$250K International Collaborative Grant Funds Liposarcoma Research
The Liddy Shriver Sarcoma Initiative and the Wendy Walk are pleased to announce the awarding of a $250,000 grant to fund global collaborative research on liposarcoma. The project builds on previous research funded by the Initiative and brings together four lead investigators in three countries.
Liposarcoma accounts for 10-20% of all soft tissue sarcomas, which are rare cancers of the connective tissues. There are currently no effective systemic therapies for the disease, which can be highly lethal.
In this study, investigators aim to learn how well-differentiated liposarcoma transforms to a more aggressive and less differentiated form, to understand how liposarcoma can adapt to treatment and become resistant, and to identify new therapeutic targets for treatment.
Dr. Meza-Zepeda describes the research: "This project involves a multinational and multidisciplinary group of oncologists and basic researchers working together to advance liposarcoma research. For rare cancers like liposarcomas, this collaboration has accelerated and made possible research projects that would take considerably longer time to perform within an individual research group. This type of project allows sharing of competence, as well as material and data, accelerating the generation of results."
The Challenge of Targeting Liposarcoma
Recently, researchers have discovered that targeted therapy may be helpful for lipsarcoma patients, but clinical trials have been challenging. Dr. Myklebost explains: "The most important targeted therapy that has appeared for this type of tumours has been inhibitors of MDM2 and CDK4, oncogenes that are activated by amplification in most of these tumours. However, the clinical trials so far have been disappointing, at least for the MDM2 inhibitor Nutlin, due to severe side effects not seen in mice."
The team plans to investigate a different inhibitor, but the difficult side effects may appear again in clinical trials. They will also go through a large number of amplified genes that are present in well-differentiated liposarcomas, seeking those that can be targeted and testing the efficacy of therapies in vitro. If new therapeutic possibilities open up, they will be communicated to the World Sarcoma Network to be considered for clinical testing.
Understanding the Development of Well-Differentiated and Dedifferentiated Tumors
By Andrew Wagner, MD, PhD
We know that both well-differentiated and dedifferentiated liposarcoma share very similar genetic alterations that contribute to tumor formation. This includes, among other processes, too many copies of the genes for CDK4 and MDM2, which in turn help drive the cells to grow and survive. Both these changes are seen in both of these forms of liposarcoma, so surely other changes must also be occurring that determine whether the tumor is aggressive or not. Why do some patients only have one form or the other? Why are some tumors mixed with both well-differentiated and dedifferentiated tumors? Why do some tumors regrow after surgery and others do not? By examining the genetic constitution of both subtypes of liposarcoma from the same patient, we can study what abnormalities have been gained or lost to lead to each type.
We hope to be able to use this information to deconvolute the biology and behavior of these tumors, and also to guide us in developing new treatments that may convert an aggressive tumor into one that is 'well-behaved,' and ideally to stop its growth or shrink it altogether.
Fruit of the International Collaborative Grant (ICG) Program
This study is the result of a similar grant awarded in 2010. Dr. Thomas explains: "The first liposarcoma ICG developed a small community, initially in Norway and Australia, but extending to colleagues in the US and France. The grant enabled the establishment of a global liposarcoma program that would have struggled to find the resources for such an enterprise from any other source. Both the initial commitment, and this subsequent investment, have been courageous and innovative, demanding that the scientific community match the vision and ambition of the Liddy Shriver Sarcoma Initiative."
The Lead Investigators
This research will be conducted by a team of clinicians and scientists with strong individual track records in liposarcoma research. Investigators include:
Ola Myklebost, PhD
"My group will focus on identification of mutations in liposarcoma tumours and cell lines and identify possible new therapeutic targets. These will then be investigated further in our cell line models to confirm if they appear to have effect in these cancers, and subsequently in some of our human tumor models transplanted to mice. The hope is that some of these experimental therapies may be tested on patients in the future, and open new opportunities."
Leonardo Meza-Zepeda, PhD
"My research group will contribute with our genomic competence, helping to reveal molecular mechanism of drug resistance to targeted therapies in liposarcomas and identify possible novel targets for therapy. We will also characterize, at the molecular level, widely used preclinical models for basic and translational studies."
Andrew Wagner, MD, PhD
"I will lead a project that examines the genetic differences between the typically slow growing well-differentiated liposarcoma and the aggressive dedifferentiated liposarcoma, which often are found within the same tumor. When we identify genes that are implicated in this transition, we will experimentally test their function in our models of liposarcoma that we use in the laboratory, and potentially will be able to test drugs that specifically target these alterations, depending on what we find."
David Thomas, FRACP, PhD
"My laboratory is focusing on a fundamental question about how the WDLPS genome changes under selective pressure—including drugs such as the MDM2 and CDK4 inhibitors. We believe that these insights into how cancer genomes react to stress will be important in addressing the clinical problem of drug resistance."
The Wendy Walk has stepped forward to fund this entire $250,000, two-year liposarcoma research effort. We are indebted to their continuing generosity and the attention they have brought, internationally, to liposarcoma.
The Wendy Walk was formed by the children of Wendy Landes, Matt, Ali and Jackie, to support her when she was diagnosed with liposarcoma. In March 2013, sadly, Wendy passed away after an incredibly brave, inspiring, and hard-fought, five-year battle against this terrible disease. Her family has been relentless in their efforts to help improve the treatment options available to liposarcoma patients and to help find a cure for these rare cancers.
V11N4 ESUN Copyright © 2014 Liddy Shriver Sarcoma Initiative.
Translational Research in Well-Differentiated
Dedifferentiated Liposarcoma - Part II
Liposarcomas are mesenchymal tumors with adipocytic differentiation that comprise approximately 20% of adult sarcomas. These tumors represent at least three biologically distinct diseases based on clinical, morphological, and cytogenetic and molecular genetic characteristics:
(1) well differentiated/dedifferentiated liposarcoma (WDLPS/DDLPS),
(2) myxoid liposarcoma, and
(3) pleomorphic liposarcoma.
The most common type is well-differentiated/dedifferentiated liposarcoma (WDLPS/DDLPS) and is the focus of our research. WDLPS is a slowly growing tumor that exhibits adipocytic differentiation, yielding a soft mass that can resemble normal fat. It is classified as a low-grade neoplasm with almost no capacity to metastasize and can be cured with complete surgical excision. However, in certain anatomic locations such as the retroperitoneum, a common site for presentation, it can be impossible to obtain negative surgical margins, and with time local recurrences can develop and cause morbidity and mortality. It is common that WDLPS over time progresses to DDLPS, which is a much more aggressive subtype.
DDLPS is an aggressive malignant spindle cell tumor with a much more rapid rate of growth in contrast to WDLPS, and a risk of both local recurrence and distant metastases, leading to an important cause of mortality. Once further surgery is no longer feasible, there are no known therapies that provide durable disease control.
Clinically, there are heterogeneous presentations and behavior of WDLPS/DDLPS. These can present as a purely well-differentiated (adipocytic) mass and never develop a de-differentiated component; as a rapidly growing dedifferentiated spindle cell tumor with little to no adipocytic component; as a mixture of slow growing well-differentiated and rapidly growing de-differentiated components; or initially as a well-differentiated tumor that later aggressively recurs as a de-differentiated tumor.
Importantly, virtually all WDLPS and DDLPS have a pathognomonic genomic amplification with hundreds of copies of a region on the long arm of chromo-some 12, encompassing a large number of genes notably including the oncogenes MDM2 (12q15) and frequently CDK4 (12q14.1). The products of these genes are useful as diagnostic markers and also are the targets of novel therapeutic approaches to management of the disease.
These findings present a paradox: why do WDLPS and DDLPS have such different clinical presentations despite having very similar gross genomic changes? Clearly additional genomic alterations are likely to occur and may explain the clinical progression of WDLPS to DDLPS. Identification of such processes will yield important insight into the development of these diseases, ability to predict risk of progression, and potentially identify other therapeutic targets.
Recently, other groups have identified changes in MAP3K5 and JUN copy number as possible causes of the transition to WDLPS (Chibon 2004; Mariani 2007; Snyder 2009). However, these alterations appear to be relatively infrequent and a detailed genomic analysis with biochemical validation has not yet been performed.
As part of this research project we will perform whole exome sequencing (WES) of triplets of samples containing DNA from normal tissue, WDLPS, and DDLPS, all obtained from 23 patients with synchronous components in their tumors, in collaboration with colleagues at the Broad Institute in Cambridge, MA. These studies will permit detection of mutations and copy number alterations in reference to patient-defined normal sequences as well as in the tumorigenesis of WDLPS and DDLPS. Genes of interest, that is genes with significant alterations in sequence or copy number and differing between the two components from the same patient, will be functionally studied in 3T3-L1 cells (that can be induced to undergo adipocytic differentiation) in an immortalized mesenchymal stem-like cell system (iMSC#3, developed in Oslo ) and in well-characterized liposarcoma cell lines and tumor models available in the consortium. Through combinations of siRNA knock down experiments, overexpression of amplified genes, or introduction of mutated alleles of genes, we will investigate the effects on cellular proliferation, migration, tumorigenic potential, and adipocytic differentiation using established assays.
Cancers may be regarded as a species, evolving and adapting to selective pressures. We want to understand the molecular mechanisms that form the basis of adaptive mechanisms in human cancer cells in response to environment fluctuations represented by pharmacological and physiological stresses.
Most novel cancer therapies produce remissions for a limited time, ultimately developing resistance and clinical relapse. Addressing this issue is fundamental to the future of cancer therapy. We believe it is important to better understand how cancer cells adapt and evolve in response to selection, whether by therapeutics, or other environmental and endogenous factors.
Selection of any species depends on the operation of environmental pressure and genetic diversity. Conventionally, evolution is thought to operate primarily at the level of selection, rather than genetic diversity. Recent studies demonstrate that genetic instability can itself accelerate adaptive evolution in stressful environments. Moreover, elevated rates of genetic and phenotypic diversity are characteristic of malignancy.
On the other hand, genomic instability comes at a fitness cost. Random mutations, particularly involving large-scale changes in copy number, are more often deleterious than advantageous. This suggests that genetic diversity itself may be selected against in stable environmental niches (endogenous selective pressure).
We hypothesize that:
- genetic diversity may not be constant, but may vary in response to environmental stress, including drug selection; and
- the mechanisms regulating genetic diversity are evolutionarily conserved, and reactivated in cancer cells.
If true, a "hard-wired" stress-regulated genetic diversity program may be a novel therapeutic target.
Preliminary Results of the Experimental Program
We have established an in vitro model system using the human WDLPS/DDLPS cell line (778) to study the adaptive response to selective pressure. 778 cells display a key feature of WDLPS: the amplification and over-expression of the Mouse Double Minute 2 (MDM2) gene, thus they represent a good target for the MDM2 inhibitor Nutlin-3a (aka "Nutlin") that binds MDM2 and interferes with its ability to inhibit the activity of p53. We have deep genetic maps of the genome of the 778 cell line, including the neochromosome that carries the amplified copies of MDM2 and CDK4.
Characterization of the genomic instability in stress-adapted cancer cells
Using spectral karyotyping, Nutlin-resistant populations of cells demonstrated karyotypic instability, with a marked doubling of the numbers of marker chromosomes/metaphase (6.90.6 in untreated to 13.30.7 in resistant clones; avSEM, P<0.01). Consistent with this, a higher rate of -H2AX foci was observed in 778 resistant cells (Figure 1B) compared to parental cells (Figure 1A). These observations are reproducible, and are also observed with other selective stresses (CDK4 inhibitor, radicicol, tunicamycin). Collectively, these data indicate that drug resistance is associated with greater genomic instability. Importantly, these agents are not thought to directly induce genetic injury.
We next sought to quantitate the rate of accelerated mutagenesis at the single nucleotide level in resistant compared to parental cells. Using individual clones, the estimated relative mutation rate is ~1.24x higher in the resistant line than in the sensitive line. This estimated mutation rate is likely to be conservative, since we expect the drug-sensitive line to have more variants due to the nature of a bottleneck resulting in loss of variation. Collectively, these data indicate that adaptability programs induce a hypermutable phenotype in cancer cells. Array expression analysis (not shown) shows drug resistance is associated with down-regulation of genes involved in DNA repair, chromosome segregation and spindle checkpoint activity.
Increased genetic diversity comes at a fitness cost. As predicted[2,3], Nutlin-resistant cells demonstrate reduced clonogenic potential in the absence of drug (Figure 2A), as well as differences in morphology and invasivenesss (Figure 2B-G). As shown in Figure 2H, colonies adapted to Nutlin demonstrate much lower cell density (P< 0.0001), with increased migratory properties using wound healing assays (not shown).
A generalized phenomenon in response to non-genotoxic stress
Critically, we have shown other non-genotoxic stressors also elicit a hypermutable phenotype associated with fitness costs. These include tunicamycin (unfolded protein response in the endoplasmic reticulum); radicicol (Hsp90 inhibitor); and importantly, PD 033299, an inhibitor of CDK4. In each case, adapted subclones generally emerge after 4-8 weeks of stress exposure, with rate of emergence of resistant clones between 1 (Nutlin-3a) and 30 colonies per 100,000 cells (Tunicamycin).
These settings provide a reproducible model system able to recapitulate the mechanisms responsible for the fitness costs, the increase of genetic instability and the acquisition of a mutator phenotype characterizing the stress response and the adaptive programs of human cancer cells.
1. A genome-wide functional screen for genetic determinants of hypermutability
Using this model system, we aim to identify the genetic determinants of adaptation to unrelated pleiotropic stresses. A population of 778 cells is transduced with a lentiviral shRNA library and exposed to five different selective pressures (defined above). We hypothesize that shRNAs for genes that are rate-limiting for emergence of resistance will be enriched in common in the resistant cell populations.
These shRNAs will fall into two classes:
1. Genes that are specific to the selective pressure. For example, shRNA to TP53 will emerge from a positive selection screen for resistance to Nutlin, and shRNA to RB1 will be enriched in populations resistant to CDK4 inhibition.
2. Genes that are shared across all selection conditions (including for example those involved in DNA repair and replication fidelity) that may be responsible for hypermutability.
We will utilize a short-hairpin RNA (shRNA) library-based approach in which lentiviral particles with a pGIPZ shRNA-mir30 backbone (OpenBiosystems) will be used to transduce cancer cells.
Cells will be infected at low multiplicity of infection to ensure single integration and adequate (1000x) shRNA representation. 72 hours after infection, cells will be selected with Puromycin for 2 days to remove the non-transduced cells. Five days later cells will be exposed to different stress conditions as described, and cultured for up to 8 weeks. Cells will be harvested for quantitation of shRNA representation. ShRNA frequencies will be performed by massively parallel sequencing.
Controls will be the unselected transduced population at time 0, and a population of cells cultured in the absence of selection for 2 weeks to measure negative selection for shRNAs. Importantly, we anticipate that hypermutability adaptation pathways will be negatively selected in the absence of stress.
The highest confidence candidates will be those for which:
- Multiple shRNAs (minimum of 5 shRNAs per gene target in the whole genome library) have the same statistically robust phenotype.
- Bioinformatic pathway analysis of candidates that have one or two hairpins showing an effect will be used to triage the list to be analysed in more detail.
- Expression is modified following exposure to drug selection (see 1. above).
Validation of candidate genes will include reproducible adaptability to the stress using independent shRNA, and cognate siRNA, and testing of shRNAs in different cell lines to determine whether these genes are specific to WDLPS, or whether they are generically implicated in adaptability to stressful conditions.
2. Mechanisms of clinical resistance to antagonists of CDK4
Introduction to MDM2 and CDK4
Amplification and overexpression of the cyclin-dependent kinase 4 (CDK4) and the mouse double minute 2 (MDM2) genes play an important role in the biology of these tumors. Increased levels of MDM2 inactivate the TP53 pathway by MDM2-TP53 binding and sequestration, as well as targeting this complex for degradation. CDK4 amplification and overexpression leads to inactivation of the RB1 pathway by phosphorylation of RB1. Inactivation of these pathways plays is crucial for driving tumor growth. A number of small molecules have been designed to inhibit the function of these proteins, some of which are in clinical testing. Unfortunately the efficacy in patients has been limited, compared to the promising preclinical studies [4,5], and new biological knowledge is required to better understand the unsatisfactory response observed in clinical trials.
Cell lines and xenografts represent important in vitro models for studying disease mechanisms at the cellular level, and have been widely used as preclinical models for preclinical evaluation of new therapeutic drugs. Therefore, characterization of preclinical cell line models represent an important step towards better understanding the biology of different cancers, as well as identifying novel therapeutic strategies and biomarkers.
We have established a xenograft and an in vitro cell line model from a patient's tumor treated with the MDM2 inhibitor Nutlin and the CDK4 inhibitor Palbociclib. The patient showed no response to these inhibitors despite the clear high-level amplification and expression of both genes (Figure 3). Our genome sequencing demonstrated deletion of the RB1 gene, explaining the lack of effect of CDK4 inhibition, since the most important target of CDK4 is the retinoblastoma protein, encoded by this gene. Blocking MDM2, on the other hand, killed these cells in vitro, showing that they were not resistant to the treatment with Nutlin, but that insufficient dose was possible clinically, due to severe side effects. We will investigate these cells for other drug targets, that may be useful also for other tumours of this kind.
Molecular and phenotypic characterization of preclinical models developed from patient's tumors treated with CDK4 and MDM2 inhibitors, including several others available from the Wagner's group, will be of crucial importance to reveal the molecular mechanisms behind drug sensitivity and resistance, and the possible use of new drugs for cancer treatment in the future.
Using an integrative approach, we will investigate genomic, transcriptomic and phenotypic changes in cell line models derived from CDK4 inhibitor exposed tumors. This analysis will aim to reveal possible mechanisms of resistance to CDK4 inhibitors. DNA copy number changes, somatically mutations and gene expression will be used to identify additional targetable aberrations. In cases where patient material is available, we will also analyze patient samples to validate these models. In addition, we will compare the response of the in vitro model to selected drugs, and also alternative inhibitors available in the Wagner group.
At the molecular level we will search for biomarkers that can indicate to which drug the tumor models may be sensitive. The study may also extend to functional testing of candidate genes identified by the Thomas' group, who have by functional screening identified a number of genes that may code for components of resistance pathways. The contribution of these genes will be investigated by different functional studies.
3. Molecular and functional characterization of preclinical LPS models
To date, only a few immortal LPS cell lines have been characterized extensively. In order to better understand the biology of liposarcomas and provide model systems for sarcoma research and preclinical studies, we will characterize, at the molecular level, an extended panel of LPS cell lines.
The Myklebost group has collected a large panel of LPS cell lines that have been characterized extensively at the phenotypic level [6,7], specifically looking at cell proliferation, migration, tumorigenic potential, and stem cell and differentiation potential. Transcriptome data from these cell lines will be used to identify specific gene expression signatures that correlate to specific cancer phenotypes. This approach has successfully been used in an osteosarcoma cell line panel, where we have identified specific mRNAs that are associated with aggressive phenotypes . In addition to transcriptome data, we will generate whole exome sequencing data using a similar approach as for the WDLPS/DDLPS study. Sequencing will be performed at high coverage and somatic mutations will be identified using our custom analysis pipeline developed for the Norwegian Cancer Genomics Consortium (see NoSarC.org). Integration of genomic (DNA copy number and exome) and transcriptomic data will be used to identify mutations or amplified genes that are expressed, and that may represent drivers of LPS, as well as possible druggable targets. Candidate genes will be further validated by Myklebost and Wagner groups.
Protein kinases represent an important group of druggable molecules in cancer, and only a limited number of kinases and kinase pathways are known to be active in LPS (e.g. [9,10]). Using the well characterized LPS cell line panel we will profile activity of phosphotyrosine and serine-threonine kinases using the PamGene system. The PamChip uses an array of peptides representing substrates for the kinases present in a given sample. The functional readout of individual or multiple kinases is based on kinetic measurement of phosphorylation of the arrayed peptides using fluorescently-labeled anti-phosphate antibodies. Kinase profile activity of each cell line will be integrated with mutational and transcriptional information in order to identify activating or silencing mutations within the studied kinase pathways. The Myklebost group using different in vitro models will perform more detailed characterization of targetable mutations.
This well-characterized collection of cell lines will be of great value for identifying suitable models for preclinical studies. In addition, upon publication, the detailed molecular and phenotypic characteristics of these models will be made available to the scientific community through our consortium web site.
This proposal arises out of the first Liddy Shriver Sarcoma Initiative International Collaborative Grant in WD/DDLPS and explores three clinically important areas of liposarcoma biology:
- Are there specific mutations that govern the transformation of the rather indolent well-differentiated tumours to the aggressive dedifferentiated type?
- Are there mechanisms of drug resistance to targeted therapies?
- Are there mutations and mechanisms that may be targeted by available drugs?
The experience from early clinical studies of targeted drugs in these patients is not as straightforward as the preclinical data would suggest. The efficacy of targeted drugs has been limited, perhaps partly due to the adverse side effects not seen in mouse studies. Thus, a better understanding of the mechanisms involved and how they are modified by other factors in these tumours and how the efficacy can be improved, while limiting adverse side effects, are critical topics to investigate for the field to progress, and will have wider implications for the treatment of cancer.
The understanding of mechanisms involved in dedifferentiation of well-differentiated liposarcomas will be important, both to develop biomarkers that may indicate the need for aggressive therapy and perhaps to suggest new treatments for the dedifferentiated variants. Knowledge of how tumours become resistant to antagonists of MDM2 and CDK4 will facilitate improved strategies to avoid resistance or counteract it with additional drugs. More knowledge of the many mutated and, in particular, amplified genes in liposarcomas may open new therapeutic options, but preclinical investigations are needed to confirm their efficacy in this tissue type. The targets and activity profiles of mutated kinases will be investigated in appropriate experimental models.
By Ola Myklebost, PhD
Department of Tumor Biology, Institute for Cancer Research, Oslo University Hospital
David Thomas, FRACP, PhD
Department of Medical Oncology, University of Melbourne
Leonardo A. Meza-Zepeda, PhD
Department of Tumor Biology, Institute for Cancer Research, Oslo University Hospital
Andrew Wagner, MD, PhD
Adult Oncology, Dana-Farber Cancer Institute
1. Skårn M, Noordhuis P, Wang MY, Veuger MJT, Kresse SH, et al. (2014) Generation and characterisation of an immortalised human bone marrow derived mesenchymal stromal cell line. Stem Cells and Development In press.
2. Das Thakur M, Salangsang F, Landman AS, Sellers WR, Pryer NK, et al. (2013) Modelling vemurafenib resistance in melanoma reveals a strategy to forestall drug resistance. Nature 494: 251-255.
3. Loeb LA (2001) A mutator phenotype in cancer. Cancer Res 61: 3230-3239.
4. Tovar C, Rosinski J, Filipovic Z, Higgins B, Kolinsky K, et al. (2006) Small-molecule MDM2 antagonists reveal aberrant p53 signaling in cancer: implications for therapy. Proc Natl Acad Sci USA 103: 1888-1893.
5. Muller CR, Paulsen EB, Noordhuis P, Pedeutour F, Saeter G, et al. (2007) Potential for treatment of liposarcomas with the MDM2 antagonist Nutlin-3A. Int J Cancer 121: 199-205.
6. Stratford EW, Castro R, Wennerstrom A, Holm R, Munthe E, et al. (2011) Liposarcoma Cells with Aldefluor and CD133 Activity have a Cancer Stem Cell Potential. Clin Sarcoma Res 1: 8.
7. Stratford EW, Castro R, Daffinrud J, Skarn M, Lauvrak S, et al. (2012) Characterization of liposarcoma cell lines for preclinical and biological studies. Sarcoma 2012: 148614.
8. Lauvrak SU, Munthe E, Kresse SH, Stratford EW, Namløs HM, et al. (2013) Functional characterization and genomic profiling of osteosarcoma cell lines identify mRNAs and miRNAs associated with aggressive cancer phenotypes. Submitted.
9. Zhang K, Chu K, Wu X, Gao H, Wang J, et al. (2013) Amplification of FRS2 and activation of FGFR/FRS2 signaling pathway in high-grade liposarcoma. Cancer Res 73: 1298-1307.
10. Jia B, Choy E, Cote G, Harmon D, Ye S, et al. (2014) Cyclin-dependent kinase 11 (CDK11) is crucial in the growth of liposarcoma cells. Cancer Lett 342: 104-112. .
V11N4 ESUN Copyright © 2014 Liddy Shriver Sarcoma Initiative.
Figure 1. Representative confocal microscopy images of untreated (A) and Nutlin-adapted 778 cells (B). (magnification 60x). Yellow, H2AX foci; red: Propidium Iodide. Average number of foci per cell shown (P=0.019).
A: Difference colony number between parental (white bar) and Nutlin-resistant (black bar) cells in clonogenic assays (P<0.001). Crystal violet staining of colonies obtained with 778 cells (B-D) and Nutlin-resistant cells (E-G), magnification 20x. H, scattered plot showing the cell density of individual colonies. Nutlin-resistant: black circles; untreated: white circles.