The CCR’s Commitment to Health Disparities Research

participated in the two-day NCI Wide Workshop on Cancer Health Disparities at the Natcher Conference Center, organized and co-chaired by CCR’s Associate Director, Dr. L. Michelle Bennett. The workshop drew broad participation from NCI staff. Attending were a diverse cadre of health disparities scientists and administrators and senior leadership from the NCI intramural and extramural divisions. I heard wide-ranging perspectives on minority health and health disparity research that helped identify potential gaps, challenges, and opportunities across the broad discovery-development-delivery continuum. What was most evident in the workshop is the breadth and depth to which NCI divisions, offices, and centers are engaged in health disparities research focused on reducing the undue burden of cancer on the minority populations in our country as well as globally. There was a general agreement among the participants that we need to communicate more effectively within and across divisions, and with the external cancer community, to share information about current efforts, develop collaborations around future opportunities, and leverage strengths to reduce health disparities.

Basic science discoveries within the CCR are having a positive impact on many health issues that disproportionately affect minority or disadvantaged populations. Highly active antiretroviral therapy (HAART), pioneered by our intramural investigators, has reduced the death rate from AIDS dramatically, and today it is used worldwide. Another discovery positively influencing global health is an effective vaccine against human papillomavirus (HPV). Original discovery and development research by CCR investigators led to a prophylactic HPV vaccine that may ultimately eliminate cervical cancer, a significant health problem among underserved women in the developing world. CCR scientists today are also studying a number of cancers that disproportionately affect minority populations in the United States (for example endometrial, liver, lung, and prostate cancers).

In addition to work in the laboratory, we are also reaching out to the local community to share our expertise, knowledge, and resources. For example, we accepted an invitation from the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) to share clinic space and provide cancer care at the Upper Cardozo Clinic in Washington, D.C., to a largely Hispanic and African American community. Through this effort, we hope to facilitate access to clinical trials, both at the NIH clinical center and in the community. Over time we would like to establish a broad network of community physicians, leaders, and universities. In another effort, we organized a program to educate patient navigators from NCI-funded Patient Navigator programs about clinical trials and opportunities for accessing them, both in their communities as well as at the NIH clinical center. Patient navigators serve as resources for medical information, including information about medical services available to the communities they serve. In conjunction with their diverse roles and responsibilities, it is hoped that this training will improve access to clinical trials in these populations. Also, the Cancer Research Interns in Residence (CRIR) program provides a training opportunity for minority researchers in our intramural program. The CRIR was inaugurated in 2004, and in two summers has recruited about 100 fellows.

Going forward, we will seek to increase awareness within the CCR community about the spectrum of health disparities research/initiatives occurring not only within our program, but also more broadly within the NCI. We will encourage collaborations and partnerships with other NCI divisions, offices, and centers. And by leveraging our strength in translational research using multidisciplinary approaches, we will contribute to reducing cancer health disparities in minority and underserved populations.

Robert H. Wiltrout, PhD
Director

Finding New Approaches to Attacking the Latent Reservoir of HIV-Infected Cells

Krishnan V and Zeichner SL. Host cell gene expression during human immunodeficiency virus type 1 latency and reactivation and effects of targeting genes that are differentially expressed in viral latency. J Virol 78: 9458–73, 2004.

he persistence of latent reservoirs of HIV-infected cells is a critical issue in HIV disease. Highly active antiretroviral therapy (HAART) can control but not eliminate HIV infection, due to the long-lived reservoirs. One potential approach to addressing this problem involves treatment with agents that activate HIV from latency, while blocking new rounds of infection with HAART. However, no clinically useful, highly effective agents capable of depleting HIV reservoirs have been developed. Most known agents are highly toxic and/or incompletely effective. To construct improved reservoir depletion strategies, we must better understand the mechanisms that control HIV latency. Recent findings suggest that model systems may enable the systematic identification of cellular genes involved in latency maintenance and that this information can inform the search for new agents that target the products of those differentially expressed genes and activate HIV replication in latently infected cells.

In an expression-profiling study for cellular genes differentially expressed during HIV replication, we found that host-cell gene expression changed in an ordered, temporally dependent pattern, and some of the differentially expressed genes had known relationships to important aspects of the viral replication cycle. However, we also unexpectedly found that the expression patterns of HIV latently infected cell lines differed from those of their uninfected parental cells. The cellular gene expression differences in the latently infected cell lines likely arose during their production and cloning. The production of cloned cell lines typically yields derivative cell lines that vary in expression patterns from the parental lines. These variations may be informative, because during production of the latently infected cells, there is likely selection for the differential expression of sets of cellular genes that help maintain the virus in latency (Figure 1, part A). We further hypothesized that targeting the products of the differentially expressed genes would activate HIV replication in the latently infected cell lines and potentially in the latently infected cells of HIV patients. The genes that were differentially expressed in the latently infected cell lines included genes that were already known to be involved in HIV replication and the control of HIV latency, supporting the hypothesis.

Figure 1. A) Model to account for the differences in gene expression in latently infected cell lines compared with their uninfected parental cell lines. During the production of an HIV latently infected cell line, HIV is added to cells, which are then maintained in culture for a long period of time. During this maintenance in culture, there is selection for cells that can maintain HIV in latency (latency-favoring cellular environment [LFCE] cells), because the large majority of cells that do not maintain HIV in latency die (replication-favoring cellular environment [RFCE] cells). The remaining, living cells are then subjected to limiting dilution cloning to produce HIV latently infected cell lines. These cell lines are further characterized to show that they contain an HIV provirus that can be induced into active viral replication, and the cells’ gene expression profiles are determined and compared with the gene expression profiles of the uninfected parental cells. Such studies have revealed significant differences in the expression profiles of the latently infected cells compared with the uninfected parental cells. Studies to determine whether different HIV latently infected cell lines, produced using the same host cell and cloned virus, have similar or different patterns of cellular gene expression, are in progress. B) Cellular genes differentially expressed in an HIV latently infected cell line and the effect on HIV latency of targeting products of the differentially expressed genes. The figure shows results for two agents, clastolactacystin-β-lactone, a proteasome inhibitor, and resveratrol, an upstream activator of transcription factor Egr1. In the top left of each panel is a color map of the expression profile of the gene(s) in the latently infected cells compared with the uninfected parental cell line before (No) and after (in hours) activation into lytic replication. (Green indicates relative expression lower than the uninfected control; red indicates relative expression higher than the uninfected control.) The proteasome genes were overexpressed, and Egr1 was underexpressed in the latently infected cells compared with their uninfected parental cells before induction. The bar graphs show the fold change in HIV p24 antigen, a marker for viral replication, following treatment with the indicated agent. Tumor necrosis factor-α (TNF-α) serves as a positive control agent. All experiments except the “no AZT” data were produced from cells that were also treated with AZT to ensure that HIV p24 antigen production resulted only from reactivated latent infection and not subsequent rounds of viral replication. Clastolactacystin-β-lactone and resveratrol activated HIV replication in the latently infected cells. These agents, therefore, represent two new classes of agents that can activate HIV replication in at least some latently infected cell lines.

To test the hypothesis that targeting cellular genes differentially expressed in HIV latently infected cells would activate HIV replication, we paid special attention to genes whose products could be targeted by available small molecule inhibitors. Figure 1, part B shows the results obtained using two different inhibitors targeting host cell genes that were differentially expressed in the latently infected cell lines. We found that genes encoding several proteasome components were upregulated in latently infected cells and showed that treating the cells with a proteasome inhibitor, clastolactacystin-β-lactone (CLC), activated HIV replication. We found that transcription factor Egr1 was downregulated in the latently infected cells and showed that targeting Egr1 with an upstream activator, resveratrol, activated HIV replication.

HIV latently infected cell lines, therefore, appear to constitute model systems that enable the discovery of host cell genes involved in the maintenance of latency; studying these host cell genes can identify new agents that eject HIV from latency. These new agents may also be able to eject HIV from latency in the latently infected cells of HIV patients, which could lead to new ways to attack the latent reservoir of HIV infected cells in vivo.

Additional Reading
Williams SA and Greene WC. Trends Microbiol 13: 137–9, 2005.

Steven L. Zeichner, MD, PhD
HIV and AIDS Malignancy Branch
Principal Investigator
NCI-Bethesda, Bldg. 10/Rm. 10S255
Tel: 301-402-3637
Fax: 301-480-8250
zeichner@nih.gov

The Slinky as a Ubiquitous Pathogen Recognition Structure

Bell JK, Botos I, Hall PR, Askins J, Shiloach J, Segal DM, and Davies DR. The molecular structure of the Toll-like receptor 3 ligand-binding domain. Proc Natl Acad Sci U S A 102: 10976–80, 2005.

hen considering antigen recognition, antibodies and T-cell receptors, the receptors of the adaptive immune system, typically come to mind. However, immunologists have known for some time that other, innate forms of antigen recognition must exist, since infectious agents are held in check prior to the development of adaptive immune responses. The most dramatic demonstration of the innate response is the ability of immunodeficient mice that lack antibodies or T-cell receptors to survive in non-sterile environments. In both immunodeficient and normal mice, pathogen invasion results in the immediate recruitment of phagocytes and other immune cells that ingest the pathogen, produce toxic substances that kill it, or both. So how, in the absence of T cells and antibodies, are these pathogens recognized? Recently, a family of homologous proteins known as the Toll-like receptors (TLRs) was shown to serve just such a pathogen-recognition function. The TLRs were discovered as homologs of the Drosophila receptor Toll, an essential component of the immune response to fungi in flies, and it is now known that similar molecules serve immune functions throughout the animal and plant kingdoms. In humans, 10 TLR paralogs recognize a wide variety of “pathogen-associated molecular patterns” (PAMPs), including lipids, proteins, carbohydrates, and nucleic acids from bacteria, parasites, and viruses. We asked how only 10 germ-line–encoded molecules are able to recognize such a wide variety of structures at the molecular level.

The TLRs are type I integral membrane receptors, each consisting of an N-terminal extracellular PAMP-binding domain, a single transmembrane helix, and a C-terminal, cytoplasmic signaling domain. Our approach was to express large amounts of the extracellular domains (ECDs) of each TLR for X-ray crystallographic analysis and ligand-binding studies. In the paper cited above, we presented the first crystal structure of a TLR ECD, the unliganded form of TLR3-ECD. TLR3 responds to dsRNA from viruses, and we found that purified TLR3-ECD protein binds pI:pC (a dsRNA surrogate) in solution.

The TLR3-ECD consists of 23 tandem repeats of a motif known as the leucine-rich repeat (LRR). In three-dimensions, each LRR forms a loop, with consensus hydrophobic residues pointing inward, forming a stabilizing hydrophobic core (Figure 1, part A). When hooked together, the LRRs create a large solenoid in the shape of a horseshoe; overall, the TLR3-ECD can be described as a 23 turn “slinky” (Figure 1, part B). The concave inner surface consists of a large parallel β-sheet, with each β-strand roughly perpendicular to the solenoid axis and linked to the next strand by an irregular loop. LRR12 and LRR20 contain large insertions (Figure 1, parts B and C, highlighted in red). Since these insertions are unique to TLR3 and are conserved in all known mammalian TLR3 orthologs, they likely play important roles in TLR3 function, perhaps in ligand binding. The molecular surface of the TLR3-ECD is abundantly and unevenly populated with N-linked carbohydrates. However, one surface of the ECD is devoid of carbohydrate and free to interact with either ligand or another protein molecule (Figure 1, part C). In the absence of a TLR3-dsRNA complex structure, we can only speculate where ligand binding occurs. However, the presence of bound sulfate molecules from the crystallization medium (Figure 1, parts B and C) provides clues. The sulfate ions mimic the phosphate residues from the nucleotide backbone of a dsRNA molecule, indicating areas of the receptor that are capable of recognizing ligand.

Figure 1. Structure of the Toll-like receptor 3 (TLR3) extracellular domain (ECD) and model of the full-length receptor. A) A single leucine-rich repeat (LRR) loop highlighting conserved hydrophobic side chains (brown spheres) that form the core of the solenoid. B) A cartoon trace showing the curved solenoid, or “slinky” shape of the ECD. β-strands are shown as arrows on the concave surface of the ECD. C) A surface rendering of TLR3. In B and C, glycans are shown in green, sulfate ions in orange, and insertions in LRRs 12 and 20 in red. Transmembrane and cytoplasmic domains, based on previously reported structures, are shown in gray.

The binding of PAMPs by TLRs triggers inflammatory processes that can have either beneficial or detrimental consequences. Understanding how the recognition of pathogens by TLRs occurs should aid in the development of TLR agonists and antagonists for use as adjuvants in vaccine development, or as anti-inflammatory drugs.

This project is a collaboration between the laboratories of David Segal, PhD, Experimental Immunology Branch/National Cancer Institute (NCI), and David Davies, PhD, Laboratory of Molecular Biology/National Institute of Diabetes & Digestive & Kidney Diseases (NIDDK), with help from an intramural biodefense award from the National Institute of Allergy and Infectious Diseases (NIAID).

David M. Segal, PhD
Principal Investigator
Experimental Immunology Branch
NCI-Bethesda, Bldg. 10/Rm. 4B36
Tel: 301-496-3109
Fax: 301-496-0887
dave_segal@nih.gov

Modifying Chromatin to Protect the Genome

Shroff R, Arbel-Eden A, Pilch D, Ira G, Bonner WM, Petrini JH, Haber JE, and Lichten M. Distribution and dynamics of chromatin modification induced by a defined DNA double-strand break. Curr Biol 14, 1703–11, 2004.

ells respond to a double-strand break (DSB) in their DNA by phosphorylating chromatin in a large region surrounding the break site. In post-replicative cells, this modification promotes the de novo deposition of cohesin, a multiprotein complex that is normally loaded onto chromosomes during replication. DSB-induced cohesin loading is likely to tether break ends close to the sister chromatid, facilitating repair and helping the cell to maintain genome integrity.

DSBs induce the rapid phosphorylation of the H2AX isoform of histone H2A to form γH2AX (Rogakou EP et al. J Biol Chem 273: 5858–68, 1998), which is thought to play an important role in break repair (Fernandez-Capetillo O et al. DNA Repair 3: 959–67, 2004). The study cited at the top of this article (Shroff R et al. Curr Biol 14, 1703–11, 2004) and a second study (Ünal E et al. Mol Cell 16: 991–1002, 2004) provide a picture of one way that γH2AX protects the genome from damage.

Shroff R et al. used chromatin immunoprecipitation (ChIP) to probe γH2AX formation and recruitment of the repair protein Mre11p at a site in the budding yeast genome where breaks can be formed in a controlled manner. The relative contribution of the two yeast damage-response kinases, Tel1p (ATM homolog) and Mec1p (ATR homolog), to H2AX phosphorylation was also determined.

A panel of mutants in DNA damage response/repair genes was used to show that both Tel1p and Mec1p phosphorylate H2AX. Mutants blocking steps further down the DNA response/repair pathway had no effect on γH2AX formation, confirming that γH2AX formation is part of the initial DNA damage response. In the G1 phase of the cell cycle, Tel1p was responsible for most γH2AX formation, a finding similar to those obtained in studies of mammalian cells.

ChIP analysis showed that γH2AX and Mre11p occupy distinct regions around the induced DSB. Mre11p, like other repair proteins, bound to sites directly adjacent to the DSB (within 1–2 kb). Conversely, γH2AX was present in a 40–50 kb region surrounding the break site. γH2AX was most abundant in a 3–5 kb band on either side of the break, with significant levels up to 25 kb from the peak site. Remarkably, very little γH2AX was detected at sites within 1–2 kb of the DSB, although ChIP analysis showed that histones were still present in this interval (Figure 1).

Figure 1. γH2AX recruits cohesin to damage sites. Top—experimental data showing the broad region where γH2AX (red) forms; in contrast, repair proteins (here Mre11p, blue) bind in a narrow region. Bottom—nucleosomes containing γH2AX (red) recruit cohesin complexes (green rings), which use the unbroken sister chromatid to “splint” broken ends together, while leaving the ends themselves free for repair. ChIP, chromatin immunoprecipitation; DSB, DNA double-strand break.

The disparity between the location of repair proteins and of γH2AX indicates that γH2AX most likely does not directly recruit repair proteins to DNA damage sites. Instead, we suggested that this large region of γH2AX creates chromosome structural changes that promote damage repair.

Ünal E et al. (Mol Cell 16: 991–1002, 2004) and Ström L et al. (Mol Cell 16: 1003–15, 2004) provide support for this idea by examining the distribution of cohesin around a DSB. Cohesin is normally loaded onto chromosomes during replication, and holds sister chromatids together until mitosis. These papers report that DSBs provoke post-replicative cohesin loading in a large region, and that this additional cohesin is important for efficient DSB repair. The distribution of cohesin closely resembles that of γH2AX, suggesting that γH2AX might play a role in this damage-induced cohesin loading. Ünal E et al. showed that this is, in fact, the case. Mutants unable to form γH2AX do not recruit cohesin to DSBs and, consequently, have defects in repairing gamma ray–induced chromosome breaks.

These findings suggest a picture where γH2AX formation and the subsequent recruitment of cohesin stabilize broken chromosomes, using the unbroken sister chromosome as a splint to hold broken ends together while leaving the actual site of damage open for repair proteins (Figure 1). This helps ensure that DSB repair occurs efficiently and with fidelity, maintaining genome integrity in the face of DNA damage and avoiding the genome rearrangements associated with cancer.

Robert Shroff, PhD
Research Fellow
Laboratory of Biochemistry
shroffr@mail.nih.gov

Michael Lichten, PhD
Principal Investigator
Laboratory of Biochemistry
NCI-Bethesda, Bldg. 37/Rm. 6124
Tel: 301-496-9760
Fax: 301-402-3095
lichten@helix.nih.gov

Studying Tumor-Host Interactions Reveals a Novel Mechanism for the Activity of TIMP-2

Feldman AL, Stetler-Stevenson WG, Costouros NG, Knezevic V, Baibakov G, Alexander HR Jr, Lorang D, Hewitt SM, Seo DW, Miller MS, O’Connor S, and Libutti SK. Modulation of tumor-host interactions, angiogenesis, and tumor growth by tissue inhibitor of metalloproteinase 2 via a novel mechanism. Cancer Res 64: 4481–6, 2004.

umor growth, invasion, and metastasis are the results of a complex series of interactions between tumor cells and the cells that make up the host microenvironment. Each of the cell types involved in this process has the potential to influence the other cell types through secreted cytokines and through alterations of the environment, such as changes in pH and oxygen content. This complex interplay is extremely difficult to model in an in vitro system. This difficulty led us to develop an in vivo model system that would allow these interactions to be studied at both a genomic and proteomic level.

The model is based on altering the expression of a single factor by the tumor cell using retroviral transduction and studying the effects of this change on the surrounding host environment and on the tumor cells themselves growing in vivo. The effects seen in vivo can then be compared to the differences seen between the altered cell line and the wild-type parental cell line in vitro; those changes unique to the in vivo observations can be ascribed to a relationship between the tumor and the host.

To test this model system, we chose to study tissue inhibitor of metalloproteinase 2 (TIMP-2). TIMP-2 is an endogenous protein present in a variety of tissues and characterized by its ability to both block metalloproteinase activation in the extracellular matrix and to inhibit the development of blood vessels. Previous work has demonstrated that these two activities may be attributable to separate domains of the TIMP-2 protein (i.e., truncated forms of the protein have different activities, with one portion inhibiting metalloproteinase activity and an alternative portion inhibiting angiogenesis). To identify the pathways through which TIMP-2 mediates its antiangiogenic activity in vivo, we applied the following experimental design.

A murine colon cancer line, MC38, was chosen because of its ability to form significant tumor neovasculature when grown as subcutaneous tumors in syngeneic BL/6 mice. The TIMP-2 gene was cloned into a retroviral vector, and MC38 cells were transduced with either a TIMP-2–expressing retrovirus or a null retrovirus control. Clones were selected, and a high-expressing TIMP-2 clone was chosen for further study, which was identified as MET-11. MET-11 and the null retrovirus–transduced tumor line, MEX, demonstrated no difference in their in vitro growth characteristics. MET-11 and MEX cells were then injected subcutaneously into BL/6 mice and were allowed to grow for 18 days. Between day 6 and day 18, MET-11 tumors were significantly smaller then their MEX or wild-type counterparts and had significantly less vascularity as determined by immunohistochemical staining of the tumors with CD31 antibody and vessel counts. This observation was consistent with the known antiangiogenic activity of TIMP-2. Tumors were harvested at day 6 and day 12. RNA was extracted from both MET-11 and MEX tumors, and cDNA microarray analysis was performed. A comparison was also made between MET-11 and MEX cells grown in vitro. Figure 1 depicts the array analysis schema.

Figure 1. Identification of genes associated with the host response to tissue inhibitor of metalloproteinase 2 (TIMP-2). A) Strategy for comparing gene expression patterns of MC38/null and MC38/TIMP-2 tumor cells in vitro and in vivo. Using cDNA microarrays, MC38/null (green) and MC38 TIMP-2 (red) were compared in vitro and in vivo after 6 or 12 days of growth. Genes associated with tumor-host interactions due to TIMP-2 might be similarly expressed in vitro (e.g., yellow spot in lower left corner of top array, arrowhead), but differentially expressed in vivo (red spot in lower left corner of bottom two arrays, arrowheads). B) cDNA microarray analysis identified 13 such genes. Each pixel represents the expression ratio on one array. Red indicates upregulation in the MC38/TIMP-2 sample, and green indicates downregulation. Color intensity is proportional to expression ratio. Black represents ratios close to 1.0. C) Microarray and qRT-PCR data for PTPN16, the murine gene for mitogen-activated protein (MAP) kinase phosphatase-1 (MKP-1). Expression ratios were close to 1.0 in vitro, but showed upregulation in MC38/TIMP-2 tumors in vivo.

To identify differentially expressed genes between MET-11 and MEX tumors grown in vivo, we chose day 6 for analysis, as this was a time point in which both MET-11 and MEX tumors were of similar size. We hypothesized that the gene differences seen here have a cause-effect relationship on the change in growth characteristics seen between day 6 and day 18. Gene expression changes were also compared at day 12 to look for those genes that were persistently altered in expression between MET-11 and MEX tumors. Those genes that were altered between MET-11 and MEX at day 6 and persisted through day 12 in vivo but were not altered in vitro were selected for further study.

We found 13 genes to be differentially expressed between MET-11 and MEX tumors that fulfilled our criteria of greater than 2-fold up- or downregulation at both day 6 and day 12 in vivo and no differential expression in vitro. Among these genes PTPN16 (MKP1) was found to be upregulated to the greatest degree in MET-11 tumors compared to MEX tumors at day 6. PTPN16 is a protein-tyrosine-phosphatase that dephosphorylates p38 MAP kinase, thus inactivating it. p38 MAP kinase is known to play an important role in both vascular endothelial grow factor (VEGF) and basic fibroblast growth factor (bFGF) signaling, and therefore, its modulation may be important with respect to TIMP-2’s angiogenic inhibitory activity.

We sectioned MET-11 and MEX tumors at day 6 and analyzed the levels of protein expression for PTPN16, p38 MAP kinase, and phosphorylated p38 MAP kinase. We found, in concordance with the RNA data, that PTPN16 expression was significantly elevated in TIMP-2–expressing tumors (MET-11) compared with their null-transduced counterparts (MEX). In addition, whereas total p38 levels were similar in both tumors, the proportion of phosphorylated p38 was significantly reduced in the MET-11 TIMP-2 overexpressers. This observation fit with the increased levels of PTPN16. To test whether increased expression of PTPN16 and therefore decreased phosphorylation of p38 led to the impaired growth we saw in TIMP-2 overexpressing tumors, we inoculated BL/6 mice with 1 × 106 MET-11 tumor cells in their flank. Tumors were allowed to grow for 14 days, at which time mice were divided into two groups: one group received systemic phosphate-buffered saline (PBS) injections from day 14 until day 25, whereas the other group received systemic injections of orthovanadate (a phosphatase inhibitor) over the same time period. Tumors growing in the mice receiving orthovanadate grew significantly larger than did those in the mice receiving the PBS control. Tumors harvested from mice receiving orthovanadate compared with tumors harvested from mice receiving PBS showed increased phosphorylation of p38 MAP kinase consistent with an inhibition of PTPN16 activity.

This model system, which allowed us to study tumor host-interactions, led us to hypothesize a new mechanism of action for TIMP-2 with respect to its effects on tumor blood vessel growth. TIMP-2 upregulates the expression of PTPN16, resulting in a decrease in the phosphorylation status of p38 MAP kinase. Inactivation of p38 MAP kinase inhibits the ability of VEGF and bFGF to signal through their receptors. Since VEGF and bFGF are important mitogens for endothelial cell proliferation, this inhibition would be expected to impair the ability of a tumor to develop a blood supply. This model system can be used to study other genes to identify their in vivo mechanisms of action and represents a technique for applying both genomic and proteomic approaches to the study of tumor-host interactions.

Andrew L. Feldman, MD
Clinical Fellow
Laboratory of Pathology

Steven K. Libutti, MD
Principal Investigator
Surgery Branch
NCI-Bethesda, Bldg. 10/Rm. 4W-5940
Tel: 301-496-5049
Fax: 301-402-1788
slibutti@nih.gov

Designing a Chemical Probe to Find a Molecular Target

Malolanarasimhan K, Lai CC, Kelley JA, Iaccarino L, Reynolds D, Young HA and Marquez VE. Synthesis and biological study of a flavone acetic acid analogue containing an azido reporting group designed as a multifunctional binding site probe. Bioorg Med Chem 13: 2717–22, 2005.

wo decades ago, a synthetic flavonoid known as flavone-8-acetic acid (FAA, Figure 1) captured a great deal of attention because of its ability to reduce tumor growth in a number of relatively refractory murine solid tumors (Corbett TH et al. Invest New Drugs 4: 207–20, 1986). Unfortunately, the extensive antitumor activity observed in these mouse models did not translate to humans, and FAA showed no activity in phase I and II clinical trials (Kerr DJ and Kaye SB. Eur J Cancer Clin Oncol 25: 1271–2, 1989). Despite extensive studies, the specific mechanism of FAA has remained undefined, and no specific biological target has been identified. Because such a promising drug candidate for solid tumors failed in humans, knowledge of FAA’s exact mechanism of action in mice remains critically important in potentially providing new leads for drug development to treat human cancers.

Figure 1. Chemical structures of flavone-8-acetic acid (FAA) and azido-FAA. The blue circle denotes the locus of chemical substitution that still retains the full activity of the parent FAA molecule. Azido-FAA can undergo efficient coupling via a modified Staudinger ligation (Saxon E and Bertozzi CR. Science 287: 2007–10, 2000) with a FLAG peptide-phosphine tag under mild, biologically relevant conditions (pH 7.0–7.4, 25°C) to generate a peptide-tagged probe for locating and isolating proteins that bind to FAA. The red portions of the peptide-phosphine tag and the azido-FAA conjugate represent the FLAG peptide epitope used for immunorecognition.

Numerous studies at the NCI and elsewhere have suggested that the antitumor activity of FAA is the result of indirect effects engaging the immune system and acting as a biological response modifier. NCI researchers found that FAA had very potent immunomodulatory activity against murine kidney tumors when combined with interleukin 2 (IL-2) (Mace KF et al. Cancer Res 50: 1742–7, 1990). These same researchers investigated the effect of FAA in mouse macrophage cell lines and found it to potently induce cytokines and interferons in primary splenocytes, including macrophages and T cells. Cytokines are thought to mediate an increase in natural killer cell activity and, through action on vascular endothelial cells, a reduction of tumor blood flow and the resultant onset of tumor necrosis. Indeed, one of the main effects of FAA was tumor necrosis factor-α (TNF-α)–mediated tumor vascular collapse followed by tumor necrosis. Species differences also apparently existed for the immunomodulatory action of FAA. In particular, FAA did not induce cytokines/interferons in human cells, and the NCI researchers who carried out this study hypothesized that the failed clinical trials were possibly due to the failure of FAA to activate cytokine genes in human cells (Futami H et al. Cancer Res 51: 6596–602, 1991). Since that time, FAA has been found to activate the transcription factor NFκB, which is critical for the transcription of many cytokine genes, and it has recently been confirmed to directly induce interferon.

It was against this backdrop of much tantalizing information but no defined and confirmable molecular target for FAA that the interests of investigators in the Laboratory of Experimental Immunology and the Laboratory of Medicinal Chemistry converged. As part of a collaborative effort, we decided to design and synthesize a compound with a close structural resemblance to FAA that would be capable of capturing a protein target. To accomplish this goal, we wanted an FAA-like molecule that was modified with a functional group capable of reacting in a biological milieu with a molecular tag. We were aided in the task by the recent development of a rapid mouse macrophage tissue culture screen with which we could dissect the molecular structure of FAA and determine which chemical variations abrogated, maintained, or enhanced its cytokine-inducing activity. We found that FAA derivatives substituted in the para-position of the phenyl ring (Figure 1, blue circle) retained biological activity. We therefore designed the compound azido-FAA (Figure 1, X = N3), which uses an azide group as an affinity label that can be activated either by conventional chemistry or photochemically. When tested, azido-FAA possessed the same capacity as FAA to induce chemokine expression in a mouse macrophage cell line and induced an identical pattern of chemokine gene expression. This indicated that the azide group did not interfere with the activity of the parent FAA.

In view of the technical difficulties expected with photochemical activation of our probe molecule, we investigated chemical approaches to reacting it with a specific peptide tag. Especially attractive was a reaction that allowed us to covalently trap our probe molecule with a reactive tag tethered to the amino terminus of a FLAG octapeptide (Figure 1, FLAG peptide-phosphine); this conjugate would then provide a way to fish out the drug-bound complex. This approach utilizing the Staudinger reaction was developed by Dr. Carolyn R. Bertozzi at University of California, Berkeley, who kindly provided us with the FLAG peptide-phosphine tag (Vocadlo DJ et al. Proc Natl Acad Sci U S A 100: 9116–21, 2003). We were able to demonstrate that azido-FAA could readily and efficiently couple with the FLAG peptide-phosphine tag in a nanoscale reaction under conditions that mimicked those of a potential biological experiment. High-resolution MALDI mass spectrometry was used to track the progress of this reaction and to unambiguously confirm the identity of the coupled product (Figure 1).

Western blot analysis or immunoprecipitation with a mouse anti-FLAG antibody are possible strategies for detecting the drug-bound complex in the murine cellular systems of interest. Indeed, in partnership with the Laboratory of Proteomics and Analytical Technology (SAIC-Frederick, Inc.) at NCI-Frederick, we have shown that the conjugate can be mixed with extracts from mouse macrophage cells to interact with potential FAA-binding proteins expressed in these macrophages. The conjugate can also be used as a competitor in incubation with extracts of FAA- or azido-FAA–treated cells. Several FAA-FLAG conjugate–bound protein complexes have been immunoprecipitated using anti-FLAG antibody immobilized on protein G-beads, and the isolated proteins are being characterized with the help of the Laboratory of Proteomics and Analytical Technology in order to identify and validate the FAA molecular target. Our approach thus represents an important model for identifying the molecular targets of novel small molecules whose mechanism of action is unknown.

Howard A. Young, PhD
Senior Principal Investigator
Laboratory of Experimental Immunology
youngh@ncifcrf.gov

Victor E. Marquez, PhD
Chief
Laboratory of Medicinal Chemistry
marquezv@dc37a.nci.nih.gov

James A. Kelley, PhD
Principal Investigator
Laboratory of Medicinal Chemistry
NCI-Frederick, Bldg. 376/Rm. 106
Tel: 301-846-5955
Fax: 301-846-6033
kelleyj@dc37a.nci.nih.gov

Susceptibility for Malignant Conversion Resides in the Target Cells

Woodworth CD, Michael E, Smith L, Vijayachandra K, Glick A, Hennings H, and Yuspa SH. Strain-dependent differences in malignant conversion of mouse skin tumors is an inherent property of the epidermal keratinocyte. Carcinogenesis 25: 1771–8, 2004.

he induction of skin tumors on mice from different genetic backgrounds provides a valuable model for studying the genetics of cancer susceptibility or resistance. Skin tumors are superficial, multiple, and display reproducible phenotypic markers as they progress to malignancy. A number of inbred mouse strains display characteristic high (SENCARA/Pt) or low (BALB/c, C57BL/6) susceptibility to skin tumor induction. Mice of the FVB/N strain have been widely used for carcinogenesis studies and for the development of transgenic mice. An important feature is their susceptibility to carcinogenesis at several organ sites including skin, lung, colon, and mammary glands. Studies of multistage chemical carcinogenesis in skin have shown that FVB/N mice are moderately susceptible to induction of benign papillomas, but the papillomas are unusually prone to undergo malignant conversion to carcinomas. In addition, transgenic mice that express human papillomavirus (HPV) type 16 E6 and E7 oncogenes in skin rapidly develop squamous carcinomas in the FVB/N strain, but develop only a few benign tumors when tested in BALB/c, SENCARA/Pt, and C57BL/6 mice. An important question is whether this unusual propensity for rapid and high-frequency malignant progression in the FVB/N phenotype is determined by systemic factors or whether it is an inherent property of the keratinocyte (Figure 1).

Figure 1. The high frequency of malignant conversion in FVB/N mice is demonstrated in vivo, in vitro, and in keratinocyte grafts. In vivo: After induction of skin tumors by an initiation-promotion protocol in FVB/N, BALB/c, and C57BL/6 mice, the conversion of papillomas to squamous cell carcinomas was compared in the mouse strains. In vitro: Primary skin keratinocytes isolated from each of the strains were used in a colony assay in cell culture to assess the frequency of carcinogen-induced progression to malignancy. Grafts: Human papillomavirus (HPV) type 16–immortalized keratinocyte cell lines from each mouse strain were grafted to the dorsal skin of athymic nude mice, and the frequency of carcinoma development in grafts was compared.

To address this question, we isolated skin keratinocytes from each of the above strains and used them in an in vitro colony assay to assess the frequency of carcinogen-induced progression to malignancy of genetically initiated keratinocytes. FVB/N keratinocytes were 10 times more sensitive to chemically induced malignant conversion than keratinocytes from other strains, consistent with their known sensitivity for premalignant progression in vivo. Furthermore, the high frequency of conversion in FVB/N keratinocytes in vitro was suppressed in keratinocytes that were derived from the offspring of a cross between C57BL/6 and FVB/N mice, indicating that this susceptibility trait is not dominant.

Keratinocytes from each strain were infected with retroviruses containing HPV-16 E6 and E7 oncogenes to compare progression to immortal foci. FVB/N keratinocytes could be immortalized by E6 or E7 proteins as easily as keratinocytes from other strains. When immortal cells were grafted to the dorsal skin of nude mice to examine tumor development in a common genetic background, HPV-16–immortalized FVB/N keratinocytes formed carcinomas in the grafts more frequently (50%) than keratinocytes from SENCARA/Pt (14%), BALB/c (1.9%), or C57BL/6 (2.5%). Although tumor stromal cells and inflammatory cells can contribute to the tumor phenotype in several cancer models, normal dermal fibroblasts from either sensitive or resistant strains did not influence the tumor outcome in this model. In addition, the frequency of immortalization of keratinocytes among the strains did not correlate with susceptibility to malignant conversion. Thus, immortalization is not the rate-limiting step in this model.

To determine whether the sensitivity of the FVB/N genotype for malignant conversion is a dominant trait, FVB/N females were crossed with SENCARA/Pt males, and skin tumors were induced in the offspring by an initiation-promotion protocol. Enhanced premalignant progression and malignant conversion in the FVB/N strain were suppressed by the SENCARA/Pt genotype, a finding consistent with the suppression detected in the F1 cross between C57BL/6 and FVB/N in the in vitro conversion assay. Thus, the frequency of malignant conversion of papillomas in F1 hybrids of FVB/N and SENCARA/Pt crosses reflects the frequency in the SENCARA/Pt parent.

Our results show that keratinocytes from FVB/N mice are significantly more susceptible to malignant progression than keratinocytes from other strains, and that this is not a dominant characteristic. Although our studies have not yet revealed the precise factors responsible for the intrinsic sensitivity of FVB/N keratinocytes to premalignant progression, they do provide some clues and a foundation for further analysis at the cellular level. Using keratinocytes from inbred strains that vary substantially in susceptibility and retain this characteristic in vitro should be helpful in defining the underlying basis for these susceptibility differences.

Craig D. Woodworth, PhD
Associate Professor
Department of Biology, Clarkson University
woodworth@clarkson.edu

Stuart H. Yuspa, MD
Chief
Laboratory of Cellular Carcinogenesis and Tumor Promotion
sy12j@nih.gov

Henry Hennings, PhD
Principal Investigator
Laboratory of Cellular Carcinogenesis and Tumor Promotion
NCI-Bethesda, Bldg. 37/Rm. 4054B
Tel: 301-435-8360
Fax: 301-496-4258
hh20v@nih.gov

How Selenium Makes Its Way into Protein as Selenocysteine, the 21st Amino Acid in the Genetic Code

Carlson BA, Xu X-M, Kryukov GV, Rao M, Berry MJ, Gladyshev VN, and Hatfield DL. Identification and characterization of phosphoseryl-tRNA[Ser]Sec kinase. Proc Natl Acad Sci U S A 101: 12848–53, 2004.

n 1970, a kinase activity that phosphorylated a minor species of seryl-tRNA to form phosphoseryl-tRNA was observed in rooster liver (Maenpaa PH and Bernfield MR. Proc Natl Acad Sci U S A 67: 688–94, 1970), and a minor species of seryl-tRNA that decoded the termination codon UGA was observed in bovine and chicken livers (Hatfield D and Portugal FH. Proc Natl Acad Sci U S A 67: 1200–06, 1970). The seryl-tRNA in both cases was subsequently identified by us as selenocysteine (Sec) tRNA[Ser]Sec, but despite many efforts, the kinase activity remained elusive. Sec is now regarded as the 21st amino acid in the genetic code, marking the first expansion to the code since it was deciphered by Marshall Nirenberg and collaborators at the NIH in the 1960s.

The biosynthesis of Sec, unlike the 20 other amino acids in the genetic code, occurs on its tRNA, and it is the pathway by which the element selenium makes its way into protein. A stem-loop structure in the 3′-untranslated region of selenium-containing (selenoprotein) genes is responsible for recoding UGA for Sec, which circumvents the normal function of UGA as a stop codon in protein synthesis. The stem-loop structure in the selenoprotein genes is recognized by a specific factor, designated SBP2, that forms a complex with Sec tRNA[Ser]Sec and its specific elongation factor and inserts Sec into protein in response to UGA. Although many factors dedicated to the insertion of selenium into protein as Sec have been identified, the biosynthesis of Sec in eukaryotes and the role of phosphoseryl-tRNA[Ser]Sec have not been resolved.

Using a comparative genomics approach that searched completely sequenced archaeal genomes for a kinase-like protein with the pattern of occurrence similar to that of components of the Sec insertion machinery, we detected a candidate gene for mammalian phosphoseryl-tRNA[Ser]Sec kinase (pstk). Mouse pstk was cloned, and the gene product (PSTK) was expressed and characterized. PSTK specifically phosphorylated the seryl moiety on seryl-tRNA[Ser]Sec and in addition had a requirement for ATP and Mg++. Proteins with homology to mammalian PSTK occur in the fruit fly, Drosophila, the nematode, Caenorhabditis elegans, and in the archaea, Methanopyrus kandleri and Methanococcus jannaschii. This suggests a conservation of its function across archaea and eukaryotes that synthesize selenoproteins, and the absence of this function in bacteria, plants, and yeast. The fact that PSTK has been highly conserved in evolution suggests that it plays an important role in selenoprotein biosynthesis and/or regulation.

The recent identification of the means by which cysteine (Cys) is synthesized on its tRNA in some archaea provides an excellent model of how Sec is biosynthesized on its tRNA. Cys RNACys is aminoacylated by phosphoserine to form phosphoseryl-tRNACys that in turn is converted to cysteyl-tRNACys by an enzyme that replaces the phosphate on serine with an activated form of sulfur (Sauerwald A et al. Science 307: 1969–72, 2005). Since phosphoserine is attached to tRNA[Ser]Sec, it would seem to be the most likely intermediate in Sec biosynthesis wherein selenium would be activated by selenophosphate synthetase, an enzyme previously identified in mammals. This pathway of Sec biosynthesis is shown in Figure 1. Interestingly, Sec tRNA[Ser]Sec has a dual role of serving as the carrier molecule for the biosynthesis of Sec and as the adaptor molecule for decoding UGA for the insertion of Sec into protein.

Figure 1. Proposed pathway of Sec biosynthesis on its tRNA in mammalian cells. Serine (Ser, shown in blue as an oblong circle) is attached to tRNA[Ser]Sec (shown in green as a cloverleaf structure) by seryl-tRNA synthetase (SRS) to form seryl-tRNA[Ser]Sec (shown in blue as serine attached to tRNA[Ser]Sec) and is then phosphorylated by phosphoseryl-tRNA kinase (PSTK) to form the intermediate phosphoseryl-tRNA[Ser]Sec (P, shown in red as a circle attached to serine). The phosphate on phosphoseryl-tRNA[Ser]Sec is then replaced by the selenium donor that is likely activated by selenophosphate synthetase (SPS), and the compound is converted to selenocysteyl-tRNA[Ser]Sec (Sec, shown in gold as an oblong circle attached to tRNA[Ser]Sec) by Sec synthase (SST).

It should also be noted that selenium is an essential micronutrient in the diet of mammals. Numerous health benefits have been associated with selenium, such as preventing cancer and heart disease, delaying the aging process, and delaying the onset of AIDS in HIV-positive patients, as well as beneficial roles in male reproduction, immune function, and development. Most, if not all, of these health benefits are due to selenoproteins. Thus, it is of paramount importance to determine how this element makes its way into protein. The identification and characterization of PSTK provides a major step in establishing the pathway of Sec biosynthesis.

Dolph Hatfield, PhD
Senior Principal Investigator
Laboratory of Cancer Prevention
NCI-Bethesda, Bldg. 37/Rm. 6032a
Tel: 301-496-2797
Fax: 301-435-4957
hatfield@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

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 (Web) *
Rob Wald, Publications Manager *
Michael Fleishman, Graphic Artist *
Yvonne Bersofsky, Web Developer *

* Palladian Partners, Inc.