Cyclin-Dependent Kinases as Tumor Initiators and Therapeutic Targets

Material Information

Cyclin-Dependent Kinases as Tumor Initiators and Therapeutic Targets
Corsino, Patrick
Place of Publication:
[Gainesville, Fla.]
University of Florida
Publication Date:
Physical Description:
1 online resource (121 p.)

Thesis/Dissertation Information

Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Medical Sciences
Physiology and Pharmacology (IDP)
Committee Chair:
Law, Brian K.
Committee Members:
Baker, Stephen P.
Rowe, Thomas C.
Ostrov, David A.
Graduation Date:


Subjects / Keywords:
Antibodies ( jstor )
Breast cancer ( jstor )
Cadherins ( jstor )
Cancer ( jstor )
Cell cycle ( jstor )
Cell growth ( jstor )
Cell lines ( jstor )
Cells ( jstor )
Cyclins ( jstor )
Tumors ( jstor )
Physiology and Pharmacology (IDP) -- Dissertations, Academic -- UF
Electronic Thesis or Dissertation
born-digital ( sobekcm )
Medical Sciences thesis, Ph.D.


Cyclin D1/cdk2 complexes are present at high frequency in human breast cancer cell lines, but the significance of this observation is not fully understood. This report demonstrates that expression of a cyclin D1-cdk2 fusion protein under the control of the MMTV promoter results in mammary gland hyperplasia and fibrosis, and mammary tumors. These tumors contain regions of spindle-shaped cells expressing both luminal and myoepithelial markers. Cell lines cultured from these tumors exhibit the same luminal/myoepithelial 'mixed-lineage' phenotype that is associated with human basal-like breast cancer, and express a number of myoepithelial markers. The MMTV-D1K2 tumor-derived cell lines form highly invasive tumors when injected into mouse mammary glands. Invasion is associated with E-cadherin localization to the cytoplasm, or loss of E-cadherin expression. Cytoplasmic E-cadherin correlates with lack of colony formation in vitro and beta-catenin and p120ctn localization to the cytoplasm. Data suggest that the invasiveness of these cell lines results from a combination of factors including mislocalization of E-cadherin, beta-catenin, and p120ctn to the cytoplasm. Similar characteristics were also observed in human basal-like breast cancer cell lines, suggesting that these results are relevant to human tumors. Together these results suggest that abnormal cdk2 activation may contribute to the formation of basal-like breast cancers. The importance of cdk2 as well as the other cell cycle specific cdks in tumor development indicates their potential usefulness in cancer therapy. For this reason, cdks have been the subject of extensive research, and consequently many inhibitors have been developed to target these proteins. However, the compounds that comprise the current list of cdk inhibitors are largely ATP competitive. This study discusses the identification of a novel structural site on cdk2, which is well conserved between the cell cycle cdks. Small molecules identified by a high throughput in silico screen of this pocket exhibit cytostatic effects and act by reducing the apparent protein levels of cell cycle cdks. Drug-induced cell cycle arrest is associated with decreased Rb phosphorylation and decreased expression of E2F-dependent genes. Multiple lines of evidence indicate that the primary mechanism of action of these compounds is the direct induction of cdk aggregation. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis (Ph.D.)--University of Florida, 2008.
Adviser: Law, Brian K.
Statement of Responsibility:
by Patrick Corsino.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Corsino, Patrick. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
LD1780 2008 ( lcc )


This item has the following downloads:

Full Text




2008 Patrick Corsino 2


To my Mom 3


ACKNOWLEDGMENTS Firstly, I would like to acknowledge my profe ssor Brian Law. For four years he has guided me through my graduate education, and has provided whatever support he could, whenever it was needed. I would like to thank the members of the Law lab (Brad Davis, Mary Law and Nicole Teoh Parker) for their assistance in the lab as well as their friendship. I also thank the members of my committee (Dr. Steven Baker, Dr. David Ostrov and Dr. Thomas Rowe) for helping me with ideas for experime nts and in many cases with the experiments themselves. Finally, I would like to thank my family, and in particular my mother, without whom I would not be where I am today, literally and figuratively. 4


TABLE OF CONTENTS page ACKNOWLEDGMENTS ...............................................................................................................4 LIST OF FIGURES .........................................................................................................................8 ABSTRACT ...................................................................................................................................10 CHAPTER 1 BACKGROUND................................................................................................................... .12 Introduction .............................................................................................................................12 Cyclin-Dependent Kinases .....................................................................................................12 Genetic Studies .......................................................................................................................14 Cdk Aberrance in Cancer ........................................................................................................15 Cyclin D1/Cdk2 Complexes ...................................................................................................17 Cdk Inhibitors, Classic and Novel ..........................................................................................18 2 TUMORS INITIATED BY CONSTI TUIVE CDK2 ACTIVATION EXHIBIT TRANSFORMING GROWTH FACTOR-BETA RESISTANCE........................................22 Introduction .............................................................................................................................22 Materials and Methods ...........................................................................................................23 Construction of MMTV-Cyclin D1-Cdk2 (MMTV-D1K2) Transgenic Mice................23 Whole-Mount Staining and Histological Analyses .........................................................23 Isolation and Culture of Cancer and Tumor-Associated Fibroblast Cell Lines ..............24 Immunoblot Analysis of Tumor Samples and Tumor-Derived Dell lines and Rb Kinase Assays ..............................................................................................................24 Transcriptional Reporter Assays and Cell Proliferation Analyses ..................................25 Results .....................................................................................................................................25 Characterization of MM TV-D1K2 Mammary Glands ....................................................25 Characterization of MMTV-D1K2 Tumors ....................................................................26 Derivation and Characterization of Can cer Cell Lines from MMTV-D1K2 Tumors .....27 Derivation and Characterization of Fi broblast Cell Lines from MMTV-D1K2 Tumors .........................................................................................................................30 Discussion ...............................................................................................................................31 3 MAMMARY TUMORS INITIATED BY CONSTITUTIVE CDK ACTIVATION CONTAIN AN INVASIVE BA SAL-LIKE COMPONENT.................................................40 Introduction .............................................................................................................................40 Materials and Methods ...........................................................................................................42 Isolation of Tumor Cell Lines .........................................................................................42 Preparation and Analysis of Tu mor and Cell Extracts by Immunoblot ..........................43 Immunofluorescence Microscopy ...................................................................................43 5


Orthotopic Tumor Growth Studies ..................................................................................44 Immunohistochemical Analysis of Tumor Tissue Sections ............................................45 Results .....................................................................................................................................45 Mouse Mammary Tumor Virus-D1K2 Hypercellular Lesions Exhibit an Invasive Phenotype .....................................................................................................................45 Mouse Mammary Tumor Virus-D1K2 Tu mor Cells Display Characteristics Consistent with Basal-Like Breast Cancer ..................................................................46 Mouse Mammary Tumor Virus-D1K2 Tumor Lines Display Mixed Luminal/Myoepithelial Character ................................................................................48 Mouse Mammary Tumor Virus-D1K2 Tumor-D erived Cell Lines Form Invasive Tumors In Vivo ............................................................................................................49 Mouse Mammary Tumor Virus-D1K2 Tumors Resemble Human Basal-Like Breast Cancers .........................................................................................................................51 Mouse Mammary Tumor Virus-D1K2 Tu mor-Derived Cell Lines Exhibit Extensive Stress Fiber Formation and Cytoplasmic E-Cadherin, p120ctn, and Catenin Localization ....................................................................................................52 Discussion ...............................................................................................................................53 Mouse Mammary Tumor Virus-D1K2 Tumors ..............................................................53 "Mixed Lineage" Characteristics of MMTV-D1K2 Tumor Cell Lines ..........................54 Mouse Mammary Tumor Virus-D1K2 Invasiveness ......................................................55 4 A NOVEL CLASS OF CYCLIN-DEP DENDENT KINASE INHIBITORS IDENTIFIED BY MOLECULAR DO CKING ACT THROUGH A UNIQUE MECHANISM........................................................................................................................72 Introduction .............................................................................................................................72 Materials and Methods ...........................................................................................................73 Molecular Docking ..........................................................................................................73 Sequence Alignment ........................................................................................................74 Cell Culture .....................................................................................................................74 Chemical Synthesis .........................................................................................................74 Western Blot Analysis .....................................................................................................75 Construction of Stable Cell Lines ....................................................................................76 Ultracentrifugation Assay ................................................................................................77 Fluorescence Microscopy ................................................................................................77 Cloning and Expression of a Cyclin D1Cdk2 Fusion Protein (D1K2) Baculoviral Construct ......................................................................................................................78 In Vitro Aggregation Assay .............................................................................................79 Results .....................................................................................................................................79 High Throughput Screening of a Novel Cdk Drug Binding Site ....................................79 The NSC Compounds Inhibit the Proliferation of Cells in Culture................................80 Cytostatic Effects of the Compounds are a Result of an Apparent Reduction in Cellular Cdk Levels .....................................................................................................81 Decrease in Cdk Levels is a Result of Protein Aggregation ...........................................82 The NSC Compounds Bind Directly to Cdks ..................................................................84 Discussion ...............................................................................................................................85 6


5 CONCLUSION................................................................................................................... ....99 Discussion ...............................................................................................................................99 Summary ..........................................................................................................................99 The Nature of Cdk Aggregation ......................................................................................99 Implications for Aggregate-Inducing Molecules in Cancer Therapy ............................101 Future Research ....................................................................................................................101 In Vivo Compound Studies ............................................................................................101 Protein Crystallization ...................................................................................................102 Alternative Structural Pocket for Screening ..................................................................102 Lead Compound Optimization ......................................................................................103 Alternative Mouse Model Systems ...............................................................................104 Conclusion ............................................................................................................................104 LIST OF REFERENCES .............................................................................................................106 BIOGRAPHICAL SKETCH .......................................................................................................121 7


LIST OF FIGURES Figure page 2-1 Mouse Mammary Tumor Virus-cyclin D1-cdk2 (MMTV-D1K2) transgenic model and mammary phenotype ...................................................................................................33 2-2 Mouse Mammary Tumor Virus-D1 K2 mammary and salivary tumors ............................35 2-3 Isolation and characterizati on of MMTV-D1K2 cancer cell lines .....................................37 2-4 Isolation and characterization of tumor-derived fibroblast cell lines ................................39 3-1 Mouse Mammary Tumor Virus-D1K2 hype rcellular lesions invade into the mammary stroma ...............................................................................................................58 3-2 Cell lines derived from MMTV-D1K2 tumo rs exhibit protein expression profiles consistent with basal-like breast cancer .............................................................................60 3-3 Mouse Mammary Tumor Virus-D1K2 tumor cell lines exhibit mixed luminal/myoepithelial lineage ............................................................................................62 3-4 Tumors formed from MMTV-D1K2 cancer cell lines exhibit stromal invasion and Ecadherin mislocalization/downregulation upon orthotopic implantation ..........................65 3-5 Basal-like breast cancers exhibit an i nvasive, mixed-lineage phenotype and tumorassociated fibrosis ..............................................................................................................67 3-6 Mouse Mammary Tumor Virus-D1K2 tumor-derived cell lines exhibit extensive stress fiber formation and E-cadherin, p120ctn, and -catenin localization to the cytoplasm ...........................................................................................................................69 4-1 Identification of a novel cdk2 binding pocket and interacting molecules by highthroughput in silico screening ............................................................................................88 4-2 Compounds inhibit cell proliferation .................................................................................90 4-3 Compounds affect cell cycle cdks ......................................................................................93 4-4 Decrease in cdk levels is a result of protein aggregation ...................................................95 4-5 Compounds act directly on cdks ........................................................................................97 8


LIST OF ABBREVIATIONS SMA alpha-Smooth Muscle Actin CDK Cyclin-dependent kinase D1K2 Cyclin D1/cdk2 Fusion Protein DAPI 4',6-Diamidino-2-Phenylindole EGFR Epidermal Growth Factor FBS Fetal Bovine Serum GFP Green Fluorescent Protein H&E Hematoxylin and Eosin HGF Hepatocyte Grwoth Factor LCS Leica Confocal Software MMTV Mouse Mammary Tumor Virus mTOR Mammalian Target of Rapamycin NCI/DTP National Cancer Institute/ Developmental Therapeutics Program NLVS NIP-leu-leu-leu-vinylsulfone NSC Nomenclature Standards Committee PAI1 Plasminogen Activator Inhibitor PCNA Proliferating Cell Nuclear Antigen Rb Retinoblastoma RCSB Research Collaboratory fo r Structural Bioinformatics STAT3 Signal Transducer and Ac tivator of Transcription-3 TGF Transforming Growth Factor-alpha TGF Transforming Growth Factor-beta 9


Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CYCLIN-DEPDENENT KINASES AS TUMOR INITIATORS AND THERAPEUTIC TARGETS By Patrick Corsino December 2008 Chair: Brian K. Law Major: Medical SciencesPhysiology and Pharmacology Cyclin D1/cdk2 complexes are present at high frequency in human br east cancer cell lines, but the significance of this observation is not fu lly understood. This report demonstrates that expression of a cyclin D1-cdk2 fusion protein un der the control of the MMTV promoter results in mammary gland hyperplasia and fibrosis, a nd mammary tumors. These tumors contain regions of spindle-shaped cells expressing both luminal and myoepithelial markers. Cell lines cultured from these tumors exhibit the same luminal/myoepithelial "mixed-lineage" phenotype that is associated with human basal-like breast cancer, and ex press a number of myoepithelial markers. The MMTV-D1K2 tumor-derived cell lines form highly invasive tumors when injected into mouse mammary glands. Invasion is asso ciated with E-cadherin localization to the cytoplasm, or loss of E-cadherin expression. Cy toplasmic E-cadherin corr elates with lack of colony formation in vitro and -catenin and p120ctn localizati on to the cytoplasm. Data suggest that the invasiveness of these cell lines resu lts from a combination of factors including mislocalization of E-cadherin, -catenin, and p120ctn to the cytoplasm. Similar characteristics were also observed in human ba sal-like breast cancer cell lines, suggest ing that these results are relevant to human tumors. Together these re sults suggest that abnor mal cdk2 activation may contribute to the formation of basal-like breast cancers. 10


The importance of cdk2 as well as the other cel l cycle specific cdks in tumor development indicates their potential usefulne ss in cancer therapy. For this r eason, cdks have been the subject of extensive research, and consequently many inhi bitors have been deve loped to target these proteins. However, the compounds that comprise the current list of cdk inhibitors are largely ATP competitive. This study discusses the identification of a novel st ructural site on cdk2, which is well conserved between the cell cycl e cdks. Small molecules identified by a high throughput in silico screen of this pocket exhibit cytostatic effects and act by reducing the apparent protein levels of cell cycle cdks. Drug-induced cell cycl e arrest is associated with decreased Rb phosphorylation and decreased expr ession of E2F-dependent genes. Multiple lines of evidence indicate that th e primary mechanism of action of these compounds is the direct induction of cdk aggregation. 11


CHAPTER 1 BACKGROUND Introduction The cellular events of DNA replication and subs equent division from a single cell into two daughter cells are of critical importance to a mu lticellular organism. When this process is well regulated and functions in a precisely timed ma nner, an organism grows properly and maintains healthy cell turnover. When the sequence of events leading to cell division become deregulated, cells can develop cancerous traits. Members of the cyclin dependent kinase (cdk) family function at the center of the ce ll cycle, acting to regulate the progression from the G0/G1 phase to mitosis. The cell cycle specific proteins in this family, namely cdk1, cdk2, cdk4, and cdk6 control cell cycle progression, and in normally divi ding cells are subject to multiple levels of regulation. The first part of th is study focused on the contribu tion of cdk2 in the process of tumorigenesis. It examined the characteristics of cancer cells that are driven by a constitutively active version of this protei n. The second part of this study was concerned with the identification and characterization of novel cdk in hibitors. As the first step in the drug development pathway, these compounds are characterized in cell culture and in vitro leaving in vivo studies yet to be completed. Cyclin-Dependent Kinases Cdks are serine/threonine protei n kinases that phosphorylate a wi de variety of proteins that are involved in cell cycle progression (1, 2). Regulation of these kinases occurs on multiple levels. In order to become en zymatically active, cdks require the binding of an appropriate cyclin, as well as the phosphorylation of the appropria te residue on their activation loop by a cyclin dependant kinase activating kinase (CAK) (3, 4). Mechanisms of inactivation of cdks include phosphorylation on an inhibi tory site by wee1 kinase (5) and binding of an inhibitory 12


protein from one of the two families of cdk inhibitors. The first of these two families consists of p27 kip (6), p21 cip (7), and p57 kip2 (8) which primarily inhibits cdks 1 and 2 by binding to the cyclin/cdk complex and preventing access of cdks to substrates for phosphorylation. The second group consists of four proteins enco ded by the Ink4a/Arf locus, namely p15 ink4b p16 ink4a p17 ink4c and p18 ink4d which inhibit cdks 4 and 6 by preventi ng cyclin association and subsequent activation (9). Cdk involvement in a normally regulated ce ll cycle begins with external stimuli converging to upregulate cyclin D1 (10, 11). Cy clin D1 then binds to and activates cdk4 and cdk6, which translocate to the nucleus and phosphor ylate members of the pocket protein family. This family consists of Rb, p107 and p130 (12-14). These proteins bind to members of the E2F family of transcription factors and inhibit their activ ity by masking their tr ansactivation domains (15, 16). Of the six members of the E2F family, E2F1, E2F2, and E2F3 associate with Rb and are activators of transcription (14). These three transcriptio n factors are essential for cell proliferation, as shown by conditional knockout studies in mice (17). Rb also associates with a histone deacetylase, HDAC1. A complex between Rb, HDAC1 and E2F1 actively represses cisacting elements and basal promoters, further in hibiting E2F-dependant tr anscription (18, 19). Once phosphorylated, Rb dissociates from E2F1, thereby leading to its activation, and the transcription of genes involved in triggering the S phase of the cel l cycle (14). Transcriptional targets of the E2F family include genes i nvolved in DNA synthesis (Thymidine Kinase, Thymidylate Synthetase, Dihydr ofolate Reductase (20, 21)), DNA replication (PCNA and DNA Polymerase-, (20, 22)), and cell cycle cont rol (cyclin A and cyclin E (2 3, 24)). Upregulation of cyclins A and E leads to the activation of cdk2, which furthe r phosphorylates Rb, leading to greater E2F dependant transcrip tion (25) and further progression through the cell cycle. At the 13


end of S-phase, cdk1 is activated by cyclins A and B leading to progression through the G2 phase of the cell cycl e and mitosis (26, 27). Genetic Studies The generalized view of the cell cycle descri bed above assumes that different cdks can only bind to and become activated by their designated cyclins (cyclin D1 with cdk4 or cdk6, cyclin E with cdk2, etc.). Recently, however, ge netic knockout studies in mice have shown that there is a significant degree of promiscuity between the cell cycle specific cdks with regards to their cyclin partners and their cellular substrates (28). Cdk2 was assumed to be an essential protein, because of its requirement for cell cycle progression. It was considered to be the only cdk able to bind to cyclins A and E, and it was k nown to be involved in critical processes in G1S phase transitions. One study revealed that a do minant negative mutant of cdk2 induced a block in the cell cycle in the G1 phase (29). Howeve r in 2003, it was discovered that mice deficient in cdk2 were not only viable, but were relatively normal with respect to development (30). The fact that these mice were sterile indi cated that cdk2 was not essential for mitosis, but only for meiosis in germ cells. Another study revealed that cdk1 was able to compensate for the loss of cdk2 in knockout mice by binding to cyclin E and phosphorylating appropriate substrates (31). This observation was crucial to the understanding of how c dks can substitute for one another. Cdk4 is also nonessential for normal cell pr oliferation (32). In cdk4 nu ll mice, cdk6 activity increases two-fold in a compensatory mechanism. These mice are more resistant to developing skin and mammary tumors (32). Cdk6 null mice also deve lop relatively normally, except for a slight impairment in hematopoiesis (33). Combination knockout studies reve aled that although mice could not develop past the late stages of embryogenesis, mouse em bryonic fibroblast cells prolifer ate normally in the absence of both cdk4 and cdk6 (33). Similarly, mice lacking both cdk2 and cdk4 die shortly after 14


embryonic development, but cell lines derived from those embryos become immortalized and exhibit regular cell cycle kineti cs (34). Finally, triple knockou t mice, deficient in cdk2, cdk4, and cdk6 develop to embryonic da y 12.5 (35). Mouse embryonic fibr oblast cells derived from these embryos undergo proliferation in vitro although with altered cell cy cle kinetics. In these cells, cdk1 was able to bind to all cyclins and phosphorylate Rb, leading to successful completion of the cell cycle. In light of these studie s, it is clear that there is a significant degree of functional redundancy between cdks. This issue becomes important when developing a strategy for halting the growth of cancer cells with the use of cdk i nhibitors. If an inhibitor has a high degree of specificity and targets only one cdk, the remaining, unaffected cdks will likely compensate for the loss in activity, and the cells w ill continue to divide. According to the results from the genetic knockout studies, an effective inhibitor must be able to act on multiple cdks in order to have a significant effect on the rate of cancer cell proliferation. Cdk Aberrance in Cancer Uncontrolled cell division is a defining featur e of cancer. As cdk activity is intimately linked with cell cycle control, it would stand to reason that an overabundance of cdk activity would lead to unrestrained cell growth. Ther e are many examples in which defects in the components of cdk regulatory machinery are linked to specific cancers. In normally dividing cells, cdk levels do not fl uctuate dramatically. Cyclins however, are subject to many fold changes in levels depending on the phase on the cell cycle. As cyclins are the regulatory partners of cdks, their presence or absence determin es the activation state of cdks. Therefore, an upregulation of cyclins can lead to overactivated cdks. Cyclin E upregulation for instance has been linked to neuroe ndocrine lung tumors (36), as well as basal-like breast tumors (37). High levels of cyclin E also correlate w ith a poor outcome in breast cancer patients (38). Moreover, a truncated, lower molecular weight fo rm of cyclin E induces a higher level of cdk2 15


activity than the full length protein. This truncated version of cyclin E has been linked to breast tumorigenesis (39, 40). Similarly, cyclin D1 upregulation, caused by overexpression (by gene amplification, for example) has been observed in 40-50% of human br east cancers (41, 42). In creases in cyclin D1 levels can affect cell cycle regulation dynamics through multiple mechanisms. Firstly, as cyclin D1 is a regulator of cdk4 and cdk6, higher cyclin D1 levels could resu lt in increased cdk4 and cdk6 dependent kinase activity. Secondly, in creased cyclin D1/cdk4 and cyclin D1/cdk6 complexes act as molecular sinks for the cdk in hibitory proteins p21 and p27. By decreasing the availability of these proteins cyclin D1 upregulation would indirectly increase cdk1 and cdk2 activity by preventing their i nhibition by p21 and p27. Thirdly, overexpression of cyclin D1 results in higher levels of cdk4 due to increased transcription (43). Other links between cyclin D1 and cancer include a chromosomal translocati on event in the cyclin D1 gene occurring in 90% of Mantle Cell lymphoma cases (44, 45). Also, a G870A polymorphism in the cyclin D1 gene has been associated with several cancers, including a squamous cell carcinoma of the head and neck, and cervical cancer (46, 47). Perturbations in the proper functioning of cdk inhibitory proteins can also lead to hyperproliferation of cells via over activated cdks. Mislocalizati on and proteolysis of p27 have been implicated in breast and other cancers (48, 49). Similarly, p21is mislocalized in cancers, as cellular staining in human prim ary breast cancer samples show a predominantly cytoplasmic rather than nuclear localization pattern (50, 51). Cdks exert their function primarily in the nucleus. By localizing to the cytoplasm, p21 and p27 can no longer interact with cdks, and inhibit their activity. Therefore, with respec t to cdk activity, imprope r localization of p21 and p27 has the equivalent effect of reducing their total levels. 16


The loss of function of the cyclin D1/cdk4 and cyclin D1/cdk6 regulator p16 is also linked to cancer. Whether due to point mutations, homozygous deletion, or promoter methylation, reduced p16 activity has is an important factor in carcinoge nesis in many cases (52). An alternative mechanism by which increased cdk activity could potentially lead to cancer is an activating mutation in the cdk itself. Tran sgenic mice engineered to express cdk4 with an R24C mutation, which renders the kinase unable to bind to the inhibitory protein p16, develop tumors with varied etiology (53). Taken together these studies re veal the potential consequences of unregulated cdk activity and hi ghlight the need for effective i nhibitors in cancer therapy and prevention. Cyclin D1/Cdk2 Complexes Although the standard view of the cell cycle involves specific pairing of cyclins and cdks (cyclin D1 with cdk4 and cyclin A with cdk2 fo r example), genetic knockout studies reveal that cdks can replace one another by binding to alterna tive cyclins. It has been known for many years that cyclin D1/cdk2 complexes can form in cells including human breast cancer cells (54-56). However, for some time, the functions of thes e complexes were unclear, and it was unknown if cdk2 in complex with cyclin D1 was catalyti cally active. Our laboratory has previously published a study examining the characteristics of such complexes (57). A fusion protein in which a His 6 -tagged cdk2 was attached to a Flag-tagge d cyclin D1 by a flexible poly-Glycine linker was generated. This fusion protein allowe d cyclin D1/cdk2 complexes to be isolated and studied. In this study, it was shown that thes e complexes are properly phosphorylated on the appropriate activating residue (Threonine 160) by CAK and phosphorylate the appropriate substrates (pRB and Histone H1) in vitro. Cell lines constructed to overexpress the fusion protein display anchorage independent growth, and their prolif eration is resistant to the inhibitory effects of th e cell signaling protein TGF. The potential importance of these 17


complexes is underscored by the f act that cyclin D1 is overexp ressed in approximately 50% of human breast cancers (41, 42). That cyclin D1 can bind to and activate cdk2 indicates that these complexes may have important roles in mammar y carcinogenesis. The following two chapters will discuss the use of the cyclin D1/cdk2 fusi on protein as a basis for tumor studies in transgenic mice. Cdk Inhibitors, Classic and Novel Unregulated cdk activity can lead to tumorigenesi s. This indicates that cdks may be ideal targets for anti-cancer therapeutics. As kinases, cdks must bind to ATP for enzyme catalysis. An obvious method of inhibiting cdk activity would be to prevent ATP binding. This premise explains the rationale behind the mechanism of action of the majority of cdk inhibitors in existence today. There are several different cl asses of these inhibitors, such as adenine derivatives, flavones, and oxindole derivatives (58). These cdk i nhibitors reversibly occupy the ATP binding site on cdks, preventing catalysis. Flavopiridol is one of the furthest developed cdk inhibitors of the ATP-competitive sort. This synthetic flavone is derived from a natural product found in the Indian plant, Dysoxyl um binectariferum (59). As a pan-cdk inhibito r, Flavopiridol affects the activity of cdks 1, 2, 4, 6, and 7, and has an IC 50 for inhibiting cell proliferation in the nanomolar range (60). It has also been used clini cally for a number of years, in both phase I and phase II clinical trials in th e treatment of cancer (61-63). While promising, this drug has displayed side effects, with the dose-limiting toxi city being severe diarrh ea (64). Flavopiridol also affects multiple cellular processes that may not be related to cdk inhibition. Flavopiridol increases apoptosis, induces differentiation, displays antiangeogenic properties, and decreases the levels of cyclin D1 through transcription inhibition (65). Other ATP-competitive cdk inhibitors in clin ical trials include Roscovitine and UCN-01. Roscovitine is relatively specific to cdk1 and cdk2 in vitro and has shown moderate success in 18


phase I trials (66-68). UCN-01 is a non-specific c dk inhibitor and also acts on multiple pathways unrelated to cdks (65). In clin ical trials it has mostly been used in combinations with other drugs, including topoisomerase inhibitors such as irinotecan and t opotecan (69, 70), and DNA damaging agents such as cisplatin (71). These drugs have had reasonable success, and demonstrate the potential of ATPcompetitive agents to be of clinical importance in the treatment of cancer. However, their mechanism of action is coupled with an inherent limit in target specificity; all kinases possess an ATP-binding site. This introduces the possibility that some of these molecules may inhibit kinases unrelated to cdks and perhaps lead to unwan ted side effects in the clinic. If, on the other hand, drugs were developed to bind to sites uniq ue to cdks, and were distinct from the ATPbinding cleft, they would have an ad vantage in terms of target specifi city. In recent years this is the approach many researchers have take n in developing new cdk inhibitors. A number of crystal structures of cdk2 ha ve been solved, including in complex with cyclin A (72), cyclin E1(73), cyclin A and p27 (74), bound to ATP-competitive inhibitors (75, 76), as well as in apo form (77). Crystal struct ures of cdk6 have also been solved, including one with a bound viral cyclin (78), and one in a tern ary complex with cyclin K and the inhibitory protein p18 (79). Unfortunately, no structures of cdk1 or cdk4 have been solved as of yet. Nevertheless, the wealth of stru ctural information obtained from the structures of cdk2 and 6 have helped to provide the basis for rational drug design. One example of this effort is the identific ation of a D-amino acid hexapeptide molecule, NBI1 that inhibits the kinase activity of cdk2 by binding to cyclin A (80). This molecule does not interfere with ATP or substrate binding. NBI1 effectively inhibits pr oliferation and induces 19


apoptosis in several cancer cell lines, and was found to inhibit cdks 1 and 2 to a significant extent. A similar approach to cdk-directed drug design was taken in identifying a series of cyclic peptides that bind to the substrate recognition site of cdk complexes (81). In this study, peptides were designed based on the sequence of p27 in an attempt to mimic the inhibitory properties of the protein. These molecules inhibit cdk2 kinase activity, block cyclin A from binding to cdk2 in vitro and are relatively potent, with IC 50 values in the low micromolar range. No other kinases were tested in this study, so it is unknown whet her these molecules have any effect on other cdks. One other example of novel approaches to c dk inhibition is the use of a small, 39 amino acid peptide based on a pRb2/p130 pocket protei n amino acid sequence, which is a spacer domain that inhibits the kina se activity of cdk2 (82). The molecules designed based on this spacer domain not only inhibit cdk2 activity, but i nduce cell cycle arrest, and exhibit effects on tumor sizes in mice in vivo Finally, the natural product Sili binin inhibits cdks through an unusual mechanism. Isolated from the milk thistle plant Silybum marianum, th is compound effectively inhibits prostate cancer in mouse model systems (83, 84). Silibinin is surprisingly non-toxic, with no dose limiting toxicity observed in mice. It appears to affect multiple cellular processes, including decreasing total levels of cdks 1, 2, 4, and 6 in cells, as well as several cyclins. The mechanism by which Silibinin reduces levels of these cell cycle regulators is as yet undetermined, however it presents a novel means by which to control aberrant cdk activity in cancer. The fact th at it is tolerated as well as it is also allows for the possibility of its use as a cancer chemopreventive agent. 20


These novel cdk inhibitors are promising, but ha ve several limitations. Firstly, in general, they do not act as pan-cdk inhibitors. In some tumor types, this may not be problematic, as in some cell lines, inhibition of a single cdk is su fficient to result in decreased cell proliferation and/or apoptosis. However, the redundancy of cdks discussed previously could reduce the clinical potential of these drugs, as in some cases, unaffected cdks would simply substitute for the inhibition of other cdks. Secondly, the nove l cdk inhibitors develo ped are largely peptidebased. These molecules may be subject to drug delivery problems. Based on the examples of currently used inhibitors, and the problems they present, an ideal cdk inhibitor would be one that would 1) bind to a site that is unique and speci fic to cdks, 2) affect multiple cell-cycle specific cdks, and 3) have characteristic s which would allow for proper drug delivery, including being cell permeable and resistant to degradation in the human digestive/circulatory system. 21


CHAPTER 2 TUMORS INITIATED BY CONSTITU IVE CDK2 ACTIVATION EXHIBIT TRANSFORMING GROWTH FACTOR-BETA RESISTANCE Introduction Cdk2 becomes activated during mammary tumo rigenesis through a number of mechanisms including cyclin E overexpression and proteolytic processing (39), p21 and p27 downregulation and mislocalization (49-51, 85-87), and cyclin A1 re-expression (88). Re cent studies indicate that cdk2 is likely to be an im portant target for anti-cancer agen ts (60, 89, 90). Therefore it is important to understand how cdk2 activation th rough various mechanisms leads to tumor formation, and the biochemical and cellular mech anisms involved. Complexes between cyclin D1 and cdk2 were shown to be present in mammary carcinoma cells some time ago (56), but the function of these complexes is unclear. Cyclin D1 overexpression occurs in approximately 50% of human breast cancers (42), thus cyclin D1/c dk2 complexes might contribute to the oncogenic effects of cyclin D1 overexpres sion. Different cyclins and cyc lin-dependent kinases interact with each other rather promiscuously making it difficu lt to ascribe specific functions to particular cyclin/cdk complexes. To circumvent this pr oblem we designed a cyclin D1-cdk2 fusion protein in which the cyclin D1 domain stimulates th e phosphorylation and kina se activity of the cdk2 domain through an intramolecular mechanism (57). We constructed a transgenic mouse model in which mammary expression of the cyclin D1-cdk2 fusion protein is driven by the mouse mammary tumor virus (MMTV) promoter (MMTV-D1K2) (Fig. 2-1A). MMTV-D1K2 transgenic mice exhibit mammary fibrosis a nd hyperplasia and develop mammary tumors associated with significant desmoplasia. Th e MMTV-D1K2 transgenic mouse model may prove useful for testing cdk2 i nhibitors and for the development and testing of novel therapeutic agents targeting tumor cells. 22


Materials and Methods Construction of MMTV-Cyclin D1-Cdk2 (MMTV-D1K2) Transgenic Mice The cDNA encoding the cyclin D1-cdk2 fusion protein was excised from the pAdTrack vector described previously (57) with EcoR I and EcoRV and subcloned into the EcoRI and BstXI sites of MMTV-TGF (91) This resulted in the replacement of the TGF transgene with the cDNA encoding the cyclin D1-cdk2 fusi on, creating the MMTV-cy clin D1-cdk2 (MMTVD1K2) vector. The MMTV-D1K2 vector was ve rified by DNA sequencing. The transgene and MMTV promoter were excised from the MMTVD1K2 vector using AatII and NruI. The purified AatII/NruI fragment was submitted to the Vanderbilt University Transgenic Mouse/Embryonic Stem Cell Shared Resource and tr ansgenic mice were created in the inbred FVB strain. Transgenic animals were initially identified by Southern blotting of genomic tail DNA and routinely screened by polymerase chain reaction (PCR) using primers complementary to the region encoding the FLAG epitope ta g (5'GACTATAAGGACGA TGATGAC-3') and the flexible linker joining the cyclin D1 and cdk2 domains (5'-CCTCCAGAACCTCCACCACC-3'). Multiple lines of transgenic mice were obtained. Lines designated #34 and #44 were selected for further study. Whole-Mount Staining and Histological Analyses Whole mount preparation and staining with he matoxylin was performed as described (92). Tissue samples were fixed with 4% paraformal dehyde in phosphate-buffered saline (PBS) for 14 hours at 4 C and switched to 70% ethanol for 24 hours, followed by an additional 24 hour incubation in 70% ethanol at 4 C. The University of Florida Molecular Pathology Core embedded the tissue samples in paraffin, prepared 5 m sections, and stained the sections with hematoxylin and eosin (H&E) or Trichrome. 23


Isolation and Culture of Cancer and Tu mor-Associated Fibroblast Cell Lines Cancer cells were cultured from the tumors as described previously (9 2). Fibroblasts were removed from colonies of tumor cells by differe ntial trypsinization and retained and cultured separately. Established cell lines were propa gated in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% fetal bovine se rum (FBS). Recombinant adenoviruses and retroviruses and the procedures used for infection and selection were described previously (57, 93). Roscovitine, a TGF Receptor I kinase inhibitor (Cat. # 616452), and rapamycin were obtained from EMD Biosciences Inc. (La Jo lla, CA). Recombinant human TGF and recombinant human HGF were obtained from Chemicon International (Temecula, CA). Immunoblot Analysis of Tumor Samples a nd Tumor-Derived Dell lines and Rb Kinase Assays Preparation of tumor and cell ly sates and subsequent immunobl ot analysis were performed as described previously (57, 93). Anti bodies to the FLAG epitope (F-3165) and -Smooth Muscle Actin (A-2547) were obtai ned from Sigma-Aldrich, Inc. (St. Louis, MO). Her2/c-neu antibody (MS-730) was obtained from Neomarkers (Fremont, CA). E-Cadherin antibody was obtained from BD Biosciences (San Jose, CA). P-Cadheri n antibody (sc-7893) was obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Antibodies to c-Met (#3127), c-Met phosphorylated on tyrosine residues 1234/ 1235 (#3126), Akt (#9272), Akt phosphorylated on threonine 308 (#9275), STAT3 (#9132), and ST AT3 phosphorylated on tyrosine 705 (#9145P) were purchased from Cell Signaling Technologies (Beverly, MA). The anti-HIRA antibody (94) was generously provided by Dr. Peter Adams (Fox Ch ase Cancer Center Philadelphia, PA). The sources of the other antibodies were listed previously (57, 93, 95, 96) Rb kinase assays of antiFlag and anti-cdk4 immunoprecipita tes were performed as descri bed previously (57, 93). 24


Transcriptional Reporter Assays and Cell Proliferation Analyses The Mv1Lu cell line with the plasminogen activat or inhibitor-1 (PAI-1) promoter driving a luciferase reporter gene (PAI-Luc cells) was prov ided by Dr. D. Rifkin (New York University, New York, NY) and has been described previous ly (97). Conditioned cell culture medium was prepared by incubating equal numbers of cells w ith the same volume of medium over a period of five days. The medium was collected and debris was removed by centrifugation, followed by passage through a 0.2 m filter. PAI-Luc cells were incubated with conditioned medium samples for twenty-four hours, luciferase assays were performed, and the results normalized to protein concentration as described (57). 3 H-thymidine incorporation assays were perfor med as described previously (93). Flow cytometry analysis of propidium iodide stained nuclei and data analysis were performed by the University of Florida Flow Cytometry Core Laboratory. Results Characterization of MMTV-D1K2 Mammary Glands Two transgenic mouse MMTV-D1K2 lines term ed #34 and #44 engineered to express the cyclin D1-cdk2 fusion protein in the mammary gl and (Fig. 2-1A) were characterized. Mammary glands from 56 week old virgin females were examined as hematoxylin stained whole mounts and by hematoxylin and eosin (H&E ) staining of tissue sections (F ig. 2-1B). Mammary glands from wild type FvB females exhibit a fully deve loped ductal tree, but l ittle side branching. Mammary glands from transgenic lines #34 a nd #44 exhibit a significant degree of side branching. H&E stained ma mmary sections indicate th at epithelial structures are sparse in wild type glands, but dense assemblies of epithelial structures are presen t in the transgenic animals. Mammary glands from aged transgenic fema les exhibit a progressively more abnormal phenotype. Glands from 102 week old line #44 MMTV-D1K2 animals exhibit hyperplastic 25


lesions. Examination of H&E stained sections of these mammary glands indicate that these lesions consist of ductal struct ures surrounded by fibroblasts (F ig. 2-1C). The strong blue staining by trichrome overlaps with the fibroblasts and is consistent with collagen deposition associated with the fibrosis. The hyperplastic le sions observed frequently progressed into tumors and by two years of age about 70% of the anim als in the MMTV-D1K2 #44 line had developed mammary tumors (Fig. 2-1D). Salivary gland tumors occurred less frequently and were not included in the data used to plot Fig. 2-1D. Several mammary a nd salivary tumors were also observed in the MMTV-D1K2 #34 line indicating that tumor formation in the MMTV-D1K2 animals is not an artifact of the transgene insertion site. No tumo rs of any kind were observed in wild type virgin female littermates. Mammary tumors were not observed in male transgenic mice. Characterization of MMTV-D1K2 Tumors MMTV-D1K2 mammary tumors exhibit a promin ent stromal reaction (Fig. 2-2A) and in many cases a large fraction of the tumor bulk wa s made up of cells of mesenchymal morphology (arrows) distributed throughout the tumors. In one mouse, L447f, a mammary gland lesion was observed that resembled human mammary sclerosing adenosis. These lesions are considered benign yet premalignant in humans and consist of at ypical glandular structures that proliferate to various extents in a fibrotic stroma with pro liferating myoepithelial cells. Tumors from MMTVneu transgenic animals were examined in parallel because the huma n homolog of the rat neu gene, Her2 is overexpressed in approximately 30% of breast cancers ( 98, 99), cyclin D1 is thought to play an important role in neu -induced tumorigenesis (100, 101), and because we had previously isolated and characterize d cancer cell lines derived from MMTVneu tumors. The phenotype of MMTV-D1K2 mammary tumors differed from the MMTVneu tumors, which exhibited a relatively homogenous center surro unded by a thin layer of stroma (arrow). 26


Examination of H&E stained tissue sections from several representative tumors indicated that the majority of the MMTV-D1K2 tumors were mamm ary gland adenocarcinomas ranging from low to high grade (Fig. 2-2B). Adenosquamous di fferentiation was observed in one tumor. Two tumors were observed in salivary glands and one of these (L34-57f) appeared to be of salivary origin. Immunoblot analysis of tumor lysates (Fi g. 2-2C) demonstrated that the FLAG-tagged cyclin D1-cdk2 fusion protein is expressed in the MMTV-D1K2 tumors. MMTV-D1K2 tumors contained similar levels of E-Cadherin and Actin as MMTVneu tumors, but on average expressed higher levels of -Smooth muscle actin. Higher levels of -Smooth muscle actin could result from either a greate r proportion of myofibroblasts or myoepithelial cells in the MMTV-D1K2 tumors than in the MMTVneu tumors. Immunoblot analyses examining a MMTV-D1K2 tumor extract with a MMTVneu tumor extract serving as the control (Fig. 2-2D) demonstrated that Rb is hyperphosphorylated in the MMTV-D1K2 tumor, and that the protein products of the E2F-dependent genes BRCA1, p107, and E2F1 are upregulated relative to the level observed in the MMTV-neu tumor lysate. Immunoprecipitation of these tumor lysates with anti-FLAG-agarose resin demonstrated that the Flag-tagged cyclin D1-cdk2 fusion protein present in the MMTV-D1K2 tumor lysate was present in complexes with p21, p27, and PCNA (Fig. 2-2E). It is likely th at the fusion protein drives tumor formation by directly phosphorylating substrates, a nd by sequestering p21 and p27. Derivation and Characteriza tion of Cancer Cell Lines from MMTV-D1K2 Tumors The large proportion of fibroblasts in the MMT V-D1K2 tumors made it difficult to study the biochemical properties of the cancer cells. We isolated a series of cancer cell lines and tumor-derived fibroblast cell lines to allow a detail ed analysis of the prope rties of each cell type 27


in isolation and to study how th ese two cell types might functionally interact in tumors. The neu T cell line was derived from an MMTV-neu tumor and serves as a reference for comparison with results obtained with cell lines derived from MMTV-D1K2 tu mors. Five cancer cell lines were derived from five diffe rent MMTV-D1K2 tumors and termed D1K2-T1, D1K2-T2, D1K2T3, D1K2-T4, and D1K2-T5. D1K2-T1 was isolated from a mammary tumor arising in the L4425f MMTV-D1K2 transgenic mouse. D1K2-T3 was isolated from a salivary tumor arising in the L34-57f MMTV-D1K2 transgenic mouse. The D1K2-T1 and D1K2-T3 cell lines exhibit the cuboidal morphology typical of luminal epithelial cells and are similar in appearance to the neu T cells (Fig. 2-3A). The D1K2-T2, -T4, and T5 cell lines exhibit myoepithelial morphology and express markers of both the luminal and myoepithe lial lineages. These cell lines are the subject of ongoing investigation and will not be described further here. We performed propidium iodide staining followed by flow cytometry analysis to determine whether expression of the cyclin D1-cdk2 fusion protein significantly altered the cell cycle profile (Fig. 2-3B). Rapidly growing neu T cells exhibit a typical cell cycle profile. In contrast, the D1K2-T1 cells and to a lesser extent the D1 K2-T3 cells, exhibited multiple peaks suggesting that these cell lines are aneuploid. Similar observations were made with the D1K2-T4 and D1K2-T5 cell lines (data not shown). In all cas es, each of the observed peaks of fluorescence intensity was shifted to exactly twice the fluores cence intensity by a 24 h treatment of the cells with the M-phase arresting agen t nocodazole (data not shown). The results suggest that the peaks represent different populations of cancer cells with different st ates of ploidy. Overall, four of the five cell lines derived from MMTV-D1K2 tumors exhibited evidence of aneuploidy by flow cytometry. This observation is consis tent with the observations that cyclin D1 overexpression induces aneuploidy (102), and that hyperactivation of c dk2 induces aneuploidy 28


(103). Thus the induction of aneuploidy may be an additional mechanism by which the cyclin D1-cdk2 fusion protein, and by extension cyclin D1/cdk2 complexes, drive tumorigenesis. Cell lines were derived from the neu T cells by transduction w ith a control retroviral vector ( neu T/Hyg) or the same vector encoding the cyclin D1-cdk2 fusion protein ( neu T/D1K2). Immunoblot analysis of lysates from these cells demonstrated that expression of the cyclin D1cdk2 fusion protein was higher in the D1K2-T1 cells than in the D1K2-T3 cells (Fig. 2-3C). The neu T/D1K2 cells express the cyclin D1-cdk2 fusion protein at levels similar to those observed in the D1K2-T3 cells. Interestingly, Rb phos phorylation levels are similar in the neu T/Hyg, neu T/D1K2, and D1K2-T3 cell lines but higher in the D1K2-T1 cells. D1K2-T1 cells also exhibit altered p130 electrophoretic mobility consistent with p130 hyperphosphorylation. Together, the results indicate that high-level expression of the cyclin D1-cdk2 fusion protein can occur in tumors and is associated with Rb hyperphosphorylation and cell cycle deregulation. However, lower levels of cyclin D1-cdk2 expr ession that do not cause obvious biochemical perturbations as observed in the D1K2-T3 cells are apparently sufficient to initiate tumor formation. Immunoblot analysis of lysates from the neu T, D1K2-T1, and D1K2-T3 cell lines demonstrated that expression of the HGF recepto r c-Met is elevated in the D1K2-T1 and D1K2T3 cell lines relative to the neu T cell line (Fig. 2-D). Cdk2 immunoblots indicated that expression of the cyclin D1-cdk2 fusion protein (*) in the D1K2-T1 cells is about twice that of endogenous cdk2 (-), while the fusion protein was ba rely detectable in the D1K2-T3 cells using the cdk2 antibody. Analysis of the phosphorylati on status of the cdk2 domain of the fusion protein on the regulatory sites Tyr15 and Thr160 using phosphospecific antibodies indicated that the level of phosphorylation of the fusion pr otein on both sites is higher than that of endogenous cdks. This is consiste nt with our previous observations indicating that the cyclin D1 29


domain of the fusion protein stimulates the phosphorylation of the cdk2 domain through an intramolecular mechanism (57). The observation that inhibitory phosphoryl ation of Tyr15 of the fusion protein is enhanced s uggested that the cyclin D1-c dk2 fusion protein might not be catalytically active. We addressed this issue by performing ki nase assays of the cyclin D1-cdk2 fusion protein immunoprecipitated using the FL AG antibody (Fig.2-3E). We also examined whether expression of the cyclin D1-cdk2 fusion protein altered the activity of endogenous cdk4. Although the fusion protein was heavily phos phorylated on both Tyr15 and Thr160, it phosphorylated Rb in vitro This result suggests that a population of the fusion protein molecules exists in which the activating site, Thr160, is phosphorylated, but the inhibitory site, Tyr15, is not phosphorylated. The levels of cdk4 ki nase activity toward Rb were similar in the neu T, D1K2-T1, or D1K2-T3 cell lin es, indicating that expression of the cyclin D1-cdk2 fusion has minimal effects on endogenous cdk4 activity. Th ese results demonstrate that the cyclin D1cdk2 fusion protein expressed in the MMTV-D1K2 cancer cell lines is enzymatically active, is overexpressed only marginally relative to endoge nous cdk2 levels, and that the fusion protein displays enhanced regulatory phosphoryl ation relative to endogenous cdks. Derivation and Characterization of Fibrob last Cell Lines from MMTV-D1K2 Tumors Given the large proportion of fibroblasts in the MMTV-D1K2 tumors and the close proximity between the epithelial cells and fibroblas ts in hyperplastic lesions (Fig. 2-1C) all the way to fully developed tumors (F ig. 2-2A), it is likely that tu mor-associated fibroblasts have a significant influence on the progr ession and growth of MMTV-D1K2 tumors. We isolated fibroblast cell lines from the MMTV-D1K2 tumors to examine the properties of these cells. The tumor-derived fibroblast lines 1 and 2 (D1K2-TDF1 and D1K2 -TDF2) exhibit the elongated morphology typical of fibroblasts and lack the extensiv e cell-cell contacts characteristic of colonies of epithelial cells (Fig. 2-4A). Immunoblot analysis demonstrated that D1K2-TDF1 and 30


D1K2-TDF2 cells did not express pr oteins typically present in can cer cells of epithelial origin such as c-Met, E-Cadherin, and P-Cadherin (Fig. 2-4B). The tumor derived fibroblast cell lines also did not express the cyclin D1-cdk2 fusion protein, but expressed very high levels of Smooth muscle actin ( -SMA). -SMA is expressed in fibroblast s that have differentiated into myofibroblasts. Such differentiation might arise as a result of cell culture in vitro However the presence of myofibroblasts in the MMTV-D1K2 tumors could explain the relatively high levels of -SMA present in the tumors and could also explain the high levels of collagen detected by trichrome staining (Figs. 2-1 and 2-2) since myofibroblasts are thought to be the major cell type responsible for collagen deposition during fibros is (104-106). The differentiation of stromal fibroblasts in the vicinity of tumors is thought to be caused by TGF secreted by tumor cells since TGF is capable of inducing the differe ntiation of mammary stromal fibroblasts to myofibroblasts in vitro (107, 108), fibroblast differentiation to myofibroblasts occurs in the vicinity of tumors in a graded manner (109) and tumors are known to secrete significant amounts of TGF (110). Discussion Cyclin D1/cdk2 complexes are present in hum an breast cancer cell lines (56), and the levels of these complexes correlate well with the degree of cyclin D1 overexpression. It is unknown whether these complexes participate in the transforming effects of cyclin D1 overexpression. We constructed a gene encoding a cyclin D1-cdk2 fu sion protein to explore in a selective fashion the potential functions of cyclin D1/cdk2 comp lexes in cell transformation and mammary tumorigenesis (57). The results presen ted here demonstrate that expression of this cyclin D1-cdk2 fusion protein (D1K2) in the mammary gland under the control of the MMTV promoter causes precocious lobuloalveolar differen tiation of mammary glands in virgin mice. 31


The extent of the phenotype increases with age and results in the formation of hyperplastic lesions and eventually breast tumors. MMT V-D1K2 tumors induce a strong desmoplastic reaction as compared with MMTVneu tumors. Biochemical analyses demonstrated cell cycle deregulation in the tumors, including Rb and p130 hyperphosphorylation and upregulation of E2F-dependent gene products. The MMTV-D1K2 mice and cell lines derived from their tumors will be valuable tools for further delineating the mechanisms involve d. A more thorough understanding of the mechanisms at work in this cycle is critical because of the high frequency with which similar desmoplastic reactions are observe d in a wide variety of human tumors, the connection between tumor-induced stromal desmoplasia and tumor i nvasiveness and metast atic potential, and because several key elements of the cycle including c-Met (111, 112), TGF receptors (113115), and cdk2 (60, 89, 90) are targets for agents under development as anticancer therapeutics. 32


A B Figure 2-1.MMTV-cyclin D1-cdk2 (MMTV-D1K2) transgenic model and mammary phenotype. A) Model depicting the MMTV promoter driving the expression of the transgene encoding the N-terminal FLAG epitope tag, the cyclin D1 domain, a flexible linker, the cdk2 domain, and the C-terminal His 6 affinity tag. B) Re presentative mammary whole mounts from 56 week old wild type female mice and the #34 and #44 lines of the MMTV-D1K2 transgenic mice stained with hematoxylin (top panel). Inset panels show branching morphology at higher ma gnification. The bottom panels are hematoxylin and eosin (H&E) stained hi stological sections at low and high magnification. Mammary lymph nodes are marked LN. C) H&E and trichrome stained mammary gland sections from 102 week old MMTV-D1K2 transgenic mice of the #44 line displaying regions of epithe lial hyperplasia associated with fibrosis. Lymph nodes are marked LN. D) Kaplan-M eier curve showing tumor incidence in wild type and line #44 mice as a function of age in months. The number of virgin female animals in each group is shown in the inset legend. 33


D C Figure 2-1.Continued. 34


A B Figure 2-2.MMTV-D1K2 mammary and salivary tumors. A) H&E and trichrome stained histologi cal sections of MMTV-D1K2 and MMTVneu mammary tumors. B) Morphological char acterization of several representative tumors arising in the #34 (L34-) and #44 (L44-) transgenic lines. C) Immunoblot analysis of extracts from four diffe rent MMTV-D1K2 and four different MMTVneu mammary tumors demonstrating expressi on of the flag-tagged cyclin D1-cdk2 transgene product (F lag (D1K2)) and cneu Tumor lysates were also analyzed for the levels of E-Cadherin, alpha-Smooth Muscle Actin ( -SMA), and Actin as a loading control. D) Immunoblot analysis demonstrating the hype rphosphorylation of Rb on multiple residues in a MMTV-D1K2 tumor extract relative to a MMTV-neu tumor extract. P-Rb and P-p130 represent the phospho-forms of Rb and p130, and Rb and p130 represent the corresponding unphosphorylated forms. Multiple products of E2F-dependent genes are upregulated in the MMTV-D1K2 tumor relative to the MMTVneu tumor including, BRCA1, p107, and E2F1. Hira serves as a loading control. E) Lysates from MMTVneu and MMTV-D1K2 tumors were subjected to immunoprecipitation with Anti-Flag-agarose to isolate complexes containing the cyclin D1-cdk2 fusion protein. Immunoblot an alysis indicated that these complexes contain the cyclin D1-cdk2 fusi on protein (Flag), p21, p27, and PCNA. 35


C D E Figure 2-2.Continued. 36


A B Figure 2-3.Isolation and characteriza tion of MMTV-D1K2 cancer cell lines A) Phase contrast micrographs of cell lines derived from an MMTVneu tumor ( neu T) and two different MMTV-D1K2 tumors (D1K2-T1 and D1K2-T3). B) Flow cytometry analysis of rapidly growing neu T, D1K2-T1, and D1K2-T3 cells stained with propidium iodide. C) Immunoblot analysis of the neu T/Hyg and neu T/D1K2 cell lines prepared by retr oviral transduction of the neu T cells with an empty retroviral vector or a vector encoding the cyclin D1-c dk2 fusion protein, respectively, and the D1K2-T1 and D1K2-T3 cell lines. The results indicate expression of the cyclin D1-cdk2 fusion protein (Flag (D1K 2)) in the appropriate samples and demonstrate Rb and p130 hyperphosphoryla tion in the D1K2-T1 cell line. ECadherin is expressed at similar levels in th e four cell lines. Actin serves as a loading control. D) Immunoblot analysis demonstrat ing the levels of expression of the cyclin D1-cdk2 fusion protein (*) relative to the levels of endogenous cdk2 (-), and phosphorylation of the cyclin D1-cdk2 transgene product on the activating Thr 160 (Pcdk2(T160)) and inhibitory Tyr 15 (P-cdk2(Y15)) phosphorylation sites of the cdk2 domain. C-Met is expressed at higher levels in the D1K2-T3 and D1K2-T1 cell lines than in the neu T cell line. E) Assay of the kina se activity of the cyclin D1-cdk2 fusion protein and endogenous cdk4. Extrac ts of the indicated cell lines were immunoprecipitated using the FLAG antibody to isolate the cyclin D1-cdk2 fusion protein, or a cdk4 antibody to isolate endogenous cdk4. Immunoprecipitates were assayed for kinase activity using GST-Rb as the substrate, and site-specific Rb phosphorylation was detected by immunoblo tting with phospho-specific antibodies. Controls included the use of neu T cell extracts in FLAG immunoprecipitations, and omission of the cdk4 antibody (No 1). 37


C D E Figure 2-3.Continued. 38


A B Figure 2-4.Isolation and char acterization of tumor-deri ved fibroblast cell lines. A) Phase contrast micrographs of fibrobl ast cell lines derived from two different MMTV-D1K2 tumors (D1K2-TDF1 and D1K2-TDF2). B) Immunoblot analysis of extracts from the tumor derived fibroblast cell lines D1K2-TDF1 and D1K2-TDF2, the MMTV-D1K2 tumor cell line D1K2-T1, and the BT549 human mammary carcinoma cell line. The results demonstrate a lack of E-Ca dherin, P-Cadherin, cMet, or the cyclin D1-cdk2 transgene (Flag (D1K2)) in the tumor derived fibroblasts. The D1K2-TDF1 and D1K2-TDF2 cells expres s high levels of alpha-Smooth Muscle Actin ( -SMA). D1K2-T1 and BT549 lysates served as positive controls for the immunoblot analysis and Actin se rves as a loading control. 39


CHAPTER 3 MAMMARY TUMORS INITIA TED BY CONSTITUTIVE CDK ACTIVATION CONTAIN AN INVASIVE BASAL-LIKE COMPONENT Introduction Microarray analyses have recently allowed breast tumors to be categorized as luminal, basal-like, normal-like, or Her2 positive, ba sed on distinct gene expression profiles, morphological characteristics, prognostic outcomes and responsiveness to currently available therapeutic approaches (116, 117). The basal-li ke subtype represents approximately 20% of human breast cancers overall, but 39% of breast tumors in premenopausal African American women (118). These tumors are associated w ith a high rate of recu rrence and poor outcome (117). The basal-like subtype of cancers is also termed "triple negative" because these tumors typically lack Estrogen Receptor (ER), Progest erone Receptor, and Her2 overexpression, but generally express a subset of myoepithelial markers, including Cytokeratin 14 (CK14), Cytokeratin 5 (CK5), -Smooth Muscle Actin ( SMA), Nestin, or p63 (119-121). Basal-like tumors lack responsiveness to Tamoxifen and Ar omatase inhibitors that target ER positive luminal tumors, and Herceptin that targets Her2 positive tumors. The mouse basal-like breast can cer models described to date involve genetic deletion of the BRCA1 and p53 tumor suppressor genes ( 122, 123). Tumors initiated by BRCA1 inactivation in mice express the Progesterone Receptor (124) and overexpress Her2 (125) and thus do not fit the "triple negativ e" clinical definition of basal breast cancer. Therefore it is likely that additional genetic lesions contribute to the formation of sporadic human basal-like breast cancers. Microarray studies have suggested several candida te "drivers" of basal breast cancer including Epidermal Growth Factor Receptor (EGFR), c-K it, c-Met, and cyclin E. However, none of these genes have yet been dem onstrated to specifically induce basal-like breast cancer when overexpressed. Interestingly, human basal-like breast tumors frequently exhibit p16 40


overexpression, low levels of Rb and cyclin D1 expression, and high levels of cyclin E expression (126). Based on these observations it was proposed that Rb inactivation is mechanistically linked to the basa l-like subtype (126). Together these results suggest that basallike tumors may have low levels of cdk4/cdk6 activity, but perhaps high levels of cdk2 activity. In the previous chapter, a novel mouse tran sgenic model of breast cancer was described, in which expression of a cyclin D1-cdk2 (D1K2) fusion protein (57) under the control of the MMTV promoter/enhancer induces mammary tumo rigenesis (MMTV-D1K2 animals) (127). Mammary tumors from these animals exhibit Rb hyperphosphorylation, high levels of cdk2 activity, and upregulation of E2F-dependent tran scription (127). Thus, MMTV-D1K2 tumors exhibit functional inact ivation of Rb tumor suppressor activity. MMTV-D1K2 tumors are heterogeneous and induce a desmoplastic reaction associated with TGF secretion by the cancer cells. As men tioned previously (127), some of the cancer cell lines derived from the MMTV-D1K2 tumors exhib it the morphological featur es of myoepithelial cells. This chapter is focused on a more exte nsive characterization of MMTV-D1K2 cell lines and demonstrates that these cells express protein markers associated with the basal/myoepithelial lineage. E-cadherin is a potent invasion suppresso r expressed in nontransformed mammary epithelial cells (128). The MMTV-D1K2 cell line s exhibit decreased or mislocalized E-cadherin expression in culture. Introduction of cell line s derived from MMTV-D1K2 tumors into the mammary glands of wild type syngeneic mice re sults in the formation of invasive tumors composed of spindle-shaped cells that exhibit E-cadherin mislocalization to the cytoplasm and the expression of basal/myoepithelial mark ers. Morphological a nd immunohistochemical analyses of the primary tumors demonstrate a biphasic morphology characteristic of adeno41


myoepithelial-type carcinoma with populations of spindle-shaped cells. These spindle-shaped cells exhibit E-cadherin downregul ation and localization to the cy toplasm and expression of the myoepithelial marker SMA. These studies indicate the pr esence of a subpopulation of invasive basal-like breast cancer cells in the primary MMTV-D1K2 tumors. In vitro analysis of multiple clonal cell lines derived from MMTV-D1K2 tumors demonstrate the expression of various subsets of myoepithelial and luminal epithelial markers, a finding consistent with the "mixed-lineage" prop erties of human basal breast cancers (129-131). In all of the cell lines isolated, E-cadherin expres sion is either low and/or mislocalized to the cytoplasm. E-cadherin mislocalization is associ ated with the inability of the cells to form colonies with normal cell-cell contacts in culture and correlates with the lack of -catenin and p120ctn staining at cell-cell junc tions. In some cell lines, decreased E-cadherin expression is associated with an increase in Nor P-cadherin expression. This "cadherin switch" from Eto Ncadherin expression is associated with increased invasiveness and part of the epithelial to mesenchymal transition (EMT) that occurs durin g the progression of some tumor types (132). EMT has been shown to occur in the basa l-like category of breas t cancers (133). Materials and Methods Isolation of Tumor Cell Lines MMTV-D1K2 cancer cell lines were isolated essentially as desc ribed (127). When differential trypsinization was perf ormed, the cells remaining adherent to the flasks as well as the detached cells were retained. This is critical be cause several of the basal breast cancer cell lines adhere very loosely to tissue cu lture flasks or flasks coated wi th rat tail collagen. The loosely adherent myoepithelial-like cancer cells were se parated from myofibroblas ts by taking advantage of their differential rates of adhesion. The cancer cells were cloned by limiting dilution. All 42


cells were maintained in Dulbecco's Modified Eagles Medium (DMEM) supplemented with 10% heat inactivated fetal bovine serum (Mediatech, Inc., 35-011-CV, Manassas, VA, USA). Preparation and Analysis of Tumor and Cell Extracts by Immunoblot Cell extracts were prepared as described (127) and immunoblot an alysis was performed using antibodies from the following sources: Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA (N-cadherin (sc-7939), P-ca dherin (sc-7893), Vimentin (s c-32322), Nestin (sc-23927), Catenin (sc-7199), Zyxin (sc-6437), p130 (sc-31 7), cdk2 (sc-163), Actin (1616), and p53 (sc100)); Sigma-Aldrich (St. Louis, MO, USA) (Flag, M2 (F-3165), -Smooth Muscle Actin (A2547)). Antibodies specific for EGFR (#2232) were obtained from Cell Signaling Technology, Inc., Danvers, MA, USA. P53 antibodies we re also obtained from Oncogene Science (Cambridge, MA, USA). Antibodies specifi c for Cytokeratin 14 (MS-115) and Her2/ neu (MS730) were obtained from LabVision/Neomarkers, Inc., Fremont, CA, USA. P120 catenin (610133) antibody was obtained from BD Transduc tion Laboratories, San Diego, CA, USA. Tumor samples contain large amounts of immunoglobulin, which interferes with subsequent immunoblot and imm unoprecipitation assays. Tumo r-associated immunoglobulin was removed by preclearing aliquots of tumor lysate containing 1 mg of protein with 100 l/tube packed Protein G-Sepharose (Invitrogen, 10-1242, Carlsbad, CA, USA). The supernatants were retained for subsequent analyses. Immunofluorescence Microscopy Cells were plated onto glass coverslips in si x-well plates. After a 24 hr incubation, cells were fixed with 1% Paraformaldehyde in phos phate-buffered saline (PBS) for 20 minutes, followed by a 10 minute incubation with quench solution (50 mM Ammoni um Chloride + 0.5% TritonX-100 in PBS). The cells were then bloc ked for one hour with antibody buffer (10% Goat Serum + 0.5% Triton X-100 in PBS). Primary staining was performed using the following 43


antibodies at a 1:100 dilution in antibody buffer for two hours: Cytokeratin 14 (Neomarkers, Inc., MS-115); E-cadherin (610181) and p120 Catenin (610133) from BD Biosciences Pharmingen; E-cadherin (24E10) from Cell Si gnaling Technology, Inc.; and Zyxin (sc-6437), Catenin (sc-7199), N-cadherin (sc-7939) from Sa nta Cruz Biotechnology, Inc. Following four washes with PBS, cells were incubated with secondary antibody for one hour using either goat anti-Rabbit Fluor 488 (Invitrogen Molecular Pr obes (Carlsbad, CA, USA), A11008), goat antiMouse Fluor 488 (Invitrogen Molecular Probes, A11001), rabbit anti-Goat Fluor 488 (Invitrogen Molecular Probes, A11078), or Goat anti-Mous e Cy3 (Zymed (Carlsbad, CA, USA), 81-6515), at either a 1:200 dilution in antibody buffer fo r single staining, or 1:300 dilution for double staining. Following four washes with PBS, coverslips were mounted onto slides with Vectashield + DAPI (Vector Laboratories, H-1200, Burlingame, CA USA). Actin was visualized using Texas-Red-X Phalloidin (I nvitrogen Molecular Probes-T7471), added during the secondary staining step at a dilution of five units per slide. Images were captured using a Zeiss (Thornwood, NY, USA) Axiopl an2 upright microscope and visualized using Openlab 5.3.0 Improvision software. Orthotopic Tumor Growth Studies Cells in log growth phase were collected by trypsin digestion, suspended in 10% FBSDMEM, and washed three times with Hepes balan ced salt solution (HBSS) (Mediatech, Inc., 21020-CV). The cells were counted and diluted to a concentration of 107 cells/ml in HBSS. The cell suspension (100 l) was injected into the #4 mammary glands of adult wild type female FvB mice just beneath the surface of the nipple. Th ree mice were injected with each polyclonal cell line and tumor formation occurred from two to six weeks in all of the injected animals. Tumors were excised at a small size (2-6 mm in diamet er) so that tumor inva sion into the surrounding stroma could be observed. 44


Immunohistochemical Analysis of Tumor Tissue Sections Two micrometer serial sections of para formaldehyde-fixed, paraffin-embedded tumor tissue were dewaxed in Tissue-Clear (Sakura Finetek Europe, Zouterwoude, The Netherlands) and hydrated through a series of diluted ethanol followed by antigen retrieval in 10 mM Tris, pH 9.0, 0.5 mM EGTA solution using microwave oven treatment (15 min.). Immunostaining was performed with a commercially available kit; Animal Research Kit ARK (DakoCytomation, Glostrup, Denmark) in accordance with the manufacturer's instructions. Additional blocking of endogenous biotin was performed with the DAKO Bi otin Blocking System (DakoCytomation) in accordance with the manufacturer's instructi ons. The following antibodies were employed: Monoclonal Mouse Anti-Human Smooth Muscle Actin, Clone 1A4, d ilution 1:200, Monoclonal Mouse Anti-Human E-cadherin, Clone NCH-38, dilution 1:25 (both DakoCytomation); Monoclonal Mouse Anti-Human Keratin 14 Ab-1, clone LL002, dilution 1:400 (LabVision/NeoMarkers Inc.). Images were captured using an Olympus BX51 microscope equipped with a Color View camera using Anal ySIS getIT version 5.0 (Soft Imaging System, Munster, Germany). The tumor analyzed by immunohistochemistry in Figure 3-5A is from a 58 year-old caucasian woman diagnosed with invasive ducta l carcinoma, grade III. Morphologically the tumor is described as a "ring carcinoma" because of its necrotic center. Results Mouse Mammary Tumor Virus-D1K2 Hyp ercellular Lesions Exhibit an Invasive Phenotype In the previous chapter it was shown that tumors arise in transgenic mice in which a cyclin D1-cdk2 fusion protein (57) is driven by the MM TV promoter/enhancer (127). Tumors arising in these animals are heterogeneous and contain ductal structures surroun ded by spindle-shaped 45


cells. The identity of these spindle-shaped cells is unclear. However these cells are of interest because they appear to invade into the surround ing mammary fat pad (Fig. 3-1A). This is in contrast to MMTVneu tumors or tumors derived from MMTVneu tumor cells that show welldemarcated boundaries (Fig. 3-1B). We isolated cell lines from different MMTV-D1K2 tumors to more fully characterize the cell types compos ing them. Several of the cell lines, including D1K2-T2 and D1K2-T4, display features in culture similar to that of primary myoepithelial cells (134-137) including an elongated morphology, multiple cellular extensions or processes, and a relative lack of cell-cell adhesi on (Fig. 3-1C). The BT549 human basal-like breast cancer cell line (138) exhibits a similar morphology. In contrast, nontransformed mouse mammary epithelial NMuMG cells have a cuboidal shape and form distinct colonies that exhibit a "cobblestone" morphology. Mouse Mammary Tumor Virus-D1K2 Tumor Cells Display Characteristics Consistent with Basal-Like Breast Cancer Because the MMTV-D1K2 tumor-derived cell lines exhibit a myoepithelial morphology, we examined whether they expressed myoepithelia l markers. Immunoblot analysis indicates that the cell lines derived from MMTV-D1K2 tu mors express varying amounts of several basal/myoepithelial markers including P-cadheri n, Epidermal Growth Factor Receptor (EGFR), CK14, SMA, and Nestin (Fig. 3-2A). In contrast, the neu T cell line derived from an MMTVneu tumor (127) expresses low or un detectable levels of these ba sal markers. Several of the D1K2 cell lines also express the luminal epithe lial marker E-cadherin. Because the polyclonal D1K2 cell lines exhibit a wide variety of cell morphologies, clonal cell lines were isolated. Specifically, the D1K2-T2,CL1 and D1K2-T2,CL6 cl onal lines exhibit diffe rent morphologies in culture and different expression pr ofiles in immunoblot analyses. Th is observation indicates that the cancer cell lines isolated are heterogeneous and is consistent with the heterogeneous nature of 46


the primary tumors (127). The D1K2-T4 cell line does not exhibit detectable transgene expression (Fig. 3-2A), although th e DNA from this cell line tests positive for the presence of the transgene by PCR analysis (not shown). Thus, expression of the cyclin D1-cdk2 fusion protein may not be required for maintenance of the ba sal-like/myoepithelial phenotype. The D1K2 tumor cell lines express high levels of the interm ediate filament protein Nestin (see Fig. 3-2A) which was shown to be a basal breast cancer/ myoepithelial marker (119). Quantitative proteomic analyses of protein extracts from the invasive D1K2 tumor cell lines and the BT549 cell line indicate that these cells express higher levels of the stru ctural protein Zyxin than the neu T and mouse mammary gland NMuMG cell lines (proteomic analyses will be described in detail elsewhere). This was veri fied by immunoblot (Fig. 3-2A). The cell lines isolated from the MMTV-D1K2 tumors might represent a small fraction of the cells present in the primary tumors and th erefore may not be typical of the overall tumor composition. Immunoblot analyses of the primary tumors (Fig. 3-2B) demonstrated that the tumor from which the D1K2-T1 cell line was deri ved expressed relatively high levels of the D1K2 fusion protein as measured by staining with Flag antibody (Flag (D1K2)). The tumor of origin of the D1K2-T1 cells also displayed low Her2 expression, high levels of E-cadherin expression, and low levels of SMA staining. The tumor from which the D1K2-T4 cell line was derived did not express the cyclin D1-cdk2 fusion protein, expressed low levels of E-cadherin, low levels of Her2, and high levels of SMA. MMTVneu tumors exhibit high Her2 expression that correlates with high E-ca dherin expression. In contrast, SMA expression is inversely related to Her2/neu transgene expression and E-cadherin ex pression. Collectively these results suggest that the cell lines isolat ed for analysis have protein ex pression patterns similar to the primary tumors. 47


The lack of expression of the cyclin D1-c dk2 transgene product in the D1K2-T4 cell line and in the primary tumor from which it was derived was unexpected because these cells resemble the other MMTV-D1K2 cell lines in their morphology and expression pattern of luminal and basal markers. Immunoblot analyses were performed to examine the possibility that other molecular changes occurred that might substitute for D1K2 expression. The results showed that the D1K2-T4 cell line exhibited seve ral features that could render D1K2 expression dispensable including cyclin A overexpressi on, (presumably mutant) p53 overexpression, and low levels of Rb and p21 expression (Fig. 3-2C). Mouse Mammary Tumor Virus-D1K2 Tumor Li nes Display Mixed Luminal/Myoepithelial Character In the normal mammary gland E-cadherin is e xpressed in luminal epithelial cells while SMA and CK14 are expressed in m yoepithelial cells. The observ ation that luminal proteins such as E-cadherin and myoepithelial markers such as SMA and CK14 are expressed in the same polyclonal cell population could be explai ned by the presence of cell subpopulations that each express different subsets of markers. Cl onal cell lines were derived from the MMTV-D1K2 tumor cell lines to determine whether these lumi nal and myoepithelial ma rkers are expressed in the same cells. Immunoblot analysis indicates th at in multiple cases the clonal cell lines express luminal markers such as E-cadherin and Cytokeratin 19 (CK19) and also express basal/myoepithelial markers su ch as P-cadherin, EGFR, CK14, SMA, and Nestin (Fig. 3-3A). Interestingly, two of the four D1K2-T 4 subclones exhibit p53 overexpression. Immunofluorescence microscopy experiments de monstrated that the D1K2-T1,CL1 cell line expresses both CK14 and E-cadherin uniformly in all of the cells (Fig. 3-3B). Together the results in Fig. 3-3A and 3-3B indicate that th e cell lines derived from the MMTV-D1K2 tumors exhibit a mixed luminal/myoepithelial protein expression pattern at the single cell level. 48


A number of human mammary car cinoma cell lines have been analyzed in microarray experiments and classified into luminal or basal-like subgroups ( 138). Interestingl y, analysis of some of these mammary carcinoma cell lines by immunoblot suggests that human breast carcinomas may also frequently exhibit a luminal/myoepithelial mixed lineage phenotype (Fig. 3-3C). The T47D, MCF7, and MDA-MB-361 cell lines, classified as luminal by microarray analysis, express the myoepith elial marker P-cadherin. Like wise, the MDA-MB-468 and BT549 cell lines, classified as basal by microarray analys is, express the luminal marker E-cadherin (Fig. 3-3C). In contrast, the nontransformed mouse mammary gland NMuMG cell line does not express P-cadherin, and the nontransformed hu man basal-like cell lines MCF10A and HBL100 do not express E-cadherin. P53 overexpressi on in the T47D, MDA-MB-468, BT549, MDA-MB231, and MDA-MB-435s carcinoma cell lines is exp ected because these cells harbor mutant p53 alleles (76, 139-141). Mouse Mammary Tumor Virus-D1K2 Tumor-D erived Cell Lines Form Invasive Tumors In Vivo Basal-like tumors are often i nvasive, therefore we examined whether the MMTV-D1K2 tumorderived cell lines would form invasive tumors in vivo The polyclonal D1K2-T1, D1K2-T2, D1K2-T4, and D1K2-T5 cell lines were injected in to the mammary glands of three wild type female FVB mice. All of the mice formed tumors from two to six weeks after injection. The tumors exhibited invasion into the surrounding mammary fat pad and muscle (Fig. 3-4A). Invasion of individual cancer ce lls between adipocytes and muscle fibers was observed at the tumor/stroma interface. The in filtrative tumors exhibited inclusions of adipocytes. The disorganized and poorly differentiate d appearance of the tumors is si milar to that of the cells in hypercellular lesions in the MMTV-D1K2 transgenic mammary glands that exhibit invasion into the fat pad (see Fig. 3-1A). The blue Trichrome st aining at the invading edge of the tumors (Fig. 49


3-4A, D1K2-T1 and D1K2-T2) coincides with cells of fibroblastic morphology and is indicative of desmoplasia and fibroblast infiltration into the tumors. The MMTV-D1K2 cell line-derived tumors were negative for Estrogen Receptor a nd Progesterone Receptor expression and Her2 overexpression by immunohistochemi stry (data not shown), cons istent with the expression pattern of the cell lines in vitro (Figs. 3-2A, 3-2C). The finding that some of the D1K2 cell lines express high levels of E-cadheri n in cell culture (see Figs. 3-2B and 3-3A) seems inconsistent with the poor ability of the cells to form distinct colonies in vitro (Fig. 3-2A), and the invasiveness of the cell lines observed in vivo (Fig. 3-4A). Immunohistochemistry experiments examining the localization of E-cadherin within the tumors demonstrated that the noninvasive tumors derived fro m neuT cells exhibit strong Ecadherin staining at cellcell junctions (Fig. 3-4B), and lack expression of SMA and CK14 (not shown). In contrast, tumors derived from the D1K2-T1, D1K2-T4, and D1K2-T5 cell lines exhibit weak, diffuse cytoplasmic E-cadherin stai ning, suggesting that E-ca dherin localization to the cytoplasm contributes to the invasiveness of these cell lines. The D1K2-T1, D1K2-T2, and D1K2-T4 cell line-derived tumors also express variable levels of CK14 (Fig. 3-4C). The positive staining of the tumors for both E-cadheri n and CK14 suggests that the tumors, like the initiating cell lines, exhibit a mixed luminal/myoepithelial phenotype. We next examined whether populations of cancer cells with a mixed lineage phenotype were present in the primary MMTV-D1K2 tumors. The primary tumor from which the D1K2T1 cell line was derived has a biphasic morphology characteristic of adeno-myoepithelial-type carcinoma, with both glandular stru ctures and clusters of less di fferentiated spindle-shaped cells (Fig. 3-4D). Immunohistochemical staining of serial tumor secti ons showed that these spindleshaped cells stain positively for E-cadherin diffusely localized to the cytoplasm, and also show 50


positive staining for SMA. In contrast, the surrounding gla ndular structures are composed of cells that express E-cadherin at their cell-cell junctions and lack SMA expression. This observation suggests that the pr imary MMTV-D1K2 tumors contai n both lumina l/myoepithelial mixed-lineage cell populations and cells that exhi bit a luminal-like staining pattern. Thus, it seems likely that the basal-like cancer cell lines isolated from the MMTV-D1K2 tumors represent the spindleshaped subpopulation of cells pr esent in the primary tumors. Mouse Mammary Tumor Virus-D1K2 Tumors Resemble Human Basal-Like Breast Cancers MMTV-D1K2 cancer cells resemble human basa l-like breast cancer cell lines in terms of their morphology, protein expressi on patterns, and invasiveness in vivo Therefore we examined how the morphology of the MMTV-D1K2 tumors compares with that of human basal-like (triple-negative) breast tumors. The representative triple-negative breast tumor shown is from a 58 year-old caucasian female with grade III invasive ductal carcinoma. This tumor was verified to lack Estrogen Receptor, Progesterone Recepto r, and Her2 overexpression in two separate immunohistochemical analyses (data not shown). The tumor contains a necrotic core and an expanding boundary that invades into the surroun ding mammary fat pad and contains inclusions of adipocytes (Fig. 3-5A). Immunohistochemical staining shows that the tumor cells express SMA, E-cadherin, and CK14. Red Van Gieson's staining in dicates that the tumor contains extensive collagen deposits (fibrosis) interspersed between clusters of cancer cells. This fibrosis colocalizes with tumorassociated fibroblasts. In many areas of the tumor, the tumor-a ssociated fibroblasts make up a larger fraction of the tumor volume than the cancer cells. We also examined whether the MDAMB-231 and MDA-MB-436 cell lines, which exhibit ba sal-like expression profiles (138), form tumors that display features similar to human basal-like breast cancers and similar to tumors 51


formed from MMTV-D1K2 cancer cell lines. Ten million cells of each line were injected orthotopically into three adult female athymic nude mice. The animals developed tumors from three to six weeks after inject ion. Tumors were excised when they reached two to six millimeters in diameter so that invasion into th e surrounding stroma could be visualized. Fig. 35B shows that MDA-MB-231 and MDA-MB-436 tu mors grown as xenografts in athymic nude mice form invasive tumors that contain inclusio ns of adipocytes, exhibit fibroblast accumulation and fibrosis at the tumor/stroma boundary, and develop a fibrotic necrotic core when tumors grow larger than approximately three millimeters in diameter. The invasive growth pattern of the MDA-MB-231 cells in vivo is similar to that observed with the MMTV-D1K2 tumor-derived cell lines. Mouse Mammary Tumor Virus-D1K2 Tumor-Deri ved Cell Lines Exhibit Extensive Stress Fiber Formation and Cytoplasmic E-Cadheri n, p120ctn, and -Catenin Localization Primary myoepithelial cells exhi bit constitutive stress fibers in culture (135, 136). Zyxin is a component of focal adhesions and associates with Actin stress fibers (142). Immunoblot experiments (Fig. 3-2B) indicate that the D1K2 tumor cell lines express higher levels of Zyxin and Nestin than the neu T cell line. Therefore we examined Zyxin localization and colocalization with Actin in immunofluorescence microscopy studies (Fig. 3-6A). The neu T cells grow as colonies and exhibit cortical actin staining but no detectable stress fibe rs. Zyxin is present primarily at the outer border of the neu T colonies. In contrast, the D1K2 cancer cell lines exhibit constitutive stress fiber formation. Zyxin colocalizes with the ends of actin stress fibers and is present around the periphery of each of the cells that is not part of a colony. It is unclear from the studies in Figs. 31 to 3-4 whether the cytoplasmic E-cadherin localization observed in the tumors formed from the MMTV-D1K2 cancer cell lines results from environmental influences in the tumor milieu or is due to intrinsic properties of the cancer cells 52


themselves. Immunofluorescence microscopy studies indicate that the noninvasive neu T cancer cells form colonies in culture and that E-cadhe rin and the E-cadherin-associated protein p120ctn localize to cell-cell contacts in th ese colonies (Fig. 3-6B). In c ontrast, in several of the D1K2 tumor cell lines E-cadherin and p120 ctn are largely localized to th e cytoplasm. This is not specific to the D1K2 cancer cell lines because the BT549 and MDA-MB-435s (Fig. 3-6B) and MDA-MB-231 and MDA-MB-436 (dat a not shown) human basal-lik e breast cancer cell lines also exhibit punctate cytoplasmic E-cadherin localization and diffuse cytoplasmic p120ctn localization. E-cadherin also functions to se quester -Catenin by local izing it to adherens junctions and preventing it from translocating to the nucleus and functioni ng as a transcriptional coactivator. E-cadherin localiza tion to the cytoplasm (or loss of E-cadherin expression as observed in the D1K2-T5, CL1 cell line) correlate s with decreased junctional -Catenin staining in the D1K2 and BT549 cell lines (Fig. 3-6C). These observations are significant because Ecadherin is thought to function as a tumor suppressor by mediating cell-cell adhesion, and by restraining the proinvasive, oncogenic effect s of -Catenin (143-145) and p120ctn (128, 146, 147). Discussion Mouse Mammary Tumor Virus-D1K2 Tumors Cyclin D1 is overexpressed in approximat ely 40-50% of human breast cancers (41, 42), but cyclin D1 overexpression typically occurs in luminal tumors rather than basal-like breast cancers (126). Cyclin E overexpre ssion has been noted in basal-lik e breast cancers (37). Cyclin E overexpression in breast cancers correlates w ith Estrogen Receptor (ER) negativity and poor prognosis, while cyclin D1 overexpression correlates with ER expression and a favorable outcome (148). Cyclin E potently activates cdk2 and the cyclin D1-cdk2 fusion protein functions as a constitutively active form of cdk2 (57, 127). 53


Whether any type of constitutive cdk2 activation is sufficient to induce basal breast cancer formation requires further study. The animal models of basal-like breast cancer constructed to date involve genetic inactivation of BRCA1 a nd p53 (122, 149). It is unknown whether the cyclin D1-cdk2 fusion protein induces the form ation of basal-like can cer cells through a mechanism distinct from BRCA1 and p53 deletion, or whether expression of the cyclin D1-cdk2 fusion protein is functionally equivalent to BRCA1 and p53 deletion. Interestingly, cdk2 has recently been shown to inhibit the ubiquitin ligase activity of the BARD1/BRCA1 complex (150), and the BARD1/BRCA1 ubiquitin ligase complex appears to mediate the tumor suppressive functions of the BRCA 1 gene (151). P53 function is frequently lost in breast cancers, and p53 inactivation is thought to cont ribute to cell invasivene ss (152). P53 also suppresses tumorigenesis in part by inducing expr ession of the cdk inhibitor p21. P21-mediated inhibition of proliferation plays a critical role in suppressing tumorigenesis in some contexts (153). We have shown previously that D1K2 can function to sequester p21 and p27 (57, 127), therefore D1K2 may partially overrid e p53 function, with respect to p21. "Mixed Lineage" Characteristics of MMTV-D1K2 Tumor Cell Lines Previously studies of transgenic mouse breas t cancer models (154)) showed that expression of cyclin D1 in the mammary gland under the control of the MMTV pr omoter resulted in adenocarcinomas in 75% of the mice, although some squamous differentiation was observed (155). MMTV-cyclin D2 transgenic mice also develop adenocarcinomas albeit at a lower frequency (19%) (156). In contrast, MMTV-cyclin D3 mice form primarily squamous cell carcinomas (157). These studies indicate that th e type of cell cycle de regulation that drives tumor formation can influence the differentiation status of the resul ting tumors. The MMTVcyclin D1-cdk2 transgenic mouse tumors desc ribed here are morphologically heterogeneous, including metaplastic, adenosquamous as well as adeno-myoepithelial-ty pe carcinomas (127). 54


The spindle cell myoepithelial-like component of these tumors stains positively for E-cadherin and SMA and exhibits E-cadherin misl ocalization to the cytoplasm. Cell lines isolated from these tumors exhibit several similarities with hum an basal-like breast cancers including: (a) the expression of myoepithelial markers such as SMA, Nestin, Cytokeratin 14, and EGFR, (b) lack of Her2 overexpression and lack of Estrogen Receptorexpression, (c) the expression of subsets of luminal markers including E-cadherin, Cytokeratin 18 (not shown), and Cytokeratin 19, consistent with the luminal/m yoepithelial "mixed lineage" natu re of human basal-like breast cancers (129-131), (d) th e cells exhibit myoepithelial-like morphology and cytoskeletal features in vitro (e) the cell lines form invasive tumors with spindle morphology in wild type mouse mammary fat pads in vivo and (f) the cells appear to have undergone EMT which has recently been shown to occur in the basal-like subtype of breast tumors (133). MMTV-cyclin D1-cdk2 transgene expression in a mammary stem-like cell might explain the observed heterogeneity in tumor morphology. Alternatively, cyclin D1-cdk2 expression may inhibit lineage specification or block differentiation. Both of these hypotheses are consistent with the observation that the markers Cytokeratin 14 and Nestin expressed in the MMTV-cyclin D1-cdk2 tumor cell lines are associated with relatively undiffere ntiated cell populations (158-163). Mouse Mammary Tumor Virus-D1K2 Invasiveness Several non-mutually exclusive mechanisms could contribute to the invasiveness of MMTV-D1K2 tumors including E-cadherin misloc alization/downregulation and expression of proteins previously correlated with invasiveness such as Nestin (164) and Zyxin (165). The role of cdk2 activation in these processes is unclear Chronic treatment with the cdk2 inhibitor Roscovitine did not revert the MMTV-D1K2 cell lines to a luminal-like morphology or increase the formation of adherens junctions in culture (data not shown). It is possible that cdk2 55


activation directly influences cell invasivene ss, perhaps by functiona lly inactivating Rb. Recently it has been shown that knock-down of Rb expression by short-interfering RNAs decreases E-cadherin expression an d induces EMT (166). However it is also possible that cdk2 activation induces irreversible changes to the ce lls that result in increased invasiveness. The mixed-lineage nature of the MMTV-D1K2 cells may result from an alteration in the normal differentiation program that causes increased inva siveness. The observation that invasive human basal-like breast cancer cell lines exhibit the same E-cadherin downregul ation/mislocalization and mixed-lineage expression pattern is consistent with this hypothesis. We have shown previously that MMTV-D1K 2 cell lines exhibit aneuploidy (127). Cdk2 activation may induce genetic inst ability that results in subpopul ations of tumor cells with increased invasive properties. The observation that the D1K2-T4 cells do not express D1K2, but exhibit E-cadherin mislocal ization, a mixed lineage phenotype, and invasiveness in vivo raises the possibility that D1K2 initiates the formation of a subpopulation of invasive cells, but is not required for its maintenance. Intriguingly, the D1K2-T4 cells exhibit alterations that might mimic D1K2 function and mediate invasiveness in its absence, including p53 mutation (152) and decreased Rb expression (166). Another potential explanation for the Ecadherin downregulation and cytoplasmic localization is the production of TGF by the MMTV-D1K2 cancer ce lls. We have previously shown that the MMTV-D1K2 cells secrete TGF (127). TGF is capable of inducing EMT associated with E-cadherin downregulation/cyt oplasmic localization (167). TGF signaling has been associated with basal-like breast cance rs (168-170), and EMT has been specifically associated with basal-like tumors (133). Chroni c treatment with a TGF type I receptor kinase inhibitor did not convert the MMTV-D1K2 cell li nes to a luminal morphology in culture (data 56


not shown). However, autocrine TGF might function to induce an aberrant luminal/myoepithelial mixed lineage differentiati on status that cannot be reversed by blocking TGF signaling. The previously observed tumo r-associated fibrosis (desmoplasia) in the MMTV-D1K2 tumors (127) is recapitulated in the tumors generated from the MMTV-D1K2 tumor-derived cell lines (Fig. 3-4A). This suggests that the desmoplasia is due to the cancer cells themselves, and rules out possible effects caused by transgene expression in the stroma. Desmoplasia initiated by the MMTV-D1K2 can cer cells could resu lt from TGF production because TGF is well known to induce fibrosis and desmoplasia (171, 172). In summary, the evidence presented here s uggests that constitutively active cdk2 in the form of a cyclin D1-cdk2 fusion protein induces tu mors that contain an invasive component that exhibits multiple features in common with human basal-like tumors and tumor-derived cell lines. Current efforts are focused on understanding th e respective roles of cdk2 hyperactivation, genetic instability, and TGF production in the form ation of the invasive basal-like cancer cells in the MMTV-D1K2 tumors. It is hoped that these studies will yield insights into the mechanisms responsible for the invasiveness of human breast tumors. 57


A B Figure 3-1.MMTV-D1K2 hypercellular lesions invade into the mammary stroma. A) Representative mammary MMTV-D1K2 tu mor section stained with H&E at low and high magnification displays areas of poorly differentiated cells invading into the surrounding mammary fat pad. B) Tumors derived from MMTVneu tumor cells stained with Masson's Trichrome exhibit distinct tumor cell compartmentalization from the surrounding stroma. C) Phase contrast microgr aphs of nontransformed mouse mammary NMuMG cells, MMTV-D1K2 tumor-derived cell lines D1K2-T2 and D1K2-T4, and the BT549 human basal-like breast cancer cell line. 58


C Figure 3-1.Continued. 59


A B Figure 3-2.Cell lines derived from MMTV-D1K2 tumors exhibit protein expression profiles consistent with basal-like breast cancer. A) Immunoblot analysis of cells de rived from MMTV-D1K2 tumors, an MMTVneu tumor ( neu T), and the BT549 basal-like human mammary carcinoma cell line, using antibodies specific for N-cadherin (N-Ca d.), E-cadherin (E-Cad.), P-cadherin (PCad.), Epidermal Growth Factor Recept or (EGFR), Vimentin, Cytokeratin 14 (CK14), Smooth muscle Actin ( SMA), Nestin, p120 Catenin (p120 ctn ), Catenin, Zyxin, p53, p130, the cyclin D1-cdk2 fusion protein (detected with a cdk2 antibody (cdk2(D1K2)), Her2/ neu ( neu ), and Actin as a loading control. D1K2-T2, CL1 and D1K2-T2, CL6 are clonal cell lines derived from the D1K2-T2 cancer cell line. B) Immunoblot an alysis of primary MMTVneu and MMTV-D1K2 tumors (arbitrarily labeled A-G) using antibodies specific for Flag (detecting the cyclin D1cdk2 fusion protein (Flag(D1K2)), Her2/ neu E-cadherin (E-Cad.), -Smooth Muscle Actin ( -SMA), Vimentin, Rb, Rb phosphorylated on residues 249/252 (PRb[249/252]), Rb phosphorylated on residue s 807/811 (P-Rb[807/811]), and Actin as a loading control C) Immunoblot analysis of the indicated cell lines with the indicated antibodies. "-D1K2" represents expression of the cyclin D1-cdk2 fusion protein detected with either cyclin D1 or cdk2 antibodies. "End. cyclin D1" and "End. cdk2" represents levels of endogenous cyclin D1 and cdk2. 60


C Figure 3-2.Continued. 61


A Figure 3-3.MMTV-D1K2 tumor cell lines exhi bit mixed luminal/myoepithelial lineage A) Immunoblot analysis of clonal MMTV-D1K2 cancer cell lines using the antibodies listed in Fig. 3-2, and a Cytokera tin 19 antibody (CK19). Clonal lines are designated by the name of the original cell line followed by the clone number. B) Immunofluorescence micrographs showing nuclear DAPI st aining (blue) in the no primary antibody control (No 1), Cytokera tin 14 staining (CK14, orange), E-cadherin staining (E-Cad., green), and a merged im age showing coexpression of E-cadherin and Cytokeratin 14 at the single cell leve l (CK14 + E-Cad. + DAPI). C) Immunoblot analysis of human mamm ary carcinoma cell lines (T47D, MCF7, MDA-MB-361, MDA-MB-468, BT549, MDA-MB-231, MDAMB-435S, and MDA-MB-436) and nontumorigenic mammary epithelial cell lines (NMuMG, MCF10A, and HBL100) was performed using th e indicated antibodies. 62


B Figure 3-3.Continued. 63


C Figure 3-3.Continued. 64


A B Figure 3-4.Tumors formed from MMTV-D1K2 cancer cell lines exhibit stromal invasion and Ecadherin mislocalization/downregulation upon orthotopic implantation. A) Masson's Trichome staining of histological sections from tumors derived from D1K2-T1 cells invading into the mammary fat pad (first panel), muscle (second panel), and a tumor derived from D1K2-T2 cells invading into the mammary fat pad (third panel). H&E stained section of a tumor derived from D1K2-T5 cells invading into the mammary fat pad (fourth panel). B) Immunohistochemical E-cadherin detection (brown staining) in histological sections of tumors generated from the indicated polyclonal cell lines. E-cadherin is localized to sites of cell-cell contact in neu T tumors (inset). In contrast, tumors formed from the D1K2-T1, D1K2-T2, and D1K2-T4 cell lines exhibit weak, mosaic E-cadherin expression. E-cadherin is localized diffusely throughout the cytoplasm (e.g. D1K2-T2, inset) in cells that express it. C) Cytokeratin 14 (CK14) im munohistochemical analysis of tumors generated from the indicated polyclonal cell li nes (red staining). D) Hematoxylin and Eosin (H&E) histochemical staining of an MMTV-D1K2 primary tumor (left panels) showing poorly differentiated spindleshaped cell populations with abundant cytoplasm surrounded by more densely pack ed cells. Immunohi stochemical staining (brown) of serial sections of the same tumor for E-cadherin (E-Cad., center panel) and -Smooth Muscle Actin (SMA, right panel) Sections were counterstained with hematoxylin. Note the diffuse cytoplasmic E-cadherin staining (inset) and positive staining for -Smooth Muscle Actin in the compartments containing the spindleshaped cells. 65


C D Figure 3-4.Continued. 66


A Figure 3-5.Basal-like breast cancers exhibit an invasive, mixed-lineage phenotype and tumorassociated fibrosis. A) The top panel shows an H&E stained hi stological section ( center) of a triplenegative human breast adenocarcinoma. Th e top-right and topleft panels show invasion of the tumor into the surrounding fat pad and inclusions of adipocytes in the tumor (inset). The second row of panels demonstrates immunohistochemical staining for the myoepithelial markers -Smooth muscle actin ( -SMA) and Cytokeratin 14 (CK14), and the luminal marker E-cadheri n (E-cad.). The bottom panel shows Van Gieson's staining of the tumor. The cancer cells stain brown. The intense red staining between the cancer cells shows extensive collagen deposition, indicative of tumor-associated fibrosis. B), MDA-MB-231 and MDA-MB -436 cell lines were grown as orthotopic xenograft tumors in athymic nude mice. Upper panels show H&E stained histological secti ons and lower panels show se rial sections stained with Masson's Trichrome. Arrows in the MDA-MB-231 panels point out fibroblasts/fibrosis at the a dvancing tumor front, and the infiltration of cancer cells between stromal adipocytes. Arrows in the leftmost MDA-MB -436 panel point out numerous blood vessels at the advancing tumor boundary that colocalize with the extensive fibrosis at tumor/stroma interf ace. Arrows in the rightmost MDA-MB-436 panel show the necrotic, fibrotic center of a four millimeter tumor. 67


B Figure 3-5.Continued. 68


A Figure 3-6. MMTV-D1K2 tumor-derived cell lines exhibit extensive stre ss fiber formation and E-cadherin, p120 ctn and -catenin localization to the cytoplasm. A) Immunofluorescence micrographs of th e indicated cell lines stained for Zyxin (green), Actin (orange), and DNA (DAPI, blue). B) Immunofluorescent staining (yellow) for E-cadherin (upper panels), and p120 ctn (p120, lower panels). The cells were counterstained for DNA (DAPI, blue). E-cadherin and p120 ctn are localized to cell-cell junctions in the neu T cells, but are largely localiz ed to the cytoplasm in the D1K2-T2,CL1, D1K2-T4,CL1, D1K2-T5,CL1, BT549, and MDA-MB-435s cell lines. C) Immunofluorescent stai ning for E-cadhe rin (orange) and -Catenin (green) in the indicated cell lines showing that in the neu T cells E-cadherin and -Catenin localize to cell-cell contacts, while in the D1K2-T2,CL1, D1K2-T5,CL1 and BT549 cell lines E-cadherin and -Cate nin do not co-localize. Th e cells were counterstained for DNA (DAPI, blue). 69


B Figure 3-6.Continued. 70


C Figure 3-6.Continued. 71


CHAPTER 4 A NOVEL CLASS OF CYCLIN-DEPDENDENT KINASE INHIBITORS IDENTIFIED BY MOLECULAR DOCKING ACT TH ROUGH A UNIQUE MECHANISM Introduction Uncontrolled cell proliferation is one of the defining features of cancer. Cdks are serine/threonine protein kinases th at play key roles in controlli ng cell cycle progression (1, 2). The concerted activities of cdks result in ch romatin condensation, nuclear envelope breakdown, and the up-regulation of genes involved in nucleotide synthesis and DNA replication, among other events. Cdks have long been considered id eal targets for anti-cance r drugs, owing to their importance in the cell cycle. As a result, many cdk inhibitors have been developed, some of which have progressed to clinical trials. Rosc ovitine (Selicilib) and Flavopiridol (Alvocidib) are examples of cdk inhibitors that have passed Ph ase I clinical trials (61, 66) and have been approved for Phase II clinical trials (62, 63, 67). These drugs as well as most other cdk inhibitors, are ATP-competitive. The disadvantage of using a therapeutic strategy involving ATP competition is that all kinases possess an ATP-binding site, leading to the potential for reduced target specificity. Recen tly, advances have been made in identifying cdk inhibitors that act through novel mechanisms. One example of this effort is the identifica tion of a D-amino acid hexapeptide molecule, NBI1, that inhibits the kinase activity of cdk2 through its binding with cyclin A (80). A similar approach was taken in id entifying a series of cyclic peptides that bind to the substrate recognition site of cdk complexes (81). Another example is the use of a small, 39 amino acid peptide that inhibits the kinase activity of cdk2 by mimi cking the inhibitory effects of the pRb2/p130 spacer domain (173). These approaches are promising, but rely on peptide-based inhibitors that have inhe rent disadvantages for use as therapeutic agents. The use of knockout mice has recently generated much information about the role of cdks with respect to cell cycle regul ation. For instance, mice lacking cdk2, cdk4 and cdk6 are viable 72


(30, 174, 175). Furthermore, cells lacking both c dk4 and cdk6 proliferate almost normally (33). More recently it has been discovered that mouse embryonic fibroblast cells are able to cycle in the absence of cdk2, cdk4 and cdk6, needing only cdk1 to complete cell division (35). In light of the fact that cdks are able to functionally replace one another, highly selectiv e cdk inhibitors that target only one type of cell cycle cdk may not be as effective anti-tum or agents as compounds that inhibit cdk1, cdk2, cdk4, and cdk6. An inhibito r that acts on multiple ce ll cycle cdks would therefore have a greater probability of inhibiting tumor cell growth by ensuring that the cell cycle is arrested. Here we report th e identification of a nove l structural pocket pr esent on cdk2 that is likely conserved on cdks 1, 4, and 6. Using a high throughput in silico screening procedure we have identified compounds that decrease the function of cdks in cells through binding to this site. Materials and Methods Molecular Docking The two protein crystal structur es used for identification and in silico screening of the structural site in question were the cyclinA/cdk2 and p27 kip1 /cyclin A/cdk2 complexes (RCSB Protein Data Bank codes: 1FIN (72) and 1JSU, respectively (74)). A molecular surface of 1JSU was prepared using the MSROLL program, which was then used as input for the sphere generating program SPHGEN. A cluster of spheres that was shown to be within the pocket of interest was then selected and edited manually to leave a cluster of 21 spheres. The SHOWBOX program was used to construct a 3-dimensional r ectangle, 4 Angstroms in every direction from the sphere cluster. The program CHIMERA was us ed to convert the PDB file of 1JSU into the appropriate mol2 format. The box file that was generated was then used as input for the GRID program, which calculates and saves the inform ation concerning the steric and electrostatic environment within the box of th e 1JSU mol2 file. DOCK was used to screen the entire National Cancer Institute/Developmental Therapeutics Pr ogram (NCI/DTP) database of small molecules 73


(which consisted of approximately 140,000 small mo lecules at the time of docking) within the 1JSU grid, with the selected spheres as theore tical binding sites. CHIMERA was subsequently used to rank the small molecule output ba sed on predicted energy scores composed of electrostatic interactions and va n der Waals forces. The top 40 compounds were obtained from the NCI/DTP for cell culture testing. All of the programs listed for this procedure were part of the DOCK5.0 suite developed at UCSF (176). Protein structure visualization and image generation were done using Pymol software (DeLano Scientific, Palo Alto, California). Sequence Alignment Human cdk1, cdk2, cdk4, and cdk6 were al igned using the program Geneious 3.8.5 (Biomatters, Auckland, New Zealand). Sequence similarity was calcula ted using the program MegAlign 3.07 (DNASTAR, Inc., Madison, WI). Cell Culture All experiments involving mammalian cell culture were performed using Dulbeccos Modified Eagles Medium (DMEM) supplemented with 10% heat inactivat ed fetal bovine serum (Mediatech, Inc., Manassas VA 35-011-CV). SF9 insect cells were cultured with SF-900 II SFM (Gibco, Invitrogen, Carlsbad, CA 10902) BT549 and HCT116 cell lines were obtained from the American Type Culture Collecti on (ATCC, Manassas VA). QBI-293A kidney cells were obtained from Quantum (Montreal, Canada). The SF9 cells were generously provided by Dr. Sergei Zolotukhin (University of Florida, Gainesville, Florida). Chemical Synthesis Chemical reagents and solvents were purchas ed from Aldrich (St. Louis, MO) and Acros (Geel, Belgium). Both synthesized compounds disp layed spectroscopic data consistent with the proposed structures. 74


NSC Compound 43067 (1-(5-Methylthiophen-2-yl)-3-phenylpropenone): Solid NaOH (1.78 g, 44.5 mM) was dissolved in 50 mL of water/etha nol (2:1, v/v) with coo ling in an ice bath. Sequential addition of 5-methyl -2-acetyl thiophene (5.00 g, 35.6 mM) and benzaldehyde (3.77 g, 35.6 mM) to the cooled solution of NaOH was foll owed by rapid stirring on ice for 2 hours. The mixture was stored overnight at 4 C, resulting in formation of an oily solid. The solid was removed by vacuum filtration, and th e filtrate was then concentrated in vacuo The residue was dissolved in hot absolute etha nol and let cool to provide 4.04 g (50 % yield) of NSC 43067 as flaky yellow crystals, MP 93-95 C. NSC Compound 63002 (2-(4-methoxy-phenyl)-3-pyridin-2-yl-acrylo nitrile): The compound was prepared in 26% yield as described in (177). MP 69-70 C (lit 69.5-70.5). Western Blot Analysis Cells were lysed with extraction buffer (0.1% Triton X-100, 20 mM HEPES pH 7.6, 1 mM EDTA, 1 mM EGTA, 0. 1% Triton X-100, 0.1% -mercaptoethanol, 5% glycerol, 10 nM microcystin, 1 mM sodium orthovanadate, a nd 40 mM sodium pyrophosphate) followed by sonication. Extracts were cl eared by centrifugation at 16,000 x g for 20 minutes. The cleared supernatant was then analyzed for total protein concentration with Bradford protein assay dye reagent (Biorad, Hercules, CA 500-0006), and a ll extracts were norma lized to the lowest protein concentration. The extracts were boiled with one-third volume of 4X SDS sample buffer (60 mM Tris pH 6.7, 24 mM EDTA 200 mM SDS, 40% glycerol, 300 M bromophenol blue, 0.4% -mercaptoethanol) for 5 to 10 minutes. Samp les were resolved on SDS-PAGE gels, and transferred to nitrocellulose membranes. Membranes were immunoblotted using antibodies specific for either Actin ( -rabbit, Santa Cruz, Santa Cruz, CA sc-1616-R), cdk1 ( -mouse, Santa Cruz sc-54), cdk2 ( -rabbit, Santa Cruz sc-163), cdk4 ( -rabbit, Santa Cruz sc-601), Flag 75


(M2 antibody, Sigma, St. Louis, Mo F-3165), E2F1 ( -rabbit, Santa Cruz sc-193), cyclin A ( rabbit, Santa Cruz sc-596), B-myb ( -rabbit Santa Cruz sc-725), p-Rb ( -mouse, Cell Signaling, Danvers, MA #9309), p-Rb-780 ( -mouse, Cell Signaling, #9307), p-Rb-807/811 ( -mouse, Cell Signaling #9308), or Erk1/2 ( -rabbit Santa Cruz sc-93). Construction of Stable Cell Lines Green Fluorescent Protein (GFP) fused to cdk4 was stably overexpressed in 293A cells. GFP cDNA was amplified from the pAcGFP-Tubu lin plasmid (Clontech, Mountain View, CA) using the following PCR primers: 5 TTTTGGATCCGATATCCCACCATGGTGAGCAAGGGCGCCGAG 3 and 5 TTTTGGATCCCTTGTACAGCTCATCCAT GCC 3. Following amplification, the GFP PCR product was purified by chloroform/phenol extraction and ethanol precipitation. The PCR product was subsequently digested with BamHI. The previously described pcDNA3 plasmid encoding cdk4-His 6 contains a BamHI site 5 to cdk4-His 6 (57). This plasmid was also digested with Bam HI, followed by treatment with calf intestinal alkaline phosphatase for 1 h at 37 o C. Vector and insert DNA were ligated for 18 h at room temperature, creating a construct that contains cDNA encoding GFP in-fra me with cDNA encoding cdk4-His 6 The orientation of the insert was confirmed with EcoRV digesti on. Ten micrograms of the GFP-cdk4/pcDNA3 construct was transfected into 293A cells us ing lipofectamine (Invitrogen 18324-020). Cells that stably retained the plasmid were first sele cted by treating the cells with medium containing 500 g/ml G418 Sulfate (Cellgro 61-234-RG), fo llowed by one round of cell sorting for GFPpositive cells using a FACSAria cell sorter (Bect on Dickinson, Franklin Lakes, NJ) and the Diva program (version 6.1). 76


Mink lung epithelial cells (Mv1Lu) stably ex pressing a cyclin D1-cdk2 fusion protein or E2F1 were described previously (57, 93). Ultracentrifugation Assay Cell extracts were prepared as described in th e previous section. Samples were centrifuged at either 16,000 x g for 20 minutes or 150,000 x g fo r 1 hour. In both cases, the supernatants were removed and normalized for protein concen tration. The pellets were extracted with 300 l extraction buffer containing 1.0% Triton X-100, 20 mM HEPES pH 7.6, 1 mM EDTA, 1 mM EGTA, 0.1% Triton X-100, 0.1% -mercaptoethanol, 5% glycerol, 10 nM microcystin, 1 mM sodium orthovanadate, and 40 mM sodium pyrophosphate and 300 l 2X SDS sample buffer. The extracted pellets were normalized using the same dilution factors as for the supernatant fraction, sonicated vigorously, and boiled for 5 to 10 minutes. Fluorescence Microscopy Cells for immunofluorescence studies were plated onto glass coverslips in 6-well plates. After a 24 hour incubation period, cells were treated with either vehicle control or the indicated NSC compounds. After a further 24 hour incubation, the cells were fixed with 1% paraformaldehyde in phosphate-buffered salin e (PBS) for 20 minutes, followed by a 10 minute incubation with the quench solution (50 mM amm onium chloride + 0.5% Triton X-100 in PBS). The coverslips were subseque ntly incubated with antibody buf fer (10% Goat Serum + 0.5% Triton X-100 in PBS) in a humidified chambe r for 1 hour. Primary antibody staining was performed using an antibody for cdk4 ( -rabbit, Santa Cruz sc-601) at a 1:100 dilution in antibody buffer, or no primary antibody as a cont rol for non-specific staining, for 2 hours. Following primary antibody incubation, the coverslip s were washed four times with PBS, and incubated with a goat anti-Rabb it Fluor 488 secondary antibody (I nvitrogen, Molecular Probes, 77


Carlsbad, CA A11008) for 1 hour at a 1:200 dilu tion in antibody buffer. Following four more washes with PBS, coverslips were mounted onto slides with Vect ashield + DAPI (Vector Laboratories, Orton Southgate, Engl and H-1200) to visualize nuclei. Slides were viewed on a Leica TCS SP2 AOBS spectral confocal microscope. Images were collected and processed using LCS (Leica Confocal Software, Leica Microsystems, Wetzlar, Germany) Version 2.61, Build 1537. GFP tagged 293A cells were plated, trea ted, and processed the same as for immunofluorescence. However, instead of an tibody incubation, the fixed and quenched cells were mounted directly onto slides with Vectashield + DAPI. Cloning and Expression of a Cyclin D1Cdk2 Fusion Protein (D1K2) Baculoviral Construct cDNA encoding D1K2 was digested from the pcDNA3 expression vector (57) using the 5 EcoRI and 3XhoI sites. DIK2 was subsequently subcloned into the 5-E coRI and 3-SalI sites of pFBDM. Baculovirus generation in SF9 insect cells was performed as described (178, 179). The plaque isolated D1K2 virus, at a concentration of 2.5 x 10 9 plaque forming units (PFU) per ml was used to infect SF9 cells at a multiplicity of infection of five. Cells were allowed to grow for two days at 27 C shaking at 100 RPM. Cells were pelleted by centrifugation and washed once with PBS, followed by extraction with 0.1% Triton X-100 extraction buffer (20 mM HEPES pH 7.6, 0.1% Triton X-100, 0.1% -mercaptoethanol, 5% glycerol, 10 nM microcystin, 1 mM sodium orthovanadate, and 40 mM s odium pyrophosphate) and sonication. The supernatant was centrifuged at 100,000 x g for one hour and filtered through a 0.45 m filter to remove remaining particulates. The lysate was th en loaded onto a 1 ml Ni-NTA agarose column (Qiagen, Gaithersburg, MD) equilibrated with 25 mM Tris-HCl (pH 7.5) and 100 mM NaCl. The column was washed with 10 ml 10 mM imidazole in the same buffer at a flow rate of 1 ml 78


per minute. The D1K2 protein was then eluted with 200 mM imidazole-containing buffer. Peak fractions were pooled and exchanged into a buffer containing 40 mM HEPES (pH 7.0), 0.2 M NaCl and 5 mM dithiothreitol by alternating concentration and dilution using a 10,000 molecular weight cutoff Amicon Ultra centrifugal filter device (Millipore, Billerica, MA UFC 901024) until a dilution factor of 1:10,000 of the original buffer had been reached. The protein was concentrated to approximately 200 g/ml and centrifuged at 150,000 x g to remove particulates and aggregated protein. In Vitro Aggregation Assay Purified, concentrated, and ultracentrifuged pr otein was mixed with an equal volume of buffer containing either DMSO or NSC compound for a final reaction concentration of 100 g/ml D1K2, 400 M NSC compound or 0.8% DMSO vehi cle in 40 mM HEPES (pH 7.0), 0.2 M NaCl, 5 mM dithiothreitol, and 0.1% Triton X-100. The reaction mixture was incubated at room temperature for 16 hours. Following in cubation, the reactions were centrifuged at 16,000 x g for 20 minutes to pellet aggreg ated protein. Supernatants we re removed and boiled with one third volume of 4X SDS sample buffer, while 2X SDS sample buffer was added to the pellets. These samples were resolved by SDS PAGE and stained with Coomassie blue. Results High Throughput Screening of a Novel Cdk Drug Binding Site We examined the differences between several cr ystal structures of c dk2 in order to identify a novel inhibitor binding site on cdks. A side-b y-side comparison of the structures of the catalytically active cyclinA/cdk2 complex a nd the catalytically in active p27/cyclinA/cdk2 complex revealed the formation of a structural pocket, present only in the inhibited, p27-bound form of cyclinA/cdk2 (Fig. 4-1A). We hypothesi zed that a small molecule could bind to this pocket and stabilize the cyclinA/cdk2 complex in an open conformation which would mimic 79


the p27-bound form of cyclinA/cdk2, and would th ereby inactivate the enzy me catalytically. A sequence alignment of cdks 1, 2, 4, and 6 revealed that the residues that comprise the pocket are conserved (between 48.3% and 86.2% sequence similar ity) between these cdks, indicating that this pocket is likely to be present in the other ce ll cycle cdks (Fig. 4-1B ). We performed a highthroughput in silico molecular docking screen on the p27cyclinA/cdk2 crystal structure using the UCSF program suite DOCK5.0. Approximate ly 140,000 small molecules from the NCI/DTP database were docked into the pocket, designat ed by the appropriate spheres and scoring grid, and ranked according to their predicted binding energies. Th e top 40 compounds were ordered from the NCI/DTP and tested in cell culture. Fi gure 4-1C shows a selection of the most active compounds as determined by assays described in la ter sections, with pred icted energy scores. The NSC Compounds Inhibit the Prol iferation of Cells in Culture 3 H-thymidine incorporation assays were performed to examine the influence of the compounds on cell proliferation. Initially, all 40 compounds orde red from the NCI/DTP were tested in cell culture at a concentration of 100 M to determine if any cyto static effect could be measured (data not shown). The most promisi ng compounds were subsequently used in more detailed dose-response 3 H-thymdine incorporation assays, as well as in further mechanistic studies. The BT549 human breast cancer cell line and the HCT116 human colon cancer cell line exhibited decreased cell pro liferation as measured by 3 H-thymidine incorporation after treatment with compounds NSC43067, NSC43042, and NSC 63002 for 24 hours (Fig. 4-2A). Cell cycle analysis of each of these two cell lines was pe rformed by flow cytometr y of propidium iodide stained cells after treatment with the same compounds for 24 hours. In both cell lines, and for all three compounds, an arrest at the G0/G1 pha se of the cell cycle is observed at lower concentrations, but an accumulation of cells in the G2/M cell cycle phase is apparent at higher 80


compound concentrations (Fig. 4-2B). This assay confirms that the decrease in DNA synthesis as determined by 3 H-thymdine incorporation is asso ciated with cell cycle arrest. Cytostatic Effects of the Compounds are a Re sult of an Apparent Reduction in Cellular Cdk Levels We analyzed the total levels of several cdks by immunoblot to determine whether the compounds affect cdk abundance. BT549 cells were treated with compounds NSC43067, NSC269621 and NSC63002 at 100 M and 200 M concentrations for 24 or 48 hours. All three compounds significantly reduced the levels of soluble cdk1, cdk2 and cdk4, especially at the highest concentrations a nd longest time points. This effect was not due to a universal effect on all cellular proteins because the levels of the protein phosphatase PP2A and the structural protein Actin did not change appreciably (Fig. 4-3A). E2F-1 is a transcription factor that plays a cr ucial role in mediating cdk-initiated cell cycle progression. Early cell cycle cdk activity (G1 to S phase) leads to th e phosphorylation of the E2F inhibitor Rb, resulting in its release from E2F family protei ns and increased E2F dependent transcription (14). Activation of E2F-dependent transcription is thought to be one of the primary cell cycle-related functions of the G1/S c dks. Therefore, E2F-1 overexpression would be expected to result in the partial reversion of the cdk reducing effects of the compounds. To examine this possibility, we used a mink lung epithelial ce ll line (Mv1Lu) engineered to overexpress E2F-1 (Mv1Lu-E2F1-11) (93). Treat ment of the parental Mv1Lu cells with 200 M NSC63002 resulted in an almost complete reduction of phosphorylation of Rb at Serine 780, as well as a decrease in the levels of several E2F-1 dependent transcription products including cyclin A, B-Myb, and E2F-1 itself (Fig. 4-3B). Levels of cdk1 and cdk4 also decreased, while the levels of the kinase Erk1/2 and Actin did not change. E2F1 overe xpression resulted in a diminished response to compound NSC63002 compared to the effect observed in the parental 81


cells. A decrease in E2F dependent gene products as well as serine 780 phosphorylation of Rb, was observed, albeit the response was much weaker than in parental cells. Similarly, an Mv1Lu cell line that overexpresses a c onstitutively active cyclinD1/cdk2 fusion protein (57) showed a dampened NSC63002-induced effect on Rb phosphor ylation and E2F-depe ndent transcription, even though NSC63002 induced a pa rtial decrease in the levels of the fusion protein. Time course experiments were performed to ensure that cdk downregulation by the compounds occurred in parallel with cell cycle arrest. BT549 cells treated with 200 M NSC43042 or NSC63002 were incubated for 4, 8, 12, or 24 hours before measuring cell proliferation by 3 H-Thymidine incorporation (Fig. 4-3D). The cytostatic effects of both compounds were observed as early as four hours afte r treatment. A decrease in cdks 1, 2, and 4 was observed as early as four hours as well (Fig. 4-3C), in accord with the cell proliferation assay. Taken together, these data suggest that the cytostatic effect s of the compounds are mediated through a decrease in soluble cdk levels. Decrease in Cdk Levels is a Result of Protein Aggregation We initially examined the possi bility that the compounds incr ease protein degradation of the cdks through proteolysis to de termine the cause of the apparent decrease in cdk levels. Cotreatment of cells with proteasom e inhibitors, such as lactacystin or NLVS, did not result in diminished cdk ablation, indicating that the e ffect was not induced by proteasomal degradation of the cdks (data not shown). Similarly, tran sfection of a degradatio n-resistant mutant of ubiquitin did not reduce the effect of the compou nds, also suggesting that the effects of the compounds were not due to ubiqu itin-dependent proteasomal degradation (data not shown). Overexposure of the higher molecular weight region of a cdk4 immunoblo t of 293A cell lysates after treatment with 200 M NSC63002 revealed the presence of cdk4 immunoreactive bands 82


exhibiting decreased electrophor etic mobility (Fig. 4-4A). Overexpression of cdk4 by transfection intensified this effect. This result is suggestive of an aggregation event involving cdk4, occurring in a compound-dependent manner. We hypothesized that if aggregat ed cdks were being formed in cells treated with the NSC compounds, and that if they were of higher molecu lar weight than free, soluble cdks, this could be indicative of protein aggr egation and that these aggr egates could be pelleted by ultracentrifugation. To test this hypothesis, BT54 9 cells were treated with either DMSO or 200 M NSC63002 for 24 hours. After harvesting and sonicating the cells, th e cellular particulate was removed by centrifugation at either 16,000 x g or 150,000 x g for 20 minutes or 1 hour, respectively. The supernatants and the pellets from both centrifugations were collected and analyzed by immunoblot (Fig. 4-4B). While a dramatic decrease in the levels of cdks 1, 2, and 4 was observed in the supernatant, a corresponding increase in c dk levels was observed in the pellet, particularly after ultracentrifugation. Erk1/2 was also in creased in the pellet. The relative ratio between the decrease in the supernatant levels of Erk1/2 and the increase in the pellet levels was, however, much less than that observed for the cdks. We next performed immunofluorescence micros copy to visualize possible cdk aggregates in cells. BT549 cells were treated with DMSO or 200 M NSC63002 for 24 hours. Cells were subsequently fixed and stained with an antibody for cdk4. Cells treated with DMSO exhibited a cdk4 staining pattern that was mostly uniform and homogenous, while compound NSC63002 treatment caused the formation of concentrated c dk4 staining at intense foci (Fig. 4-4C). The presence of these foci is consistent with cdk ag gregation. In order to en sure that the apparent aggregation observed was not due to an artifact of the immuno fluorescence staining procedure, we generated a cell line that st ably expressed GFP fused to cdk4 (293A/GFP-cdk4). These cells 83


were treated in the same manner as in the previous experiment, except that after fixation the cells were mounted and viewed directly. A similar result was observed in the 293A/GFP-cdk4 cells as in the BT549 cells, where treatment with th e NSC compounds induced the formation of aggregated foci of cdk4 (Fig. 4-4D). To c onfirm that the GFP fluorescence observed was from full length cdk4, we analyzed the 293A/GFP-cdk4 cell extract by immunoblot. We observed no significant bands of a lower molecular weight, indicating that the foci viewed by microscopy were not degradation products of full length cdk4 -GFP (Fig. 4-4E). Thes e results indicate that the apparent decrease in cdk levels in cells is due to the formati on of insoluble protein aggregates that would presumably not have access to cellular substrates, and would be functionally inactive. The NSC Compounds Bind Directly to Cdks In order to determine whether the effect of the compounds on the cdks was direct or indirect, we tested the ability of the compounds to induce aggregation of a cyclin D1-cdk2 fusion protein (D1K2) in vitro Purified D1K2 was ultracentrifuged for 2 hours at 150,000 x g to ensure that the starting material was fr ee of aggregated material. The supernatant from this spin was incubated with DMSO (at a final concentrati on of 0.8%), the NSC compounds, or Roscovitine, an ATP-competitive cdk inhibitor (a ll at a final concentration of 400 M) overnight at room temperature. Triton X-100 at a final concentration of 0.1% was us ed in the assay in order to inhibit non-specific colloidal aggregation of the small molecules. After the incubation period, the samples were centrifuged for 20 minutes at 16,000 x g to pellet the aggregated material. The soluble supernatant was removed from the pelleted material and both fractions were analyzed separately by SDS-PAGE. Figure 4-5A demonstrates that NSC compounds 43042, 269621, 63002, and 43067 decreased the amount of soluble D1 K2 and increased the amount of insoluble pelleted material, in comparison with DMSO or Roscovitine. 84


In order to ensure that the effects of th e NSC compounds were speci fic to D1K2, and not other proteins, the assay described above was repeated with purifie d bovine serum albumin (BSA). Figure 4-5B shows that none of the treatment conditions re sulted in a decrease in BSA in the soluble fraction, and that no pr otein was visible in the pellet fraction. These results suggest that the compounds act directly on cdk proteins to induce their aggregation and that this is a protein-specific phenomenon. Discussion Cyclin D1 is overexpressed in approximately 50% of human primary breast cancers (180). Cdk4 is also up-regulated by gene amplification in several types of cancers (181, 182). The use of cdk inhibitors as anti-cancer agents has been considered a valuable approach for many years, owing to aberrant cdk activity in cancers combined with the fact that cdks are key controllers of the cell cycle. Furthermore, cdk inhibitors that do not act through ATP competition have the potential to be more selective drugs. We have identified a nove l site on cdk2, distinct from the ATP binding site, which can be exploited for drug targeting. Moreover, this site is relatively well conserved among cell cycle cdks and is ther efore ideal for decreasi ng overall cdk activity, as there is significant functi onal redundancy among cell cycle c dks. The compounds identified from a screen of this pocket poten tly inhibit cell prolifer ation. Flow cytometric analysis of cells treated with increasing doses of the compounds i ndicate that a G1 arrest occurred at lower concentrations, but a G2 arrest resulted after trea tment with higher doses. This may be due to the fact that the compounds affect the activity of cdks 1, 2, and 4; most likely with different IC 50 s for each kinase. Possibly, the G1 arrest observe d was due to the compounds displaying a higher affinity for the G1-S phase regulators, cdk2 and 4, and therefore affecting them at lower doses, while the G2 arrest observed at high doses was due to inhibition of cdk1 activity. 85


Our studies suggest that the NSC compounds bind directly to cdks 1, 2, and 4 and result in a decrease in the levels of soluble cdk protei n in cell culture experiments. The mechanism through which these compounds act in a cell culture system is novel for a small molecular agent. A possible explanation for the ability of the compounds to reduce the solubility of th e cdks to the point of aggregation could be the properties of the residues that line the structural pocket in question. The crystal structure of cdk2 used to screen for interacting molecules includes p27 in the structural complex, however we have no evidence to indicate that p27 is a necessary prerequisite for compou nd binding. Artificial removal of p27 fr om the crystal structure reveals that a series of hydrophobic residue s are exposed to the solvent. If the co mpounds bind to this pocket in the absence of p27, it is possible that these residue s would be exposed to the intracellular milieu resulting in aggregation. An other potential explanation for the induced aggregation is that the cdks ma y be inherently prone to aggreg ation themselves. Cdk1/cyclinB complexes localize as aggregates in Xenopus oocytes before resolubili zing and becoming active for the initiation of mitosis (183, 184). If c dks are inherently prone to aggregation, the stabilization by a small molecule of an alte rnate conformation in which hydrophobic residues become more exposed may be sufficien t to trigger cdk precipitation. There has been much research conducted on the practicality of hi gh throughput molecular screens, including work on the prevalence of false positives in large molecular databases. With regards to the mechanism of action of the comp ounds presented in this paper, several studies have addressed the issue of promiscuous inhibi tors that act through nonspecific aggregation (185-187). Small molecules identified in high-throughput in vitro screens may exhibit apparent inhibition by decreasing enzymatic activity, but only due to the formation of small molecular colloidal aggregates which adsorb proteins, inhibiting enzymes nonspecifically. The addition of 86


non-ionic detergents such as Triton X-100 at a concentration of 0.01-0.1% to the assay buffers inhibits the formation of these aggregates or even disrupts them after their formation (185-187). In all of our biochemical studies examining the aggregation of cdks, Triton X-100 was present at a concentration of either 0.1% or 1.0%. Furt hermore, our screening procedure was based on in silico docking rather than high-throughput in vitro kinase assays, in which non-specific small molecular aggregates might have presented themse lves as false-positives more frequently. The cdk aggregates observed in this st udy were formed in live cells in culture, indicating that the compounds had to be sufficiently soluble to pass through the cell membrane. A ll of this suggests that the cdk aggregates reported in this st udy are not due to non-specific small molecular colloidal aggregates, but rather due to specifi c small molecule-induced protein aggregation. There is precedent for cdk ablative agents, a lthough not for those that induce aggregation. The natural product Silibinin inhibits the prog ression of prostate can cer in mouse models concomitantly with the decrease of levels of cdk1, 2, 4, and 6 in prostate tissue (83, 84). These studies contribute to the idea that decreasing cdk levels can be effec tive in anti-cancer therapies. The novel concept of a small molecule being able to decrease total prot ein levels in live cells can potentially open the door fo r other targets as well. Through in silico screening, we were able to take advantage of the fact that th e RCSB protein structure da tabase contains crystal structures of cdk2 in an active and an inactiv e conformation. By examining key differences between two conformations of the same protein, it may be possible in other circumstances to stabilize proteins in inacti ve conformations, or to enc ourage protein aggregation. 87


A Figure 4-1.Identification of a novel cdk2 bindi ng pocket and interacting molecules by highthroughput in silico screening A) Visualization of the cyclinA/cdk2 and cyclinA/cdk2/p27 crystal structure with Pymol (Yellow-cdk2, Salmon-cyclin A, Pu rple-p27, Tyrosine 15 and Arginine36 are colored red and blue respectively as a point of reference). B) Sequence alignments of human cdk1, cdk2, cdk4, and cdk6. Shaded regions represent amino acid residues lining the pocket targeted by molecular docking. C) A selection of the most effective compounds identified by molecu lar docking as determined by 3 H-thymidine incorporation. 88


B C Figure 4-1.Continued. 89


A Figure 4-2.Compounds inhib it cell proliferation A) BT549 breast cancer, and HCT116 colon can cer cells were treated with increasing doses of NSC43067, NSC43042, and NSC63002. After a 24 hour incubation, the cells were pulsed with 3 H-thymidine for 2 hours to measure DNA synthesis. All compounds displayed a dose-dependent decreas e in cell proliferation in both cell lines. B) BT549 cells were treated with increasing doses of NSC43067, NSC43042, and NSC63002. After 24 hours, the nuclei of the cells were stai ned with propidium iodide and subjected to flow cytometric analysis. Cells a rrested in the G1, S, and G2 phases of the cell cycle were obser ved depending on the compound and the concentration. C) HCT116 cells were tr eated with increasing doses of NSC43067, NSC43042, and NSC63002. After 24 hours, the nuclei of the cells were stained with propidium iodide and subjected to flow cytometric analysis. Cells arrested in the G1, S, and G2 phases of the cell cycle were observed depending on the compound and the concentration. 90


B BT549 Figure 4-2.Continued. 91


C HCT116 Figure 4-2.Continued. 92


A B Figure 4-3.Compounds affect cell cycle cdks A) BT549 cells were treated with 100 M or 200 M NSC43067, NSC269621 or NSC63002 for 24 or 48 hours. Protein from the cells was extracted with 0.1% Triton X-100 extraction buffer, normalized, and subjected to immunoblot analysis. The levels of cdk1, cdk2, and cdk4, but not Actin or Protein Phosphatase 2A (PP2A), decrease sharply after treatment with the NSC compounds. B) Mv1Lu mink lung epithelial cells overexpressing D1 K2, E2F1, or empty vector were treated with either 0.2% DMSO or 200 M NSC63002 for 24 hours. The presence of higher levels of E2F1 or a constitutively active form of cdk2 (D1K2) partially reverses the effects of the compounds on levels of E2F dependent gene products. C) BT549 cells were treated with 0.2% DMSO or 200 M NSC43042 or NSC63002 for 1, 2, 4, and 24 hours. Immunoblot analysis reveals a decrease in levels of cdk1 as early as 4 hours, with cdk2 and cdk4 decreasing by 24 hours. D) BT549 cells were treated with 0.2% DMSO or 200 M NSC43042 or NSC63002 for 4, 8, 12, and 24 hours. 3 HThymidine incorporation analysis reveals that the decrease in cdk levels occurs on the same time scale as the inhi bition of cell proliferation. 93


C D Figure 4-3.Continued. 94


A B Figure 4-4.Decrease in cdk levels is a result of protein aggregation A) 293A cells were transiently transf ected with either pcDNA3 vector or cdk4/pcDNA3 and treated with 0.2% DMSO or 200 M NSC63002 for 24 hours. Immunoblot analysis reveals a marked in crease in high molecular mass bands in samples treated with NSC63002. B) BT549 cells were treated with 0.2% DMSO or 200 M NSC63002 for 24 hours. Cell extracts were centrifuged at either 16,000 x g for 20 minutes or 150,000 x g for one hour. Immunoblot analysis reveals that more cdk 1, 2, and 4 pelleted by high speed centr ifugation from samples treated with 200 M NSC63002 than the DMSO treated samp les or the lower speed centrifugation. C) BT549 cells were treated with 0.2% DMSO or 200 M NSC 63002 for 24 hours. Immunofluorescence analysis reveals the fo rmation of cdk4 aggregates in samples treated with NSC 63002. D) 293A cells st ably expressing a G FP-cdk4 fusion protein were treated with 0.2% DMSO or 200 M NSC 63002 for 24 hours. Fluorescence microscopy reveals the formation of cdk4 a ggregates in NSC compound treated cells. E) Immunoblot analysis of the 293A cells overexpressing GFP-cdk4 confirms the presence of full-length GFP-cdk4 a nd no significant degradation products. 95


E C D Figure 4-4.Continued. 96


A Figure 4-5.Compounds act directly on cdks A) A purified cyclin D1-cdk2 fusion protein was incubated with 400 M of Roscovitine, NSC 43042, NSC269621, NSC63002, or NSC43067, or 0.8% DMSO in a buffer containing 40 mM HEPES (pH 7.0), 200 mM NaCl, 5 mM DTT and 0.1% Triton X-100 overnight at r oom temperature. The incuba ted samples were centrifuged and the supernatant and pelle t fractions were processed separately for SDS-PAGE analysis. The NSC compounds, but not DMSO or Roscovitine, caused a decrease in D1K2 in the soluble fraction, and an increase in the pellet fraction. B) The assay in part A) was repeated with bovine serum albumin (BSA). No differences in the supernatant fractions are observed, and no pe lleted protein is detected by Coomassie staining. 97


B Figure 4-5.Continued. 98


CHAPTER 5 CONCLUSION Discussion Summary Chapters 2 and 3 described the generation and characterization of a novel transgenic mouse tumor model system with which to study basal-like breast cancer etiology. The results from these two sections have provided new evidence for the role of cdk2 and cyclin D1 in breast tumorigenesis. The mouse tumor model describe d in this study will be a useful tool in examining invasiveness and differe ntiation of cancer cells in ma mmary tissue. It will also provide a means with which to test the ability of potential novel therapeutics to inhibit tumor growth. Furthermore, it presents an opport unity to examine the effects of potential chemopreventive drugs as prophylactic agents. Chapter 4 described the char acterization of a novel class of cdk inhibitors identified by high throughput in silico molecular screening. The screen wa s designed to target a site on cdk2 that was distinct from the ATP-binding site. Through an apparent direct interaction with cdks, the compounds identified by this screening method d ecrease total soluble levels of cdks in cells through a mechanism of aggregation. This decrease in cdk leve ls correlates with a potent inhibition of cell proliferation. This is a uni que mechanism of action for cdk inhibition. Taken together, these studies describe the ro le of cdk2 in breast tumorigenesis, followed by a novel means by which to inhibit cdks and halt the progression of cancer. It examines the entire process of cancer from initiation to potential cure. The Nature of Cdk Aggregation At this point, there is insufficient data to identify the precise m echanism for the observed cdk aggregation induced by trea tment with the compounds. As mentioned in the discussion 99

PAGE 100

section of chapter 4, one possible explanation for the aggregation is that a structural change in the cdks occurs, exposing hydrophobic side chains of residues in the pocket. Originally, the crystal structure used for the high throughput in silico screen contained p27. However, immunoprecipitation studies indica te that there is no increa se in p27 binding to cdk2 upon compound treatment. This suggests that the compounds bind to cdk2 regardless of p27 presence. If this is the case, th en the compounds may be able to induce a similar conformational change in cdk2 to that induced by p27. By artificially re moving p27 from the original crystal structure using the protein structure viewer PYMOL, we can visualize what the pocket may look like in the presence of one of the com pounds, and in the absence of p27. In the absence of p27, the side chains of a tyrosine, lysine, and two valine resi dues are exposed to the surrounding solvent. If this is any indication of what the structure of cdk2 bound to a compound in the absence of p27 would be, then these residues could serve as th e basis of aggregation by interacting with the surrounding intracellular milieu. Several studies have been published that describe the aggregation of cyclinB1/cdk1 complexes in starfish oocytes (183, 184). These studies discuss the dynamic aggregation and resolublization of these cyclin /cdk complexes with relation to their activation and regulation. These aggregates, which exist in the cytoplasm, disperse after treatment with maturation hormone, thereby responding to cell cycle stimulati on. As of yet, such aggregates have not been described in other cellular system s. Although it is unclear whethe r or not these complexes cycle between states of solubility and insolubility in mammalian cells, these papers set a precedent for the aggregation of cdks in cells. Cdk2 is a client of the chaperone proteins hsp90 and cdc37 (188). Furthermore, when cells are treated with an inhib itor of hsp90, cdk2 levels decrease dramatically (188), indicating a 100

PAGE 101

requirement for chaperone assistan ce in protein stability. Taken together, these studies indicate that the cdks are relativity unstabl e. In the context of treatmen t with the compounds described in this study, these proteins may be sensitive to slight changes in conformations that could lead to the aggregation observed. Implications for Aggregate-Inducing Molecules in Cancer Therapy As discussed in the introduction, compounds that reversibly inhibit the enzymatic activity of cdks have proven to be somewhat successful as anti-cancer therapeutics. However, as with all drugs, there is a finite half-life within cells, a nd once the drug is metabolized or excreted, cdk activity will recover. In this respect, the compounds identified in this study have an added advantage, affecting total levels of cdks. The outcome is that when the drug is metabolized, the cell will need to replenish the levels of the cdks by protein synthesis, re sulting in a lag period, in which cell cycle activity will be low. This c ould allow the compounds to have effects that extend beyond their natural half-life, a potential a dvantage with respect to dosage intervals in the clinical setting. Future Research In Vivo Compound Studies As the in vitro data collected on the NS C compounds indicate that they induce cell cycle arrests, the next logical step in the course of drug developmen t would be to test the compounds on mice in vivo Mice will be injected into the mammary fat pad with tumor cells to initiate tumorigenesis. After the tumors grow to a palpab le size, intraperitoneal injections of several of the NSC compounds will be administered, and tumo r growth will be monitored. As mentioned above, the transgenic mouse tumor model described in chapters 2 and 3 will be a useful tool for examining the effect of novel cancer therapeutics on tumor growth. As this model is driven by a 101

PAGE 102

constitutively active form of cdk2, and the compounds are directed towards cdks, the two projects will overlap well. Protein Crystallization Although chapter 4 describes multiple lines of evidence that the identified compounds directly affecting cdks, it is as of yet still undetermined whethe r or not these compounds bind to the exact pocket to which they were directed. In order to address this i ssue, it will be necessary to solve the crystal structures of a cdk protein either alone or in complex with a cyclin, with one of the compounds in the crystallization solution. A crystal structure of a cdk with an NSC compound will be definitive evidence that the com pounds interact directly with cdks, and will validate the study. Furthermore, it will be interes ting to see whether or not the solved crystal structures reveal a conformational change in the cdk induced by one or more of the NSC compounds. Any changes observed will be informa tive in understanding the regulation of cdks. Alternative Structural Pocket for Screening As mentioned above, p27 can be artificial ly removed from the cylinA/cdk2/p27 crystal structure, revealing a possible conformation of cdk2 in the absence of p27. In an attempt to develop another class of small mo lecular inhibitors of cdk2, we ha ve used this structure as a basis for a second round of high throughput in silico molecular screening. The second pocket chosen (pocket 2) partially contains the ATP binding site of cdk2, which interacts with p27 in the original structure. In this se nse, compounds directed to this site could act as p27 mimetics, by stabilizing the conformation of cdk2 in an inhibited form. The ATP binding s ite is altered in this structure, which would mean that compounds designe d to bind to this site would have different characteristics to other compounds targeted to this site, giving them an advantage in terms of kinase specificity. 102

PAGE 103

Lead Compound Optimization Several of the compounds identified by th e high-throughput molecular docking screen described in this study have IC 50 values for inhibiting cell prolif eration in the low micromolar range. While these are promising results for lead compounds, to become marketable as potential cancer therapeutics, a dramatic increase in potenc y would be desirable. Of course not all drugs must have IC 50 values in the low nanomolar range to be effective therapeutics; salicylate for example, has an IC 50 value of greater than 1.5 mM for inhibiting cyclooxgenase-1 (189). Nevertheless, to improve the chances of having successful clinical trials with any of the compounds described in this study, some degree of optimization will most likely be necessary. There are several means through which this can be pursued. Firstly, rational modifications can be made to the original compounds based on their modeled orientation in the structural pocket of the protein. Software tools sim ilar to the DOCK program used to screen the cdk2 pocket can be used to make logical structural modifications to the lead compounds. The program RACHEL, for example, will add molecular groups one at a time to selected attachment sites on the compound. This process of iterative refinement is designed to increase the affinity of the compounds for the protein. A second means of lead compound optimiza tion could involve bridging two compounds that occupy two separate protein pockets that are in close prox imity to one another. Some compounds directed to the original pocket (poc ket 1) are only 3.5 angstroms away from some pocket 2 compounds. As these two pockets are within such close proximity, this approach is entirely feasible. Once again there is computer software available that is designed to build scaffolds to link separate compounds together. The program CHARLIE (a module within the RACHEL program) iteratively adds small molecular groups between two compounds, outputting the highest scoring linker groups based on predicted molecular inte ractions. By combining two 103

PAGE 104

compounds, the IC 50 values could be decreased by orders of magnitude. For e ither of these two approaches, the ideal situation would involve obtaining the crys tal structures of the compounds bound to the pocket of cdk2. By determining the precise orientation of the compounds in the pocket, the process of optimization will be much more likely to yield effective drugs. Alternative Mouse Model Systems The cyclin D1/cdk2 fusion protein is a usef ul tool in understanding the precise function of this complex in cells. The transgenic mous e model based on this protein has further allowed us to examine the contribution of cyclin D1/cdk2 complexes in tumor development. An extension of this project will involve the generation of alternative cyclin/cdk complexes including cyclin E/cdk2 and cyclin D1/cdk4 by fusing them together. A thorough characterization of these complexes to determin e their phosphorylation state and kinase activity will be conducted in vitro Upon establishing that these fu sion proteins function in a manner comparable to their endogenous counterparts, they will be used to generate transgenic mouse lines for tumor studies. It will be informative to determine how the tumors generated by these alternative fusion proteins compare to the cancer etiology of the original D1K2 mouse model, or if they generate tumors at all. This appro ach is a unique way with which to understand the contribution of specific cyclin/cdk complexes in breast tumorigenesis in vivo It will help to better understand whether the characteristics of th e D1K2 tumors are brought about by cdk2, or cyclin D1, or whether both are necessary for the specific etiology observed. Conclusion This study was an effort in further determining the role of cdks, and in particular cdk2, in cancer. It examines the effects of overactive cd k2 in tumorigenesis, and the effect of cdk inhibition on cell cycle dynamics. Together, the work here provi des new tools in understanding 104

PAGE 105

the pathways involved in tumor development, as well as identifying promising new compounds that may eventually be used as anticancer agents in the clinical setting. 105

PAGE 106

LIST OF REFERENCES 1. Deshpande, A, Sicinski, P, and Hinds, PW Cy clins and cdks in development and cancer: a perspective. Oncogene, 2005; 24(17): 2909-2915. 2. Sanchez, I and Dynlacht, BD New insights in to cyclins, CDKs, and cell cycle control. Semin Cell Dev Biol, 2005; 16(3): 311-321. 3. Desai, D, Gu, Y, and Morgan, DO Activation of human cyclin-dependent kinases in vitro. Mol Biol Cell, 1992; 3(5): 571-582. 4. Desai, D, Wessling, HC, Fisher, RP, and Morgan, DO Effects of phosphorylation by CAK on cyclin binding by CDC2 and CDK2. Mol Cell Biol, 1995; 15(1): 345-350. 5. Parker, LL and Piwnica-Worms, H Inactiva tion of the p34cdc2-cyclin B complex by the human WEE1 tyrosine kinase. Science, 1992; 257(5078): 1955-1957. 6. Toyoshima, H and Hunter, T p27, a novel inhi bitor of G1 cyclin-Cdk protein kinase activity, is related to p2 1. Cell, 1994; 78(1): 67-74. 7. Xiong, Y, Hannon, GJ, Zhang, H, Casso, D, Koba yashi, R, and Beach, D p21 is a universal inhibitor of cyclin kinases. Nature, 1993; 366(6456): 701-704. 8. Lee, MH, Reynisdottir, I, and Massague, J Cloning of p57KIP2, a cyclin-dependent kinase inhibitor with unique domain structure and tissue distribution. Ge nes Dev, 1995; 9(6): 639649. 9. Canepa, ET, Scassa, ME, Ceruti, JM, et al INK4 proteins, a family of mammalian CDK inhibitors with novel biological f unctions. IUBMB Life, 2007; 59(7): 419-426. 10. Aktas, H, Cai, H, and Cooper, GM Ras links growth factor signa ling to the cell cycle machinery via regulation of cyclin D1 a nd the Cdk inhibitor p27KIP1. Mol Cell Biol, 1997; 17(7): 3850-3857. 11. Sherr, CJ Mammalian G1 cyclin s. Cell, 1993; 73(6): 1059-1065. 12. Harbour, JW, Luo, RX, Dei Santi, A, Po stigo, AA, and Dean, DC Cdk phosphorylation triggers sequential intramolecular interactions that progressively block Rb functions as cells move through G1. Cell, 1999; 98(6): 859-869. 13. Hatakeyama, M, Brill, JA, Fink, GR, and Weinbe rg, RA Collaboration of G1 cyclins in the functional inactivation of the retinoblastoma protein. Genes Dev, 1994; 8(15): 1759-1771. 14. Dyson, N The regulation of E2F by pRB-fa mily proteins. Genes Dev, 1998; 12(15): 22452262. 106

PAGE 107

15. Flemington, EK, Speck, SH, and Kaelin, WG, Jr. E2F-1-mediated transactivation is inhibited by complex formation with the reti noblastoma susceptibility gene product. Proc Natl Acad Sci U S A, 1993; 90(15): 6914-6918. 16. Lees, JA, Saito, M, Vidal, M, et al. The re tinoblastoma protein binds to a family of E2F transcription factors. Mol Cell Biol, 1993; 13(12): 7813-7825. 17. Wu, L, Timmers, C, Maiti, B, et al. The E2 F1-3 transcription factors are essential for cellular proliferation. Na ture, 2001; 414(6862): 457-462. 18. Magnaghi-Jaulin, L, Groisman, R, Naguibneva, I, et al. Retinoblastoma protein represses transcription by recruiting a histone deacetylase. Nature, 1998; 391(6667): 601-605. 19. Classon, M and Harlow, E The retinoblastoma tumour suppressor in development and cancer. Nat Rev Cancer, 2002; 2(12): 910-917. 20. DeGregori, J, Kowalik, T, and Nevins, JR Cellular targets for activation by the E2F1 transcription factor include DNA synthesisand G1/S-regulatory genes. Mol Cell Biol, 1995; 15(8): 4215-4224. 21. Nevins, JR E2F: a link between the Rb tu mor suppressor protein and viral oncoproteins. Science, 1992; 258(5081): 424-429. 22. Lee, HH, Chiang, WH, Chiang, SH, Liu, YC, Hwang, J, and Ng, SY Regulation of cyclin D1, DNA topoisomerase I, and proliferating cell nuclear antigen promoters during the cell cycle. Gene Expr, 1995; 4(3): 95-109. 23. Ohtani, K, DeGregori, J, a nd Nevins, JR Regulation of the cyclin E gene by transcription factor E2F1. Proc Natl Acad Sci U S A, 1995; 92(26): 12146-12150. 24. Schulze, A, Zerfass, K, Spitkovsky, D, et al Cell cycle regulation of the cyclin A gene promoter is mediated by a variant E2F s ite. Proc Natl Acad Sci U S A, 1995; 92(24): 11264-11268. 25. Akiyama, T, Ohuchi, T, Sumida, S, Mats umoto, K, and Toyoshima, K Phosphorylation of the retinoblastoma protein by cdk2. Proc Na tl Acad Sci U S A, 1992; 89(17): 7900-7904. 26. Doree, M and Hunt, T From Cdc2 to Cdk1: wh en did the cell cycle kinase join its cyclin partner? J Cell Sci, 2002; 115(Pt 12): 2461-2464. 27. Fang, F and Newport, JW Evidence that the G1 -S and G2-M transitions are controlled by different cdc2 proteins in highe r eukaryotes. Cell, 1991; 66(4): 731-742. 28. Santamaria, D and Ortega, S Cyclins and C DKS in development a nd cancer: lessons from genetically modified mice. Front Biosci, 2006; 11(1164-1188. 29. van den Heuvel, S and Harlow, E Distinct role s for cyclin-dependent kinases in cell cycle control. Science, 1993; 262(5142): 2050-2054. 107

PAGE 108

30. Berthet, C, Aleem, E, Coppola, V, Tessaro llo, L, and Kaldis, P Cdk2 knockout mice are viable. Curr Biol 2003; 13(20): 1775-1785. 31. Aleem, E, Kiyokawa, H, and Kaldis, P Cdc2-c yclin E complexes regulate the G1/S phase transition. Nat Cell Biol, 2005; 7(8): 831-836. 32. Rodriguez-Puebla, ML, Miliani de Marval, PL, LaCava, M, Moons, DS, Kiyokawa, H, and Conti, CJ Cdk4 deficiency inhibits skin tumor development but does not affect normal keratinocyte proliferation. Am J Pathol, 2002; 161(2): 405-411. 33. Malumbres, M, Sotillo, R, Santamaria, D, et al. Mammalian cells cy cle without the D-type cyclin-dependent kinases Cdk4 and Cdk6. Cell, 2004; 118(4): 493-504. 34. Barrire, C, Santamara, D, Cerqueira, A, et al. Mice thrive without Cdk4 and Cdk2. Molecular Oncology, 2007; 1(1): 72-83. 35. Santamaria, D, Barriere, C, Cerqueira, A, et al. Cdk1 is sufficient to drive the mammalian cell cycle. Nature, 2007; 448(7155): 811-815. 36. Salon, C, Merdzhanova, G, Brambilla, C, Brambilla, E, Gazzeri, S, and Eymin, B E2F-1, Skp2 and cyclin E oncoproteins are upregulat ed and directly correlated in high-grade neuroendocrine lung tumors. On cogene, 2007; 26(48): 6927-6936. 37. Foulkes, WD, Brunet, JS, Stefansson, IM, et al. The prognostic implication of the basallike (cyclin E high/p27 low/p53+/glomeruloid-m icrovascular-prolifer ation+) phenotype of BRCA1-related breast cancer. Cancer Res, 2004; 64(3): 830-835. 38. Keyomarsi, K, Tucker, SL, Buchholz, TA, et al. Cyclin E and survival in patients with breast cancer. N Engl J Med, 2002; 347(20): 1566-1575. 39. Akli, S and Keyomarsi, K Cyclin E and its low molecular weight forms in human cancer and as targets for cancer therapy. Cancer Biol Ther, 2003; 2(4 Suppl 1): S38-47. 40. Harwell, RM, Mull, BB, Porter, DC, and Ke yomarsi, K Activation of cyclin-dependent kinase 2 by full length and low mo lecular weight forms of cyclin E in breast cancer cells. J Biol Chem, 2004; 279(13): 12695-12705. 41. Buckley, MF, Sweeney, KJ, Hamilton, JA, et al Expression and amplification of cyclin genes in human breast cancer. Oncogene, 1993; 8(8): 2127-2133. 42. Sutherland, RL and Musgrove, EA Cyclin D1 and mammary carcinoma: new insights from transgenic mouse models. Breast Cancer Res, 2002; 4(1): 14-17. 43. Parker, MA, Deane, NG, Thompson, EA, et al Over-expression of cyclin D1 regulates Cdk4 protein synthesis. Ce ll Prolif, 2003; 36(6): 347-360. 108

PAGE 109

44. Bosch, F, Jares, P, Campo, E, et al. PRAD-1/cyclin D1 gene overexpression in chronic lymphoproliferative disorders: a highly spec ific marker of man tle cell lymphoma. Blood, 1994; 84(8): 2726-2732. 45. Knudsen, KE, Diehl, JA, Haiman, CA, and Knudsen, ES Cyclin D1: polymorphism, aberrant splicing and cancer ris k. Oncogene, 2006; 25(11): 1620-1628. 46. Zheng, Y, Shen, H, Sturgis, EM, et al. Cyclin D1 polymorphism and risk for squamous cell carcinoma of the head and neck: a case-con trol study. Carcinogenesis, 2001; 22(8): 11951199. 47. Satinder, K, Chander, SR, Pushpinder, K, Indu, G, and Veena, J Cyclin D1 (G870A) polymorphism and risk of cervix cancer: a cas e control study in nor th Indian population. Mol Cell Biochem, 2008; 315(1-2): 151-157. 48. Alkarain, A, Jordan, R, and Slingerland, J p27 deregulation in breast cancer: prognostic significance and implications for therapy. J Mammary Gland Biol Neoplasia, 2004; 9(1): 67-80. 49. Blain, SW and Massague, J Breast cancer banishes p27 from nucleus. Nat Med, 2002; 8(10): 1076-1078. 50. Winters, ZE, Hunt, NC, Bradburn, MJ, et al. Subcellular localis ation of cyclin B, Cdc2 and p21(WAF1/CIP1) in breast cancer. association with prognosis. Eur J Cancer, 2001; 37(18): 2405-2412. 51. Xia, W, Chen, JS, Zhou, X, et al. Ph osphorylation/cytoplasmic localization of p21Cip1/WAF1 is associated with HER2 /neu overexpression and provides a novel combination predictor for poor prognosis in breast cancer patients. Clin Cancer Res, 2004; 10(11): 3815-3824. 52. Liggett, WH, Jr. and Sidransky, D Role of the p16 tumor suppressor gene in cancer. J Clin Oncol, 1998; 16(3): 1197-1206. 53. Rane, SG, Cosenza, SC, Mettus, RV, and Reddy, EP Germ line transmission of the Cdk4(R24C) mutation facilitate s tumorigenesis and escape fr om cellular senescence. Mol Cell Biol, 2002; 22(2): 644-656. 54. Dou, QP, Molnar, G, and Pardee, AB Cyclin D1/cdk2 kinase is pr esent in a G1 phasespecific protein complex Yi1 that binds to the mouse thymidine kinase gene promoter. Biochem Biophys Res Comm un, 1994; 205(3): 1859-1868. 55. Dulic, V, Drullinger, LF, Lees, E, Reed, SI, and Stein, GH Altered regulation of G1 cyclins in senescent human diploid fibroblas ts: accumulation of inactive cyclin E-Cdk2 and cyclin D1-Cdk2 complexes. Proc Na tl Acad Sci U S A, 1993; 90(23): 11034-11038. 109

PAGE 110

56. Sweeney, KJ, Swarbrick, A, Sutherland, RL, and Musgrove, EA Lack of relationship between CDK activity and G1 cyclin expressi on in breast cancer cells. Oncogene, 1998; 16(22): 2865-2878. 57. Chytil, A, Waltner-Law, M, West, R, et al. Construction of a cyclin D1-Cdk2 fusion protein to model the biological functions of cyclin D1-Cdk2 complexes. J Biol Chem, 2004; 279(46): 47688-47698. 58. Canduri, F and de Azevedo, F Structural basis for interaction of inhi bitors with Cyclindependent kinase 2. Curr Comp-Aid Drug Design, 2005; 1(53-64. 59. Noble, ME, Endicott, JA, and Johnson, LN Prot ein kinase inhibitors: insights into drug design from structure. Science, 2004; 303(5665): 1800-1805. 60. Shapiro, GI Preclinical and c linical development of the cyc lin-dependent kinase inhibitor flavopiridol. Clin Cancer Res, 2004; 10(12 Pt 2): 4270s-4275s. 61. Christian, BA, Grever, MR, Byrd, JC, and Lin, TS Flavopiridol in the treatment of chronic lymphocytic leukemia. Curr Opin Oncol, 2007; 19(6): 573-578. 62. Schwartz, GK, Ilson, D, Saltz, L, et al. Ph ase II study of the cyc lin-dependent kinase inhibitor flavopiridol administered to patients with advanced gastric carcinoma. J Clin Oncol, 2001; 19(7): 1985-1992. 63. Shapiro, GI, Supko, JG, Patterson, A, et al. A pha se II trial of the cyclin-dependent kinase inhibitor flavopiridol in patie nts with previously untreated stage IV non-small cell lung cancer. Clin Cancer Res, 2001; 7(6): 1590-1599. 64. Kahn, ME, Senderowicz, A, Sausville, EA, and Barrett, KE Possible mechanisms of diarrheal side effects associ ated with the use of a nove l chemotherapeutic agent, flavopiridol. Clin Cancer Res, 2001; 7(2): 343-349. 65. Senderowicz, AM Small-molecule cyclin-dep endent kinase modulat ors. Oncogene, 2003; 22(42): 6609-6620. 66. Benson, C, White, J, De Bono, J, et al. A phase I trial of the selective oral cyclindependent kinase inhibitor se liciclib (CYC202; R-Roscovitine ), administered twice daily for 7 days every 21 days. Br J Cancer, 2007; 96(1): 29-37. 67. Belani, C Efficacy study of oral Selicicl ib to treat Non-Small Cell Lung Cancer. identifier: NCT00372073; 2006. 68. Hahntow, IN, Schneller, F, Oelsner, M, et al. Cyclin-dependent kinase inhibitor Roscovitine induces apoptosis in chronic lymphocytic leukemia cells. Leukemia, 2004; 18(4): 747-755. 110

PAGE 111

69. Jimeno, A, Rudek, MA, Purcell, T, et al. Ph ase I and pharmacokinetic study of UCN-01 in combination with irinotecan in patients with solid tumors. Cancer Chemother Pharmacol, 2008; 61(3): 423-433. 70. Hotte, SJ, Oza, A, Winquist, EW, et al. Ph ase I trial of UCN-01 in combination with topotecan in patients with advanced solid ca ncers: a Princess Margaret Hospital Phase II Consortium study. Ann Oncol, 2006; 17(2): 334-340. 71. Lara, PN, Jr., Mack, PC, Synold, T, et al. The cyclin-dependent kinase inhibitor UCN-01 plus cisplatin in advanced solid tumo rs: a California cancer consortium phase I pharmacokinetic and molecular correlative tr ial. Clin Cancer Res, 2005; 11(12): 44444450. 72. Jeffrey, PD, Russo, AA, Polyak, K, et al. Mechanism of CDK activ ation revealed by the structure of a cyclinA-CDK2 co mplex. Nature, 1995; 376(6538): 313-320. 73. Honda, R, Lowe, ED, Dubinina, E, et al. The structure of cyclin E1/CDK2: implications for CDK2 activation and CDK2-independe nt roles. Embo J, 2005; 24(3): 452-463. 74. Russo, AA, Jeffrey, PD, Patten, AK, Massague, J, and Pavletich, NP Cr ystal structure of the p27Kip1 cyclin-dependent-kinase inhibi tor bound to the cyclin A-Cdk2 complex. Nature, 1996; 382(6589): 325-331. 75. De Azevedo, WF, Leclerc, S, Meijer, L, Havli cek, L, Strnad, M, and Kim, SH Inhibition of cyclin-dependent kinases by pur ine analogues: crystal struct ure of human cdk2 complexed with roscovitine. Eur J Biochem, 1997; 243(1-2): 518-526. 76. Wang, NP, To, H, Lee, WH, and Lee, EY Tumor suppressor activity of RB and p53 genes in human breast carcinoma cells. Oncogene, 1993; 8(2): 279-288. 77. Wu, SY, McNae, I, Kontopidis, G, et al. Di scovery of a novel family of CDK inhibitors with the program LIDAEUS: structural basis for ligand-induced disordering of the activation loop. Structur e, 2003; 11(4): 399-410. 78. Schulze-Gahmen, U and Kim, SH Structural basis for CDK6 activation by a virus-encoded cyclin. Nat Struct Biol, 2002; 9(3): 177-181. 79. Jeffrey, PD, Tong, L, and Pavletich, NP Stru ctural basis of inhibition of CDK-cyclin complexes by INK4 inhibitors. Genes Dev, 2000; 14(24): 3115-3125. 80. Canela, N, Orzaez, M, Fucho, R, et al. Identif ication of an hexapeptide that binds to a surface pocket in cyclin A and inhibits the catalytic activity of the complex cyclindependent kinase 2-cyclin A. J Biol Chem, 2006; 281(47): 35942-35953. 81. Andrews, MJ, McInnes, C, Kontopidis, G, et al. Design, synthesis, biological activity and structural analysis of cyclic peptide inhibitors targeting the substrate recruitment site of cyclin-dependent kinase complexes. Org Biomol Chem, 2004; 2(19): 2735-2741. 111

PAGE 112

82. De Luca, A, MacLachlan, TK, Bagella, L, et al. A unique domain of pRb2/p130 acts as an inhibitor of Cdk2 kinase activity. J Biol Chem, 1997; 272(34): 20971-20974. 83. Raina, K, Blouin, MJ, Singh, RP, et al. Dietary feeding of silibinin inhibits prostate tumor growth and progression in transgenic adenocarci noma of the mouse prostate model. Cancer Res, 2007; 67(22): 11083-11091. 84. Singh, RP, Deep, G, Blouin, MJ, Pollak, MN, and Agarwal, R Silibinin suppresses in vivo growth of human prostate carcinoma PC-3 tumor xenograft. Carcinogenesis, 2007; 28(12): 2567-2574. 85. Catzavelos, C, Bhattacharya, N, Ung, YC, et al Decreased levels of the cell-cycle inhibitor p27Kip1 protein: prognostic implications in primary breast cancer. Nat Med, 1997; 3(2): 227-230. 86. Han, S, Park, K, Kim, HY, Lee, MS, Ki m, HJ, and Kim, YD Reduced expression of p27Kip1 protein is associated w ith poor clinical outcome of breast cancer patients treated with systemic chemotherapy and is linked to cell proliferation and differentiation. Breast Cancer Res Treat, 1999; 55(2): 161-167. 87. Viglietto, G, Motti, ML, Bruni, P, et al. Cytoplasmic reloca lization and inhibition of the cyclin-dependent kinase i nhibitor p27(Kip1) by PKB/Aktmediated phosphorylation in breast cancer. Nat Med, 2002; 8(10): 1136-1144. 88. Coletta, RD, Christensen, K, Reichenberger, KJ, et al. The Six1 homeoprotein stimulates tumorigenesis by reactivation of cyclin A 1. Proc Natl Acad Sci U S A, 2004; 101(17): 6478-6483. 89. Deans, AJ, Khanna, KK, McNees, CJ, Merc urio, C, Heierhorst, J, and McArthur, GA Cyclin-dependent kinase 2 func tions in normal DNA repair and is a therapeutic target in BRCA1-deficient cancers. Can cer Res, 2006; 66(16): 8219-8226. 90. Wesierska-Gadek, J and Schmid, G Dual actio n of the inhibitors of cyclin-dependent kinases: targeting of the cel l-cycle progression and activati on of wild-type p53 protein. Expert Opin Investig Dr ugs, 2006; 15(1): 23-38. 91. Matsui, Y, Halter, SA, Holt, JT, Hogan, BL and Coffey, RJ Development of mammary hyperplasia and neoplasia in MMTV-TGF alpha transgenic mice. Cell, 1990; 61(6): 11471155. 92. Methods in Mammary Gland Biology and Breast Cancer Research. In. New York, New York 10013: Kluwer Academic/Plenum Publishers; 2000. p.317. 93. Law, BK, Chytil, A, Dumont, N, et al. Ra pamycin potentiates transforming growth factor beta-induced growth arrest in nontransfor med, oncogene-transformed, and human cancer cells. Mol Cell Biol, 2002; 22(23): 8184-8198. 112

PAGE 113

94. Hall, C, Nelson, DM, Ye, X, et al. HIRA, th e human homologue of yeast Hir1p and Hir2p, is a novel cyclin-cdk2 substrate whose expression blocks S-phase progression. Mol Cell Biol, 2001; 21(5): 1854-1865. 95. Brown, KA, Roberts, RL, Arteaga, CL, and Law, BK Transforming growth factor-beta induces Cdk2 relocalization to the cytoplasm coincident with dephosphorylation of retinoblastoma tumor suppressor protein. Breast Cancer Res, 2004; 6(2): R130-139. 96. Law, M, Forrester, E, Chytil, A, et al Rapamycin disrupts cyclin/cyclin-dependent kinase/p21/proliferating cell nu clear antigen complexes and cy clin D1 reverses rapamycin action by stabilizing these complexes. Cancer Res, 2006; 66(2): 1070-1080. 97. Abe, M, Harpel, JG, Metz, CN, Nunes, I, Loskutoff, DJ, and Rifkin, DB An assay for transforming growth factor-beta using cells transfected with a plasminogen activator inhibitor-1 promoter-luciferase construct. Anal Biochem, 1994; 216(2): 276-284. 98. Bacus, SS, Gudkov, AV, Esteva, FJ, and Yarden, Y Expression of erbB receptors and their ligands in breast cancer: implications to biological behavior and therapeutic response. Breast Dis, 2000; 11(63-75. 99. Hulit, J, Lee, RJ, Russell, RG, and Pestel l, RG ErbB-2-induced mammary tumor growth: the role of cyclin D1 and p27Kip1. Biochem Pharmacol, 2002; 64(5-6): 827-836. 100. Bowe, DB, Kenney, NJ, Adereth, Y, and Ma roulakou, IG Suppression of Neu-induced mammary tumor growth in cyclin D1 deficient mice is compensated for by cyclin E. Oncogene, 2002; 21(2): 291-298. 101. Lee, RJ, Albanese, C, Fu, M, et al. Cyclin D1 is required for tr ansformation by activated Neu and is induced through an E2F-depende nt signaling pathway. Mol Cell Biol, 2000; 20(2): 672-683. 102. Nelsen, CJ, Kuriyama, R, Hirsch, B, et al. Short term cyclin D1 overexpression induces centrosome amplification, mitotic spindle abnormalities, and aneuploidy. J Biol Chem, 2005; 280(1): 768-776. 103. Duensing, A, Liu, Y, Tseng, M, Malumbres, M, Barbacid, M, and Duensing, S Cyclindependent kinase 2 is dispensable for nor mal centrosome duplication but required for oncogene-induced centrosome overdup lication. Oncogene, 2006; 25(20): 2943-2949. 104. Desmouliere, A, Guyot, C, and Gabbiani, G The stroma reaction myofibroblast: a key player in the control of tumor cell be havior. Int J Dev Bi ol, 2004; 48(5-6): 509-517. 105. Ohtani, H, Kuroiwa, A, Obinata, M, Ooshim a, A, and Nagura, H Identification of type I collagen-producing cells in human gastrointe stinal carcinomas by non-radioactive in situ hybridization and immunoelectron micros copy. J Histochem Cytochem, 1992; 40(8): 1139-1146. 113

PAGE 114

106. Walker, RA The complexities of breast cancer desmoplasia. Breast Cancer Res, 2001; 3(3): 143-145. 107. Ronnov-Jessen, L and Petersen, OW Induc tion of alpha-smooth muscle actin by transforming growth factor-beta 1 in qui escent human breast gland fibroblasts. Implications for myofibroblast generation in breast neoplasia. La b Invest, 1993; 68(6): 696-707. 108. Ronnov-Jessen, L, Van Deurs, B, Nielsen, M, and Petersen, OW Identification, paracrine generation, and possible function of human breast carcinoma myofibroblasts in culture. In Vitro Cell Dev Biol, 1992; 28A(4): 273-283. 109. Ronnov-Jessen, L, Petersen, OW, Kotelians ky, VE, and Bissell, MJ The origin of the myofibroblasts in breast cancer. Recapitulation of tumor environment in culture unravels diversity and implicates conve rted fibroblasts and recruite d smooth muscle cells. J Clin Invest, 1995; 95(2): 859-873. 110. Beauchamp, RD, Coffey, RJ, Jr., Lyons, RM, Perkett, EA, Townsend, CM, Jr., and Moses, HL Human carcinoid cell produc tion of paracrine growth fa ctors that can stimulate fibroblast and endothelial cell growth Cancer Res, 1991; 51(19): 5253-5260. 111. Christensen, JG, Burrows, J, and Salgia, R c-Met as a target for human cancer and characterization of inhibitors for therapeuti c intervention. Cancer Lett, 2005; 225(1): 1-26. 112. Christensen, JG, Schreck, R, Burrows, J, et al. A selective small molecule inhibitor of cMet kinase inhibits c-Met-dependent phenotypes in vitro and e xhibits cytoreductive antitumor activity in vivo. Cancer Res, 2003; 63(21): 7345-7355. 113. Ge, R, Rajeev, V, Ray, P, et al. Inhibition of growth and metastasis of mouse mammary carcinoma by selective inhibitor of transforming growth facto r-beta type I receptor kinase in vivo. Clin Cancer Res, 2006; 12(14 Pt 1): 4315-4330. 114. Halder, SK, Beauchamp, RD, and Datta, PK A specific inhibitor of TGF-beta receptor kinase, SB-431542, as a potent an titumor agent for human cancers. Neoplasia, 2005; 7(5): 509-521. 115. Matsuyama, S, Iwadate, M, Kondo, M, et al. SB-431542 and Gleevec inhibit transforming growth factor-beta-induced prol iferation of human osteosarcoma cells. Cancer Res, 2003; 63(22): 7791-7798. 116. Sorlie, T, Perou, CM, Tibshirani, R, et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Pr oc Natl Acad Sci U S A, 2001; 98(19): 10869-10874. 117. Rakha, EA, El-Rehim, DA, Paish, C, et al Basal phenotype identifies a poor prognostic subgroup of breast cancer of clinical importance. Eur J Cancer, 2006; 42(18): 3149-3156. 114

PAGE 115

118. Carey, LA, Perou, CM, Livasy, CA, et al. Race breast cancer subtypes, and survival in the Carolina Breast Cancer Study. Jama, 2006; 295(21): 2492-2502. 119. Li, H, Cherukuri, P, Li, N, et al. Nestin is expressed in the basal/myoe pithelial layer of the mammary gland and is a selective marker of basal epithelial breast tumors. Cancer Res, 2007; 67(2): 501-510. 120. Abd El-Rehim, DM, Pinder, SE, Paish, CE, et al. Expression of luminal and basal cytokeratins in human breast carcinoma. J Pathol, 2004; 203(2): 661-671. 121. Dabbs, DJ, Chivukula, M, Carter, G, and Bhargava, R Basal phenotype of ductal carcinoma in situ: recognition and immunohist ologic profile. Mod Pathol, 2006; 19(11): 1506-1511. 122. McCarthy, A, Savage, K, Gabriel, A, Naceur, C, Reis-Filho, JS, and Ashworth, A A mouse model of basal-like breast carcinoma with metaplastic elements. J Pathol, 2007; 211(4): 389-398. 123. Weaver, Z, Montagna, C, Xu, X, et al. Ma mmary tumors in mice conditionally mutant for Brca1 exhibit gross genomic instability a nd centrosome amplifi cation yet display a recurring distribution of genomic imbalances that is similar to human breast cancer. Oncogene, 2002; 21(33): 5097-5107. 124. Poole, AJ, Li, Y, Kim, Y, Lin, SC, Lee, WH, and Lee, EY Prevention of Brca1-mediated mammary tumorigenesis in mice by a progester one antagonist. Science, 2006; 314(5804): 1467-1470. 125. Brodie, SG, Xu, X, Qiao, W, Li, WM, Cao, L, and Deng, CX Multiple genetic changes are associated with mammary tumorigenesis in Brca1 conditional knockout mice. Oncogene, 2001; 20(51): 7514-7523. 126. Gauthier, ML, Berman, HK, Miller, C, et al. Abrogated response to cellular stress identifies DCIS associated with subsequent tumor events and defines basal-like breast tumors. Cancer Cell, 2007; 12(5): 479-491. 127. Corsino, P, Davis, B, Law, M, et al. Tu mors initiated by constitutive Cdk2 activation exhibit transforming growth factor beta resistance and acquire paracrine mitogenic stimulation during progression. Ca ncer Res, 2007; 67(7): 3135-3144. 128. Sarrio, D, Perez-Mies, B, Hardisson, D, et al. Cytoplasmic localization of p120ctn and Ecadherin loss characterize lobular breast carcinoma from preinv asive to metastatic lesions. Oncogene, 2004; 23(19): 3272-3283. 129. Livasy, CA, Karaca, G, Nanda, R, et al. Phenotypic evaluation of th e basal-like subtype of invasive breast carcinoma. M od Pathol, 2006; 19(2): 264-271. 115

PAGE 116

130. Kim, MJ, Ro, JY, Ahn, SH, Kim, HH, Kim, SB, and Gong, G Clinicopathologic significance of the basal-lik e subtype of breast cancer: a comparison with hormone receptor and Her2/neu-overexpressing phenotypes. Hum Pathol, 2006; 37(9): 1217-1226. 131. Laakso, M, Tanner, M, Nilsson, J, et al. Basoluminal carcinoma: a new biologically and prognostically distinct entity between basal and luminal breast cancer. Clin Cancer Res, 2006; 12(14 Pt 1): 4185-4191. 132. Maeda, M, Johnson, KR, and Wheelock, MJ Cadhe rin switching: essential for behavioral but not morphological changes during an epith elium-to-mesenchyme transition. J Cell Sci, 2005; 118(Pt 5): 873-887. 133. Sarrio, D, Rodriguez-Pinilla, SM, Hardisson, D, Cano, A, Moreno-Bueno, G, and Palacios, J Epithelial-mesenchymal transition in breast cancer relates to the basal-like phenotype. Cancer Res, 2008; 68(4): 989-997. 134. Warburton, MJ, Ormerod, EJ, Monaghan, P, Fern s, S, and Rudland, PS Characterization of a myoepithelial cell line derived from a ne onatal rat mammary gland. J Cell Biol, 1981; 91(3 Pt 1): 827-836. 135. Zavizion, B, Politis, I, and Gorewit, RC Bovine mammary myoepithelial cells. 2. Interactions with epithelial cells in vitro. J Dairy Sci, 1992; 75(12): 3381-3393. 136. Zavizion, B, Politis, I, and Gorewit, RC Bovine mammary myoe pithelial cells. 1. Isolation, culture, and characterization. J Dairy Sci, 1992; 75(12): 3367-3380. 137. Zavizion, B, van Duffelen, M, Schaeffer, W, and Politis, I Establishment and characterization of a bovine mammary myoepith elial cell line. In Vitro Cell Dev Biol Anim, 1996; 32(3): 149-158. 138. Neve, RM, Chin, K, Fridlyand, J, et al. A co llection of breast cancer cell lines for the study of functionally distinct cancer subtyp es. Cancer Cell, 2006 ; 10(6): 515-527. 139. Casey, G, Lo-Hsueh, M, Lopez, ME, Vogelstein, B, and Stanbridge, EJ Growth suppression of human breast cancer cells by the introduction of a wild-type p53 gene. Oncogene, 1991; 6(10): 1791-1797. 140. Katayose, D, Gudas, J, Nguyen, H, Srivas tava, S, Cowan, KH, and Seth, P Cytotoxic effects of adenovirus-mediated wild-type p53 protein expression in normal and tumor mammary epithelial cells. Clin Cancer Res, 1995; 1(8): 889-897. 141. Schafer, JM, Lee, ES, O'Regan, RM, Yao, K, and Jordan, VC Rapid development of tamoxifen-stimulated mutant p53 breast tumors (T47D) in athymic mice. Clin Cancer Res, 2000; 6(11): 4373-4380. 142. Crawford, AW, Michelsen, JW, and Beckerle, MC An interaction between zyxin and alpha-actinin. J Cell Biol 1992; 116(6): 1381-1393. 116

PAGE 117

143. Maruyama, K, Ochiai, A, Nakamura, S, Ba ba, S, and Hirohashi, S [Dysfunction of Ecadherin-catenin system in invasion and meta stasis of colorectal cancer]. Nippon Geka Gakkai Zasshi, 1998; 99(7): 402-408. 144. Nollet, F, Berx, G, and van Roy, F The role of the E-cadherin/catenin adhesion complex in the development and progression of cancer. Mol Cell Biol Res Co mmun, 1999; 2(2): 7785. 145. Wong, AS and Gumbiner, BM Adhesion-indepe ndent mechanism for suppression of tumor cell invasion by E-cadherin. J Cell Biol, 2003; 161(6): 1191-1203. 146. Bellovin, DI, Bates, RC, Muzikansky, A, Rimm, DL, and Mercurio, AM Altered localization of p120 catenin during epithelial to mesenc hymal transition of colon carcinoma is prognostic for aggressive di sease. Cancer Res, 2005; 65(23): 10938-10945. 147. Shibata, T, Kokubu, A, Sekine, S, Kanai, Y, and Hirohashi, S Cytoplasmic p120ctn regulates the invasive phenot ypes of E-cadherin-deficient breast cancer. Am J Pathol, 2004; 164(6): 2269-2278. 148. Loden, M, Stighall, M, Nielsen, NH, et al. Th e cyclin D1 high and cyclin E high subgroups of breast cancer: separate pathways in tu morogenesis based on pattern of genetic aberrations and inactivation of the pRb node. Oncogene, 2002; 21(30): 4680-4690. 149. Brodie, SG and Deng, CX BRCA1-associated tu morigenesis: what have we learned from knockout mice? Trends Genet, 2001; 17(10): S18-22. 150. Hayami, R, Sato, K, Wu, W, et al. Down -regulation of BRCA1-BA RD1 ubiquitin ligase by CDK2. Cancer Res, 2005; 65(1): 6-10. 151. Shakya, R, Szabolcs, M, McCarthy, E, et al The basal-like mammary carcinomas induced by Brca1 or Bard1 inactivation implicat e the BRCA1/BARD1 heterodimer in tumor suppression. Proc Natl Acad Sci U S A, 2008; 105(19): 7040-7045. 152. Junk, DJ, Vrba, L, Watts, GS, Oshiro, MM, Martinez, JD, and Futscher, BW Different mutant/wild-type p53 combinations cause a spectrum of increas ed invasive potential in nonmalignant immortalized human mammary epithelial cells. Neoplasia, 2008; 10(5): 450461. 153. Willenbring, H, Sharma, AD, Vogel, A, et al. Loss of p21 permits carcinogenesis from chronically damaged liver and kidney epithelia l cells despite unchecked apoptosis. Cancer Cell, 2008; 14(1): 59-67. 154. Sutherland, RL and Musgrove, EA Cyclins and breast cancer. J Mammary Gland Biol Neoplasia, 2004; 9(1): 95-104. 155. Wang, TC, Cardiff, RD, Zukerberg, L, Lees, E, Arnold, A, and Schmidt, EV Mammary hyperplasia and carcinoma in MMTV-cyclin D1 transgenic mice. Nature, 1994; 369(6482): 669-671. 117

PAGE 118

156. Kong, G, Chua, SS, Yijun, Y, et al. Functional analysis of cyclin D2 and p27(Kip1) in cyclin D2 transgenic mouse mammary gland during development. Oncogene, 2002; 21(47): 7214-7225. 157. Pirkmaier, A, Dow, R, Ganiatsas, S, et al. Alternative mammary oncogenic pathways are induced by D-type cyclins; MMTV-cyclin D3 transgenic mice develop squamous cell carcinoma. Oncogene, 2003; 22(28): 4425-4433. 158. Asano-Miyoshi, M, Hamamichi, R, and Emori, Y Cytokeratin 14 is expressed in immature cells in rat taste buds. J Mol Histol, 2008; 39(2): 193-199. 159. Carriere, C, Seeley, ES, Goetze, T, Longnecker, DS, and Korc, M The Nestin progenitor lineage is the compartment of origin for pancreatic intraepithelial neoplasia. Proc Natl Acad Sci U S A, 2007; 104(11): 4437-4442. 160. Laakso, M, Loman, N, Borg, A, and Isola, J Cytokeratin 5/14-positiv e breast cancer: true basal phenotype confined to BRCA1 tu mors. Mod Pathol, 2005; 18(10): 1321-1328. 161. Li, L, Mignone, J, Yang, M, et al. Nestin expr ession in hair follicle sheath progenitor cells. Proc Natl Acad Sci U S A, 2003; 100(17): 9958-9961. 162. Wiese, C, Rolletschek, A, Kania, G, et al Nestin expression--a property of multi-lineage progenitor cells? Cell Mol Life Sci, 2004; 61(19-20): 2510-2522. 163. Wu, PC, Lai, VC, Fang, JW, Gerber, MA, Lai, CL, and Lau, JY Hepatocellular carcinoma expressing both hepatocellular and biliary markers also expresses cytokeratin 14, a marker of bipotential progenitor cells. J Hepatol, 1999; 31(5): 965-966. 164. Kleeberger, W, Bova, GS, Nielsen, ME, et al. Roles for the stem cell associated intermediate filament Nestin in prostate cancer migration and metastasis. Cancer Res, 2007; 67(19): 9199-9206. 165. Sy, SM, Lai, PB, Pang, E, et al. Novel identif ication of zyxin upregulations in the motile phenotype of hepatocellular carcinom a. Mod Pathol, 2006; 19(8): 1108-1116. 166. Arima, Y, Inoue, Y, Shibata, T, et al. Rb depletion result s in deregulation of E-cadherin and induction of cellular phenot ypic changes that are characte ristic of the epithelial-tomesenchymal transition. Can cer Res, 2008; 68(13): 5104-5112. 167. Xie, L, Law, BK, Chytil, AM, Brown, KA, Aakre, ME, and Moses, HL Activation of the Erk pathway is required for TGF-beta1-indu ced EMT in vitro. Neoplasia, 2004; 6(5): 603610. 168. Honeth, G, Bendahl, PO, Ringner, M, et al The CD44+/CD24phenotype is enriched in basal-like breast tumors. Breast Cancer Res, 2008; 10(3): R53. 118

PAGE 119

169. Sheridan, C, Kishimoto, H, Fuchs, RK, et al. CD44+/CD24breast cancer cells exhibit enhanced invasive properties: an early step necessary for metastasis. Breast Cancer Res, 2006; 8(5): R59. 170. Shipitsin, M, Campbell, LL, Argani, P, et al. Molecular definition of breast tumor heterogeneity. Cancer Cell, 2007; 11(3): 259-273. 171. Lohr, M, Schmidt, C, Ringel, J, et al Transforming growth factor-beta1 induces desmoplasia in an experimental model of human pancreatic carcinoma. Cancer Res, 2001; 61(2): 550-555. 172. Roberts, AB, Sporn, MB, Assoian, RK, et al. Transforming growth factor type beta: rapid induction of fibrosis and angiogenesis in vivo and stimulation of collagen formation in vitro. Proc Natl Acad Sc i U S A, 1986; 83(12): 4167-4171. 173. Bagella, L, Sun, A, Tonini, T, et al. A small molecule based on the pRb2/p130 spacer domain leads to inhibition of cdk2 activity, cell cycle arrest and tumor growth reduction in vivo. Oncogene, 2007; 26(13): 1829-1839. 174. Rane, SG, Dubus, P, Mettus, RV, et al. Lo ss of Cdk4 expression cau ses insulin-deficient diabetes and Cdk4 activation results in beta-i slet cell hyperplasia. Nat Genet, 1999; 22(1): 44-52. 175. Malumbres, M, Hunt, SL, Sotillo, R, et al. Driving the cell cycle to cancer. Adv Exp Med Biol, 2003; 532(1-11. 176. Moustakas, DT, Lang, PT, Pegg, S, et al. Development and valid ation of a modular, extensible docking program: DOCK 5. J Comput Aided Mol Des, 2006; 20(10-11): 601619. 177. Magarian, EO and Nobles, WL New compounds : acrylonitrile deri vatives as potential antineoplastic agents. J Pharm Sci, 1969; 58(9): 1166-1167. 178. Berger, I, Fitzgerald, DJ, and Richmond, TJ Baculovirus expression system for heterologous multiprotein complexes. Nat Biotechnol, 2004; 22(12): 1583-1587. 179. Fitzgerald, DJ, Berger, P, Schaffitzel, C, Yamada, K, Richmond, TJ, and Berger, I Protein complex expression by using multigene baculovi ral vectors. Nat Methods, 2006; 3(12): 1021-1032. 180. Bartkova, J, Lukas, J, Muller, H, Lutzhoft, D, Strauss, M, and Bartek, J Cyclin D1 protein expression and function in human breas t cancer. Int J Cancer, 1994; 57(3): 353-361. 181. An, HX, Beckmann, MW, Reifenberger, G, Bender, HG, and Niederacher, D Gene amplification and overexpression of CDK4 in sporadic breast carcinomas is associated with high tumor cell pr oliferation. Am J Pathol, 1999; 154(1): 113-118. 119

PAGE 120

182. Kanoe, H, Nakayama, T, Murakami, H, et al. Amplification of the CDK4 gene in sarcomas: tumor specificity a nd relationship with the RB ge ne mutation. Anticancer Res, 1998; 18(4A): 2317-2321. 183. Slepchenko, BM and Terasaki, M Cyclin aggr egation and robustness of bio-switching. Mol Biol Cell, 2003; 14(11): 4695-4706. 184. Terasaki, M, Okumura, E, Hinkle, B, a nd Kishimoto, T Localization and dynamics of Cdc2-cyclin B during meiotic re initiation in starfish oocytes Mol Biol Cell, 2003; 14(11): 4685-4694. 185. McGovern, SL, Caselli, E, Grigorieff, N, and Shoichet, BK A common mechanism underlying promiscuous inhibitors from vi rtual and high-throughput screening. J Med Chem, 2002; 45(8): 1712-1722. 186. McGovern, SL, Helfand, BT, Feng, B, a nd Shoichet, BK A specific mechanism of nonspecific inhibition. J Med Chem, 2003; 46(20): 4265-4272. 187. Feng, BY, Simeonov, A, Jadhav, A, et al. A hi gh-throughput screen fo r aggregation-based inhibition in a large compound library. J Med Chem, 2007; 50(10): 2385-2390. 188. Prince, T, Sun, L, and Matts, RL Cdk2: a genuine protein kinase client of Hsp90 and Cdc37. Biochemistry, 2005; 44(46): 15287-15295. 189. Amann, R and Peskar, BA Anti-inflammatory ef fects of aspirin and s odium salicylate. Eur J Pharmacol, 2002; 447(1): 1-9. 120

PAGE 121

BIOGRAPHICAL SKETCH Patrick Corsino was born in Port of Spain, Trinidad and Tobago on October 30 th 1982. He received his high school education at St. Georges British Interna tional School in Rome, Italy. He graduated with a B.Sc. in medical biochemist ry from the University of Birmingham, UK, in 2004. After graduation he immediately began the Interdisciplinary Program in the College of Medicine at the University of Fl orida and received his doctoral tr aining in the lab of Brian Law. 121