Cancer Cell Phenotypes, in Fifty Shades of Grey

9 downloads 0 Views 337KB Size Report
Cancer Cell Phenotypes, in Fifty Shades of Grey. This copy is for your personal, non-commercial use only. clicking here. colleagues, clients, or customers by.

Cancer Cell Phenotypes, in Fifty Shades of Grey Andriy Marusyk and Kornelia Polyak Science 339, 528 (2013); DOI: 10.1126/science.1234415

This copy is for your personal, non-commercial use only.

If you wish to distribute this article to others, you can order high-quality copies for your colleagues, clients, or customers by clicking here.

The following resources related to this article are available online at (this information is current as of February 4, 2013 ): Updated information and services, including high-resolution figures, can be found in the online version of this article at: A list of selected additional articles on the Science Web sites related to this article can be found at: This article cites 8 articles, 3 of which can be accessed free: This article appears in the following subject collections: Medicine, Diseases

Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by the American Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright 2013 by the American Association for the Advancement of Science; all rights reserved. The title Science is a registered trademark of AAAS.

Downloaded from on February 4, 2013

Permission to republish or repurpose articles or portions of articles can be obtained by following the guidelines here.


Intratumor heterogeneity is a major obstacle in successfully eradicating tumors.

Cancer Cell Phenotypes, in Fifty Shades of Grey Andriy Marusyk and Kornelia Polyak

Department of Medical Oncology and Center for Functional Cancer Epigenetics, Dana-Farber Cancer Institute and Department of Medicine, Harvard Medical School, Boston, MA 02215, USA. E-mail: [email protected]


High proliferation

High proliferation

Downloaded from on February 4, 2013

Changes in clonal dynamics during tumor growth


High proliferation

Distinct clonal populations Changes in clonal dynamics due to therapeutic interventions Reduced tumor growth Chemotherapy

Low proliferation (reduced sensitivity to chemotherapy)

Increased proportion of population

Differences in clonal behavior. (Top) A tumor that has not been exposed to therapies consists of clones that display distinct behaviors (shown by different colors). The tumor is generally dominated by clones with high proliferative output that can be detected through serial xenograft transplantations into immunodeficient mice. In the absence of therapy, the proportion of different types of clones can change randomly. (Bottom) By contrast, chemotherapy increases the proportion of clones with lower proliferative output, which in turn translates into reduced overall tumor growth even after the treatment is stopped. This change in clonal landscape likely reflects the reduced sensitivity of slower-proliferating clones to chemotherapy. The differences in proliferative output among clones are unlikely to be linked to genetic differences or position in differentiation hierarchies.

observations did not necessarily mean that all distinct clonal behavior patterns reflect meaningful biological diversity, mathematical modeling and statistical analyses suggested clonal differences in proliferative outputs. Notably, treatment of xenograft-bearing mice with oxaliplatin—a chemotherapeutic agent commonly used in colorectal cancer patients—preferentially eliminated persistent clones and increased the proportion of clones that were initially below the detection threshold (see the figure). What mechanism could explain these differences in clonal behavior? Although it is formally possible that the different clones were genetically distinct, the stability of copy number alteration profiles and concordance for mutational hot spots renders this possibil-

ity unlikely, at least for the nuclear genome. Nor can the differences in proliferative outputs between the clones be explained on the basis of “traditional” distinctions between stem cells and non–stem cells, because tumor engraftment and maintenance over multiple passages fulfill the definition of cancer stem cells. Cell differentiation hierarchy is not the only type of nongenetic heterogeneity, as genetically identical cells can display phenotypic variability due to stochasticity in gene expression and signaling pathways—a phenomenon known as cellular heterogeneity (5). However, phenotypic differences associated with cellular heterogeneity are relatively transient, and it is not clear whether they could account for the apparently stable differences in clonal behavior.

1 FEBRUARY 2013 VOL 339 SCIENCE Published by AAAS



ntratumor heterogeneity refers to biological differences between malignant cells originated within the same tumor. Possible explanations for this include genetic heterogeneity (resulting from the inherent genetic instability of cancer combined with evolutionary dynamics) and cell differentiation hierarchies in tumor cell populations (1–3). However, the results of Kreso et al. (4) on page 543 of this issue strongly suggest that biological differences between tumor cells can be due to additional mechanisms. To explore functional heterogeneity within tumors, Kreso et al. traced the fates of single cell–derived clones from 10 different human primary colorectal tumors over multiple serial transplantations in mouse xenografts. Analysis of copy number alterations and deep sequencing for mutational hotspots in 42 cancer genes revealed that a number of xenografts retained the genomic profile of the primary tumor, whereas in some cases the first transplant and the parental tumor displayed substantial genetic differences indicative of clonal selection during xenograft growth, but in subsequent transplants the tumors remained genetically fairly stable. Deep sequencing of the mutational hotspots in distinct single cell–derived clones also showed high concurrence, implying relative genetic homogeneity within xenografts. Despite this genetic homogeneity, the different cancer clones that Kreso et al. observed in a tumor—made distinguishable through unique genomic integration sites of a lentiviral vector as a marker—displayed notable differences in behavior during serial transplantation. In addition to persistent clones that were observed through multiple passages and clones that became extinct with serial passages, the authors identified clones with rather unexpected behaviors. Some of these unusual clones remained below the detection threshold during the initial passages but reemerged with later transplantations; another pattern was the disappearance below threshold in mid-passages and reappearance in subsequent transplants. Although these

PERSPECTIVES That therapeutic resistance can be caused by distinct epigenetic states (7), which can be prevented by targeting epigenetic regulators, opens up new opportunities for improved cancer treatment. The conclusions of Kreso et al. also warrant more careful assessments of studies interpreted through the cancer stem cell paradigm. Quantitative analysis of such studies frequently reveals numerical inconsistencies that can invalidate the conclusions (8). Refraining from mapping the differences in cancer cell phenotypes into differentiation hierarchies would lead to more accurate scientific interpretation of the data, which is critical for clinical translation. For example, despite relatively limited experimental evidence, it is widely assumed that therapeutic resistance is fueled by the preferential survival of cancer stem cells, whereas the results of Kreso et al. suggest that a therapy-resistant state might be unrelated to “stemness” per se. Cancer cell phenotypes are not black and white but rather display a continuum of many different colors and shades. Kreso et al. analyzed colorectal cancers that are fairly homogeneous for tumor-driving genetic alterations (i.e., mutations in the APC-Wnt–β-catenin signaling pathway are observed in nearly all human colorectal cancer) and these tumors were further genetically “homogenized” by passing through a selec-

tion bottleneck created by xenotransplantation. However, phenotypic heterogeneity in the majority of human cancers is likely to be more complex, as it represents the integration of both genetic and nongenetic inputs. Therefore, adequate understanding and eradication of cancers requires a comprehensive picture that accounts for all major inputs that dictate tumor cell behaviors, including their response to current and future therapies. Technological advances that enable the complete molecular and functional profiling of individual cancer cells, as well as improved mathematical models built on actual clinical and experimental observations, will likely allow us to construct these pictures in the not-so-distant future.

Downloaded from on February 4, 2013

A more likely explanation is the involvement of one or more distinct semistable epigenetic states that cannot be directly mapped to a differentiation hierarchy. In bacterial cell populations, a fraction of cells randomly assumes a distinct phenotype that is characterized by resistance to stress, including antibiotic treatment, at the expense of reduced proliferation rates (6). The reemergence of previously minor clones after oxaliplatin treatment and their ability to initiate new tumors (although at smaller size) suggest the presence of such slow-growing dormant clones. Similarly, a distinct phenotypic transition characterized by altered chromatin organization has been described recently as a mechanism of resistance to anti–epidermal growth factor receptor therapy in human lung cancer (7). Although the cells are locked in this epigenetic state for multiple population doublings, they can eventually revert back to their original drug-sensitive state. Kreso et al. do not reveal mechanisms responsible for the variability in clonal behavior. Still, the study has several important implications. It highlights the need to adequately understand epigenetic mechanisms that underlie cancer cell phenotypes, including studies of cellular heterogeneity and more stable distinct epigenetic states (stochastic and deterministic heterogeneity).

References and Notes 1. L. L. Campbell, K. Polyak, Cell Cycle 6, 2332 (2007). 2. M. Shackleton, E. Quintana, E. R. Fearon, S. J. Morrison, Cell 138, 822 (2009). 3. A. Marusyk, V. Almendro, K. Polyak, Nat. Rev. Cancer 12, 323 (2012). 4. A. Kreso et al., Science 339, 543 (2013); 10.1126/ science.1226670. 5. S. J. Altschuler, L. F. Wu, Cell 141, 559 (2010). 6. N. Q. Balaban, J. Merrin, R. Chait, L. Kowalik, S. Leibler, Science 305, 1622 (2004). 7. S. V. Sharma et al., Cell 141, 69 (2010). 8. S. E. Kern, D. Shibata, Cancer Res. 67, 8985 (2007).

Acknowledgments: Supported by the Breast Cancer Research Foundation. 10.1126/science.1234415


Toward Molecular-Scale MRI

Use of a diamond-based nanomagnetometer reduces the detection volume of MRI to the level of individual protein molecules.

Philip Hemmer


agnetic resonance imaging (MRI) is a mainstay of medical diagnostics, allowing nondestructive imaging inside opaque objects with high resolution. There have been many attempts to use MRI to image small objects such as living cells, because the resolution can be well below the optical diffraction limit. However, the detection sensitivity of conventional MRI falls rapidly for smaller feature sizes, making it impossible to resolve features smaller than a few micrometers with this method. Two reports in this issue, by Mamin et al. on page 557 (1) and Staudacher et al. on page 561 (2), demonstrate the ability to detect volumes of a few cubic nanometers, comparable to the size of large Texas A&M University, Electrical and Computer Engineering Department, College Station, TX 77843, USA. E-mail: [email protected]

protein molecules. These independent studies are a crucial step toward molecular-scale magnetic resonance imaging. In conventional MRI and the related chemical diagnostic nuclear magnetic resonance (NMR), magnetic nuclei such as protons are excited by a radiofrequency field, plus a large magnetic field gradient in the case of MRI. The resulting magnetic signal emitted by the excited nuclei is collected by a magnetic induction coil, much like the antenna loop used in ultrahigh-frequency TV reception, and digitally processed to produce a spectrum or three-dimensional image. However, as the image voxel size is shrunk to allow imaging of ever smaller objects, the magnetic signal becomes too weak to be extracted from the background magnetic noise in the environment. To mitigate this noise, the antenna loop can be shrunk, but so far this approach has not

allowed MRI to achieve a resolution beyond a few micrometers. To circumvent this problem, nonconventional detection techniques are needed. The first of these was magnetic resonance force microscopy (MRFM), developed by Rugar et al. nearly two decades ago (3). This approach takes advantage of the force between two magnets: the nuclear or electron spins being imaged on the one hand and a powerful nanomagnet attached to the tip of a scanning probe on the other. This approach allowed detection and imaging of single electrons (4) and imaging of ensembles of nuclei in a virus particle (5). However, it required ultralow temperatures and could therefore not be used to image dynamic samples in ambient environments, like living cells. Another approach uses a diamondbased nanomagnetometer to sense the magnetic fields generated by small ensembles of SCIENCE VOL 339 1 FEBRUARY 2013 Published by AAAS