The CCR’s Commitment to Partnerships and Sharing of Scientific Information

he CCR is committed to developing and sharing molecular research tools and scientific information within the NCI as well as the wider cancer research community. During the past 2 years, the CCR’s Office of Science and Technology Partnerships (OSTP) has developed new programs in collaboration with the private sector to advance these goals. Several new CCR-wide initiatives are identifying and characterizing cancer-relevant proteins and their binding partners. Many of these protein interactions regulate important cellular processes, including proliferation, differentiation, apoptosis, metastasis, cell cycle control, and angiogenesis, which are the key areas of investigation within the Intramural Research Program.

The OSTP has forged partnerships with private-sector organizations, such as Myriad Genetics, to facilitate the discovery of novel protein-protein interactions and cell-signaling networks and to generate high-affinity reagents to serve as research tools for functional characterization and molecular interrogation. The partnership with Myriad also enables CCR investigators to access Myriad’s automated process for identification of novel protein interactions based on the yeast two-hybrid (Y2H) methodology. In the first year of the program, 60 cancer-relevant genes of interest to CCR investigators were analyzed by Myriad’s unique, high-throughput system. To date, this large-scale effort between CCR and Myriad has led to the identification of more than 1,000 novel protein interactions, many of which are now being validated and confirmed as functionally relevant in mammalian cell systems.

A partnership with Becton Dickinson PharMingen and Rockland Immunochemicals was spearheaded to develop monoclonal and polyclonal antibodies against hundreds of key cancer-related targets of interest to CCR—at no cost to the investigator. As part of the partnership agreement, CCR investigators characterize the antibodies using biological systems available in their laboratories and provide the data to the companies for further development. The companies will market the antibodies to the public and cite the CCR investigators who validate them. To date, the program has accepted more than 200 targets for antibody development. Over 50 antibodies have been delivered to various groups, and the rest are at various stages of development. This example illustrates how the CCR can leverage the breadth and diversity of the Intramural Program to capitalize on relationships with the private sector that ultimately benefit the entire research community.

The integration of protein modification and interaction data is critical for the characterization of biochemical and genetic pathways, the elucidation of functional relationships, and the identification of novel molecular targets of cancer. The CCR is collaborating with the Advanced Biomedical Computing Center in Frederick (ABCC) and the NCI Center for Bioinformatics (NCICB) to develop a data-sharing, Web-based tool to integrate these data. This system will incorporate data related to genes and proteins being studied within the CCR. For example, it will integrate data generated from the Y2H and antibody partnerships, including confirmation of novel protein interactions in mammalian systems, validation of antibodies raised against newly identified targets, and protein interaction and modification data generated by mass spectrometry or other analytical methods. Future plans for this Web tool include links to a variety of CCR-generated and publicly available databases, such as RNAi resources, gene expression profiles, and pathways. The sharing of novel findings among CCR laboratories through the database will facilitate both the establishment of new collaborations and the exchange of technical expertise and resources. The resulting synergy and cooperation between various NCI groups will expedite the translation of basic research findings into new therapies, diagnostics, and preventative agents. Information related to all of the programs managed by the OSTP can be found on the CCR Web site.

Robert H. Wiltrout, PhD
Director

Fusion Gene Transcripts in Expressed Sequence Tags Database

Hahn Y, Bera TK, Gehlhaus K, Kirsch IR, Pastan IH, and Lee B. Finding fusion genes resulting from chromosome rearrangement by analyzing the expressed sequence databases. Proc Natl Acad Sci U S A 101: 13257–61, 2004.

he creation of fusion genes by chromosome translocation is a common feature of human cancer cells. The gene fusion often disrupts the normal regulation of the genes involved. It may result in overexpression of an oncogene, inactivation of a tumor suppressor gene, or production of altered protein with modified function. Several specific fusion genes are known to be responsible for hematologic disorders. The BCR/ABL1 fusion gene, for instance, is found in more than 90% of patients with chronic myelogenous leukemia. Evidence is emerging that fusion genes are also important in epithelial carcinogenesis.

Chromosome translocations can be discovered by cytogenetic experiments, but it is difficult to tell if a fusion gene has been created by the translocation and, if so, to identify it. Here we describe a procedure for identifying fusion genes by an analysis of the expressed sequence tags (EST) database. ESTs are short (~500 bp) sequences of randomly selected cDNAs prepared from a variety of tissues. The current database holds more than 6 million human ESTs, about half of which are from cancer tissues or derived cancer cell lines. The ESTs from fusion genes in this database can be identified because they map to two different locations in the human genome. A complicating factor is that many such chimeric transcripts in the EST database are cloning artifacts generated during the cDNA library construction process. However, these can be separated from genuine fusion gene transcripts because the fusion point usually occurs in an exon for the former, whereas it usually occurs at an exon-exon boundary for the latter.

We developed a semi-automatic procedure for systematic identification of fusion gene transcripts in the mRNA and EST databases based on these principles. Using this procedure, we could identify 118 mRNAs and 196 ESTs as fusion gene transcript sequences, from a total of 237 putative fusion genes. Among the mRNA sequences, 96 were previously annotated as fusion transcripts, including most of the BCR/ABL1 fusion transcript sequences.

The procedure also identified 177 novel fusion gene candidates. We experimentally verified one of these, the IRA1/RGS17 fusion, which was supported by three independent EST clones (Figure 1). A reverse transcriptase (RT)-PCR experiment using an mRNA sample from the MCF7 breast cancer cell line yielded a clear band with the correct size. A fluorescence in situ hybridization (FISH) experiment using two BAC clones containing IRA1 and RGS17 genes, respectively, detected a derivative chromosome, most likely the previously identified t(3;6)(q26;q25)del(3)(p14). The 5´-UTR exon 1 of IRA1 on 3q26.32 is fused with the start codon–bearing exon 2 of RGS17 on 6q25.2. The RGS17 protein is a member of the GTPase-activating proteins that act as regulators of G-protein signaling. Components in the G-protein–coupled receptor-signaling pathways, including RGS proteins, are known to be involved in many cancers and considered as potential therapeutic targets in cancer therapy.

Figure 1. Prediction and verification of the IRA1/RGS17 fusion resulting from a chromosome translocation. A) Schematic representation of the IRA1/RGS17 fusion. Boxes represent the exons, and broken lines the introns. The fusion event is indicated by an arc. Arrows indicate the transcription start sites. Exons are numbered as they occur in the original genes. Primers for the reverse transcriptase (RT)–PCR reaction are indicated (T530 and T531). ORFs (open reading frames) are marked with grey boxes. B) RT-PCR detection of the fusion transcripts in MCF7 cells. The fusion gene transcripts for the previously known BCAS4/BCAS3 and the predicted IRA1/RGS17 fusions were detected in the cells. The β actin (ACTB) was used as the positive control. The product sizes of ACTB, BCAS4/BCAS3, and IRA1/RGS17 are 600, 328, and 367 bp, respectively. C) Detection of the 3;6 translocation in MCF7 cells by a fluorescence in situ hybridization (FISH) experiment. A representative result is presented. The IRA1 gene (red) and the RGS17 gene (green) are on the chromosomes 3 and 6, respectively. Besides two copies each of chromosomes 3 and 6, a 3;6 translocation was detected (white arrow).

We expect to collect more fusion gene candidates in the future as the EST database continues to expand. A large collection of cancer-related gene fusions, attained through a combination of computational prediction and experimental verification, should present a new opportunity to uncover novel molecular mechanisms of carcinogenesis.

Yoonsoo Hahn, PhD
Visiting Fellow
hahny@mail.nih.gov

Byungkook Lee, PhD
Principal Investigator
Laboratory of Molecular Biology
NCI-Bethesda, Bldg. 37/Rm. 5120A
Tel: 301-496-6580
Fax: 301-480-4659
bk@nih.gov

Why Is DNA Like a Plumber’s Snake?

Kouzine F, Liu J, Sanford S, Chung HJ, and Levens D. The dynamic response of upstream DNA to transcription-generated torsional stress. Nat Struct Mol Biol 11: 1092–100, 2004.

he double helix must be untwisted to decode or copy the genetic information embedded in DNA sequences. As one strand is locally rotated about the other, torsional stress is inevitably generated. In topological domains where the total number of helical turns is fixed, which may occur by a number of means such as protein-protein interactions between DNA-bound factors, torsional stress accumulates as DNA is unwound, causing the double helix to coil back upon itself to form “supercoils” (Figure 1, part A). Because double-stranded DNA is intrinsically very stiff, unless dissipated, the supercoiling forces stored in DNA may rise to levels that alter its structure or impede the enzymatic machineries that conduct genetic business (i.e., transcription, replication, recombination, and repair). Therefore, special enzymes, termed “topoisomerases,” cut one or both DNA strands to allow rotations that release torsional stress and then reseal the cuts (Figure 1, part A, middle).

Figure 1. A) When a DNA fiber is wrapped around itself, with its ends restrained (in this case, the ends are fixed to each other forming a circle), supercoils are formed and trap torsional stress (top). The only way to remove the stress is to break the DNA and let the ends counter-rotate (middle). We wondered what happens dynamically if stress is applied to an open DNA fiber. If the DNA is rigid enough, the whole molecule would rotate as a unit and no supercoils would form, or if the DNA is flexible enough, the whole fiber would writhe around itself in response to applied torque (bottom). B) Linear, open molecules with divergent promoters were transcribed in vitro; concurrent recombination between loxP sites bracketing the interposed DNA, which was excised as closed circles, trapped any supercoils residing in or transiting through the interposed segment at the instant of recombination. Transcription generates torque as the double helical template is threaded through the RNA polymerase active site. Without transcription (Trx), no stress would be captured; the number of transient supercoils captured was expected to reveal how rigid or flexible the DNA was. C) A sample result: ongoing transcription (+) traps a large number of supercoils. Without transcription (–) very few supercoils are trapped. D) Factors recognizing dynamic changes in DNA structure resulting from transcriptional torque provide the necessary effector components to construct a molecular “cruise control.” FUSE, far upstream element; FBP, FUSE binding protein; FIR, FBP interacting repressor; loxP, target sequences for the site-specific Cre recombinase. T3 and T7 indicate bacteriophage T3 and T7 promoters.

Although supercoiling forces often modify gene expression in prokaryotes, in metazoans, the capacity of vast stretches of non-coding DNA to absorb this stress and abundant topoisomerase activity have been presumed to mitigate the influence of supercoiling on gene regulation. Moreover, attempts to measure the stable level of supercoils per unit length of DNA (the superhelical density, σ) suggest that torsional stress does not accumulate to high levels in the DNA of higher eukaryotes.

We wondered, however, what does DNA (in this case linear DNA that cannot hold onto supercoils) look like dynamically while it is being transcribed? Does it writhe like a plumber’s snake being whipped about, or do the supercoils run off the ends of the DNA so rapidly that the template is relaxed and unstressed (Figure 1, part A, bottom)?

The experimental problem was to trap the evanescent stresses propagating from an activated promoter in linear DNA. These dynamic supercoils had to be captured and preserved during transcription for future study, because active RNA polymerase translocating along the template was the kinetic engine cranking the DNA. The trick was to convert the dynamic supercoils into stable, conventional supercoils. To accomplish this, a 1-kilobase (kb) (100 double helical turns) segment of DNA was interposed between two similarly oriented loxP sites, target sequences for the site-specific Cre-recombinase (Figure 1, part B). This loxP-bracketed segment was in turn placed upstream of a single phage RNA polymerase promoter or was inserted between two divergently transcribed phage promoters, all in linear DNA fragments. With the latter arrangement, the dynamic supercoils from each promoter would be expected to be mutually reinforcing. Upon addition of Cre, site-specific recombination between the loxP sites was expected to excise a 1-kb, covalently closed DNA circle, trapping any supercoils residing in or transiting through the segment at the instant of recombination (Figure 1, part B). Two-dimensional electrophoresis of the 1-kb circles recovered from these reactions to separate the DNA rings into a series of spots, each differing from its neighbor by a single supercoil, promised to give an accurate accounting of dynamic supercoils generated during transcription (Figure 1, part C).

In the absence of transcription, recombination trapped no more than the two supercoils explained by thermal wriggling of DNA. As the transcription intensity was increased, the circles trapped more and more supercoils. As many as 14 supercoils were captured in the 1-kb segment; σ = 0.14, an incredibly high number (Figure 1, part C). What are the biological consequences of σ rising transiently to such a high level?

At high levels of σ, DNA melts at susceptible sites. This melting does not occur gradually, but at critical thresholds, the susceptible segments, “soft spots” in the DNA, buckle. In fact, from 2-dimensional gel electrophoresis, chemical modification, and nuclease hypersensitivity assays, we show that the far upstream element (FUSE) from the human c-myc gene pops open during transcription initiated at downstream promoters. Depending on the σ, melted FUSE binds an activator, the FUSE binding protein (FBP), and a repressor, the FBP interacting repressor (FIR). Thus, these proteins superimpose effector function on the stress-sensor properties of FUSE and, in principle, create a mechanical device for the real-time regulation of transcription (Figure 1, part D). Real-time regulation is likely to be especially important for genes yielding short half-life, low abundance transcripts, such as from c-myc and perhaps other protooncogenes, tumor suppressors, and cell cycle regulators.

Focal melting of DNA may have several other important consequences: (1) Melted DNA is much more flexible than duplex, so a melted segment may help to juxtapose widely separated elements and their associated factors. (2) Dynamic supercoiling may energetically assist chromatin remodeling and modification. (3) Propagation of dynamic supercoils from one gene to another, in principle, allows the activity of one gene to modulate directly and immediately the activity of a closely situated promoter. Knowledge of the transmission of mechanical forces through DNA may help us to understand chromosome architecture and to devise strategies for the precise control of genetic processes.

Fedor Kouzine, PhD
Visiting Fellow
kouzinef@mail.nih.gov

David Levens, MD, PhD
Senior Investigator
Laboratory of Pathology
NCI-Bethesda, Bldg. 10/Rm. 2N105
Tel: 301-496-2176
Fax: 301- 594-5227
levens@helix.nih.gov

Structural Studies of Rio2, an Atypical Serine Kinase Required for Ribosome Biogenesis

LaRonde-LeBlanc N and Wlodawer A. Crystal structure of A. fulgidus Rio2 defines a new family of serine protein kinases. Structure 12: 1585–94, 2004.

ibosome production is fundamental to cellular proliferation and therefore to tumorigenesis. Increased nucleolar size, which corresponds to increased ribosomal RNA (rRNA) production, has long been accepted as a hallmark of tumor cells. rRNA processing is a stepwise process that requires several non-ribosomal factors. Yeast Rio1, one such factor, is the founding member of the RIO kinase family. Yeast Rio1 is an essential gene for proper cell cycle progression and chromosome maintenance in addition to rRNA processing (Angermayr M et al. Mol Microbiol 44: 309–24, 2002; Vanrobays E et al. Mol Cell Biol 23: 2083–95, 2003). Sequence alignments have demonstrated that members of two RIO subfamilies, Rio1 and Rio2, are represented in organisms across the biological spectrum. The two subfamilies are distinct in several highly conserved regions of the catalytic domain, known as subdomains—which have been shown to be important for the fold of the domain, or for ATP binding and phosphoryl transfer catalysis. For example, Rio1 and Rio2 each contain a distinct nucleotide-binding loop in what is known as subdomain I. In addition, the Rio2 subfamily is characterized by a conserved N-terminal domain that is not present in Rio1.

The RIO kinases are interesting molecules for several reasons. They are ancient essential proteins, and it has been suggested that they represent an evolutionary link between prokaryotic lipid kinases and eukaryotic protein kinases (ePKs). They are very divergent in sequence from known protein kinases and lack many of the sequence features required for the function of ePKs. They are rRNA processing factors with kinase activity and, thus, are attractive potential targets for therapeutic intervention. To understand the nature of these enzymes and their relationships to ePKs, we determined the X-ray crystal structure of Rio2 from Archaeoglobus fulgidus, a hyperthermophilic archaeal organism. Despite a lack of significant sequence homology to ePKs, the RIO catalytic domain bears striking structural homology to the canonical kinase domain (Figure 1). The catalytic domain of kinases such as protein kinase A (PKA) contains 11 subdomains. The catalytic domains of members of the RIO kinase family contain only eight, and these subdomains contain variations that produce differences in phosphate binding, and perhaps substrate recognition and catalysis as well. The most significant difference is the complete absence of subdomain VIII, which is known as the APE (Ala-Pro-Glu), or activation loop. In ePKs, the APE loop is often phosphorylated to regulate the activity of the kinase and provides much of the surface for binding to substrate peptide. The lack of this region in the catalytic domain of Rio2 and other RIO kinases leaves open the question of how RIO kinases bind their substrates. In addition, we found that the N-terminal Rio2–specific domain adopts a winged helix fold, which is commonly used by proteins for nucleic acid interactions.

Figure 1. The structure of atypical serine kinase Rio2. A ribbon illustration of the three-dimensional structure of the Rio2-ATP-Mn complex showing the secondary structure elements and the various domains. DFG, Asp-Phe-Gly.

Our studies provided a detailed description of the ATP-binding pocket of the Rio2 proteins. To determine the mode of ATP binding for the Rio2 kinase, we soaked crystals of the Rio2 protein in a solution containing an ATP analog, AMPPNP, and Mn2+ ions. Divalent cations are known to be required for catalytic activity, and RIO kinases retain the metal-binding residues. Although we were able to observe AMPPNP bound in the active site, we did not find any metal ions. We therefore concluded that the conformation of the complex was not a productive one, and that the confines of the crystal did not allow for the conformational changes required for proper binding of ATP and metal ions. More recent studies confirmed this conclusion. Crystals of the Rio2-ATP-Mn2+ complex assembled before crystallization contain the ATP molecule as well as two Mn2+ ions in the active site. We now know that ATP binds in the RIO kinase domains in a conformation that is different from canonical ePKs. Due to this difference, the γ-phosphate is located in a position that would require further conformational changes in the Rio2 protein for substrate access. These findings support our conclusion that the peptide substrate binding by RIO kinases must differ significantly from that of canonical serine kinases.

Because the RIO kinases lack the classical peptide substrate–binding loops and bind ATP in a distinct conformation, it is likely that their inhibitors will be very distinct from the inhibitors of classical protein kinases. Our investigations should be helpful in guiding the efforts to develop such inhibitors, based on the structures of the enzymes from each RIO subfamily with bound nucleotides, as well as on the planned structural studies of their complexes with peptides. These discoveries have implications for the role of RIO proteins in rRNA cleavage and provide a framework for experiments to determine their target and function. In addition, this work defines, in structural terms, a novel family of atypical protein serine kinases.

Nicole LaRonde-LeBlanc, PhD
Postdoctoral Fellow
Macromolecular Crystallography Laboratory
NCI-Frederick, Bldg. 539/Rm. 145
Tel: 301-846-5326
Fax: 301-846-5326
nicole@ncifcrf.gov

Telomere Protection Without a Telomerase: The Role of Drosophila ATM and Mre11 in Telomere Maintenance

Bi X, Wei SC, and Rong YS. Telomere protection without a telomerase; the role of ATM and Mre11 in Drosophila telomere maintenance. Curr Biol 14: 1348–53, 2004.

he conserved ATM checkpoint kinase and Mre11 DNA repair protein participate in telomere maintenance for both yeast and mammalian cells. We studied their roles at Drosophila (Drosophila melanogaster) telomeres, which are not maintained by a canonical telomerase, in the hope that our studies would shed light on their telomerase-independent function in telomere protection.

In most organisms studied, the ends of a chromosome are elongated by telomerase, which adds short repeats. In yeast, atm or mre11 mutants have shortened telomeric repeats, implicating abnormal telomerase activity. Previous studies have been largely focused on how ATM and Mre11 regulate telomerase activity. If their function is solely to regulate telomerase activity, one would predict that they have minimal telomere maintenance function for organisms that employ alternative telomere-elongating mechanisms, such as D. melanogaster. Telomeres of Drosophila are enriched with retrotransposons. It is believed that new copies of these transposons elongate Drosophila chromosomes. In addition, artificial chromosome ends that lack all transposons have been created and stably maintained in the lab. It is thus apparent that not only are Drosophila telomeres not elongated by a telomerase, the normal telomere function in Drosophila does not even contain a specific DNA element. We tested whether Drosophila ATM or Mre11 participates in telomere maintenance.

We created knockouts of the fly atm and mre11 by using a novel homologous gene targeting method. We showed that both mutations disrupt development and cause lethality in Drosophila. We examined mutant tissues cytologically and discovered a severe telomere defect for both mutants: multiple telomere fusions were observed in all the mitotically active cells. On average, about 20% of the telomeres in a nucleus were engaged in fusions, an extent that is greater than any non-Drosophila case reported. Telomere fusions led to a vicious “fusion-bridge-breakage” cycle in both mutants, which is similar to the one first described in maize by McClintock: telomere fusions lead to the formation of chromosome bridges that connect the separating sister nuclei. These bridges sometimes break creating new broken ends that can subsequently fuse with other ends, including telomeres. In both mutants, this devastating cycle led to widespread genome instability in the forms of chromosome breakage, genome rearrangements, and gross aneuploidy. By double mutant analyses, we suggest that Drosophila ATM and Mre11 function in the same telomere-protecting pathway.

Therefore, our findings soundly disprove the hypothesis that Drosophila ATM and Mre11 have minimal telomere-protecting function. Furthermore, we propose the existence of a telomerase-independent function for these proteins that is likely conserved from yeast to human. Our study raises at least two new questions. First, given that similar mutants in yeast cause no telomere fusion, and thus no loss of cell viability, and that human and mouse atm mutants are also viable with much milder telomere dysfunction, why are Drosophila telomeres especially susceptible to telomere fusion? We propose that “popular telomeres” have one additional layer of protection than Drosophila telomeres, which is conferred by the telomerase itself. Not only can telomerase serve as a physical barrier to prevent fusion, it installs multiple binding sites for various telomeric repeat binding proteins, which can also cap the telomeres, preventing them from being repaired as double strand breaks (DSBs). Since Drosophila lack this layer of telomere protection, the function of ATM and Mre11 at the telomeres becomes essential for organismal survival. Secondly, what is the nature of this telomerase-independent mechanism that we propose as conserved throughout evolution? As a first step toward definitely answering these questions, we identified a separate telomere-protecting pathway that is regulated by the ATM-related ATR kinase. This pathway is partly redundant to the one controlled by ATM so that in cells deficient for both ATM and ATR, all telomeres become unprotected and susceptible to fusion (Bi X et al. Proc Natl Acad Sci U S A 102: 15167–72, 2005). Our current model is that ATM and ATR protect telomere integrity by safeguarding chromatin architecture that favors the loading of telomere-elongating and capping proteins.

Yikang S. Rong, PhD
Principal Investigator
Laboratory of Molecular Cell Biology
NCI-Bethesda, Bldg. 37/Rm. 6056
Tel: 301-451-8335
Fax: 301-435-3697
rongy@mail.nih.gov

Scientific Advisory Committee

If you have scientific news of interest to the CCR research community, please contact one of the scientific advisors (below) responsible for your areas of research.

CCR Frontiers in Science—Staff

Center for Cancer Research

Robert H. Wiltrout, PhD, Director
Lee J. Helman, MD, Acting Scientific Director for Clinical Research
Frank M. Balis, MD, Clinical Director
L. Michelle Bennett, PhD, Associate Director for Science

Deputy Directors

Douglas R. Lowy, MD
Jeffrey N. Strathern, PhD
Lawrence E. Samelson, MD
Mark C. Udey, MD, PhD

Editorial Staff

Tracy Thompson, Editor-in-Chief
Sue Fox, BA/BSW, Senior Editor
Lamont Williams, Managing Editor *
Ave Cline, Editor
Terry Taylor, Copy Editor *
Emily R. Krebbs, MA, Copy Editor *
Amy Schneider, Copy Editor *
Rob Wald, Publications Manager *
Michael Fleishman, Graphic Artist *
Yvonne Bersofsky, Web Developer *

* Palladian Partners, Inc.