transplantation assays involve dissociation of the tumor mass.Yet,cells in solid tumors rely heavily on cell-to-cell contacts,attachment to the extracellular matrix,and signals from the microenvironment.Thus,tumor-initiating potential in transplantation assays may,at best,serve as a surrogate for CSC-autonomous properties,or even simply reflect the adaptation of particular tumor-cell populations to the experimental conditions.Genetic-lineage tracing enables the identification and study of stem cells in solid tissues in situ while avoiding mechanical perturbation.This technique rests crucially on the identification of a single marker gene that allows the expression of an inducible version of a recombinase(for example,Cre),which,in turn,allows for the stable activation of a reporter for the pertinent recombinase(for example,the R26R LacZ reporter149)in the cell population of interest,ultimately resulting in the labeling of the cells of interest.Importantly,stable reporter expression is maintained in all daughter cells of the marked cell.Persistence,size,and composition of cell clones generated over time are used to evaluate stemness potential.We refer the reader to excellent reviews on the advantages and technical limitations of this approach17,150152.The behavior of normal adult stem cells assessed by lineage tracing differs substantially from that inferred through transplantation experiments.Hair-follicle stem cells give rise to all epidermal lineages upon transplantation153,154,but upon lineage tracing,generate only hair-follicle lineages155.Similarly,although mammary basal cells are multipotent in transplantation assays156,157,these cells are unipotent when interrogated by lineage tracing158,159.Of note,the use of distinct drivers and recombination strategies to study mammary stem cells in lineage-tracing experiments has generated controversy regarding the identity of this particular cell type160,161.Similar discrepancies were encountered when HSCs were analyzed by transplantation as compared to lineage-tracing approaches.In one study,a transplant-free tagging strategy was developed using a transposon-based approach to study hematopoiesis in mice.Steady-state blood cell production seemed to be maintained by the successive recruitment of thousands of clones,each with a minute contribution162.These findings contradicted the classical model of a hierarchy supplied by few HSCs inferred from transplantation assays.In a different study,the same question was addressed by genetic lineage tracing based on the HSC-specific TIE2 gene163.Authors similarly found that steady-state adult hematopoiesis is maintained largely by multipotent progenitor cells,which are capable of reconstituting the blood lineages only transiently in classical transplantation assays,yet seem to be long-lived according to analysis of intact hematopoiesis by lineage tracing.Overall,these observations imply that transplantation-based approaches may reveal the potential of stem cells,but may not necessarily unveil the fate of these cells under steady-state conditions.REVIEW 2017 Nature America,Inc.,part of Springer Nature.All rights reserved.1126 VOLUME 23|NUMBER 10|OCTOBER 2017 NATURE MEDICINEcells in papillomas were lost after terminal differentiation,some survived long term and generated large clones within the growing benign tumors,indicating the existence of actual CSCs.Mathematical models built from the lineage-tracing data indicated that the papilloma CSC population divides asymmetrically to give rise to CSCs and progenitors that are committed to differentiate.Yet,at the individual level,CSCs undergo both symmetric(i.e.,producing two stem cells)and asymmetric divi-sions in stochastic patterns11.This mode of division is similar to that of normal stem cells in the epidermis25,26.A different pattern was observed in invasive squamous cell carcinomas,consistent with the expansion of a single CSC population that has limited potential for terminal dif-ferentiation11.Similar findings support the presence of stem cells in primary intes-tinal mouse adenomas,the precursors to intestinal cancer12.Tumors were induced by conditional deletion of the tumor-suppressor gene Apc,a negative regulator of the WNT pathway that is mutated in most colon cancers.In these experiments,Apc deletion was targeted in intestinal stem cells(ISCs)by exploiting the specific expression of the intestinal stem cellmarker gene Lgr5(ref.27)while concomitantly labeling indi-vidual Apc-mutated stem cells in red12.After the resulting single-colored tumors had grown to a substantial size,cells expressing Lgr5 in these tumors were induced to switch their color from red to blue.Blue cells generated large clonal patches in the red tumor,providing evidence for a hierarchical organization of adenoma growth in vivo12.An unbiased lineage-tracing approach based on a mutation-induced mark during DNA replication confirmed the appearance of large dominant clones from adenoma CSCs,and it refined the model by postulating that only a small fraction of Lgr5+adenoma cells act as CSCs13.Another stud