7th Principal Investigators Meeting Agenda

Ruth Angeletti
Albert Einstein College of Medicine, Bronx, New York

This applications project focuses on discovery of targets and biomarkers in three stages of the Solt-Farber rat model of hepatic neoplasia, which has well-characterized, synchronously developed stages. For analysis of tissue, laser-capture microdissection is being used to afford discrimination among hyperplastic cells and nodules, and surrounding stromal tissues. Both a global analysis of proteins and a targeted, tubulin proteomics approach are being used for the studies of laser-captured tissue. While the amount of protein for analysis of the serum from these animals is much larger, the challenge of dynamic range is an issue for each subproject. Tubulin isotypes may change during the neoplastic process, and the proteins on their cytoskeletal architecture (signaling, motor, etc.) may also be altered and provide mechanistic insights. Separation of the tubulin polypeptides from the associated proteins by a modified procedure permits the direct analysis of the tubulin isotypes by nano LC-MS, and the associated proteins, after tryptic digestion, by nano-LC-MALDI-TOF/TOF. Global analysis of the proteins in the tissue samples is limited by the presence of abundant proteins and by the losses incurred during fractionations of small amounts of protein. A selective approach to analyze only one-two peptides from each protein limits the impact of the most abundant proteins. Novel, open-source software has also been developed to analyze the MS signals from these experiments.

James Baker, Jr., Jing Yong Yee, Thommy Thomas, Ted Norris
Michigan Nanotechnology Institute for Biology and the Medical Sciences and Michigan Center for Ultrafast Optics, University of Michigan, Ann Arbor, Michigan

Conventional fluorescence quantification in tissues and whole body fluorescence imaging techniques do not provide accurate quantitative information on the distribution of fluorescently tagged molecules in tissues. Owing to the limited tissue penetration of light, these methods also lack sensitivity for detection of biomarkers using fluorescence labels. We have developed a two-photon optical fiber fluorescence (TPOFF) probe as a minimally invasive technique for quantifying fluorescence in solid tumors in vitro and in vivo on a real-time basis [1,2,3]. In these studies, we used a single mode optical fiber (SMF) through which femtosecond laser pulses were delivered into a tumor, which enabled us to measure low micromolar concentrations of targeted fluorescent nanoparticles. For detection of nanomolar levels of fluorescently labeled biomarkers, it is essential that a more sensitive TPOFF device be developed. We now are attempting to adapt double-clad optical fibers that keep high excitation rates by propagating ultrashort laser pulses down an inner single mode core, while improving the collection efficiency by using a high-aperture multimode outer core confined with a second clad.

The first fiber we investigated is a double-clad photonic crystal fiber (DCPCF, the end surface shown in Figure 1a). Air holes provide the cladding and need to be sealed before this fiber can be used for in vivo measurements to avoid entry of liquid from biological samples. This would reduce the light confinement capability of the fiber and introduce artifacts in the measurements. We developed a solution to this by splicing a multimode fiber to the end of the DCPCF, then cleaving off all but a thin piece of the multimode fiber (Figure 1b). It is critical to keep the thickness of the multimode fiber tip down to a few microns to optimize the fluorescence collection efficiency.

We also used a solid, double-clad fiber (DCF) to avoid the air hole problems. The solid DCF used has an aperture of 0.46, which is smaller than that of a DCPCF. Although it is not as efficient as the DCPCF, it still provides more than an order of magnitude improvement over traditional single-clad fibers. We have tested the two-photon fluorescence detection efficiency of using this DCF using FITC dye solutions. In comparison with the best commercially available single-mode optical fiber (SMF), the detection efficiency has been improved by a factor of about 12, if the total output power from the DCF is considered as the excitation power, and more than 20 if the inner core power of the DCF is considered as useful excitation power for two-photon fluorescence excitation (Figure 2). The efficiency of this DCF is therefore about two times lower when compared to the 40 times enhancement achieved with the DCPCF, but lacks the technical probes of the holes. In summary, we have developed two options for optical detection of biomarkers that should have detection capabilities within the range needed for biological samples.

1. Ye J.Y., Myaing M.T., Norris T.B., et al. Biosensing Based on Two-Photon Fluorescence Measurements Through Optical Fibers. Optical Letters 27:1412, 2002.
2. Thomas T.P., Myaing M.T., Ye J.Y., et al. Detection and Analysis of Tumor Fluorescence Using a Two-Photon Optical Fiber Probe. Biophys J 86:3959, 2004.
3. Thomas T.P., Ye J.Y., Yang C., et al. Tissue Distribution and Real-Time Fluorescence Measurement of a Tumor-Targeted Nanodevice by a Two Photon Optical Fiber Fluorescence Probe. Proceedings of the SPIE 6095:60950Q, 2006.
4. Ye et al., Optical Letters 27:1412, 2002; Thomas et al., Biophys J 86:3959, 2004; Thomas et al., Proceedings of the SPIE, Vol 6095, 2006.

Brian Balgley
Calibrant Biosystems, Gaithersburg, Maryland

We have developed a prototype proteome instrument that is capable of capillary electrophoresis-based separation, multiplexed capillary reversed-phase liquid chromatography and interfacing between chromatography fraction deposition and matrix-assisted laser desorption/ionization (MALDI) target preparation. The first implementation of the instrument is focused on improving the degree of automation for interfacing multiplexed liquid chromatography to MALDI-MS without sacrificing the throughput and resolution achieved during chromatography separations. In addition to interfacing multiplexed liquid chromatography with MALDI-MS in an offline approach, we have further implemented an automated column-switching scheme for online coupling of dual chromatography separations with electrospray ionization (ESI)-MS. The position of the dual columns with respect to the orifice of MS is controlled by a stepping motor for directly analyzing eluted peptides and proteins using MS measurements.

Besides the development of an automated and high-throughput prototype proteome instrument, our collaborative research efforts with Professor Eric H. Baehrecke’s laboratory at the Center for Biosystems Research, University of Maryland Biotechnology Institute, have contributed to further understanding of the regulation of programmed cell death at the proteome-wide level using the fruit fly Drosophila melanogaster as an experimental model system. A total of 5,661 proteins are identified from stages just before and after the onset of cell death. Analyses of these data enable us to identify proteins from a number of interesting categories, including regulators of transcription, the apoptosis, autophagy, lysosomal, and ubiquitin proteasome degradation pathways and proteins involved in growth control. Several of the identified proteins, including the serine/threonine kinases, are not detected using whole-genome microarrays, providing support for the importance of such high-throughput proteomic technology. These kinases regulate cell cycle arrest and apoptosis, and, significantly, mutations in kinases prevent destruction of salivary glands [1].

In addition to analyzing the soluble fraction of cell lysates, we have further expanded and demonstrated our bioanalytical capability toward the analysis of complex membrane proteomes. Although proteomics technologies have made significant progress in the analysis of soluble proteins in recent years, membrane proteins have lagged behind and are typically underrepresented in datasets. However, the importance of membrane proteins in drug discovery cannot be overemphasized—membrane proteins currently account for approximately 70% of all known pharmaceutical drug targets. In this study, the membrane protein fraction of yeast cell lysates is extracted from pellets using a 2% sodium dodecyl sulfate (SDS) solution. A total of 10,088 distinct peptides are identified, leading to the identification of 2,084 distinct yeast proteins in the membrane fraction of yeast cell lysates using the combined top-down/bottom-up proteome platform. By using the trans-membrane hidden Markov model (http://www.cbs.dtu.dk/services/TMHMM), 624 yeast proteins are predicted to contain 3 or more transmembrane domains. We have identified 258 proteins with 3 or more predicted integral membrane domains from the pellets treated with SDS, corresponding to 40% membrane proteome coverage. It should be emphasized that these membrane proteins are identified with high confidence with a false-positive rate of less than 1%. By combining with the proteome results obtained from the soluble fraction, the collective analysis yields the identification of 3,523 distinct proteins, with an average of 4.5 fully tryptic peptides per protein. By comparison with the reported yeast proteome studies, our results present the most comprehensive analysis of the yeast proteome, including low-abundance membrane proteins [2].

1. Martin D.N., Balgley B.M., Chen J., et al. Proteomic Analysis of Steroid-Triggered Autophagic Programmed Cell Death in Drosophila. Curr Biol, in press.
2. Wang W., Guo T., Rudnick P.A., et al. Comprehensive Yeast Proteome Analysis Including the Identification of Low Abundance Membrane Proteins, manuscript in preparation.

Kathlynn Brown
University of Texas Southwestern Medical Center, Dallas, Texas

Lung cancer is the largest cancer killer of both men and women. The correct histopathological diagnosis of a tumor is critical in determining the appropriate treatment. However, precise classification of tumors remains a significant biomedical challenge. Furthermore, tumors of similar histology can have different clinical outcomes, stressing the need for more detailed molecular classifications. Our overall goal is to generate a panel of cell-specific molecules that can be used to classify tumor types and utilize these cancer-specific reagents in a high-throughput diagnostic assay. Using phage display technologies, my laboratory has developed methodologies to isolate peptides that bind to specific cell lines, even in the absence of knowledge of the cell surface receptor profile. By this method, we have identified cell-specific targeting peptides for five lung cancer cell lines and have determined their binding profile to a large panel of lung cancer cell lines. The isolated peptides display remarkable cell specificities and are able to discriminate between normal and cancerous cells, as well as different lung tumor cells. Furthermore, these peptides can be conjugated to fluorescent nanoparticles (quantum dots, Qdots) in order to address cellular binding. This fluorescent-based assay can be multiplexed so that multiple binding events can be examined on a single sample and is amenable to the use of fresh or fixed cells. Our approach solves two problems in the use of molecular markers for cancer diagnosis. First, we can generate cell-specific ligands rapidly with no prior information about the cell surface. In effect, ligand binding can be thought of as a unique biomarker. Second, we are developing a multiplexed binding assay that allows multiple binding events to be monitored simultaneously on a small number of cells. At the end of this project, we will have developed a novel diagnostic platform that can be expanded to clinical samples.

Han Cao
BioNanomatrix, Philadelphia, Pennsylvania

We are developing a nanochip device for manipulating long genomic DNA for high-resolution (kilobase), whole-genome analysis of cancer biomarkers, such as gene amplifications, deletions, and translocations. These chromosome structural aberrations are strongly implicated in the process of malignant transformation and are important diagnostic, prognostic, and therapeutic indicators for many types of cancer. Although sequencing and PCR offer the ultimate (single-base) resolution for detecting and analyzing these anomalies, it is impractical for scanning the entire genome in a comprehensive, linear fashion. Techniques that rely on probing chromosomes, such as metaphase FISH, while providing a pan-genomic view, cannot resolve structures below the Mb range. By probing uncompressed interphase DNA, resolution can be improved, but spatial organization of the genome is lost, so multiplexed and quantitative information is difficult to obtain. By stretching out (linearizing) interphase DNA, using techniques such as “molecular combing” or “optical mapping,” it is possible to probe specific loci in a spatially significant way, and with resolutions in the kb range. However, techniques for mechanically linearizing DNA are inherently variable, leading to inconsistent stretching of molecules, which often cross over and retract upon themselves. This makes it difficult to standardize such techniques as high-throughput methods for the biomedical community.

We are developing an innovative alternative to mechanical stretching of DNA. We have found that individual DNA molecules, because of the self-avoiding nature of the DNA polymer, will elongate and straighten in a consistent manner when streamed into confining nanometer-scale channels (nanochannels). We have used a novel nanoimprint lithography technique to reliably manufacture nanochannel structures in silicon chips and have demonstrated that DNA in these nanochannels can be visualized and their dimensions measured. We now ask the question, can we quantitatively interrogate this linearized DNA with locus-specific probes for the detection of chromosome structural aberrations associated with cancer? Our product, the nanochannel array chip, will comprise part of an integrated platform for the routine and standardized quantitative analysis of DNA structure that will enable archiving and cross-laboratory comparison of data.

Songming Chen¹, Lin Li², Richard Hay¹, William Catalona⁴, Ziding Feng², Robert Vessella³, Brian B. Haab¹

¹Van Andel Research Institute, Grand Rapids, Michigan; ^2Fred Hutchinson Cancer Research Center, Seattle, Washington; ³University of Washington Medical Center, Seattle, Washington; ⁴Northwestern University, Chicago, Illinois

Angiogenic factors in serum, such as VEGF, IL-8, are strongly related to cancer development and metastasis; however, they showed poor sensitivity and specificity as a biomarker used in cancer diagnosis or prognosis. This is mainly the result of the heterogeneity in the baseline levels of the marker proteins and heterogeneity in the tumors and patients. We have proposed that a biomarker discovery strategy using individualized thresholds, based on longitudinal measurements, would more precisely define abnormal levels for each individual. Multiplexed, specific detection of serum markers using antibody arrays makes a systematic exploration of this hypothesis possible. We are using the detection of prostate cancer recurrence as a model system. Serum samples have been assembled from 17 patients who had recurrence and 17 who did not, with up to 5 samples from each patient, from a pre-surgery sample up to the time of recurrence or no recurrence. We have developed array-based sandwich assays for 13 potential markers, representing pro- and anti-angiogenesis factors. The detection limit of the assay can be as low as 1 ng/mL with a CV less than 20%, and 500 pg/mL for PSA with a CV less than 20%. We are also using a two-color rolling-circle amplification method to measure the relative levels of 75 different cancer-associated proteins, and we are using a novel glycan-detection assay to measure the levels of specific glycans on multiple glycoproteins. Several approaches toward modeling the longitudinal changes of the markers are being pursued. We are comparing the sensitivity and specificity of cancer detection between the conventional single-timepoint method and the longitudinal strategies. By testing many markers, we can evaluate the generality and range of marker improvement using longitudinal information.

Kevin Claffey
University of Connecticut Health Center, Farmington, Connecticut

Immune-dependent responses are selective in defining non-self or aberrant antigen presentation. Unfortunately, antibody production to human cancer antigens is limited. In an innovative and novel approach, we have developed a method to utilize primary immune reactions in tumor draining lymph nodes to provide the means to identify biologically active breast tumor antigens. This novel approach involves the construction of antibody variable region cDNA libraries, synthesis of recombinant VH monobodies, and identification of tumor antigens with a sensitive “antigen-trap” assay followed by LC-MS/MS protein identification. One key element of this process is to produce expression-ready cDNA libraries that can be shuttled into baculoviral vectors for efficient protein synthesis. These libraries are screened for orientation by multiplex PCR and in-frame by direct sequencing. The utility of this process provides correctly folded recombinant VH monobodies with a V5 epitope tag as well as a 6x histidine C-terminal domain for metal affinity purification. The purified recombinant VH proteins can identify tumor antigens from primary tumor extracts as well as on histological slides. The library synthesis and baculovirus shuttling platform is an effective way to shuttle cDNA sequences into a protein synthesis method suitable for biochemical, diagnostic, and therapeutic applications. Future directions will include screening additional antibody libraries, development of highly sensitive cancer antigen diagnostic screens, and plans for therapeutic assessment of human VH monobodies targeted to tumor antigens in vivo.

Shoshana Frankenburg
Sharett Institute of Oncology, Hadassah Hebrew University Medical Center, Jerusalem, Israel

The long-term objective of this research project is to develop a novel treatment and vaccination approach for cancer in general, and melanoma in particular, based on immunotherapy. Ex vivo antigen delivery for immunotherapy is laborious and expensive and is thus not affordable to many of those in need. We propose to develop an antigenic entity that can be applied on the skin, with direct antigen delivery to skin dendritic cells and without the need for in vitro cell manipulations. Thus, the major practical objective of this study is to establish the proof of principle that topically delivered tumor-associated antigens can elicit effective antitumor responses and can be used for cancer immunotherapy.

The study will be based on two antigenic proteins derived from melanoma: the first is a hydrophilic recombinant gp100 protein, and the second is a multiepitope polypeptide that comprises three repeats of 4 HLA-A2 melanoma peptides derived from three different melanoma proteins. In order to allow and to improve topical transdermal delivery, the antigens will be genetically fused to potential carrier molecules. One of these is E. coli heat-labile enterotoxin, a molecule recently shown to act as carrier and adjuvant. Another is a novel haptotactic C-terminal fibrinopeptide (Haptide). During the first phase of the project, the new antigenic entities will be cloned, expressed, and purified. Novel in vitro models using human skin will be used to evaluate transcutaneous passage of molecules, activation and mobilization of Langerhans’ cells, and stimulation of specific cytotoxic T cells. During the first phase of the project, we expect to demonstrate efficacy of antigen delivered transcutaneously to stimulate the immune system in human in vitro models and will allow for the selection of the molecules that will be further evaluated in depth in in vivo models. During the second phase, specific immune responses of splenic T cells from vaccinated mice will be evaluated, tumor models will be established in mice, and the response to vaccination will be determined. Finally, the most effective molecule/s will be produced under GMP or GMP-like conditions for Phase I/II clinical trials in a subsequent study.

The success of this project would allow topical application of an immunostimulant for treatment of melanoma and other cancers and would thus significantly simplify treatment, eliminating the need for hospitalization and even day care and the need for a specialized laboratory. As a result, one could treat a much larger number of patients, with the potential to clinically evaluate new antigens and immunotherapeutic modalities, improving the life quality and expectancy of metastatic melanoma patients.

Martin Guthold
Wake Forest University, Winston-Salem, North Carolina

We are developing a novel, single-cycle method to affinity-select individual aptamer species from a small pool of random oligonucleotides (oligos) using combined fluorescence and atomic force microscopy (AFM). Target molecules immobilized on a cover slip are exposed to a library of fluorescently labeled oligos. High-affinity, target-specific aptamers bind tightly to the target for prolonged periods and resist subsequent washes, resulting in a strong fluorescence signal. The AFM tip is then directed to the coordinates of the fluorescence signal for high-resolution aptamer imaging and extraction. The extracted aptamer species is PCR-amplified, analyzed, and sequenced. Our method is designed to identify individual target-binding oligos in a single cycle of detection, isolation, amplification, and sequencing. Here, we demonstrate proof of principle of all steps of our method, namely instrumentation development, sample preparation, and selection of the known thrombin aptamer GGTTGGTGTGGTTGG from a binary pool of random oligonucleotides.

Tomasz Heyduk
Saint Louis University, Saint Louis, Missouri

The overall goal of this project is to develop novel nucleic acid-based sensors designed to produce fluorescence signal upon recognition and binding to the target protein. In the first phase of the project, model protein systems were used to provide “proof of principle” verification of molecular beacon design for proteins exhibiting natural sequence-specific DNA binding activity. In the next step, we generalized our molecular beacon design to include detection of proteins lacking natural DNA-binding activity. In this design, a pair of aptamers recognizing two distinct epitopes of the protein were used to prepare the beacon. Here, we present the data illustrating development of molecular beacons detecting NF-kB, a transcription factor involved in regulation of genes important for inflammatory response, cancer cell invasion, metastasis, angiogenesis, and resistance to chemotherapy. Two versions of the assay (homogeneous and heterogeneous) were developed. Performance of the assay was tested using purified NF-kB and cellular extracts spiked with various amounts of purified protein. Both assays were able to detect NF-kB with desired specificity. Heterogeneous assay provides an opportunity for large-scale multiplexing by adapting it to a microarray format. Proof-of-principle experiments demonstrated the feasibility of this concept.

Daniel Jay
Tufts University School of Medicine, Boston, Massachusetts

Customized optimal treatment based on a precise classification of each patient’s tumor would have a marked health benefit. For example, if a diagnostic set of biomarkers applied to biopsy could predict whether a patient’s cancer was susceptible to a particular chemotherapy, this would facilitate treatment and improve prognosis. It remains a major challenge to classify cancers presented by patients, which are remarkably diverse due to individual differences in both tumor and patient. Molecular classification using expression microarrays shows promise, but much of the complexity is engendered by proteomic differences. To begin to address proteomic complexity, we have developed surface proteome signatures (SPS) as a means of molecular classification of cancer and identification of potential biomarkers. This approach is based on differential immunocytochemistry using a library of single-chain antibodies (scFvs) derived from phage display that recognize the surface proteome of HT-1080 fibrosarcoma cells. We have developed and tested SPS doxorubicin using human breast cancer lines that are either sensitive or resistant to doxorubicin: MCF7, MCF7-Adr, which is resistant because of overexpression of the multidrug resistance P glycoprotein, and MCF7-2R, a spontaneous mutation from the parent MCF7 line for which proteome changes are not known.

SPS using 70 scFvs was not able to distinguish between two near identical cell types derived from the same parent line (MCF7 vs. MCF7-2R) but could distinguish between MCF7-Adr and its parent line and clearly differentiated between breast cancer and fibrosarcoma cells. Importantly, SPS was able to identify a small subset of putative biomarkers that best distinguished between cell types. These scFvs are candidates for developing a diagnostic for breast cancer classification. The first biomarker analyzed was shown to be CD44, and we have implicated this protein in a key role in MCF7-Adr resistance to doxorubicin. We developed human breast tumors from these lines in a xenograft model and developed SPS from resected tumor cells. Based on these SPS, we can predict the efficacy of doxorubicin treatment in tumor progression in these mice. We are able to acquire SPS rapidly with small numbers of cells (derived from passaged tumors [500 per assay] and predict whether doxorubicin would be effective or not.

Together, our findings establish a proof of principle of the SPS technology. This new technology has several important applications. First, it provides a molecular classification of tumors that is complementary to expression microarray and SAGE analysis and has the potential to address the additional complexity engendered by the proteome. For example, SPS analysis could be applied to identify class descriptors for tumor aggressiveness or homing to other tissues during metastasis. Second, the analysis also provides a rapid identification of potential biomarkers that may also be tested for functional importance by using the scFv for knockdown of the antigen bound by that scFv using fluorophore-assisted light inactivation (FALI). The limited patient material required (500 cells/assay) means that SPS may be compatible with analysis of resected tumors or large biopsies from patients. Third and finally, the technology can be generalized and applied to other health-relevant questions such as identifying new and better stem cell biomarkers.

Jingfang Ju
University of South Alabama Mitchell Cancer Institute, Mobile, Alabama

Translational control plays a key role during development, cell cycle control, and mechanisms related to acute drug resistance. In particular, non-coding miRNAs can potentially regulate over 30% of gene expression at the translational level. Gene expression analysis on actively translated mRNA transcripts provides a unique approach to study posttranscriptional regulation. Current method relied on a traditional sucrose gradient ultracentrifugation procedure to isolate polysome complexes and requires a large number of cells (up to 500 million cells). As a result, this still remains a major bottleneck for the investigation of posttranscriptional regulation with limited quantities of clinical samples. Therefore, there is an urgent need to develop a novel approach to isolate actively translated polysomes from a small number of cells. The new approach will allow us to systematically study translational regulation with limited clinical samples. It has been shown that actively translated mRNAs are associated with multiple units of ribosomes and the newly synthesized polypeptides are closely associated with molecular chaperones such as hsp73. These molecular chaperones assist in the proper folding of nascent polypeptides into higher ordered structures. These chaperones will provide the anchor to separate actively translated mRNAs associated with polysomes from free mRNAs. Affinity antibody capture beads will be developed to capture hsp73 chaperones associated with the polysome complexes so that all polysomes can be separated from monosomes and free mRNAs. The isolated actively translated mRNAs will be used for high-throughput gene expression analysis.

Eugene S. Kandel, Tao Lu, Youzhong Wan, Maupali Dasgupta, Mukesh K. Agarwal, Mark W. Jackson, Patrick Varley, George R. Stark
Cleveland Clinic Foundation, Cleveland, Ohio

Genetic dissection of signaling pathways in mammalian cells involves screening or selecting phenotypic mutants obtained by a variety of techniques. The concerns about current methods include inadequate genome coverage and difficulty in validating the link between mutation and phenotype. We showed that the ability to induce mutations increases greatly if a randomly inserted promoter directs transcription into the host DNA. We predict that either gain-of-function or loss-of-function mutants could be obtained depending on the position and orientation of an insert. The mutant phenotype is due to the expression of a hybrid transcript derived from the vector and the insertion site. Since other alleles of the affected gene remain intact, the phenotype is dominant but reversible by inactivating the promoter, for example, by site-specific recombination.

Previously, we have used reversible insertional mutagenesis by modified retroviral vectors to search for factors that may cause hyperactivity of k;B, as frequently seen in cancerous cells. Reversible mutants with hyperactive NF-kB were readily obtained and the relevant integration targets were unambiguously identified, even when multiple inserts were found within the same cell. The regulatable nature of the mutant phenotype allowed us to observe new properties even for well-known components of NF-kB pathway. For example, overexpression of such components, caused by promoter insertion into the 5’ end of the corresponding genes, leads to k;B-dependent secretion of factors that further activate k;B through cell-surface receptors, establishing an autocrine loop.

Vectors based on Sleeping Beauty transposon also were successful in delivering the mutagenic promoter cassette. Similar to retroviral vectors, transposons generated mutants with constitutive NF-kB activation that were useful for investigating the functions of the targeted genes. For example, an insert in the middle of the RIPK1 gene resulted in the production of an apparently hyperactive C-terminal fragment of the classical RIPK1 protein. Bioinformatic analysis predicted the presence of a natural promoter in the vicinity of the integration site. This promoter should produce a shorter form of RIPK1 identical to the one generated in the mutant. Initial experimental data are consistent with the presence of the shorter form of RIPK1 in some cancer cells.

Relative simplicity and robust target validation make reversible insertion of a promoter suitable for a broad range of applications. Taking advantage of the cell-autonomous and regulatable nature of transposition, we are currently adopting our mutagenesis strategy for use in animal models of cancer.

Shana Kelley
Boston College, Chestnut Hill, Massachusetts

We have developed a platform for electronic protein detection. The assay relies on a robust electrocatalytic process that reports on biomolecular complexation events and is executed on a nanostructured electrode platform that provides high sensitivity. The electrocatalytic method was originally developed by the Kelley laboratory for the sequence-specific analysis of nucleic acids and was used to detect bacterial genes with single-base resolution [1] and attomole sensitivity [2,3]. Current efforts to implement this system with protein targets have produced a new system for the detection of DNA repair factors and other cancer-related proteins. The eventual outcome of this project will be multiplexed detection of panels of different biomolecular targets on an electronic chip.

1. Lapierre M.A., O’Keefe M., Taft B.J., et al. Electrocatalytic Detection of Pathogenic DNA Sequences and Antibiotic Resistance Markers. Anal Chem 15;75(22):6327-33, 2003.
2. Gasparac R., Taft B.J., Lazareck A.D., et al. Ultrasensitive DNA Detection at 2D and 3D Nanoelectrodes. J Am Chem Soc 126(39):12270-1, 2004.
3. Lapierre-Devlin M.A., Asher C., Gasparac R., et al. Amplified Electrocatalysis at DNA-Modified Nanowires. Nano Let 5(6):1051-5, 2005.

David Krizman
Expression Pathology, Inc., Gaithersburg, Maryland

In the United States, breast cancer is the second leading cause of cancer-related deaths, in women. HER2 is an important oncoprotein, the expression level of which has been shown to be directly indicative of aggressive breast cancer, and is the protein target of immuno-based therapy. Diagnostic detection and quantification of HER2 in cancerous tissue are routinely carried out by standard immunohistochemistry. However, this methodology lacks sensitivity and relies on subjective data analysis. We have developed and evaluated a method for direct detection and absolute quantification by selected reaction monitoring (SRM) of HER2 directly in formalin-fixed paraffin-embedded (FFPE) breast cancer tissue using a stable isotope standard (SIS) peptide derived from HER2. A tryptic peptide from HER2 was synthesized with 13C to serve as an internal standard for quantification. The linear range and limits of detection and quantitation were evaluated by nanoflow reversed-phase liquid chromatography (nanoRPLC) coupled online with a linear ion trap mass spectrometer (MS) operation in tandem MS mode. Breast cancer cells were procured from formalin-fixed tissue culture cells expressing known amounts of HER2 protein and from standard fixed breast cancer tissue sections showing a broad range of HER2 expression (by IHC). Soluble protein extracts were prepared using Liquid Tissue® protocol/reagents and spiked with known concentrations of the HER2 SIS peptide followed by nanoRPLC-MS/MS SRM analysis. Quantitation was accomplished by comparison of peak areas generated from reconstructed ion chromatograms. Results indicate the ability to quantitate HER2 expression in Liquid Tissue® extracts from fixed tissue sections and demonstrate the ability to quantify HER2 in IHC negative cells. Results will be discussed within the context of a novel, highly sensitive cancer diagnostic technology designed to impart absolute quantification of cancer biomarkers in formalin-fixed cancer tissue.

Dale Larson
Harvard Medical School, Boston, Massachusetts

A system is being developed to enable the extraction of aliquots from frozen serum samples without thawing the samples. Currently, the samples experience at least two freeze-thaw cycles before being analyzed. This system will require only one freeze-thaw cycle before analysis, and, because it is automated, it will increase sample processing throughput thereby reducing the lead time associated with acquisition of these samples from biobanks such as the Nurses’ Health Study. The ability to extract frozen cores has been demonstrated, and the force profile and needle geometry have been optimized.

Dale Larson
Harvard Medical School, Boston, Massachusetts

Surface plasmon-enhanced illumination (SPEI) is a technology that enables the creation of an array of nanoscale beams of semi-collimated propagating light. A scanning probe microscope based on the scanning of an SPEI device across a sample and recording the fluorescence excited in the sample has been created. The geometric characteristics of these beams have been determined by exposing photoresist and measuring the resulting patterns in the photoresist. These beam characteristics and initial scanning results will be presented.

Richard Levenson
Cambridge Research and Instrumentation, Inc., Woburn, Massachusetts

Background: Multi-analyte immunohistochemistry has potential applications in drug development, tumor biology, and clinical assessments, including (1) detection of target colocalization, (2) evaluation of spatial relationships between cell types, (3) signal transduction studies, (4) reduced sample depletion, and (5) decreased reagent utilization. However, analysis of two or more colocalized antigens by brightfield (nonfluorescence-based) microscopy is difficult because overlapping absorbing colored dyes (chromogens) are hard to resolve visually. Multispectral imaging, however, can “unmix” such signals to generate quantitative images of individual analytes. This has immediate application in the study of estrogen-receptor (ER) and progesterone (PR) expression and colocalization in breast cancer, as well as in double-immunophenotyping in hematopathology.

Design: Staining and imaging parameters were evaluated to determine optimal procedures. These factors included spectral analyses and comparisons of commercially available chromogen combinations, order of chromogen development, and chromogen development time. Imaging was accomplished using CRi-developed spectral imaging systems and software that can easily be integrated into any microscope. Automated software tools were developed to quantitate nuclear percent-positivity and degree of colocalization of dual nuclear markers, and assessment of immunophenotype for multiple membrane markers. Additional tools that allowed linking of nuclear to membrane phenotype were also developed and applied. Implementation of these tools on an automated, whole slide-scanning microscope system is under way, with application to tissue-microarray samples.

Results: Once staining protocols were optimized, quantitative spectral unmixing of multiple chromogens from the hematoxylin counterstain was achieved and automated. Colocalization of ER and PR in breast cancer was notably higher in breast cancer lesions compared to corresponding normal breast tissue. Identifying MIB1-(+) B-cells by linking nuclear and membrane markers automatically was also accomplished.

Conclusions: Automated spectral tools can easily separate and quantitate multiple chromogens and counterstain. Colocalization and associations between specific membrane and nuclear markers are readily determined without the use of fluorescence-based techniques. These multispectral microscopy approaches will allow for multiplexed molecular markers to be assessed on a cell-by-cell basis in brightfield with conventional chromogenic stains and will be useful for further characterizing and subtyping cancers and other pathological processes.

Qi Liang¹, Zhenqing Zhu1,², Minqi Wei¹, Randy King¹, Jo-Ann Andriko³, Wei-Sing Chu1,²

¹Department of Scientific Laboratories, Armed Forces Institute of Pathology, Washington, D.C.; ²Bio-Quick, Inc., Silver Spring, Maryland; ³Department of Pathology, Walter Reed Army Medical Center, Washington, D.C.

The first crucial step in cancer management is to ensure timely and accurate pathological diagnoses. Formalin fixation and paraffin embedding (FFPE) has been a standard tissue preservation method employed in over 90% of the cases for clinical histology diagnosis. However, FFPE has three disadvantages: (1) slow fixation (16-24 hours) hinders intra-operative decision-making, (2) slow quenching of enzymatic activity causes RNA degradation, and (3) extensive molecule modification affects protein antigenicity.

We have developed the ultrasound-accelerated FFPE (US-FFPE) technology for rapid and multiple-tissue fixation and processing. Applying high-frequency and high-intensity ultrasound to formalin and other reagents cuts fixation and processing time from over 16 hours to within 1 hour. Compared to conventional FFPE, US-FFPE provides similar preservation in tissue morphology with similar long-term storage stability, improved preservation of protein structure, antigen properties, and mRNA integrity. US-FFPE tissues present less alteration in protein antigenicity, so rapid immunohistochemical reactions occur with higher sensitivity and intensity, reducing the need for antigen retrieval pretreatment. Stronger and more uniformed signals in in situ hybridization suggest better RNA preservation. In addition, US-FFPE tissues better support downstream molecular analyses because proteins and nucleic acids extracted from US-FFPE tissues are of greater integrity and quantity compared to those from conventional FFPE tissues.

We also found that during fixation, tissue displays physical changes that can be monitored and reflected as changes in transmission ultrasound signals. To our knowledge, this is the first effort to monitor tissue physical changes during fixation. Further study of this phenomenon may provide a method to control and monitor the level of fixation for quality controls.

Overall, ultrasound-facilitated tissue preservation can provide rapid and improved morphological and molecular preservation to better accommodate both traditional and molecular diagnoses.

Mike Makrigiorgos
Dana-Farber Cancer Institute/Brigham and Women’s Cancer Center, Boston, Massachusetts

New technologies for analyzing gene expression, genomic, and epigenetic DNA changes are expected to lead to profound changes in the detection, prognosis, and therapy of cancer. However, prospective clinical studies to validate the new molecular knowledge take several years to yield their conclusions and can be very expensive. Retrospective studies on existing formalin-fixed, paraffin-embedded (FFPE) specimens for which complete clinical data are available are a useful alternative. However, the efficient use of FFPE specimens for research is hindered by technical issues. Because cancer specimens are heterogeneous and often contain substantial amounts of normal tissue, identification of cancer-specific genetic abnormalities from FFPE samples requires microdissection and whole-genome amplification before screening. Major hurdles to this process are the introduction of amplification bias and the inhibitory effects of formalin fixation on DNA/RNA amplification.

The aim of this sample preparation project is to enable screening of FFPE specimens via advanced genome analysis technologies, thus accelerating discovery of genes and clinical biomarkers relevant to early detection, prognosis, and therapy of cancer. We continue evaluating RCA-RCA, a rolling-circle amplification-based technology for whole-genome amplification that is tolerant to sample degradation, for recovery of molecular information from degraded FFPE samples.

In the first year of the R21, we have evaluated the effect of a variety of commonly used conditions for formalin fixation of surgical samples on the ability to (a) produce efficient whole-genome amplification across the genome, (b) identify polymorphic sites and mutations in diverse chromosomes, and (c) apply array-CGH to identify genome-wide copy number changes. FFPE samples at least 10 years old have been successfully recovered using RCA-RCA. A simple and sensitive molecular test was developed (Index of Sample Degradation) that can be used before performing molecular assays in order to predict with confidence the probability of successful outcome of downstream assays depending on sample condition. Finally, the RCA-RCA methodology was shown to be very useful in the amplification of DNA collected from bodily fluids (e.g., plasma), which is often enzymatically degraded at the time of collection. We anticipate a major impact of this technology in the identification of biomarkers from clinical samples and bodily fluids.

Timothy Jansen, John Laterra, Alessandro Olivi
Johns Hopkins University, Baltimore, Maryland

Recent advances in surgery, radiotherapy, and chemotherapy have made the clinical management of malignant gliomas more successful. However, the prognosis in patients diagnosed with these tumors remains poor. Malignant gliomas appear to proliferate in the brain without significant inhibition from the immune system. The immune privilege demonstrated by these tumors is partly attributed to tumor expression of immune-suppressive agents, such as the Fas ligand (FasL). Immunotherapy using interleukins, particularly IL-2 and IL-12, has been shown to be a promising strategy for the treatment of experimental gliomas but is impeded by T cell apoptosis. We propose that experimental gliomas impair the T cell-mediated antitumor responses by expressing FasL, which induces apoptosis of peritumoral T cells and decreases tumor cell killing by cytokine-activated T cells.

The hypothesis of this work is that immunotherapy of experimental gliomas can be enhanced by inhibiting FasL expression by glioma cells using RNA interference. We have provided ample evidence that the FasL counterattack exists in malignant gliomas using the 9L rat model. We first characterized the expression of FasL in the rat glioma cell lines 9L, C6, and F98 using immunoblot analysis and the real-time polymerase chain reaction. Immunohistochemistry was used to show the presence of apoptotic tumor-infiltrating lymphocytes (TILs). Inhibition of FasL expression in 9L, C6, and F98 glioma cell lines was also accomplished in vitro using siRNA. These data suggest that by using RNA interference techniques, FasL expression by glioma cells can be inhibited and the subsequent effects on survival and function of TILs determined. To further test our hypothesis in vivo, stable FasL knockdown and overexpressing glioma clones have been created using the rat 9L, C6, and F98 cell lines and are being implanted into immune-competent rats. This approach has the potential to improve existing and future immunotherapy strategies utilizing interleukins or vaccines.

Profiling Glycosylation in Secreted Proteomes
Lewis Pannell

Glycoproteins are becoming of increasing interest as better biomarkers of disease. The analysis of the glycosylation pattern on a single glycosylation site in a protein can be a time-consuming process requiring highly purified proteins. We have developed a method for the computer-assisted characterization of glycosylation structures using accurate mass gaps between the glycoforms [1]. An MS-only data file of the protein digest is searched for sets of these gaps, which correlate in both the mass and time domains. While this works well for the single proteins, we have made major improvements for the analysis of protein mixtures, making the isolation of proteins a much less stringent requirement [2]. We have also optimized the methods for obtaining secreted proteomes and can now obtain highly enriched samples [3]. The method for the secreted proteome will be discussed along with the approaches used for the isolation and analysis of glycosylation profiles. Our approach to analyzing mixtures of glycoproteins using sub-database development of glycol-predictive software will be elaborated. Using these approaches, the glycosylation profile of galectin 3 binding protein, highly associated with metastatic cancer cells, will be presented.

• 1. Pannell L.K., Walp E.R., Ruiz, J.C., et al. GlycoMatic: An Automated Approach to Glycoprotein Detection and Characterization. Proceedings of the 53rd American Society for Mass Spectrometry Annual Conference, San Antonio, TX, June 2005, abstract # WP258.
• 2. Pannell L.K., Ruiz J.C., Zuzo A., et al. Towards Automated Glycoanalysis of Proteomes: Partial Fractionation and Limited Database Deployment. Proceedings of the 54th American Society for Mass Spectrometry Annual Conference, Seattle, WA, May 2006, abstract # TP320.
• 3. Mbeunkui F., Fodstad O., Pannell L.K. Secretory Protein Enrichment and Analysis: An Optimized Approach Applied on Cancer Cell Lines Using 2D LC-MS/MS. J Proteome Res 5(4):899-906, 2006.

Ramila Philip¹, Sid Murthysup>1, Jonathan Krakoversup>1, Sinnathamby Gomathinayagamsup>1, Lorraine Keller², Kim Lyerly²

¹Immunotope, Inc., Doylestown, Pennsylvania; ²Duke University, Durham, North Carolina

The goal of this project is to identify autoantibody-reactive tumor-associated antigens for the development of early-stage diagnostics and immunotherapeutics for ovarian cancer. The serum of cancer patients frequently contains circulating levels of antibodies that recognize autoantigens present in the patients’ tumor tissues. We have analyzed 5 composite samples consisting of sera from 20 different ovarian cancer patients and 3 composites representing 12 normal donor serum samples.

Method: Tumor antigens were prepared from extracts of two ovarian cell lines, which yielded consistent, homogeneous and high-concentration protein preparations. Normal and ovarian cancer patient serum composites were mixed with tumor lysate, antigen-antibody complexes were immunoprecipitated, and immunoreactive proteins were isolated and subjected to tryptic digestion. Autoantibody-reactive antigens were separated from the antibodies, fractionated first by SEC, and then further fractionated by reverse-phase HPLC. Each HPLC fraction was individually treated with trypsin. Trypsin-digested samples were analyzed using an LC/MS/MS system. The mass spectral data were analyzed using two different human protein databases, PIR/UniProt and SwissProt, using Sequest and Mascot software. The search results were analyzed to identify autoantigens found in all five cancer composites and not in normal serum.

Results: Differential analysis identified more than 30 candidate autoantibody-specific autoantigens associated with cell growth, structure, DNA replication, and cell adhesion that were present in all five ovarian cancer serum composites and not in normal serum composites. Additional proteins associated with various tumor functions and known tumor markers were immunoprecipitated by both normal and tumor sera in four or fewer composites.

Discussion: We have developed and validated a high-sensitivity immunoproteomics technology for the identification of autoantibody-reactive proteins. The study has revealed a number of interesting candidates involved in cell division, cell cycle regulation, DNA synthesis and repair, and other essential tumor functions. Several novel proteins were also identified in the screen. Additional patient and normal serum sample analyses are under way to confirm the data. In addition, identification of a specific epitope recognized by the autoantibody is in progress with the goal of developing a sandwich ELISA-based early-stage diagnostic test for ovarian cancer.

Suzie Pun
University of Washington, Seattle, Washington

Nanoparticulate delivery vehicles offer distinct advantages to anticancer treatment approaches, including selective accumulation in tumor areas due to the enhanced retention and permeability effect and delivery of macromolecules, such as nucleic acids and proteins. In order to effectively treat a solid tumor, it is necessary for the nanoparticles to reach as many cells within the tumor as possible, but significant penetration into tumor tissue is not seen for relatively large particles such as liposomes and viruses that are directly injected in solid tumors.

The major goal of this project is to develop a noninvasive approach to overcome transport barriers that prevent nanoparticle penetration of solid tumors. We are investigating transcellular methods used by bacterial pathogens to invade tissues. In addition, we are evaluating methods to increase paracellular transport of nanoparticles. In this study, we systematically investigate the effects of nanoparticle size and extracellular matrix integrity on nanoparticle penetration in a 3D, multicellular spheroid tumor model. The penetration of nanoparticles ranging from 20 to 200 nm in MCS was assessed. In addition, the effect of ECM-modulating enzymes on nanoparticle penetration was investigated.

Niroshan Ramachandran
Harvard Medical School, Cambridge, Massachusetts

Protein microarrays empower investigation into a broad range of biochemical properties and activities of the target proteins on the array. The strength of this approach is that by spotting many proteins on a single array, it is possible to test the functions of all of the proteins simultaneously. Without protein microarrays, these applications are tedious and expensive. They require high-throughput methods to produce proteins, automation using large liquid-handling robots, and highly trained personnel. Yet, despite their immense potential and strong demonstrations of feasibility, protein microarrays are not widely used. In large part, this is due to the labor and technical issues associated with producing the arrays. Traditionally, arrays are assembled by producing, purifying, and printing proteins onto the array surface.

We developed a novel approach to generate protein microarrays by printing cDNAs onto glass slides and then translating target proteins with mammalian reticulocyte lysate. This robust method obviates the need to purify proteins, avoids protein stability problems during storage, and captures sufficient protein for functional studies. This technology, nucleic acid programmable protein array (NAPPA), is capable of producing thousands of different proteins including transmembrane proteins all in a single step. NAPPA has already been demonstrated to be robust for the detection of protein-protein interactions by fluorescence detection. As we previously reported, in protein interaction studies, NAPPA recapitulated 85% of the known protein interactions, demonstrating that proteins produced on NAPPA are highly functional.

We are now adapting NAPPA to several more complex functional-proteomics applications, to assess the binding selectivity of small molecules to a family of related proteins (e.g., kinases) or to a mutant series of a single protein, or to screen for substrates for an active enzyme. Current efforts are focused on adapting emerging technologies to facilitate real-time, label-free detection of biological interactions using high-density surface plasmon resonance (SPR). The amalgamation of the two HT technologies, NAPPA and SPR, promises to provide high information content regarding protein interactions.

Apart from functional proteomics, NAPPA is being used in several projects to screen for immune responses to a large panel of antigens. The goal is to discover vaccine candidates in infectious diseases (cholera and tularemia) and to discover biomarkers in various cancers and autoimmune diseases. For example, patients with cancer are known to produce antibodies against tumor antigens. In some cases, immune response to the tumor antigen has been detected before the disease clinically presents itself. Although the specificity for these responses is high, typically only 5% to 20% of patients demonstrate a response to any given antigen, limiting the usefulness of single-antigen responses as biomarkers and demanding high-throughput techniques like NAPPA.

Arfaan Rampersaud¹, Kristie Melnik¹, Jeffrey Chalmers²

¹Columbus NanoWorks, Columbus, Ohio; ²Department of Chemical and Biomolecular Engineering, Ohio State University, Columbus, Ohio

Cell separation is becoming increasingly commonplace for researchers and clinicians, who isolate rare cells from complex cell populations to understand how cells function in different environments or to diagnose disease occurrence, recurrence, or progression. Complex mixtures and searches for small variations in cell samples require new techniques that provide exquisite sensitivity with high throughput; therefore, advances in cell separation technology must allow the researcher to find a low percentage of cells of interest as well as purify them to allow for accurate assessment of biological relevance. The aim of this project was to develop high-resolution magnetic nanoparticles for use in quadrupole magnetic sorting (QMS). QMS is a flow-through immunomagnetic separation system that can provide sensitive enrichment of circulating cancer cells in blood, as well as other biological fluids. Optimal cellular separation by QMS requires immunomagnetic particles having high magnetic susceptibility, narrow particle size distribution, and high-density attachment sites for antibodies. Commercial immunomagnetic beads are too large, lack size uniformity, or have low magnetic susceptibility. As part of the Phase I goals, Columbus NanoWorks (CNW) demonstrated the manufacturing capability of magnetic nanoparticles that were uniform in size within a range of 50 to 70 nm, had a ligand density of 2 to 5 µg/mL Fe particles, and elicited a magnetophoretic mobility of immunomagnetically labeled cells an order of magnitude greater than commercially available magnetic reagents. The aforementioned criteria, defining magnetic particles achieved in the Phase I proposal, allowed the enrichment of 1 ovarian tumor cell in 107 total nucleated blood cells with an average recovery of 74.1% of spiked tumor samples, as well as an average 3.21 log10 depletion of contaminating cells in the QMS. The Phase II research will call for the design and commercialization of magnetic reagent kits specific for QMS applications involving cancer cell enrichment. CNW will (1) scale up processing efforts to obtain a fiftyfold increase over current process techniques, (2) establish procedures and practices that assure consistent control of nanoparticle size and magnetic properties, (3) develop protocols for CNW cell enrichment kits for use in the commercial prototype QMS, and (4) beta testing of protocols using breast cancer patients and head and neck cancer patients. The proposed Phase II end items are customized magnetic reagent kits specific for the QMS system that will allow the enrichment of cancer cells from various blood sources such as bone marrow and peripheral blood. Specifications of the particles will provide a greater than 70% recovery of desired cells, as well as a 4 log10 depletion of contaminating cells. The resultant enriched fraction using CNW nanoparticles in the QMS system will provide a quality-purified product for further downstream processing and subsequent molecular analysis. The goal is for CNW-customized reagents to become an integral part of gold standard testing in the current pathology repertoire.

Bing Ren
Ludwig Institute for Cancer Research, La Jolla, California

Insulators, also known as boundary elements, play crucial roles in establishing the regulatory domains in the genome by preventing the encroachment of heterochromatin and restricting the action of transcriptional enhancers. To date, our understanding of insulators’ role in global gene regulation is still limited because of a lack of comprehensive knowledge of the locations of these elements in the genome. In mammalian cells, virtually all the known insulators interact with the zinc-finger protein CTCF, and loss of CTCF binding to these regions has been implicated in loss of insulator function that results in human diseases. In order to systematically define the insulators in the human genome, we have identified the CTCF binding sites in the genome of primary human fibroblasts using chromatin immunoprecipitation combined with high-resolution genome-tiling arrays analysis. Our study identified 13,804 potential insulators and an insulator motif that is highly conserved in vertebrate species. Using this motif, we predicted over 4,000 additional potential insulators in the human genome. Together, these CTCF sites defined more than 7,300 gene domains, some of which contain as many as 50 genes. Our results provide the first genome-wide map of insulators, reveal cell type-dependent usage of these elements, and will aid the systematic analysis of transcriptional regulation in human cells.

Cleo Salisbury
The Scripps Research Institute, La Jolla, California

Proteases are suspected of playing major roles in cancer, including the activation/inactivation of growth factors and the degradation of extracellular matrix components to promote cancer cell migration and invasion. Consistent with this premise, transcript and protein levels for many proteases are up-regulated in cancer cell lines and tumors. However, whether these changes in protease expression correlate with changes in protease activity remains a critical, but largely unanswered, question. Indeed, proteases are regulated in a complex series of posttranslational events, meaning that their expression levels, as measured by conventional techniques, fail to accurately report on the activity of these enzymes. To address this problem, we have introduced a chemical proteomics technology, referred to as activity-based protein profiling (ABPP), which utilizes active site-directed probes to determine the functional state of large numbers of proteases directly in whole cell, tissue, and fluid samples. We have applied ABPP to identify several serine protease activities up-regulated in human cancer cells and primary tumors. Recently, we have created an advanced antibody-based platform for ABPP that enables profiling of protease activities while requiring only minute amounts of proteome. We are now developing methods to further increase the sensitivity of the microarray platform by developing signal enhancement tools and generating optimized micorarray-compatible antibodies for key cancer proteases. Further, we have developed new probes for the sensitive detection of metalloproteases and related enzymes, which are known to be misregulated in cancer. By combining the new probes with the enhanced-sensitivity microarray platform, we are in a unique position to study the misregulation of protease activities in cancer samples, with an unprecedented combination of sensitivity, resolution, and throughput.

Shan-Rong Shi, Cheng Liu, Clive R. Taylor
Department of Pathology, University of Southern California Keck School of Medicine, Los Angeles, California

Standardization of immunohistochemistry (IHC) for archival formalin-fixed, paraffin-embedded (FFPE) tissue sections has become increasingly important as a result of the emergence of a new field of pathology that requires demonstration of the differential expression of various prognostic markers for individualized cancer treatment. From a practical point of view, one of the most difficult issues in the standardization of IHC for FFPE tissue is the adverse influence of formalin upon antigenicity, and the great variation in fixation/processing procedures. Our previous studies have demonstrated a potential approach to standardization of IHC for FFPE tissue based on optimal antigen retrieval (AR), to achieve a maximal degree of retrieval that provides a comparable level of IHC staining among various FFPE tissue sections that have been fixed in formalin for 4 hours to 7 days. On this basis, the following hypothesis is proposed: “The use of optimized AR protocols permits retrieval of specific proteins (antigens) from FFPE tissues to a defined and reproducible degree (expressed as R%), with reference to the amount of protein present in the original fresh/unfixed tissue.” This hypothesis may be explained mathematically. Suppose the amount of a protein in a fresh cell/tissue = Pf, and that Pf produces an IHC signal in fresh tissue of ò (Pf). When the IHC signal of FFPE is ò (Pffpe), then the retrieved rate of AR (R%) is calculated as AR rate (R%) = ò (Pffpe)/ò (Pf) 100%, the amount of protein in the FFPE tissue of Pffpe = Pf x R%. In a situation where optimized AR is 100% effective, then Pffpe = Pf if the IHC signal is of equal strength in fresh tissue and FFPE tissue. It is proposed that this calculation applied to selected ubiquitous antigens (analytes) may form the basis for developing a system if there are quantifiable internal reference standards.

Further studies are required to test this hypothesis. A research design using cell/tissue models is presented to test limitations of this hypothesis, based on correlation of accurate quantitative biochemical measurements with IHC staining results.

Darryl Shibata
University of Southern California Keck School of Medicine, Los Angeles, California

The time between initiation and tumor removal is uncertain for most human cancers because serial observations are impractical. In this study, cancer mitotic ages (numbers of divisions) are retrospectively estimated by counting numbers of somatic microsatellite mutations in colorectal cancers deficient in DNA mismatch repair. The greater the numbers of divisions, the greater the numbers of mutations (molecular clock hypothesis). Preliminary data indicate sporadic cancers, and “interval” cancers that arise within years after a negative clinical examination, have similar mitotic ages, which suggests many cancers frequently physically grow to detectable sizes within a small period of time (years). It appears that cancers’ histories are surreptitiously written within their genomes.

Robert H. Singer¹, Rossanna Pezo¹, Melissa Lopez-Jones¹, Shailesh Shenoy¹, Paola Capodieci²

¹Albert Einstein College of Medicine, Bronx, New York; ²Aureon Laboratories, Yonkers, New York

We have adapted the application of fluorescence in situ hybridization (FISH) to investigate gene expression profiles in paraffin-embedded tissue samples. We have used this approach to evaluate prostatectomy material from patients. The characterization of the gene expression profiles indicates that a prognostic or diagnostic test is feasible using FISH on paraffin-embedded samples using five informative genes. One of the important results is that archival samples (22 years) were as easily profiled as new samples. Currently,we are analyzing the expression profiles from patients where outcomes are known using tissue microarrays and have found a number of predictive genes.

This technology has also been utilized to investigate the spatial coordination of mouse cyclin D1 transcription in the intestinal mouse mucosa as cells migrate along the crypt/villus axis. The majority of human colon tumors are associated with the elevation of ƒÒ-catenin/TCF signaling due to the loss of APC. It has been well established that ƒÒ-catenin/TCF activity is linked to several genes associated with cell cycle regulation, including the positive regulation of c-myc and cyclin D1. In the intestinal mucosa, these events are modulated along the crypt/villus axis. Using p27 +/+ APC +/+ paraffin-embedded mouse duodenum tissue samples, we can determine a transcriptional profile of cyclin D1 and c-myc in normal intestinal mouse mucosa. To evaluate percent transcription in the crypt and the villus, experiments were performed using probes labeled with spectrally distinctive fluorophores recognizing the cyclin D1 transcript. Preliminary results indicate that there is a 40% reduction in cells expressing cyclin D1 in the villus (1.98%) versus the crypt (3.29%). Using this powerful technology, we can further elucidate specific developmental and signaling pathways important in the intestinal homeostasis and potential perturbations leading to tumorigenesis.

Using Single-Cell Expression Profiling To Identify Markers of 5-FU Resistance in Colorectal Cancer

We are examining the transcriptional profile of individual colorectal tumor cells to better define the indications for chemotherapy using 5-fluorouracil (5-FU). Our goal is to develop a biomarker composed of several key genes that would be predictive of response to 5-FU-based chemotherapy. This approach would also enable us to identify novel genes involved in mediating sensitivity to 5-FU and to further elucidate the mechanisms by which tumor cells acquire resistance to 5-FU. Preliminary results in human colorectal tumor cell lines suggest that a set of four genes can be used to predict resistance to 5-FU-based chemotherapy in colorectal cancer. Transcription sites for specific genes are identified using probes directly labeled with multiple fluorophores for FISH. This method enabled us to simultaneously visualize many nascent mRNA transcripts and provided a gene expression profile for each single cell. Application of this technique to clinical samples allows us to preserve tissue architecture while circumventing the issue of tissue heterogeneity. Gene expression profiles can then be applied to clinical samples and correlated with outcomes. Single cell expression profiling of tumors would allow us to identify subpopulations of cells which may only represent a minor percentage of the sample but may have important phenotypes such as chemosensitivity or metastatic potential. These techniques will provide valuable data for correlating the transcriptional profile of cells with tumorigenesis, metastasis, and drug response. Using this information, chemotherapy treatments can be tailored to the individual tumor and patient.

Clive R. Taylor, Shan-Rong Shi, Cheng Liu
Department of Pathology, University of Southern California Keck School of Medicine, Los Angeles, California

Protein extraction from formalin-fixed, paraffin-embedded (FFPE) tissue sections is critical for the application of advanced proteomics-oriented technology in cell biology and medical research, paralleling the use of the heat-induced antigen retrieval (AR) technique that has revolutionized routine immunohistochemistry (IHC) for FFPE tissue sections worldwide. We have previously documented a protocol by boiling FFPE tissue sections in Tris-HCl containing 2% sodium dodecyl sulfate (SDS), which provides promise as a method to extract protein from FFPE tissue. To improve the efficiency of protein extraction from FFPE tissue, it is necessary to analyze each potential factor that may influence the effect of heat-induced retrieval procedure, including the pH value of retrieval solution, the heating condition (temperature and the time of heating treatment), and the chemicals used in the retrieval solution. In addition to the primary purpose of establishing a practical efficient protocol for protein extraction from FFPE tissue, we intend to test our previous conclusion concerning the AR basic principle: heating under the influence of pH.

Liver tissues taken from seven white mice (strain CD-1) were processed routinely by formalin fixation and paraffin-embedding. One rodent liver tissue was frozen and embedded in OCT for control. For each batch of experiments, 50 tissue sections (10 microns each) were used to extract a sufficient quantity of total protein for 12 to 15 different AR conditions. Protein extraction was performed based on the heat-induced AR principle, using a universal buffer of Britton and Robinson solution, and Tris-HCl buffer solution. Both buffer solutions were prepared at pH values ranging from 1 to 12 (1 to 10 for Tris-HCl buffer). Three heating times, 2, 10, and 30 minutes, were used for boiling FFPE tissue sections in each retrieval solution. SDS-PAGE was used to display the total proteins extracted from FFPE and from frozen tissue sections to compare the efficiency of extraction under different heating conditions and pH values. Results show the following: (1) Proteins extracted from FFPE and frozen tissue sections showed comparable SDS-PAGE patterns, albeit the former showing much weaker bands (due in part to lack of 2% SDS in the retrieval solutions tested in this experiment), (2) higher pH values combined with longer time of heating provided better results, and (3) low pH solutions (at 1.0) showed much poorer results when heating for longer times. Evaluation is ongoing to determine whether different categories of proteins respond differently at various pH values, as demonstrated previously for AR-IHC on FFPE sections.

Paul Tempst
Memorial Sloan-Kettering Cancer Center, New York, New York

As cancer involves transformation and proliferation of altered cell types that produce high levels of specific proteins and enzymes such as proteases, it will modify not only the array of existing serum proteins (serum “proteome”) but also their metabolic products, i.e., peptides (serum “peptidome”). However, it was unclear whether this complex peptidome may provide a correlate of biological events occurring in the entire organism.

We have developed an automated procedure for the simultaneous measurement of peptides in serum that utilizes magnetic, reversed-phase beads for analyte capture and a MALDI-TOF MS readout. This system is more sensitive than surface capture on chips as spherical particles have larger combined surface areas, and therefore higher binding capacity, than small-diameter spots. Coupled to high-resolution MS and MS/MS, hundreds of peptides can be detected in a single droplet of serum, many of which can be readily identified without further fractionation. Automation facilitates throughput and ensures reproducibility. We then use a minimal entropy-based algorithm that simplifies and improves alignment of spectra and subsequent statistical analysis.

MS-based sequence characterization indicated that a large part of the human serum “peptidome,” as detected by MALDI-TOF MS, is produced ex vivo (i.e., after the blood sample collection) by degradation of abundant substrates by endogenous proteases. Polypeptide fragments are generated during the proteolytic cascades that take place in the intrinsic pathway of coagulation and complement activation. Once generated, the “founder peptides” are pared down by exoproteases into ladder-like clusters. By correlating identified proteolytic patterns with disease groups and controls, we have shown that exoprotease activities, superimposed on the ex vivo coagulation and complement-degradation pathways, contribute to generation of not only cancer-specific but also “cancer type”-specific serum peptides. None of the signature peptides were derived from cancer cells, which implies that different tumor types secrete and/or shed distinct sets of proteases that, through their catalytic activity, generate unique serum peptide profiles. The small number of blood proteins that are the source of nearly all the peptides in the prostate, bladder, and breast cancer signatures are not biomarkers in a strict sense but simply serve as an endogenous substrate pool for the real ones, i.e., tumor-derived proteases. However, there is no actual relationship between the precursor substrate concentrations and the MS-ion intensities of many of the degradation products. For instance, highly abundant serum proteins such as albumin and immunoglobulins were not represented.

Direct MALDI-TOF MS-based serum peptide profiling is thus a form of activity-based proteomics, monitoring proteome “metabolomic” products. It makes this approach particularly well suited for detection of cancer as proteases are well-established components of cancer progression and invasiveness. Focused mass spectrometric analysis of key peptides, derived from endogenous or custom-synthetic substrates, and utilization of isotopically labeled standards to quantify all pattern-contributing peptides, may lead to functional assays and facilitate future introduction of this technology into clinical practice. Identification and characterization of the protease panels could also lead to direct immunoassay-based, quantitative diagnostic tests that may be better suited to a clinical environment.

Derek Thirstrup¹, Peter Wiktor¹, Jim Mcgill¹, Rob Sullivan², Helmut Zarbll,²

¹Engineering-Arts, Mercer, Washington; ²Fred Hutchinson Cancer Research Center, Seattle, Washington

Our research follows two distinct yet complementary paths. One path involves developing and testing novel piezoelectric pipetting, dispensing, sensing, and housing technology, and integrating this into a fully automated piezoelectric pipetting system. The other path involves cancer-related genomics research in collaboration with the Fred Hutchinson Cancer Research Center (FHCRC). The ultimate goal of the proposed research is to have the two paths come together, resulting in an automated, fully functional, and general-purpose piezoelectric fluid pipetting system with the reliability and performance to empower cancer-related, genomic, and functional genomics research. We have demonstrated the feasibility of using our piezoelectric pipettes in application to eGFP expression cell microarrays. Furthermore, we have collaborated with GE Healthcare to develop piezoelectric pipetting technology to construct their whole genome expression arrays.

Our current focus is shRNA loss-of-function arrays, which should allow for the screening of cellular function and will make it possible to assay phenotypic silencing effects of specific genes on a genome-wide scale. Our current work has also focused on controlling cell adhesion over the arrayed spots. We have data showing controlled cell adhesion over individual spots and regions devoid of cells between adjacent spots. We are also working on enhancing the transfection efficiency of RNAi complexes using recombinant Ad5 penton capsid as a facilitator of receptor-mediated endocytosis. We will subsequently generate an apoptosis microarray screen by targeting two oncogenes: Bcl-2 and HPV E7 oncogenes associated with colorectal and cervical cancer, respectively. Our surface transfection array will result in an RNAi-dependent apoptosis screen of two human epithelial cancer cell lines with apoptosis index quantitation using a commercial apoptosis screening assay.

Luke Tolley
Southern Illinois University, Carbondale, Illinois

Dynamic isoelectric focusing is a new technique for protein separation. It is related to capillary isoelectric focusing, but uses additional high-voltage power supplies to provide control over the shape of the electric field within the capillary. Manipulation of the electric field changes the pH gradient, enabling both the location and width of the focused protein bands to be controlled. The proteins can be migrated to a designated sampling point, while remaining focused, where they can be collected for further analysis. This ability to collect and isolate the protein bands while maintaining a high peak capacity demonstrates that dynamic isoelectric focusing has great potential for proteomic analyses. Dynamic isoelectric focusing is demonstrated to achieve a peak capacity of over 1,000 in less than half an hour, as shown by both mass spectrometry analysis and direct imaging.

Paul O. Ts'o¹, Stephen A. Lesko¹, Zhao-Ping Lum¹, Scott Deamond¹, Enzhi Shan¹, David Zhou¹, John Daniel1, Katherine Tkaczuk², Nancy S. Tait2, Olga Ioffe², David Van Echo³

¹CCC Diagnostics, LLC, Baltimore, Maryland; ²University of Maryland Greenebaum Cancer Center, Baltimore, Maryland; ³Harbor View Cancer Center, Baltimore, Maryland

Drug response indicators test (DRIT) is a diagnostic assay performed on fixed and embedded breast tumor sections or circulating cancer cells in order to quantitate the chemoresistance of tumors in individual patients. The DRIT may be used to provide information for predicting patient response to seven U.S. Food and Drug Administration-approved, National Comprehensive Cancer Network-recommended anticancer drugs based on the expression of seven DRIs in the tumor. The results are expressed in a probability percentage of chemoresistance according to the correlation of clinical response with drug response indicator (biomarker) expression.

DRIT involves three components: (1) Quantitative measurement of several DRIs simultaneously in the tumor. The expression of DRIs in tumor cells can be evaluated by immunofluorescence utilizing labeled monoclonal antibodies and a computerized fluorescence microscope. Numerical values can be derived and normalized to a fluorescent reference standard for comparison. (2) Establishment of a statistically significant correlation of DRI expression with the cytotoxic response of tumor cell lines to a related drug. An in vitro indexing system was established to correlate the cytotoxic effect of each drug to the corresponding DRI measurements. Seven breast cancer cell lines with different sensitivities to various anticancer drugs were utilized. The effect of tamoxifen, paclitaxel, trastuzumab, and doxorubicin was correlated with the DRI expression of each drug in these cell lines. The DRIs are estrogen receptor, beta tubulin III, HER2/neu, and topoisomerase II, respectively. (3) Clinical studies designed to correlate the level of DRI expression in tumor cells of individual patients with the clinical response [complete response (CR), partial response (PR), stable disease (SD), progressive disease (PD)] to mechanistically related drugs. Initially, retrospective studies will be conducted on the basis of patients’ archival tumor tissues and related records of treatment. This retrospective clinical study will be carried out with 200 patients, in conjunction with a consortium of five hospitals in the Baltimore-Washington, D.C., area. In this correlation study, a probability percentage of chemoresistance (leading to PD) will be established for each breast cancer patient’s tumor, providing information as to which drug is likely not to be effective.

The utilization of DRIT to provide advisory information to physicians can lead to cost savings, reduction in side effects, and better utilization of the time allowed in the treatment window. Thus, the DRIT can result in a greater efficacy of current drugs through personalized anticancer chemotherapy.

Marc Vidal
Center for Cancer Systems Biology, Dana-Farber Cancer Institute, Boston, Massachusetts

For over half a century, it has been conjectured that macromolecules form complex networks of functionally interacting components and the molecular mechanisms underlying most biological processes correspond to particular steady states adopted by such cellular networks. However, until recently, systems-level theoretical conjectures remained largely unappreciated, mainly because of lack of supporting experimental data.

To generate the information necessary to eventually address how complex cellular networks relate to biology, we initiated, at the scale of the whole proteome, an integrated approach for modeling protein-protein interaction or “interactome” networks. Our main questions are the following: How are interactome networks organized at the scale of the whole cell? How can we uncover local and global features underlying this organization? How are interactome networks modified in human disease, such as cancer?

Binghe Wang
Georgia State University, Atlanta, Georgia

The boronic acid moiety is a versatile functional group useful in carbohydrate recognition, glycoprotein pull-down, inhibition of hydrolytic enzymes, and boron neutron capture therapy. The incorporation of the boronic acid group into DNA could lead to molecules of various biological functions. We have successfully synthesized a boronic acid-modified thymidine triphosphate (B-TTP) and incorporated it into DNA. The synthesis was achieved using the Huisgen cycloaddition as the key reaction. We have demonstrated that DNA polymerase can effectively recognize the boronic acid-modified DNA (BA-DNA) as the template for DNA polymerization, which allows PCR amplification of boronic acid-modified DNA. DNA polymerase recognition of the B-TTP as a substrate and the boronic acid-modified DNA as a template are critical issues for the development of DNA-based lectin mimics via in vitro selection.

Yue Wang
Virginia Tech, Arlington, Virginia

The challenge of cancer treatment has been to target specific therapies to pathogenetically distinct cancer types or subtypes, to maximize efficacy and minimize toxicity. However, cancers with similar histopathological appearance can follow significantly different clinical courses and show different responses to therapy. The recent development of gene microarrays provides an opportunity to take a genome-wide approach to predicting clinical heterogeneity in cancer treatment and potentially discover new diagnostic and therapeutic targets. This IMAT project aims to develop a bioinformatics tool suite for data modeling and analysis consisting of most major computational tasks in cancer research.

This presentation will first report our experience in developing caBIGTM Silver-Maturity data clustering software, VISDA. The software has been applied to sample and gene, as well as unsupervised and supervised data clustering, with rigorous comparisons to major existing software. Second, we will report our recent effort in developing adaptive hierarchical subspace expert (AHSE) software. Different from most existing predictive classification algorithms, AHSE provides the ability to learn the tree of phenotype and hierarchical subspaces and perform adaptive multicategory cancer classifications. Third, we will report our preliminary study on decomposing the spatial-temporal images of tumor angiogenic vasculature into permeability factor images and associated perfusion time activity curves. We conducted clustered-component analysis (CCA) on dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). The objective is to characterize the response of tumor angiogenic vasculature to antiangiogenesis therapies. Last, we will summarize the algorithm development for data normalization, biomarker discovery and validation, and tissue heterogeneity correction.

Yingming Zhao
University of Texas Southwestern Medical Center, Dallas, Texas

More than 200 known posttranslational modifications have been reported, most of which have been shown to be intimately associated with cancer. However, only a few of them have been extensively investigated. The chemical nature of the protein modifications offers very limited insights into their roles in biochemical pathways. The vast number of uncharacterized protein modifications suggests that new technologies remain to be developed for their functional studies. Largely because of recent funding from the IMAT program, our laboratory was able to develop a series of novel technologies to address these challenges. In this presentation, we will highlight four approaches, which we developed in the last few years, for detection, isolation, and proteomics of posttranslational modifications. These methods take advantage of highly sensitive mass spectrometry and efficient isolation of posttranslationally modified proteins/peptides, which is based on immunoaffinity purification, tagged substrates, ionic interactions, and pI difference. We applied these technologies to global analysis of several posttranslational modifications, including lysine acetylation, sumoylation, prenylation, O-GlcNAc modification, and phosphorylation. Our proteomics analysis of lysine acetylation found that acetyllysine is present in diverse non-nuclear proteins, therefore revealing previously unappreciated roles for lysine acetylation in the regulation of diverse cellular pathways outside the nucleus. Many of the identified acetyllysine substrates are involved in the regulation of cell growth, survival, and apoptosis, shedding new light into its roles in cancer. Integration of these methods with isotope labeling promises dynamic studies of protein modifications under diverse cellular environments involved in the progression of diverse cancer.

Lara S. Collier, Eric P. Rahrmann, Laura E. Green, Paul C. Marker, David A. Largaespada
University of Minnesota, Minneapolis, Minnesota

Human prostate cancer progresses through a series of pathologically defined stages: prostatic intra-epithelial neoplasia, carcinoma in situ, invasive adenocarcinoma, and metastatic disease. These pathological changes are hypothesized to be accompanied by the accumulation of genetic changes that allow adult prostatic epithelial cells to first proliferate and then invade surrounding tissues, and to metastasize to distant sites. Clinically, advanced prostate cancer is usually treated by hormone ablation therapies. Although tumors usually initially respond to this treatment, many eventually relapse due to androgen insensitivity. The specific genetic pathways that are disregulated in each step of prostate cancer progression vary from tumor to tumor. In principle, these differences could allow medical intervention for prostate cancer to be tailored to the genetic makeup of a patient’s particular cancer. To elucidate the genetic changes that accompany each step in prostate cancer progression, we are using Sleeping Beauty transposon-mediated insertional mutagenesis [1-3] to identify new prostatic oncogenes and tumor suppressors in mouse models of prostate cancer initiation and progression. Mobilization of highly mutagenic transposons via ubiquitous expression of transposase leads to highly penetrant lymphocytic leukemia [2]. Cloning transposon insertion sites from a limited number of leukemias revealed the presence of common integration sites (CISs) at both known and candidate cancer genes. Closer examination of 4- to 6-month old male mice that succumbed to leukemia has revealed abnormal proliferative lesions in the prostatic epithelium. Work on methods to clone transposon integration sites from laser-captured lesions identified by immunohistochemical staining for cell cycle markers is progressing. We are also developing new transgenic lines that express transposase specifically in the luminal epithelial cells of the prostate so that mutagenesis will be restricted to the proper cell type. We will use this tissue-specific transposon mobilization to accelerate prostate tumor progression. Specifically, we propose to screen for genes involved in prostate cancer metastasis and progression by mobilizing transposons in a murine prostate cancer model induced by prostate-specific loss of the PTEN tumor suppressor gene. PTEN-deficient murine prostate tumors are adenocarcinomas but rarely, if ever, metastasize [4,5]. In addition, the tumors are responsive to androgen ablation therapy [5]. We hypothesize that transposon mobilization in these tumors will “tag” and mutate genes involved in prostate cancer progression, leading to widespread tumor metastasis. Furthermore, we hypothesize that in animals receiving hormone ablation therapy, transposon-mediated mutagenesis will identify genes responsible for progression to a hormone refractory state. In conclusion, our data suggest that Sleeping Beauty is a useful tool for cancer gene discovery in epithelial tumors, such as prostate cancer.

1. Collier L.S., Carlson C.M., Ravimohan S., et al. Cancer Gene Discovery in Solid Tumours Using Transposon-Based Somatic Mutagenesis in the Mouse. Nature 436(7048):272-6, 2005.
2. Dupuy A.J., Akagi K., Largaespada D.A., et al. Mammalian Mutagenesis Using a Highly Mobile Somatic Sleeping Beauty Transposon System. Nature 436(7048):221-6, 2005.
3. Ivics Z., Hackett P.B., Plasterk R.H., et al. Molecular Reconstruction of Sleeping Beauty, a Tc1-Like Transposon From Fish, and Its Transposition in Human Cells. Cell 91(4):501-10, 1997.
4. Trotman L.C., Niki M., Dotan Z.A., et al. Pten Dose Dictates Cancer Progression in the Prostate. PLoS Biol 1(3):E59, 2003.
5. Wang S., Gao J., Lei Q., Rozengurt N., et al. Prostate-Specific Deletion of the Murine Pten Tumor Suppressor Gene Leads to Metastatic Prostate Cancer. Cancer Cell 4(3):209-21, 2003.

Roland Green¹, Nan Jiang¹, Tae Hoon Kim², Bing Ren²

¹NimbleGen Systems, Inc., Madison, Wisconsin; ²Ludwig Institute for Cancer Research, Department of Cellular and Molecular Medicine, University of California, San Diego, School of Medicine, La Jolla, California

We have performed genome-wide ChIP chip experiments to map the binding sites of TCF4 in a colorectal carcinoma cell line, HT29. TCF4 is a transcription factor in the Wnt/beta-catenin signaling pathway, and mutations that inappropriately activate this pathway have been implicated in approximately 90% of colorectal carcinoma cases. This study determined a set of target genes that are regulated by Wnt/beta-catenin signaling. This set of target genes should lead to a better understanding of the cellular pathways controlled by the oncogenes and help in the development of therapeutic strategies. In addition to the genome-wide ChIP chip study, we have several projects under way to reduce the cost of genome-wide microarray studies. We have developed a new microarray substrate that allows for robust array reuse. This new substrate provides superior reuse properties and will greatly reduce the cost of arrays in genome-wide studies. We are also building a new microarray-manufacturing tool that will produce microarrays with 2.1 million features. This is compared to our current arrays, which contain 385,000 features. These new higher density arrays will allow us to cover the entire human genome on 7 arrays compared with 38 arrays. This alone will produce a fivefold reduction in array costs for a genome-wide study. These reusable, higher density arrays will make genome-wide microarray studies economically feasible for any laboratory interested in these types of experiments.

William Cleveland
St. Luke’s-Roosevelt Hospital Center, University Hospital of Columbia University, New York, New York

The Cleveland and Yao laboratories are developing an autonomous system for single-cell-level studies that will combine multidimensional optical microscopy with massively parallel multiparameter molecular methods such as microarrays, mass spectrometry, and RT-PCR. A critical requirement for this system has been the development of computer algorithms that facilitate the elimination of a human observer/operator during the routine operation of the system. Progress toward this goal has led to the development of binary classification algorithms that recognize and localize unstained viable cells in microscope images obtained with brightfield illumination. Even without the use of a viability stain, these algorithms can distinguish between viable and nonviable cells.

In many practical culture systems, there are often multiple types of cells. In order to use our robotic system with these cultures, multiclass classification algorithms will be needed to distinguish between the different types of cells in the cultures. Our recent progress has led to a multiclass classification strategy that uses composite images formed by combining images of the same field that are acquired with brightfield contrast, phase contrast, and Hoffman Modulation Contrast. This strategy takes advantage of the fact that cell types that look similar with one contrast technique may look different with another. We find that multicontrast composite images yield better results than monocontrast images and provide a multiclass classification accuracy that is sufficient for many practical applications.

The success of our binary and multiclass cell recognition algorithms raises the possibility of a highly autonomous robotic system for microscopy, manipulation, and molecular analysis of single living cells.

Steven A. Soper, Robin L. McCarley, Jost Goettert, Michael C. Murphy, David Spivak, Robert P. Hammer, Feng Xu, Andre Adams, Suying Wei, Proyag Datta, Subramanian Balamurugen, Matsuez L. Hupert
Departments of Chemistry and Mechanical Engineering, Center for BioModular Multi-Scale Systems, Louisiana State University, Baton Rouge, Louisiana

We are developing polymer-based, modular microfluidic systems that can be produced in a high-production mode and at low cost using replication micro- and nanotechnologies and assembled into 3D architectures to carry out multistep bioassays for clinical diagnostics. Because of the low cost production of these systems, they can be configured into disposable formats appropriate for point-of-care (POC) testing and, when fabricated into multichannel formats, high-throughput applications in clinical laboratories. In this presentation, a general description will be offered concerning the use of polymer-based microfluidics for building highly integrated molecular diagnostic systems for monitoring the presence of a variety of biomarkers, such as rare point mutations, proteins, and mRNAs. In addition, transport phenomena, both electrokinetically and hydrodynamically actuated flows, in these polymer-based systems will be presented. The individual modules we have developed consist of high-speed thermal cyclers (for PCR and allele-specific ligation assays [LDRs]), solid-phase extraction chips (for nucleic acids and proteins), and microarrays assembled into polymer microfluidic chips. The system was evaluated by analyzing point mutations in oncogenes, which provided high diagnostic value for breast cancer. The construction of the fully integrated flow-through bioprocessor consisted of a module to isolate and extract genomic DNA from whole cell lysates, PCR amplify the genomic DNA, score the presence of point mutations using LDR, and monitor the presence of the LDR products using a universal array microchip. The microsystem comprised three different modules containing passive elements: a polycarbonate (PC) chip for cell lysis and DNA preconcentration/isolation from a crude whole cell lysate, a PC chip for performing thermally driven PCRs and LDRs, and a poly(methyl methacrylate) chip for the detection of the LDR products using a medium density DNA universal microarray. The microarray was patterned on a polymer planar waveguide to allow imaging the entire array onto a CCD camera. With this fluidic system, a molecular assay could be completed in less than 10 minutes starting from whole cell inputs.

Peter Gascoyne
Department of Molecular Pathology, University of Texas M.D. Anderson Cancer Center, Houston, Texas

To isolate rare tumor cells from peripheral blood and cancer cell subpopulations from tumor biopsies, we developed and refined a chamber system for dielectric flow fractionation (DFF) by combining dielectrophoresis and field-flow fractionation. This approach harnesses the sensitive discrimination of field-flow fractionation to exploit cell dielectric differences (which sensitively reflect cell morphological properties) and cell density. Sorting by these intrinsic cell properties does not require the tumor or blood cells to be labeled. Using DFF for rare cell isolation experiments, tumor cells salted into peripheral blood were recovered with 80% efficiency regardless of the starting tumor cell concentration. Ten milliliters of peripheral blood specimens (containing approximately 2.5 x 107 PBMNCs) were seeded with tumor cells at ratios from 103:1 to 105:1. After fractionation, the tumor cell-bearing fractions contained approximately 103 residual PBMNCs mixed with 80% of the starting tumor cells from the specimen. With different settings, the DFF was able to fractionate tumor cells from trypsin-digested tumors and cultures to provide isolates of tumor cell subpopulation having distinct cell morphologies. The isolated tumor subpopulations ranged from fractions having small, putative tumor stem morphology, through modal tumor cells, to tumor cells having large, aberrant morphology and nuclei. We also developed an enhanced form of the fractionation method that incorporates a periodic magnet to allow trapping of magnetic antibodies in addition to the cell morphological and density discrimination capabilities of the basic device. This capability allowed residual PBMNCs to be removed from rare tumor cell isolates via CD45 pan-leukocyte antibodies and might be used to target tumor cell-specific or other cell antibodies if needed. An important feature of our method is that it has the capacity to sort specimens ranging from very small numbers to tens of millions of cells within a single chamber in the same timespan and provide viable fractions for analysis and, under sterile conditions, for culture. In this presentation, the theory and implementation of the technology will be shown together with results for rare cell isolations and tumor cell subpopulation analysis.

Sebastian Oltean^1,4, Brian S. Sorg², Todd Albrecht1,⁴, Vivian I. Bonano1,^4,5, Robert M. Brazas1,⁴, Mark W. Dewhirst², Mariano A. Garcia-Blanco1,^3,4

¹Department of Molecular Genetics and Microbiology, ²Department of Radiation Oncology, ³Department of Medicine, ⁴Center for RNA Biology, 5University Program in Genetics and Genomics, Duke University Medical Center, Durham, North Carolina

In epithelial cells, alternative splicing of FGFR2 transcripts leads to the expression of the FGFR2(IIIb) isoform, whereas in mesenc